100%(1)100% нашли этот документ полезным (1 голос)

2K просмотров574 страницыJun 23, 2013

© Attribution Non-Commercial (BY-NC)

PDF, TXT или читайте онлайн в Scribd

Attribution Non-Commercial (BY-NC)

100%(1)100% нашли этот документ полезным (1 голос)

2K просмотров574 страницыAttribution Non-Commercial (BY-NC)

Вы находитесь на странице: 1из 574

CRUX with MAYHEM

We happily herald Shawn Godin of Cairine Wilson Secondary School,

Orleans, ON, Canada as the next Editor-in-Chief of Crux Mathematicorum with

Mathematical Mayhem effective 21 March 2011.

With this issue Shawn takes the reins, and though some of you know Shawn

as a former Mayhem Editor let me provide some further background.

Shawn grew up outside the small town of Massey in Northern On-

tario. He obtained his B. Math from the University of Waterloo in

1987. After a few years trying to decide what to do with his life Shawn

returned to school obtaining his B.Ed. in 1991.

Shawn taught at St. Joseph Scollard Hall S.S. in North Bay from

1991 to 1998. During this time he married his wife Julie, and his two

sons Samuel and Simon were born. In the summer of 1998 he moved

with his family to Orleans where he has taught at Cairine Wilson S.S.

ever since (with a three year term at the board office as a consultant).

He also managed to go back to school part time and earn an M.Sc. in

mathematics from Carleton University in 2002.

Shawn has been involved in many mathematical activities. In 1998

he co-chaired the provincial mathematics education conference. He

has been involved with textbook companies developing material for

teacher resources and the web as well as writing a few chapters for

a grade 11 textbook. He has developed materials for teachers and

students for his school board, as well as the provincial ministry of

education. Shawn is involved with the problem committees at the

Centre for Education in Mathematics and Computing at the

University of Waterloo where he currently works on the committee

for the Fryer, Galois and Hypatia contests. He is a frequent presenter

at local, regional and provincial math education conferences.

In 1997 Shawn met Graham Wright, the former executive

director of the CMS, at a conference. When Shawn moved to Ottawa

Graham put him to work as Mayhem editor from 2001 to 2006 as well

as helping with math camps in Ottawa.

In his free time, Shawn enjoys hanging out with his wife and kids,

reading (mainly science fiction and mathematics), and “playing”

guitar.

Shawn has already been receiving your submissions, and please send any

future submissions to the addresses and e-mail addresses listed inside the back

cover, with your full name and affiliation on each page.

2

EDITORIAL

Shawn Godin

some point I would come back to the editorial board, but I thought it would be

when I retired and not as Editor-in-Chief. Sometimes life throws you a surprise

and you have to roll with it.

This is an interesting time for the journal. The readership is down a bit

and Robert Woodrow is stepping down as the editor of the Olympiad corner after

serving Crux for many, many years, so we are looking to make some changes.

Ultimately, we want a publication that meets the needs and wants of its readers,

so we will be looking to you for some feedback. In a future issue we will look to

get some ideas from you. What do you like about CRUX with MAYHEM?

What would you like changed? Are there things that we are not currently doing

that we should be doing? Are there things that we are doing that we need to

discontinue? Are there alternate formats that we should explore for CRUX with

MAYHEM? We will need your input, so please start thinking so that you can

give us some feedback later.

It took quite a while for me to be officially appointed the Editor-in-Chief,

and as a result we are a few months behind schedule. We will be doing our best to

make up some of that time and get us closer to our real time line. To help facilitate

this, all deadlines for solutions will appear as if we are on time. Having said that

you will have some extra time before the material will be processed. We will post

the status of the problems processing on the CMS web site, cms.math.ca/crux.

I want to say a quick thank you to Vazz Linek, the rest of the members of

the board and the staff at the CMS for all their help getting me on my feet. The

learning curve is steep but I am really looking forward to working on CRUX with

MAYHEM. I know we can work together and continue the great work that Vazz,

Jim, Bruce, Robert, Bill, Leo, Fred and all the other members of the CRUX with

MAYHEM board past and present, have done. Let’s get started!

Shawn Godin

3

Please send your solutions to problems in this Skoliad by September 15, 2011.

A copy of CRUX with Mayhem will be sent to one pre-university reader who

sends in solutions before the deadline. The decision of the editors is final.

Our contest this month is the Niels Henrik Abel Mathematics Contest, 2009–

2010, Second Round. Our thanks go to Øyvind Bakke, Norwegian University of

Science and Technology, Trondheim, Norway, for providing us with this contest and

for permission to publish it. We also thank Rolland Gaudet, University College

of Saint Boniface, Winnipeg, MB, for translating this contest from English into

French.

2009–2010

2nd Round

100 minutes allowed

two digits is twice as large as the number formed by the rightmost two digits. (For

example, 2010 is interesting.) Find the largest whole number, d, such that all

interesting numbers are divisible by d.

2. A calculator performs this operation: It multiplies by 2.1, then erases all digits

to the right of the decimal point. For example, if you perform this operation on

the number 5, the result is 10; if you begin with 11, the result is 23. Now, if you

begin with the whole number k and perform the operation three times, the result

is 201. Find k.

3. The pentagon ABCDE consists of a square, ACDE, B.

.

........

with side length 8, and an isosceles triangle, ABC, such that . ...... ....

. .

. . ....

AB = BC. The area of the pentagon is 90. Find the area ... ... ...

.

... .... ...

of 4BEC. .. ..

.. ...

...

...

A .. .. .. . .C

4. In how many ways can one choose three different integers ... .... ........

.

... .. .

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

between 0.5 and 13.5 such that the sum of the three numbers .... .... .. ....

.....

is divisible by 3? ... ... ........ ...

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

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

5. If a and b are positive integers such that a3 − b3 = 485, ............. ...

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

find a3 + b3 . E D

a−2 b2 + a2 b−2 .

4

length of AB is 12, then find the square of the length of BD.

8. If x, y, and z are whole numbers and xyz+xy+2yz+xz+x+2y+2z = 28

find x + y + z.

9. Henrik’s math class needs to choose a committee consisting of two girls and

two boys. If the committee can be chosen in 3630 ways, how many students are

there in Henrik’s math class?

10. Let S be 1!(12 + 1 + 1) + 2!(22 + 2 + 1) + 3!(32 + 3 + 1) + · · · +

S+1

100!(1002 + 100 + 1). Find . (As usual, k! = 1 · 2 · 3 · · · · · (k − 1) · k.)

101!

2009–2010

2ième ronde

100 minutes sont accordées

1. Un entier à quatre chiffres est intéressant si l’entier formé par les deux chiffres

à l’extrême gauche est deux fois plus grand que l’entier formé par les deux chiffres

à l’extrême droite. (Par exemple, 2010 est intéressant.) Déterminer le plus grand

entier, d, tel que tous les nombres intéressants sont divisibles par d.

2. Une calculatrice effectue cette opération : elle multiplie par 2,1, puis elle efface

tous les chiffres à droite de la décimale. Par exemple, si on effectue cette opération

à partir de 5, le résultat est 10 ; à partir de 11, le résultat est 23. Or, si on

commence avec un entier k et qu’on effectue cette opération trois fois, le résultat

est 201. Déterminer k.

3. Le pentagone ABCDE consiste d’un carré, ACDE, de B

..

côtés de longueur 8, puis d’un triangle isocèle, ABC, tel que ......

.... ...

AB = BC. La surface du pentagone est 90. Déterminer la ........ .....

.

... ... ...

...

surface de 4BEC. .... .... ...

... ... ...

A ... .. . . .. C

4. De combien de façons pouvons-nous choisir trois entiers .. .. ....

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

. ..

.

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

différents entre 0, 5 et 13, 5, tels que la somme des trois en- .. .. .... ..

.. . ..... ...

tiers soit divisible par 3 ? ... ... ........ ...

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

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

.........

5. Si a et b sont des entiers positifs tels que a3 − b3 = 485, E

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

D

déterminer a3 + b3 .

a−2 b2 + a2 b−2 .

longueur de AB est 12, déterminer le carré de la longueur de BD.

déterminer x + y + z.

5

filles et de deux garçons. Si ce comité peut être formé de 3630 façons, combien

d’étudiants y a-t-il dans la classe de mathématiques d’Henri ?

S+1

· · · + 100!(1002 + 100 + 1). Déterminer . (Comme d’habitude, k! =

101!

1 · 2 · 3 · · · · · (k − 1) · k.)

Next follow solutions to the selected problems from the 10th Annual

Christopher Newport University Regional Mathematics Contest, 2009, given in

Skoliad 124 at [2010:129–131]. (Note: Problems 1, 2, 3 and 4 first appeared on the

2009 Calgary Junior Math Contest. – Ed.)

1. Elves and ogres live in the land of Pixie. The average height of the elves

is 80 cm, the average height of the ogres is 200 cm, and the average height of the

elves and the ogres together is 140 cm. If 36 elves live in Pixie, how many ogres

live there?

Solution by Lena Choi, student, École Dr. Charles Best Secondary School,

Coquitlam, BC.

Let x be the number of ogres in Pixie. Then the total height of all the ogres

is 200x, and the total height of all the elves is 36 · 80 = 2880. Therefore the

total height of all the creatures in Pixie is 2880 + 200x. On the other hand, the

average height of the 36 + x creatures in Pixie is 140, so their total height is

140(36 + x) = 5040 + 140x. Thus 2880 + 200x = 5040 + 140x, so x = 36.

Also solved by WEN-TING FAN, student, Burnaby North Secondary School, Burnaby,

BC; GESINE GEUPEL, student, Max Ernst Gymnasium, Brühl, NRW, Germany; and LISA

WANG, student, Port Moody Secondary School, Port Moody, BC.

2. You are given a two-digit positive integer. If you reverse the digits of your

number, the result is a number which is 20% larger than your original number.

What is your original number?

Solution by Gesine Geupel, student, Max Ernst Gymnasium, Brühl, NRW,

Germany.

Let x be the given two-digit number. Increasing x by 20%, that is, by 51 x,

yields an integer, so x must be divisible by 5. Thus x ends in 0 or in 5. If the

ones digit of x is 0, reversing the digits would decrease the number, so x must end

in 5. If the tens digit is larger than 5, reversing the digits would again decrease

the number. Thus only the numbers 15, 25, 35, 45, and 55 remain to be checked.

Only 45 works out.

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; WEN-TING FAN, student, Burnaby North Secondary School, Burnaby, BC;

and LISA WANG, student, Port Moody Secondary School, Port Moody, BC.

Alternatively, let 10a+b be the two-digit number, where a and b are digits. Increasing by

6 6

20% is the same as multiplying by 5 , so 10b+a = 5 (10a+b). Thus 5(10b+a) = 6(10a+b),

6

so 44b = 55a, so 4b = 5a. Clearly b is divisible by 5, and since b = 0 = a does not yield a

two-digit number, b must be 5. Hence a = 4 and the number is 45.

inside a right-angled triangle as shown .

.....

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

...... .

in the diagram. The side length of the .

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

....... .. ..

smallest of the three squares is 16. The

....

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

side length of the largest of the three

....

.. .

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

. .. ..

squares is 36. What is the side length ....

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

. ..

.

of the middle square? .......................................................................................................

Solution by Wen-Ting Fan, student, Burnaby North Secondary School, Burnaby,

BC.

Impose a coordinate system as in the figure. If the middle square has side

length n, then the coordinates are as indicated. Since the slanted line passes

through (0, 16), the equation of the line is y = mx + 16 for some slope, m.

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

......

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

.. (16 + n, 36) ................... ...

...

... .

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

.. .. .. . .. . .

.

r . .. . . .. . .. . . .. . .. . .. . . .. . .. . . .. . .. . .. . . .. . . .. . .. . .. . . . .. .. .

. .. .. .. . .

... (16, n)................. .... . . .

...

...

.. ...........r..................................................... ...

(0, 16) ............... .... . ... . .. . .

... ..

..

. . . .. ... . . .

.

.

.

.

..

r..

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

.

.

.

.

.

.

.

...

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

.

.

.

.

...

.

.

.

.

.

.

.

. ....

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

. .

...

.

...

. ...

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

.... . . .

. . . .

.

Using the two other points yields

n = 16m + 16 and 36 = m(16 + n) + 16 .

Therefore 36 = m(16 + 16m + 16) + 16 = 16m2 + 32m + 16, so

0 = 16m2 + 32m − 20 = 4(2m + 5)(2m − 1), so x = − 25 or x = 21 .

In the figure, the slope is clearly positive, so m = 12 , and n = 16m + 16 = 24.

Also solved by LISA WANG, student, Port Moody Secondary School, Port Moody, BC.

You can also solve this problem using similar triangles.

4. Friends Maya and Naya ordered finger food in a restaurant, Maya ordering

chicken wings and Naya ordering bite-size ribs. Each wing cost the same amount,

and each rib cost the same amount, but one wing was more expensive than one

rib. Maya received 20% more pieces than Naya did, and Maya paid 50% more in

total than Naya did. The price of one wing was what percentage higher than the

price of one rib?

Solution by Lisa Wang, student, Port Moody Secondary School, Port Moody, BC.

Say Naya gets n pieces at $r each. Then Maya gets 1.2n pieces at, say, $w

each. Then Naya pays $nr and Maya pays $1.2nw. Since Maya pays 50% more

1.5

than Naya, 1.2nw = 1.5nr, so w = 1.2 r = 1.25r, so one wing is 25% more

expensive than one rib.

7

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; and WEN-TING FAN, student, Burnaby North Secondary School, Burnaby,

BC.

opposite corners coincide. What is the length of the crease?

BC.

If you fold the paper as instructed and unfold it again, you obtain the figure

below where the section outlined with thick lines used to overlap and the dashed

line is the crease. The Pythagorean Theorem now yields that

12 − x x

...

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

... .. ...

.. .. ...

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

.... .. ...

92 + x2 = (12 − x)2 ... ..

... ... ..

.... ...

...

..

... .. ...

2

= 144 − 24x + x2 ... .. ...

81 + x 9

. ...

...12 − x 12 − x....

... ..

.. ...

..

... 9

... .. ...

... ..

24x = 63 ... ...

....

... ..

..

.. ...

..

...

63 ... ..

... . ..

.

. ...

...

x = ... .. ..

...

24 .............................................

...

....

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

..

.. ...

..

.

x 12 − x

Then redraw the diagram as in the figure below. Use the Pythagorean Theorem

again:

12 − x x

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

. ...

... .. ..

... .. . ....

... .. .. ....

.... .. .

.. ...

.. .. ...

... .. .. ..

c2 = 92 + (12 − 2x)2 . . .

... .. . ...

.. ...

... .. ....

... .. ..

...

...

c ..

.. 9 ..

.

...

.....

9

63 2025 .. .

Since x = 24

, c2 = 16

, so ...

...

... . ..

.. .

.

..

..

...

...

...

.

.... ...

c = 45

..

4

. ..

... ..

..

..

.. ...

...

... .. .. ...

... .. .. ..

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

x 12 − 2x x

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School, Coquit-

lam, BC; and LISA WANG, student, Port Moody Secondary School, Port Moody, BC.

6. In calm weather, an aircraft can fly from one city to another 200 kilometres

north of the first and back in exactly two hours. In a steady north wind, the round

trip takes five minutes longer. Find the speed (in kilometres per hour) of the wind.

The airspeed of the plane is 4002

= 200 kilometres per hour. Let w denote

the speed of the wind. Then, if you fly with the wind, the ground speed is 200+w;

and if you fly against the wind, the ground speed is 200 − w. Therefore, the plane

200 200

takes hours to fly with the wind and hours to fly against the

200 + w 200 − w

wind. If you add these two expressions, you get the total time for the round trip,

8

5

but this time is given to be 2 + 60

hours, so

200 200 25

+ =

200 − w 200 + w 12

8(200 + w) + 8(200 − w) 1

=

(200 − w)(200 + w) 12

3200 1

=

2002 − w2 12

w = ±40 .

chalk line is drawn from one corner to the diagonally opposite corner. How many

tiles have a chalk line segment on them?

Solution by Gesine Geupel, student, Max Ernst Gymnasium, Brühl, NRW, Ger-

many.

Since 248

= 3 and 408

= 5, consider instead the 3 × 5 rectangle on the left

in the figure. You can easily count that the diagonal crosses seven squares.

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

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

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

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

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

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

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

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

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

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

....

...

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

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

....

...

.... ....

..

..

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

....

. .

.

.

....

....

...

...

...

...

..

...

..

...

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

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

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

..

...

..

...

..

..

..

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

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

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

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

..

...

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

...

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

. ...

. .

.

....

..

.

.

....

..

.

.

....

..

...

...

...

..

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

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

..

..

..

..

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

..

..

..

..

..

..

..

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

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

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

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

side of the figure. The diagonal of the 24 × 40 rectangle is also the diagonal of

each of eight of the 3 × 5 rectangles. Therefore the diagonal crosses 56 of the

1 × 1 squares.

This issue’s prize of one copy of Crux Mathematicorum for the best solu-

tions goes to Wen-Ting Fan, student, Burnaby North Secondary School, Burnaby,

BC.

As Skoliad editors we are quite pleased to see envelopes with “exotic” stamps

in the mail, but receiving more Canadian solutions would be wonderful. The

address is on the inside of the back cover. You do not have to solve the entire

featured contest; a well-presented solution to a single problem is enough. The test

for “well-presented” is that your classmates at school can understand it. You can

send your solution(s) by mail or electronically—even if that means that we miss

out on the exotic stamps.

9

MATHEMATICAL MAYHEM

Mathematical Mayhem began in 1988 as a Mathematical Journal for and by

High School and University Students. It continues, with the same emphasis,

as an integral part of Crux Mathematicorum with Mathematical Mayhem.

The interim Mayhem Editor is Shawn Godin (Cairine Wilson Secondary

School, Orleans, ON). The other staff member is Monika Khbeis (Our Lady of Mt.

Carmel Secondary School, Mississauga, ON).

Mayhem Problems

Please send your solutions to the problems in this edition by 15 August 2011.

Solutions received after this date will only be considered if there is time before publication

of the solutions.

Each problem is given in English and French, the official languages of Canada. In

issues 1, 3, 5, and 7, English will precede French, and in issues 2, 4, 6, and 8, French

will precede English.

The editor thanks Jean-Marc Terrier of the University of Montreal for translations

of the problems.

solutions past the posted due date. Solutions will be accepted until we process

them for publication. Currently we are delayed by about four months. Check the

CMS website, cms.math.ca/crux, for our status in processing problems.

Vazz needs to buy desks and monitors for his new business. A desk costs

$250 and a monitor costs $260. Determine all possible ways that he could spend

exactly $10 000 on desks and monitors.

Square based pyramid ABCDE has a square base ABCD with side length

10. Its other four edges AE, BE, CE, and DE each have length 20. Determine

the volume of the pyramid.

Buzău, Romania

Suppose that x is a real number. Without using calculus, determine the

2x2 − 8x + 17

maximum possible value of and the minimum possible value of

x2 − 4x + 7

x2 + 6x + 8

.

x2 + 6x + 10

10

Buzău, Romania

Determine all pairs (a, b) of positive integers for which a2 +b2 −2a+b = 5.

Let a, b and x be positive integers such that x2 − bx + a − 1 = 0. Prove

that a2 − b2 is not a prime number.

ON

Let bxc denote the greatest integer not exceeding x. For example,

b3.1c = 3 and b−1.4c = −2. Let {x} denote the fractional part of the real

number x, that is, {x} = x − bxc. For example, {3.1} = 0.1 and

{−1.4} = 0.6. Show that there exist infinitely many irrational numbers x such

that x · {x} = bxc.

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

Vazz doit acheter des pupitres et des écrans pour son nouveau commerce. Un

pupitre coûte 250$ et un écran 260$. Trouver de combien de manières possibles

il pourrait dépenser exactement 10 000$ en pupitres et écrans.

Une pyramide ABCDE a une base carrée ABCD de côté 10. Les quatre

autres arêtes AE, BE, CE et DE sont toutes de longueur 20. Trouver le volume

de la pyramide.

M472. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie

Supposons que x soit un nombre réel. Sans utiliser le calcul différentiel,

2x2 − 8x + 17

déterminer la valeur maximale possible de et la valeur minimale

x2 − 4x + 7

x2 + 6x + 8

possible de .

x2 + 6x + 10

M473. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie

Déterminer toutes les paires (a, b) d’entiers positifs tels que

a2 + b2 − 2a + b = 5.

Soit a, b et x trois entiers positis tels que x2 − bx + a − 1 = 0. Montrer

que a2 − b2 n’est pas un nombre premier.

11

M475. Proposé par Edward T.H. Wang, Université Wilfrid Laurier, Waterloo,

ON

Notons bxc le plus grand entier n’excédant pas x. Par exemple, b3,1c = 3

et b−1,4c = −2. Notons {x} la partie fractionnaire du nombre réel x, c.-à-d,

{x} = x − bxc. Par exemple, {3,1} = 0,1 et {−1,4} = 0,6. Montrer qu’il

existe une infinité de nombres irrationnels x tels que x · {x} = bxc.

Mayhem Solutions

M432. Proposed by the Mayhem Staff.

Determine the value of d with d > 0 so that the area of the quadrilateral

with vertices A(0, 2), B(4, 6), C(7, 5), and D(d, 0) is 24.

B(4, 6)

E(0, 6) ........... F (7, 6)

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

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

... ...

Let E = (0, 6), F = (7, 6), G = (7, 0) ..

.. ..

...

...

... .. ...

... C(7, 5)

and Ω denote the area function. Then

.

. ....

...

....

... ....

...

Ω(AOD) = 21 (d × 2) = d; ... ..

. ..

.....

... ...

. ..

..

... ...

A(0, 2) .............

...

Ω(BEA) = 21 (4 × 4) = 8; .............

...

.

....

...

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

...

... ....

Ω(BF C) = 12 (3 × 1) = 23 ; ... ...

....

.......

...

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

G(7, 0)

and Ω(CDG) = 12 (7−d)×5 = 25 (7−d). O D(d, 0)

Since Ω(OEF G) = 7 × 6 = 42, we have

3 5 3

24 = Ω(ABCD) = 42 − d + 8 + + (7 − d) = 15 + d.

2 2 2

Solving we find d = 6.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; NATALIA DESY,

student, SMA Xaverius 1, Palembang, Indonesia; SAMUEL GÓMEZ MORENO, Universidad

de Jaén, Jaén, Spain; AFIFFAH NUUR MILA HUSNIANA, student, SMPN 8, Yogyakarta,

Indonesia; GEOFFREY A. KANDALL, Hamden, CT, USA; WINDA KIRANA, student,

SMPN 8, Yogyakarta, Indonesia; JOSHUA LONG, Southeast Missouri State University, Cape

Girardeau, MO, USA; DRAGOLJUB MILO ŠEVIĆ, Gornji Milanovac, Serbia; RICARD

PEIRÓ, IES “Abastos”, Valencia, Spain; CAO MINH QUANG, Nguyen Binh Khiem High

School, Vinh Long, Vietnam; BRUNO SALGUEIRO FANEGO, Viveiro, Spain; NECULAI

STANCIU, George Emil Palade Secondary School, Buzău, Romania(2 solutions); and JOHN

WYNN, student, Auburn University, Montgomery, AL, USA;

Two incorrect solutions were received.

St. John’s, NL.

In triangle ABC, AB < BC, L is the midpoint of AC, and M is the

midpoint of AB. Also, P is the point on LM such that M P = M A. Prove

that ∠P BA = ∠P BC.

12

Ridgewood, NJ, USA.

B .

..........

.. .. ......

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

.. .. .......

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

........ ..

..

......

.......

.

.. .. ......

... ..

..

......

......

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

... .. ......

M .

.

...........

......

....

..

......

.......

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

......

......

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

... ......

.......

..

.......

.

.... .

.........

.

.

P .......

......

... ....

. ...

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

. .. . ....

..

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

...

.......

.......

. .......

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

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

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

A L C

then BC k M L so ∠P BC = ∠BP M . Also, since P M = AM = BM ,

∆BM P is isosceles. Therefore ∠ABP = ∠BP M which, when combined with

∠P BC = ∠BP M , we conclude that ∠ABP = ∠P BC, as desired.

Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain(two

solutions); GEORGE APOSTOLOPOULOS, Messolonghi, Greece; RICHARD I. HESS,

Rancho Palos Verdes, CA, USA; GEOFFREY A. KANDALL, Hamden, CT, USA; WINDA

KIRANA, student, SMPN 8, Yogyakarta, Indonesia; DRAGOLJUB MILOŠEVIĆ, Gornji

Milanovac, Serbia; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain; BRUNO SALGUEIRO

FANEGO, Viveiro, Spain; HUGO LUYO SÁNCHEZ, Pontificia Universidad Católica del Peru,

Lima, Peru; NECULAI STANCIU, George Emil Palade Secondary School, Buzău, Romania;

EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, ON; and the proposer.

One incorrect solution was received. Several readers pointed out that ∆AP B is right

angled with the right angle at P .

Determine all eight-digit positive integers abcdef gh which satisfy the

relations a3 − b2 = 2, c3 − d2 = 4, 2e − f 2 = 7, and g 3 − h2 = −1.

Since 2 ≤ a3 ≤ 92 + 2 = 83 ⇔ 2 ≤ a ≤ 4 and a3 − 2 for such a can

only be square for a = 3, then a = 3, b = 5.

Since 4 ≤ c3 ≤ 92 + 4 = 85 ⇔ 2 ≤ c ≤ 4 and c3 − 4 for such c can only

be square for c = 2 then c = 2, d = 2.

Since 7 ≤ 2e ≤ 92 + 7 = 88 ⇔ 3 ≤ e ≤ 6 and 2e − 7 for such e can

only be square for e = 3, e = 4 and e = 5 then (e, f ) = (3, 1), (4, 3), (5, 5).

Since 0 ≤ g 3 ≤ 92 − 1 = 80 ⇔ 0 ≤ g ≤ 4 and g 3 + 1 for such g can

only be square for g = 0 and g = 2 then (g, h) = (0, 1), (2, 3).

Thus abcdef gh = 35 223 101, 35 224 301, 35 225 501, 35 223 123,

35 224 323, 35 225 523.

Also solved by RICHARD I. HESS, Rancho Palos Verdes, CA, USA; RICARD PEIRÓ,

IES “Abastos”, Valencia, Spain; CAO MINH QUANG, Nguyen Binh Khiem High School, Vinh

Long, Vietnam; BRUNO SALGUEIRO FANEGO, Viveiro, Spain; EDWARD T.H. WANG,

Wilfrid Laurier University, Waterloo, ON; and the proposer. Seven incomplete solutions were

submitted. Most of the incomplete solutions missed the case where g = 0.

13

Prove that

Ê

X

n

1 1 n(n + 2)

1+ + = .

k=1

k2 (k + 1)2 n+1

Solution by Cao Minh Quang, Nguyen Binh Khiem High School, Vinh Long,

Vietnam.

For any k > 0 we have

2

1 1 1 1 2 2 2

1+ − = 1+ + + − −

k k+1 k2 (k + 1)2 k k+1 k(k + 1)

1 1 2 2

= 1+ 2 + + −

k (k + 1)2 k(k + 1) k(k + 1)

1 1

= 1+ 2 + .

k (k + 1)2

È

1 1 1 1

Hence 1+ k2

+ (k+1)2

= 1 + k

− k+1

. Therefore if we let

Pn È

1 1

S= k=1 1+ k2

+ (k+1)2

then

n

X

1 1

S = 1+ −

k=1

k k+1

1 1 1 1 1 1

= 1+ − + 1+ − + ··· + 1 + −

1 2 3 2 n n+1

n(n + 2) 1

= n+1− = ,

n+1 n+1

Also solved by ARKADY ALT, San Jose, CA, USA; MIGUEL AMENGUAL

COVAS, Cala Figuera, Mallorca, Spain; GEORGE APOSTOLOPOULOS, Messolonghi,

Greece; SAMUEL GÓMEZ MORENO, Universidad de Jaén, Jaén, Spain; G.C. GREUBEL,

Newport News, VA, USA; RICHARD I. HESS, Rancho Palos Verdes, CA, USA;

GEOFFREY A. KANDALL, Hamden, CT, USA; WINDA KIRANA, student, SMPN 8,

Yogyakarta, Indonesia; HUGO LUYO SÁNCHEZ, Pontificia Universidad Católica del Peru,

Lima, Peru; DAVID E. MANES, SUNY at Oneonta, Oneonta, NY, USA; PEDRO HENRIQUE

O. PANTOJA, student, UFRN, Brazil; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain;

PAOLO PERFETTI, Dipartimento di Matematica, Università degli studi di Tor Vergata

Roma, Rome, Italy; DRAGOLJUB MILO ŠEVIĆ, Gornji Milanovac, Serbia; NATALIA DESY,

student, SMA Xaverius 1, Palembang, Indonesia; BRUNO SALGUEIRO FANEGO, Viveiro,

Spain; NECULAI STANCIU, George Emil Palade Secondary School, Buzău, Romania;

EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, ON; and the proposer.

14

Buzău, Romania.

Determine the smallest possible value of x + y, if x and y are positive

2008 x 2009

integers with < < .

2009 y 2010

Let y = x + d then 1 − 2009 1

< y−dy

1

< 1 − 2010 1

so 2009 > yd > 2010

1

. If

d = 1 there is no solution. If d = 2, y = 4019 is a solution so x = 4017 and

x + y = 8036. If d > 2 then x, y > 6000 thus x + y > 12000. Therefore the

minimum value of x + y is 8036.

Also solved by ARKADY ALT, San Jose, CA, USA; SAMUEL GÓMEZ MORENO,

Universidad de Jaén, Jaén, Spain; CAO MINH QUANG, Nguyen Binh Khiem High School,

Vinh Long, Vietnam; DAVID E. MANES, SUNY at Oneonta, Oneonta, NY, USA; EDWARD

T.H. WANG, Wilfrid Laurier University, Waterloo, ON; and the proposer. Five incorrect so-

lutions were submitted.

Alt, Manes and Wang proved in general that if n+1 n

< xy

< n+1

n+2

then the solution with

the smallest sum corresponds to x = 2n + 1, y = 2n + 3 and thus x + y = 4n + 4. In general

for any two fractions of non-negative integers, in lowest terms, ab

c

< d the value a+c

b+d

is called

a a+c c

the mediant and it satisfies < < . If we also have bc − ad = 1 then the mediant is

b b+d d

a c

the fraction with the lowest denominator in the interval ,

b d

.

Let bxc denote the greatest integer not exceeding x. For example,

b3.1c = 3 and b−1.4c = −2. Let {x} denote the fractional part of the real

number x, that is, {x} = x−bxc. For example, {3.1} = 0.1 and {−1.4} = 0.6.

Determine all rational numbers x such that x · {x} = bxc.

The only rational number x such that x · {x} = bxc is x = 0.

If n is an integer, then n · {n} = 0 = bnc = n has the only solution

n = 0. Therefore, 0 is the only integer solution to the equation.

Assume x is a rational number different from an integer such

that

x x

x · {x} = bxc = x − {x}, then {x} = x+1 . Therefore, x x+1 = bxc

implies x2 = bxc(x + 1). Assume x = m n

where m and n are relatively prime

integers and n > 1. Then

m

m2 m+n

= bxc + 1 = bxc .

n2 n n

As a result, m2 = bxc(m + n) · n so that n is a divisor of m2 , a contradiction

since m and n are relatively prime and n > 1 .

Also solved by ARKADY ALT, San Jose, CA, USA; RICHARD I. HESS, Rancho Palos

Verdes, CA, USA; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain; EDWARD T.H. WANG,

Wilfrid Laurier University, Waterloo, ON; and the proposer. Three incorrect solutions were

submitted.

15

Ian VanderBurgh

contests (Look at the solution to question 1 of Skoliad on page 5 – Ed.). This

month and next, we will look at a few of these problems, at least one of which

uses averages in a very subtle way.

Problem 1 (2008 Small c Contest) The average of three numbers is 13. Two

numbers are added to this list so that the average of all five numbers is 17. What

is the average of the two new numbers?

(A) 21 (B) 25 (C) 23 (D) 30 (E) 15

One of the things about average problems that I like is that there are really

only about 1 21 things that you need to know about averages in order to be able

to do almost all of such problems. (That’s not to say that there isn’t a plethora

of tricks of the trade that can be useful...)

The first of these 1 12 important things is how to calculate an average: add

up the given numbers, count the given numbers, and divide the sum by the count

to get the average. The extra 12 thing to remember is that the sum of the numbers

equals the count times the average. Expressing these facts algebraically, we see

that if there are n numbers whose sum is S, then the average, a, satisfies the

S

equation a = . Rearranging this gives S = na. (I concede that occasionally

n

S

we might use the fact that n = as well.)

a

Let’s solve Problem 1 using these properties and then look at our answer to

see what we can observe.

Solution to Problem 1. Since the average of the original three numbers is 13,

then their sum is 3 × 13 = 39. Since the average of all five numbers is 17, then

the sum of the five numbers is 5 × 17 = 85.

The sum of the additional two numbers equals the sum of all five numbers

minus the sum of the original three numbers, or 85 − 39 = 46. Therefore, the

46

average of these two numbers is = 23.

2

This problem is particularly nice, in my opinion, because it doesn’t require

us to use any algebra. Let’s look at the data that we have:

16

Do you notice anything about the position of the overall average relative

to the averages of the first and last numbers? You might have noticed that the

overall average splits these averages in the ratio 4 : 6 which equals 2 : 3, which

happens to be the ratio of the count of numbers in each partial average (arranged

in reverse from what you might quickly guess).

If this rule works in general, then if we had 5 numbers with average 22 and

3 numbers with average 46, the average of all 8 numbers should split 22 and 46

3 3

in the ratio 3 : 5. In other words, the average is = of the way from 22

3+5 8

3

to 46, and so equals 22 + × (46 − 22) = 31. Try solving this problem using

8

the method that we used above to confirm the answer.

Putting this in a more general way, if m numbers have an average of a and

n numbers have an average of b with a < b, then the average of the m + n

numbers splits a and b in the ratio n : m (not m : n). Can you prove this? We’ll

look at another problem next month where this approach is really useful.

Problem 2 (2010 Pascal Contest) In the diagram, each of the five boxes is to

contain a number. Each number in a bold outlined box must be the average of

the number in the box to the left of it and the number in the box to the right of

it. What is the value of x?

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

qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq.q.q..........................................

...

...

qq qq qq qq ....

.

..

qq 26 qqq x .......

qq qq qq q

...

... 8 qq qq

... .

qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq..q.............................................

.

qq qq qq qq

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

Special cases often produce interesting facts. For example, if two numbers

x+y

x and y have an average of a, then = a or x + y = 2a. Try to use this

2

to solve the following problem algebraically.

so the numbers in the boxes are thus 8, y, z, 26, x.

Since the average of z and x is 26, then x+z = 2(26) = 52 or z = 52−x.

We rewrite the list as 8, y, 52 − x, 26, x.

Since the average of 26 and y is 52 − x, then 26 + y = 2(52 − x) or

y = 104 − 26 − 2x = 78 − 2x. We rewrite the list as 8, 78 − 2x, 52 − x, 26, x.

Since the average of 8 and 52−x is 78−2x, then 8+(52−x) = 2(78−2x)

or 60 − x = 156 − 4x and so 3x = 96 or x = 32.

Especially while writing a contest, it’s very tempting to take the answer that

we get and not think about it at all. But let’s actually take this a moment to use

this answer and go back to the list in terms of x (written as 8, 78 − 2x, 52 −

x, 26, x) and substitute to get the list 8, 14, 20, 26, 32.

17

Do you recognize what kind of sequence this list forms? This is an arithmetic

sequence. (Look up this term if you’ve never seen it before.) Do you think that

this is a coincidence? (Hint: The answer to this question is almost always no.)

Let’s think about this by going back to the list 8, y, z, 26, x. Let’s avoid

using algebra, but we’ll keep these labels to make things a little clearer. We are

told that y is the average of 8 and z. The important fact to recognize here is that

y is halfway between 8 and z. In other words, the difference y − 8 equals z − y.

Similarly, z is the average of 26 and y, so 26 − z equals z − y. But there is a

common difference in these two sentences! (And it’s no coincidence that I used

the phrase common difference...)

Since there is this common difference, then all three differences must be

equal. Since 26 − 8 = 18, then each of these differences equals 18 ÷ 3 = 6,

and so the numbers in the sequence are 8, 14, 20, 26, x. Can you extend this

argument another step to explain why x = 32?

So what is the connection between averages and arithmetic sequences? An

arithmetic sequence is a sequence with the property that each term after the first

is the average of the term before and the term after. This is pretty neat, if you’ve

never seen it before. One last thing to think about – the average is sometimes

called the arithmetic mean. Coincidence?

Adams, Douglas (1952 - 2001) The first nonabsolute number is the number

of people for whom the table is reserved. This will vary during the course of the

first three telephone calls to the restaurant, and then bear no apparent relation to

the number of people who actually turn up, or to the number of people who sub-

sequently join them after the show/match/party/gig, or to the number of people

who leave when they see who else has turned up. The second nonabsolute number

is the given time of arrival, which is now known to be one of the most bizarre of

mathematical concepts, a recipriversexcluson, a number whose existence can only

be defined as being anything other than itself. In other words, the given time of

arrival is the one moment of time at which it is impossible that any member of

the party will arrive. Recipriversexclusons now play a vital part in many branches

of math, including statistics and accountancy and also form the basic equations

used to engineer the Somebody Else’s Problem field. The third and most myste-

rious piece of nonabsoluteness of all lies in the relationship between the number

of items on the bill, the cost of each item, the number of people at the table and

what they are each prepared to pay for. (The number of people who have actually

brought any money is only a subphenomenon of this field.) “Life, the Universe

and Everything.” New York: Harmony Books, 1982.

18

No. 291

R.E. Woodrow

This number we begin by looking at the files of solutions by readers to

problems given in the February 2010 number of the Corner, and “A” problems

proposed but not used at the 2007 IMO in Vietnam, given at [2010 : 18–19].

A2. Let n be a positive integer, and let x and y be positive real numbers such

that xn + y n = 1. Prove that

! !

X

n

1 + x2k X

n

1 + y 2k 1

< .

k=1

1 + x4k k=1

1 + y 4k (1 − x)(1 − y)

Solved by Mohammed Aassila, Strasbourg, France; Arkady Alt, San Jose, CA,

USA; Oliver Geupel, Brühl, NRW, Germany; and Titu Zvonaru, Cománeşti,

Romania. We give the solution of Geupel.

From the identity

+ = + =

1+ x4k xk (1 + x4k ) 1+ x4k xk (1 + x4k ) xk

X

n

1 + x2k X

n

1 1 − xn

< = . (1)

k=1

1 + x4k k=1

xk xn (1 − x)

Similarly we have

X

n

1 + y 2k 1 − yn

< . (2)

k=1

1 + y 4k y n (1 − y)

The hypothesis xn + y n = 1 yields

(1 − xn )(1 − y n ) 1 − xn − y n + xn y n

= = 1. (3)

xn y n xn y n

The desired inequality follows immediately from the relations (1), (2), and (3).

f (x + f (y)) = f (x + y) + f (y)

for all x, y ∈ R+ . (Here R+ denotes the set of all positive real numbers.)

19

France. We give Bataille’s version.

We show that the unique solution is the function φ : x 7→ 2x.

solution. Conversely, let f be any solution and x, y be any positive real numbers.

On the one hand, adding x on both sides of the given equation and taking the

images under f , we successively have

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

= f (2x + 2y) + f (y) + f (x + y).

= f (2x + y) + f (y) + f (x + y) + f (y).

It follows that

f (2x + 2y) = f (2x + y) + f (y). (1)

Now, suppose that 0 < a < b. We prove that f (a) < f (b).

, y = b − a in (1) and obtain

f (b) = f (a) + f (b − a), hence f (b) > f (a).

b−2a

• If b > 2a, with x = 2

, y = a, (1) gives f (b) = f (a) + f (b − a) and

f (b) > f (a) again.

• If b = 2a, f (b) = f (2a) = f 2( a2 + a2 ) = f a+ a

+f a

>

2 2

a

f (a) + f 2

> f (a). Thus, f is strictly increasing on (0, ∞) and, as

such, is injective.

f (a + b) = f (a) + f (b) (for f a + 2 · b−a 2

+ f (a) = f 2a + 2 · b−a 2

).

Lastly, let y > 0. Since f (y) 6= y (otherwise f (x + f (y)) = f (x + y) in

contradiction with the given functional equation), we may write

f (y + f (y)) = f (y) + f (f (y)) as well as f (y + f (y)) = f (2y) + f (y).

It follows that f (f (y)) = f (2y) and since f is injective, f (y) = 2y, as desired.

Next we look at the “C” problems proposed but not used at the 2007 IMO

in Vietnam given at [2010: 19–20].

C2. A unit square is dissected into n > 1 rectangles such that their sides are

parallel to the sides of the square. Any line, parallel to a side of the square and

20

intersecting its interior, also intersects the interior of some rectangle. Prove that

one of the rectangles has no point on the boundary of the square.

The proof is by contradiction.

Assume the contrary and consider a counterexample where n is minimal.

Let A be one of the vertices of the square, let E be the locus of the sides of

the rectangles of the dissection, and let ABCD be the rectangle that contains

the vertex A. Since the square is covered by the rectangles, at least one of the

segments BC and DC has an extension in E beyond the point C. Without loss of

generality assume that DC can be extended beyond C where the longest possible

extension in E is up to a point E. Since the line DE intersects the interior of

some rectangle, the point E is an interior point of the square. Since the square is

covered by the rectangles, a segment EF orthogonal to CE where the points A

and F are in the same half-plane relative to the line DE, also belongs to E.

If there were a second rectangle beside ABCD which contains the side

BC, then it could be glued together with the rectangle ABCD obtaining a

counterexample with n−1 rectangles, which contradicts our minimum hypothesis.

Hence, there is a segment GH in E where G is an inner point of BC, and GH

is parallel to CE. The rectangle with vertex C and sides on the lines CE and

CG is now separated from the boundary of the square by the four segments HG,

GC, CE and EF . This is a contradiction which completes the proof.

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

... ...

.... ...

.... ...

... ....

... ..

.. ...

.... C E ...

... ...

D ... ................

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

.... ....

....

..

..

.... .. ... .

.

....

.

.

...

...

...

...

... .. ... ...

...

..

F ....

..

...

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

.. .

.

.... .... ...

... ... ...

... ... ....

... ... ..

.. . ...

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

....

...

... G ..

.

H

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

..

.. . ...

.... .... ...

... ... ...

... .... ....

... .. .

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

A B

for each integer n, and paint each region Sn either red or blue. Prove that any

rectangle whose side lengths are distinct positive integers may be placed in the

plane so that its vertices lie in regions of the same colour.

We generalize that any rectangle ABCD with distinct real sides AB = a

and BC = b may be placed so that its vertices lie in regions of the same colour.

Firstly consider the case where a ∈/ Z. Without loss of generality, we

may assume that S0 and S1 have distinct colours. We place ABCD so that

A and D have the common x-coordinate 1 − {a} while B and C have the

21

common x-coordinate dae. We then translate the rectangle with offset {a} in the

positive x-direction. With this translation, A and D move from S0 to S1 , while

B and D remain in Sdae . Consequently, in one of these two positions the vertices

A, B, C, D lie in regions of the same colour.

It remains to consider a, b ∈ Z. Let d = gcd(a, b), and let a0 , a1 , b0 , b1

be integers such that a = a0 d, b = b0 d, and a0 a1 + b0 b1 = 1. The proof is

by contradiction. Suppose ABCD cannot be placed properly. Then S0 and Sa

have distinct colours, and S0 and Sb have also distinct colours. By induction, S0

and Sua+vb , where u and v are integers, have distinct colours if and only if u + v

is odd. By b0 a − a0 b = 0, we see that b0 − a0 is even. Since a0 and b0 are

coprime, both are odd. Hence, a1 + b1 is odd, too. Thus, S0 and Sa1 a+b1 b = Sd

have distinct colours. We conclude that S0 and S2d have the same colour.

Assume a < b, which implies b0 ≥ 3. We pitch the rectangle so that the

x-coordinates of A and D as well as the x-coordinates of B and C have distance

2d. It suffices to prove that the we can translate it so that A and B lie in regions

of the same colour. If the angle between the x-axis and the line AB is ϕ and

A0 and B 0 are the projections of A and B, respectively, onto the x-axis, then we

2d 2 aÈ 2

obtain sin ϕ = = and A0 B 0 = a cos ϕ = b0 − 4 ∈

/ Z. By the first

b b0 b0

0 0

part of the proof, we can place A and B so that they lie in regions of the same

colour. This completes the proof.

C

..... ....

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

........

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

... ...

...

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

...

.

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

... ..

....

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

... ..

....

......

. ...

... ..

..

..

......

. ...

.... D .....

.. .. .

...

..

...

.. .. ... .

... ...

.. ... ..

... .. .. . b ...

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

.... .. .. ..

...

.. .. .... ...

... .. .

.. ..

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

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

.. . ...

... .. ... ...

.. ... ..

... .. .. ...

... ..

... ...

... ... .

... ..

..

..

... . ..

..

..... ..

.

. B

.........

... ... .

...

..

.

... .. . a .

..

...

....... ..

... ..... ..

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

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

.... .. ..

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

. ..

. . . ϕ ..........

. ...

. .. .

. .

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

..

.

.... 2d

..

A B 0

C7. A convex n-gon P in the plane is given. For every three vertices of P , the

triangle determined by them is good if all its sides are of unit length. Prove that

P has at most 32 n good triangles.

This is not an original problem. It first appeared in J. Pach and

R. Pinchasi, How many unit triangles can be generated by n points in convex

position?, American Math. Monthly 110 (5) 2003 : 40–406.

Next we move to the “G” problems proposed but not used at the 2007 IMO

22

sides BC, CA, AB, respectively. Let P be a variable point on the circumcircle

of ABC. Let lines P A1 , P B1 , P C1 meet the circumcircle again at A0 , B 0 , C 0

respectively. Assume that the points A, B, C, A0 , B 0 , C 0 are distinct, and that

the lines AA0 , BB 0 , CC 0 form a triangle. Prove that the area of this triangle

does not depend on P .

We denote by Γ the circumcircle of 4ABC and by A0 , B0 , C0 the points of

intersection of the lines BB 0 and CC 0 , CC 0 and AA0 , AA0 and BB 0 ,

respectively. We will use areal coordinates relatively to (A, B, C). The

equation of the circle Γ is known to be a2 yz + b2 zx + c2 xy = 0 where A = BC,

b = CA, c = AB. Let P (x0 , y0 , z0 ) and A0 (x0 , y 0 , z 0 ); note that the line AA0

has equation yz 0 − zy 0 = 0 and that (x0 , y0 , z0 ), (x0 , y 0 , z 0 ) are solutions to the

system a2 yz + b2 zx + c2 xy = 0, x(y0 − z0 ) = x0 (y − z). It readily follows

y0 y 0

that , are solutions of an equation (with unknown U )

z0 z 0

(c2 x0 )U 2 + λU + (−b2 x0 ) = 0

y0 y 0 b2 y0 z0

for some real λ. From · 0 = − 2 , we deduce = so that the

z0 z c −b2 z0 c2 y0

equation of AA0 is (c2 y0 )y + (b2 z0 )z = 0. Similarly, the equations of BB 0 and

CC 0 are (c2 x0 )x + (a2 z0 )z = 0 and (b2 x0 )x + (a2 y0 )y = 0. Then, it is easily

obtained that

A0 (−a2 y0 z0 , b2 x0 z0 , c2 x0 y0 ), B0 (a2 y0 z0 , −b2 x0 z0 , c2 x0 y0 ),

C0 (a2 y0 z0 , b2 x0 z0 , −c2 x0 y0 ).

Now, recall that if Mi (xi , yi , zi ) with xi + yi + zi = 1 for

i = 1, 2, 3, the

ratio

x 1 y2 x 3

area(M1 M2 M3 )

is the absolute value of the determinant y1 y2 y3 . Here,

area(ABC) z1 z2 z3

area (A0 B0 C0 )

we have = |4| where

area (ABC)

a2 y0 z0 b2 x0 z0

4 = ·

−a2 y0 z0 + b2 x0 z0 + c2 x0 y0 a2 y0 z0 − b2 x0 z0 + c2 x0 y0

−1 1 1

2

c x 0 y0

· 2 · 1 −1 1 .

a y0 z0 + b2 x0 z0 − c2 x0 y0 1 1 −1

(a2 y0 z0 )(b2 x0 z0 )(c2 x0 y0 )

4 = ·4

(−2a2 y0 z0 )(−2b2 x0 z0 )(−2c2 x0 y0 )

23

and

1

area (A0 B0 C0 ) = area (ABC) ,

2

independent of P .

G5. Triangle ABC is acute with ∠ABC > ∠ACB, incentre I, and

circumradius R. Point D is the foot of the altitude from vertex A, point K

lies on line AD such that AK = 2R, and D separates A and K. Finally, lines

DI and KI meet sides AC and BC at E and F , respectively.

Prove that if IE = IF then ∠ABC > 3∠ACB.

The claim is false. We prove instead that ∠ABC ≤ 3∠ACB.

We write a = BC, b = CA, c = AB, 2s = a + b + c, h = AD,

α = ∠BAC, β = ∠ABC, γ = ∠ACB. We denote by r the inradius, by

J the point on the segment AD such that DJ = 2r, and by M and N the

perpendicular projections of I onto AC and BD, respectively.

A

.

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

.... ....

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

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

.....

.. ... .....

.. ..

........ E

.

.. .

... ....

. ..

... ....

.....

.....

.....

.....

.

... ......

J . ..

... ... M .. ..... .........

. b

c .. ...

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

.....

.

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

.. ... . .....

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

... ...

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

.. h ... ..

.

... .....

.....

.

.

.

...

. I .... ...

...

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

.

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

.....

..

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

.....

.....

.....

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

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

...

...

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

...

...

..

..

....

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

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

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

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

.....

.....

...

...

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

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

B

.

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

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

.

C

D .F. N a

... ..

... ...

.....

......

..

K

By standard formulas, we have

r2 (s − a)(s − b)(s − c) bc (s − a)bc

AI 2 = = · = ,

sin(A/2) s (s − b)(s − c) s

abc [ABC] abc

2Rh = bc, 4Rr = · = .

[ABC] s s

AI AK

We obtain AI 2 = 2R(h − 2r) = AK · AJ; hence = . By

AJ AI

∠IAJ = ∠KAI, it follows that 4AIJ ∼ 4AKI. Thus,

∠AIJ = ∠AKI = ∠IKD. Recognizing the isosceles 4IJD, we deduce

∠AJI = 180◦ − ∠DJI = 180◦ − ∠IDJ = ∠IDK.

We obtain 4AIJ ∼ 4IKD and consequently

α β−γ

∠DIK = ∠JAI = ∠BAI − ∠BAD = − (90◦ − β) = .

2 2

24

is an inner point of the segment CF . Now, if M is between C and E, then

β−γ

γ = 180◦ − ∠M IN = ∠DIN + ∠EIM > ∠DIK = ; consequently

2

β < 3γ. On the other hand, if the point E is between C and M , or E = M ,

β−γ

then γ = 180◦ − ∠M IN = ∠DIK = which implies β = 3γ. We are

2

done.

Next we move to the “N” problems proposed but not used at the 2007 IMO

in Vietnam, given at [1020 : 21].

N1. Find all pairs (k, n) of positive integers for which 7k − 3n divides k4 + n2 .

We will demonstrate that there is only one such pair, namely

(k, n) = (2, 4). Suppose then that k, n are positive integers such that

(7k − 3n ) | (k4 + n2 ); which means that

8 9

< k4 + n2 = r · (7k − 3n ), =

for some nonzero integer r (1)

: ;

k, n ∈ Z+ .

First we show that both k and n must be even. We do so by ruling out the

other three possibilities: both k and n being odd, k odd and n even, or (third

possibility) k even and n odd.

Possibility 1. both k and n are odd: k ≡ n ≡ 1 mod 2.

We then have k4 ≡ n2 ≡ 1 (mod 8) (the square of an odd integer is

congruent to 1 modulo 8. And so,

k4 + n2 ≡ 1 + 1 ≡ 2 (mod 8) . (2)

Thus

≡ 1 · 7 − 1 · 3 ≡ 7 − 3 ≡ 4 (mod 8) .

dividing k4 + n2 ; is 21 = 2.

This renders equation (1) contradictory or impossible. Hence possibility 1

is ruled out.

Possibility 2. k is odd and n is even; k ≡ 1 (mod 2), n ≡ 0 (mod 2).

25

1 (mod 2), which renders (1) contradictory modulo 2: the left-hand side is

congruent to 0 modulo 2. So this possibility is ruled out as well.

Possibility 3. k is even and n odd; k ≡ 0 (mod 2), n ≡ 1 (mod 2).

Same argument as in Possibility 2; the left-hand side is congruent to 1 but

the right-hand side is zero modulo 2. So this possibility is eliminated as well.

We conclude that both positive integers k and n must be even:

§ ª

k = 2K and n = 2N,

(4)

for some positive integers K and N.

other hand,

75 = 16807 > 16 · 54 = 10, 000

(while 7K < 16K 4; for K = 1, 2, 3, 4). An easy induction shows that

7K > 16 · K 4 for K ≥ 5, we omit the details. Similarly we have

81 = 34 > 4 · 42 = 64 (while 3N < 4N 2 for N = 1, 2, 3). And an easy

induction establishes that 3N > 4N 2 , for N ≥ 4.

Therefore for K ≥ 5 and N ≥ 4 we have 7K + eN > 16K 4 + 4N 2 ; which

implies that

(7K + 3N ) · |7K − 3N | · |r| > 16K 4 + 4N 2 (6)

since |r| · |7K − 3N | is a positive integer.

Clearly (6) condtradicts (5). We have demonstrated that a necessary

condition for (5) to hold true is K ≤ 4 and N ≤ 3; which means that there

is only up to 12 possible pairs (K, N ) that may satisfy (5). We form the

following table:

7K + 3N 24 · K 4 + 22 · N 2

K = 1, N = 1 10 20 = 16 + 4

K = 1, N = 2 16 32 = 16 + 16

K = 1, N = 3 34 52 = 16 + 36

K = 2, N = 1 52 256 + 4 = 260

K = 2, N = 2 58 256 + 16 = 272

K = 2, N = 3 76 256 + 36 = 296

K = 3, N = 1 346 (16)(81) + 4 = 1300

K = 3, N = 2 352 (16)(81) + 16 = 1312

K = 3, N = 3 370 1296 + 36 = 1332

K = 4, N = 1 2404 4096 + 4 = 4100

K = 4, N = 2 2410 4096 + 16 = 4112

K = 4, N = 3 2428 4096 + 36 = 4132

The above table shows that the only pairs (K, N ) for which 7K + 3N

divides 24 · K 4 + 22 · N 2 are (K, N ) = (1, 1), (1, 2). However, the pair (1, 1)

26

does not satisfy (5) since 20 = r · 4 · 10, is impossible with r ∈ Z. The other

pair, (K, N ) = (1, 2) does: 32 = r · (−2)(16), satisfied with r = −1. Thus

(K, N ) = (1, 2) is the only pair; and so (k, n) = (2K, 2N ) = (2, 4).

N2. Let b, n > 1 be integers. Suppose that for each k > 1 there exists an

integer ak such that b − an n

k is divisible by k. Prove that b = A for some integer

A.

To show that b is the nth power of an integer, it suffices to show that for

every prime number p in the prime factorization of b, if pe is the power of p that

appears in the prime factorization of b, then the exponent e is a multiple of n.

We set b = pe · r, r a positive integer such that p does not divide r; (r, p) = 1,

e a positive integer.

We apply the hypothesis of the problem with k = (pe )n = pen : k | b − an k

means that there exists a positive integer λ such that

b − an

k = k · λ;

b − an

k = p

e·n

· λ;

and since b = pe · r, we obtain

§ ª

pe · r − a n

k = p

e·n

· λ,

(1)

e, r, ak , λ positive integers such that (r, p) = 1

§ ª

ak = pt · bk ,

(3)

t, bk positive integers such that (bk , p) = 1

pe · r − pn·t · bn

k = p

e·n

· λ. (4)

Possibility 1. n · t < e, or,

Possibility 2. e < n · t holds.

If Possibility 1 holds then, by (2) we have:

1 ≤ n · t < e , e · n. (5)

p(e−n·t) · r − bn

k = p

c·n−n·t

· λ, (6)

27

(3).

If possibility 2 holds, then by (2),

r − pn·t−e · bn

k = p

e·n−e

·λ (8)

Thus (7) and (8) imply that p | r, contrary to (1).

We have proved that e = n · t; and so b must be the nth power of an integer.

N4. For every integer k ≥ 2, prove that 23k divides the number

2k+1 2k

−

2k 2k−1

but 23k+1 does not.

This is not an original problem. It appeared first in D.B. Fuchs and M.B. Fuchs,

Arithmetic of binomial coefficients, KVANT 6 (1970).

N5. Find all surjective functions f : N → N such that for every m, n ∈ N and

every prime p, the number f (m + n) is divisible by p if and only if f (m) + f (n)

is divisible by p. (N is the set of all positive integers.)

The identity f (n) = n is such a function. We prove that it is the unique

solution. Suppose that f is any solution.

We show by contradiction that f is injective. Assume the contrary. Consider

the equivalence relation on N defined by m ∼ n if and only if f (m) and f (n)

have the same prime divisors. By assumption, there are numbers m < n such

that f (m) = f (n). For every k ∈ N and every prime p, we have

is equivalent to s − (n − m). Let p be a prime such that the least number s

with the property p | f (s) is greater than n. We have s ∼ s − (n − m), but

p - f (s − (n − m)), a contradiction. This completes the proof that f is injective.

We show by contradiction that

f (1) = 1. (1)

were a prime p such that p | f (1). Then, for every n ∈ N, we

Xn

would have p f (1) and, by Mathematical Induction, p | f (n). Therefore, f

k=1

is not surjective, a contradiction, which completes the proof of (1).

28

|f (m + 1) − f (m)| = 1 (2)

Suppose contrariwise that for any number m ∈ N there is a prime p such that

p | f (m + 1) − f (m). Since f is surjective, there is a number n ∈ N such

that p | f (m) + f (n). We obtain p | f (m + n) and p | f (m + 1) + f (n);

hence p | f (m + n + 1). Thus, p | f (1), which contradicts (1). This proves that

f (m + 1) = f (m) + 1.

From (1) and (2) and the injectivity of f , it follows by Mathematical

Induction that f (n) = n for every n ∈ N.

2006 given at [2010 : 22].

1. A circle is divided into 2n congruent sectors, n of them coloured black and the

remaining n sectors coloured white. The white sectors are numbered clockwise

from 1 to n, starting anywhere. Afterwards, the black sectors are numbered

counter clockwise from 1 to n, again starting anywhere.

Prove that there exist n consecutive sectors having the numbers from 1 to

n.

This problem appeared in the 20th Tournament of the Towns, Spring 1999,

A-level, problem 4. The author is V. Proizvolov.

2. Let Q+ (resp. R+ ) denote the set of positive rational (resp. real) numbers.

Find all functions f : Q+ → R+ that satisfy

f (xy)

f (x) + f (y) + 2xyf (xy) = for all x, y ∈ Q+ .

f (x + y)

of Pittsburgh, Pittsburgh, PA, USA. We give Bataille’s solution.

1

We show that the only solution is x 7→ 2 .

x

Let f satisfy the given functional equation, denoted by (E) in what follows.

With x = y = 1, (E) readily gives f (2) = 41 . Let a = f (1). With y = 1 in

(E), we obtain

f (x)

f (x + 1) = , (1)

(2x + 1)f (x) + a

1

which successively yields f (3) = f (2 + 1) = and f (4) = f (3 + 1) =

4a + 5

1

.

4a2 + 5a + 7

29

1

However, from (E) with x = y = 2, we also deduce f (4) = 16 . Comparing

2

with (1), we see that 4a + 5a − 9 = 0, and, since a > 0, it follows that

a = f (1) = 1.

1 1

Now, (1) rewrites as = + 2x + 1 and an easy induction

f (x + 1) f (x)

shows that

1 1

= + 2nx + n2 (2)

f (x + n) f (x)

1

for all positive integers n and rationals x. Thus, f (n) = 2 for all n in N (x = 1

n

in (2)) and

1 1

1 = + 2 + n2 .

f(n + n) f (1/n)

1

Comparing with the result given by (E) with x = n and y = n

, we have

2

1 1 1

f − n2 − f −1 =0

n n2 n

1

and f n

= n2 follows.

1

Lastly, if m, n ∈ N, taking x = m and y = n

in (E) and using (2) lead

to

1 22m m m 1 2m 2

+ n + f = f + +m

m2 n n n n2 n

m n 2 1

and a short calculation gives f n

= m

. As a result, f (x) = for all

x2

rational numbers x.

1

Conversely, it is easily checked that x 7→ satisfies (E) for all rational

x2

numbers x, y.

3. The point P lies inside the acute-angled triangle ABC and C 0 , A0 and B 0

are the feet of the perpendiculars from P to AB, BC, CA. Find all positions of

P such that ∠BAC = ∠B 0 A0 C 0 and ∠CBA = ∠C 0 B 0 A0 .

A

......

... .....

.... ......

.. ....

.. ... ....

....

. ....

.... ....

0 .. . ....

C .. ....

............................ .... B 0

.

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

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

. . ...

.

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

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

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

.

. . . .

....

. ........ ..

. ..

. .. .....

.

.

.

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

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

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

. .. ..

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

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

B ......................................................................................................... . .

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

0 C

A

30

∠BAC = ∠B 0 A0 C 0 = ∠B 0 A0 P + ∠P A0 C 0

= ∠P CA + ∠P BA.

It results that

∠BP C = 180◦ − ∠P BC − ∠P CB

= 180◦ − (∠ABC − ∠P BA) − (∠BCA − ∠P CA)

= 180◦ − ∠ABC − ∠BCA + ∠P BA + ∠P CA

= ∠BAC + ∠BAC = 2∠BAC,

We deduce that the point P lies on an arc of a circle which passes through

B and C.

Similarly, we deduce that the point P lies on an arc of a circle which passes

through A and C.

It follows that there exists at most one point P satisfying the given condition.

Since it is easy to see that the circumcentre satisfies the problem (the medial

triangle is similar to the given triangle), we conclude that the desired point is the

circumcentre.

Mathematik 2007 given at [2010 : 22].

1. Show that one can distribute the integers from 1 to 4014 on the vertices and

the midpoints of the sides of a regular 2007-gon so that the sum of the three

numbers along any side is constant.

Solved by Chip Curtis, Missouri Southern State University, Joplin, MO, USA.

Starting at any vertex number the vertices counterclockwise from v1 to

v2007. Setting v2008 = v1 , label as mk the midpoint of side vk vk+1 . For

k = 1, 2, . . . , 2007, assign the integer 2k − 1 to mk . For k = 1, 2, . . . , 1004,

assign the integer 2k to v2009−2k , and for k = 1005, 1006, . . . , 2007, as-

sign the integer 2k to v4016−2k. Then mj has the value 2j − 1, v2j has the

value 4016 − 2j, and v2j+1 has the value 2008 − 2j. Hence, the sum on side

v2j m2j v2j+1 is (4016 − 2j) + (4j − 1) + (2008 − 2j) = 6023, and the sum

on side v2j−1 m2j−1 v2j is (20080(2j − 2)) + (4j − 3) + (4016 − 2j) = 6023.

(a) The sum of three (not necessarily different) red numbers is red.

(b) The sum of three (not necessarily different) green numbers is green.

(c) There is at least one green number and one red number.

31

Solved by Chip Curtis, Missouri Southern State University, Joplin, MO, USA; and

by Titu Zvonaru, Cománeşti, Romania. We give Zvonaru’s response.

Let R be the set of red numbers and G be the set of green numbers.

Let a, b be two positive integers such that a 6= b, a ∈ R, b ∈ G. It is easy

to see that

and

It results that

positive integers.

If we suppose that the even integer 2k ∈ R, then an easy induction shows

that every integer t, with t ≥ 2k belongs to R.

Since 1 ∈ R and 2k ∈ R, then 1 + 1 + 2k = 2(k + 1) ∈ R and so on.

We deduce that {t, t ≥ 2k} ⊂ R, hence the set G is finite — a contradiction

with (1).

It results that we have two possibilities:

• (i) R is the set of all odd positive integers; G is the set of all even positive

integers.

• (ii) R is the set of all even positive integers; G is the set of all odd positive

integers.

3. In triangle ABC the points E and F lie in the interiors of sides AC and BC

(respectively) so that |AE| = |BF |. Furthermore, the circle through A, C and

F and the circle through B, C and E intersect in a point D 6= C.

Prove that the line CD is the bisector of ∠ACB.

Solved by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; and by Michel

Bataille, Rouen, France. We give the solution of Amengual Covas.

32

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

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

.........

...........

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

......... C

.......

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

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

..... ....

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

......

.

. ........ ..

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

.... .....

.

.....

. . . . .....

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

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

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

....

...

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

... ...... .

. ...... ..

.

. .. .. ...

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

... .. . ...

...

. .. .... ..

... .

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

... ... . .. .

.

. ... ...

...

... ..

. .. .... .

. . ..

. .... .

. ... . ..

.. . ... . .

... .... ..

.... ...

.

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

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

.. ....

.

..

..

....

.. . ........

. ... q F .. ..

...

... Eq . ......

.

......

. .

.

.

.

. ....

.

.

. ...

.

...

.

... . .

..... . .

. ..

. . ...

.. . . .

..

.. ..

.

.... ... ..

. .. ... ..

.

. . . .

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

. ...

. ..

. ...

... .. .

.

.

.. .. .. .

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

.

..

..

..

.. ... ...

.. ....

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

... ......... q . ....... ..

.

... 0 ............ .

. 0 ...

.

..... .. 0 .....

A ...

...

...

E ... .....

... .....

..

..

..

D .... ...

..... .....

F ...

. B

.... . . .

. ..... . ..

.

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

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

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

.... .....

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

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

.....

..... . . . .

..

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

......

........

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

........

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

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

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

Let the circle through A, C, and F intersect AB at F 0 6= A.

We denote by D 0 the point where the line CD intersects AB.

Observing that E, E 0 , B and C are concyclic, as are F , F 0 , A and

C, we have AE · AC = AE 0 · AB and BF · BC = BF 0 · BA implying

AE · AC AE 0 · AB

= which simplifies to

BF · BC BF 0 · BA

AC AE 0

= (1)

BC BF 0

because AE = BF by hypothesis.

Also, since E, E 0 , B and C are concyclic, as are F , F 0 , A, and C,

then DD 0 · D 0 C = BD 0 · D 0 E 0 and DD 0 · D 0 C = AD 0 · D 0 F 0 implying

AD 0 · D 0 F 0 = BD 0 · D 0 E 0 .

Hence

AD 0 D0E 0

=

BD 0 D0F 0

AD 0 − D 0 E 0

=

BD 0 − D 0 F 0

AE 0

=

BF 0

AC

= by (1).

BC

bisector of ∠ACB. That is, the line CD bisects ∠ACB.

jxk

x

= ?

a a+1

33

Solved by Michel Bataille, Rouen, France; Chip Curtis, Missouri Southern State

University, Joplin, MO, USA; Konstantine Zelator, University of Pittsburgh,

Pittsburgh, PA, USA; and Titu Zvonaru, Cománeşti, Romania. We give the

solution of Bataille.

a(a + 1)

We show that the required number is .

2

By long division, we may write any nonnegative integer x as qa(a + 1) + r

for some nonnegative integers q and r such that r < a(a +1). Using the fact that

bm + uc = m + buc for m ∈ Z and u ∈ R, the proposed equation becomes

jrk

r

q+ = . (1)

a a+1

j k j k

r r r r

Now, ≥ , hence ≥ and (1) cannot be satisfied if q ≥ 1.

a a+1 a a+1

Thus, the desired number is also the number of elements r ∈ A with

A = {0, 1, 2, . . . , a(a + 1) − 1} such that

jr k

r

= . (2)

a a+1

Any rj ∈kA satisfies ka ≤ r < (k+1)a for some unique k ∈ {0, 1, 2, . . . , a} and

r r (k + 1)a

then = k. Observing that such an r satisfies < < k + 1,

a a+1 a+1

r

we see that this r is a solution if and only if ≥ k, that is r ≥ ka + k. As

a+1

a result, solutions r with ka ≤ r < (k + 1)a exist if and only if k ≤ a − 1, in

which case the solutions are ka + k, ka + k + 1, . . . , ka + a − 1. Thus, for

each k ∈ {0, 1, 2, . . . , a − 1}, we obtain (a − 1) − (k − 1) = a − k solutions

and no other solution exists. In conclusion the number of solutions is

a(a + 1)

(a − 0) + (a − 1) + · · · + (a − (a − 2)) + (a − (a − 1)) = .

2

Next we move to the March 2010 number of the Corner and solutions to

problems of the Republic of Moldova Selection tests for BMO 2007 and IMO 2007

given at [2010 : 81–83].

1. In triangle ABC the points M , N and P are the midpoints of the sides

BC, AC and AB, respectively. The lines AM , BN and CP intersect the

circumcircle of ABC at A1 , B1 and C1 , respectively. Prove that the area of

the triangle ABC does not exceed the sum of the areas of the triangles BA1 C,

AB1 C and AC1 B.

Bataille, Rouen, France; by Prithwijit De, Homi Bhabha Centre for Science

Education, Mumbai, India; and by Titu Zvonaru, Cománeşti, Romania. We give

the solution of De.

34

side of a triangle with length a.

We see at once that

[BA1 C] M A1 [AB1 C] N B1 [AC1 B] P C1

= , = , = .

[ABC] AM [ABC] BN [ABC] CP

Let BC = a, CA = b and AB = c. Chords AA1 and BC of the circumcircle

of triangle ABC intersect at M . Therefore AM · M A1 = BM · M C. Also M

a2

is the midpoint of BC. Therefore BM = M C = 12 a and hence M A1 = .

4AM

Thus 2

M A1 a2 a

= = .

AM 4AM 2 2ma

2 2

N B1 b P C1 c

Similarly we can show that = and = . Therefore

BN 2mb CP 2mc

[BA1 C] [AB1 C] [AC1 B] 1 a2 b2 c2

+ + = + + ... (1)

[ABC] [ABC] [ABC] 4 m2a m2b m2c

Recall that

4

a2 = (2(m2b + m2c ) − m2a )

9

4

b2 = (2(m2c + m2a ) − m2b )

9

4

c2 = (2(m2a + m2b ) − m2c ).

9

Using these in (1) we obtain

1 a2 b2 c2 1 1 1 1

+ + = 2 x+y+z+ + + − 3 . . . , (2)

4 m2a m2b m2c 9 x y z

where

m2a m2b m2c

x = , y = , and z = .

m2b m2c m2a

Now, 2(x + y + z + x1 + y1 + 1z ) − 3 ≥ 9 because x + 1

x

≥ 2, y + 1

y

≥ 2 and

z + z1 ≥ 2. Thus from (1) and (2) we can conclude that

integers, and σ a permutation of the set A = {1, 2, . . . , p}. Prove that the set A

contains 2 distinct numbers k and l such that p divides mk · mσ(k) − ml · mσ(l) .

Solved by Prithwijit De, Homi Bhabha Centre for Science Education, Mumbai,

India.

35

C = {0, 1, 2, . . . , p−1}. Let mk ≡ 0 (mod p) for some k ∈ A and σ(k) 6= k.

Then there exists some l ∈ A, l 6= k such that σ(l) = k. Thus we obtain two

distinct positive integers m and l such that p | (mk mσ(k) − ml mσ(l) ).

Now suppose for some positive integer j ∈ A we have mj ≡ 0 (mod p)

and σ(j) = j. Consider the set A0 = A − {j}.

Claim. There exist distinct positive integers k and l in A0 such that

p | (mk mσ(k) − ml mσ(l) ).

To prove it assume on the contrary that no such pair of positive integers

exists. Then {mr mσ(r) (mod p) : r ∈ A0 } = {1, 2, . . . , p − 1}. Now observe

that Y

mr mσ(r) ≡ (p − 1)! (mod p) . . . . (1)

r∈A0

! !

Y Y Y

mr mσ(r) ≡ mr mσ(r) ≡ ((p − 1)!)2 (mod p)) . . . .

r∈A0 r∈A0 r∈A0

(2)

From (1) and (2) we conclude that (p − 1)! ≡ 1 (mod p) and this contradicts

Wilson’s Theorem. Thus our assumption is wrong and the claim is correct.

m(∠AT B) = m(∠BT C) = m(∠CT A) = 120◦ .

Prove that the Euler lines of the triangles AT B, BT C and AT C are concurrent.

A

.......

..........

Let M be the midpoint of BC, and let Ga be .. ...... .....

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

. .. .

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

the centroid of the triangle BT C. Let A0 be the ... .. ..

.. .. ..

....

....

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

....

point on the other side of BC as A such that ..

...

.. ...

.. ..

.. ..

....

....

....

.. .. ...

4BA0 C is equilateral. .. ... .. ....

.. ... ....

... .. ..

....

....

.. .. . ....

It is known that the point T lies on AA0 . ....

..

..

..

..

...

..

...

....

....

....

....

... ....

Since ∠BT C + ∠BA0 C = 180◦ , the

..

... ..

..

.

G ..

.... ....

....

..

... T ...

. ..

.. . .

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

....

0 ... ... ..... .......................

quadrilateral BA CT is cyclic; it results ...

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

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

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

....

....

....

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

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

.

that the circumcentre Oa of 4BA0 C is the .. ...... G

.........................................................................a

..

.

...... ..

.. ......

.. ....

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

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

B ... ... ........................................................................................................... C

circumcentre of 4BT C. We deduce that ...

...

...

..

..

.. M

.. ..

.. ..

.

.. .. . ...

..

... .. ..

.. .. ...

Oa Ga is the Euler line of 4BT C. Since ...

...

...

..

..

...

..

.. ..

....

.

.

...

.

.

.. .... ...

Ga M 1 M Oa ... ..

..

....

.. ..

...

= = , we obtain that ...

...

...

..

.. O ..

.

....

a

...

..

Ga T 3 Oa A

0 ...

...

..

... ..

..

...

.. ....

Oa Ga kA0 T . ...

...

...

..

..

.. ...

.. ..

.

...

.. ...

... ..

Now, denoting G = Oa Ga ∩ AM , by similarity, ...

...

.. ..

.. ...

.. .

. . .....

...

... . .

GM O M 1 ... .... .... ...

it results that = a 0 = , hence G is ... .. ..

... ... ...

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

...

...

GA Oa A 3 ... .. .. ...

........ ..

the centroid of 4ABC. ........ ...

.........

A0

36

Similarly, we deduce that the Euler lines of 4CT A and 4AT B pass

through the centroid of 4ABC, hence the three Euler lines are concurrent.

perfect cube;

perfect square.

Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA; and Titu

Zvonaru, Cománeşti, Romania. We use Manes’ solution.

For part (a), m = 1 since there exist three consecutive positive integers

whose cubes sum to a perfect cube; namely 33 + 43 + 53 = 63 . For part (b), we

will show that k = 5.

Let s(n, 2k + 1) = n2 + (n + 1)2 + · · · + (n + 2k)2 be the sum of the

squares of 2k + 1 consecutive integers, the smallest of which is n. If k = 1, then

modulo 3 for any integer x.

If k = 2, then

= 5(n2 + 2) ≡ 2 or 3 (mod 4) .

every integer x.

If k = 3, assume that s(n − 3, 7) = r 2 for some integer r. This equation

reduces to 7(n2 + 4) = r 2 . Hence, 7 divides r so that r = 7t for some integer

t. Therefore, n2 + 4 = 7t2 and so, n2 + 4 ≡ 0 (mod 7) or n2 ≡ 3 (mod 7),

a contradiction since 3 is not a quadratic residue of 7. Thus, s(n − 3, 7) is not a

perfect square.

If k = 4, assume that s(n − 4, 9) = r 2 for some integer r. This equation

reduces to 3(3n2 + 20) = r 2 . Therefore, 3 divides r so that r 2 = 9t2 for some

integer t, whence 3n2 + 20 = 3t2 . Thus, 3 divides 20, a contradiction that shows

s(n − 4, 9) is not a perfect square.

If k = 5, assume that s(n − 5, 11) = m2 for some integer m. Then

s(n−5, 11) = 11(n2 +10) = m2 implies that 11 divides m so that m2 = 112 t2

for some integer t. Therefore n2 + 10 = 11t2 or n2 − 1 ≡ 0 (mod 11). Thus,

n ≡ ±1 (mod 11), and so n = 11j ± 1 for some integer j. Then

= 112 [11j 2 ± 2j + 1] = 112 [10j 2 + (j ± 1)2 ].

37

The problem now reduces to finding the smallest value of j so that 10j 2 +(j ±1)2

is a perfect square. The value j = 1 is easily dispensed with. However, for j = 2,

10j 2 + (j + 1)2 = 72 and n = 11j + 1 = 23. Accordingly, s(18, 11) = 772 or

6. Let I be the incenter of triangle ABC and let R be the circumradius. Prove

that AI + BI + CI ≤ 3R.

Solved by Arkady Alt, San Jose, CA, USA; Miguel Amengual Covas, Cala Figuera,

Mallorca, Spain; George Apostolopoulos, Messolonghi, Greece; José Luis Dı́az-

Barrero, Universitat Politècnica de Catalunya, Barcelona, Spain; and Titu Zvonaru,

Cománeşti, Romania. We give the version of Amengual Covas.

We write a, b, c respectively for the lengths of the sides BC, CA, AB, and

a+b+c

s= for the semiperimeter. Let r be the radius of the incircle.

2 È

Note that cos A s−a

may be expressed as , and also as s(s−a) . Equating

2 È AI bc

bc(s−a)

these and solving for AI, we get AI = s

, with symmetric results for BI

and CI.

Applying the Cauchy’s inequality −

→uÈ · −

→

v È ≤ k−

→

u k · k−→

v k with

√ √ √ È

−

→u = bc, ca, ab and v = →

− s−a

, s−b

, s−c

, we now get

s s s

È È È

bc(s−a)

AI + BI + CI = + ca(s−b) + ab(s−c)

(1)

√ s s s

≤ ab + bc + ca

√

Using the relations ab + bc + ca = r 2 + 4Rr + s2 , r ≤ R

2

and s ≤ 3

2

3

R,

we have

2 √ 2

R R 3 3

ab + bc + ca ≤ + 4R · + R = 9R2 (2)

2 2 2

By (1) and (2), we obtain the desired inequality. Since equality in (1) and

(2) holds if and only if a = b = c, it holds in the required inequality if and only

if 4ABC is equilateral.

Comment. Also solved, by using the Erdős-Mordell inequality, on page 38 of the

book Experiences in Problem Solving: A W. J. Blundon Commemorative, Atlantic

Provinces Council on the Sciences, Canada, (1994).

7. Let U and V be two points inside the angle BAC such that

m(∠BAU ) = m(∠CAV ) .

Y2 respectively. Let W be the intersection of the lines X2 Y1 and X1 Y2 . Prove

that U , V , W are collinear.

38

A

..

........

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

.... .. ...

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

. . .. ....

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

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

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

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

.

... . ... ....

.. .. ... ....

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

....

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

... .. ...

...

....

....

... ..

... ....

... ..

... ... ....

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

... .. .. ....

....

... .

... ....

...

..

.. ... X .... 1

... .. . ..

....

... ........

..

..

...

...

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

.. ....

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

.

..

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

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

... .. ..... .... Y ....

1

... .. .

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

Y 2 ... ..

. ..

. .... .. ...

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

...

...

........

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

.. ....

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

... . .... .

X 2 ... ............................................................................................

. ...... ...

... ....

.. ....

....

W . .......... .. ...... ....................... ... ......... ....

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

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

.. ....

... ....

M .. U V N ....

....

.... ....

... ....

....

... ....

... ....

... ....

... ....

....

.... ..

B C

intersect at N . We denote

Suppose that the lines X2 Y1 and M N intersect at the point W . We will prove

that the points W , Y2 , X1 are collinear.

By Menelaus’ theorem we obtain:

WM Y1 N X2 A WM V N cos γ AV cos ϕ

· · = 1 ⇔ · · = 1

WN Y1 A X2 N WN AV cos ϕ M U cos β

WM AV · M U cos β

⇔ = (1)

WN AU · V N cos γ

WM X1 N Y2 A WM U N cos γ AV cos(α − ϕ)

· · = 1 ⇔ · ·

WN X1 A Y2 M WN AU cos(α − ϕ) M V cos β

WM AU · M V · cos β

⇔ = .

WN AV · U N · cos γ

AV · M U AU · M V AU 2 MU · UN

= ⇔ = ,

AU · V N AV · U N AV 2 NV · V M

39

MU UN [AM U ] [AU N ]

· = ·

NV VM [AV M ] [AN V ]

AM · AU · sin ϕ AU · AN sin(α − ϕ)

= ·

AN · AV sin(α − ϕ) AV · AM · sin γ

AU 2

= .

AV 2

S = a31 + a32 + · · · + a3n . Prove that

X

n

ai 1

≤ .

i=1

2n + 1 + S − a3i 3

Spain; and by Titu Zvonaru, Cománeşti, Romania. We give the solution of Dı́az-

Barrero.

Since ai , 1 ≤ i ≤ n, lie in the interval [0, 1], then 1 − a3i ≥ 0 for

1 ≤ i ≤ n, and

Xn

ai Xn

ai

3

≤ .

i=1

2n + 1 + S − ai i=1

2n + S

X

n

ai 1

≤

i=1

2n + S 3

or equivalently,

X

n

3(a1 + a2 + . . . + an ) ≤ 2n + S = (1 + 1 + a3i )

i=1

X

n X

n È X

n

(1 + 1 + a3i ) ≥ 3 3

1 · 1 · a3i = 3 ai .

i=1 i=1 i=1

10. Find all polynomials f with integer coefficients, such that f (p) is a prime

for every prime p.

40

Note that the polynomial f (x) = x and the constant polynomials f (x) = c

where c is a prime satisfy the requirements in the problem. They are the only such

polynomials that do.

Assume that f is a polynomial with integer coefficients such that if p is a

prime, then f (p) is a prime and also assume that f (x) 6= x and f (x) 6= c. Then

there exists a prime π such that gcd(π, f (π)) = 1 since otherwise p divides f (p)

for all primes p and it would follow that either f (x) = 0 or f (p) = p so that

f (x) = x, both contradictions. By Dirichlet’s Theorem, there exist infinitely

many integers ni such that ni f (π) + π is a prime, say ni f (π) + π = qi ,

i = 1, 2, 3, . . . . Since qi − π divides f (qi ) − f (π) we get that f (π) divides

f (qi ) for each i ≥ 1. Hence f (x) = f (π), a contradiction. As a result, the only

polynomials f with integer coefficients for which f (p) is a prime for every prime

p are f (x) = x and f (x) = c where c is a prime.

11. Let ABC be a triangle with a = BC, b = AC, c = AB, inradius r and

circumradius R. Let rA , rB and rC be the radii of the excircles of the triangle

ABC. Prove that

2 r 2 r 2 r

a2 − + b2 − + c2 − = 4(R +3r) .

rA rB rC rB rC rA rC rB rA

Solved by Arkady Alt, San Jose, CA, USA; George Apostolopoulos, Messolonghi,

Greece; Michel Bataille, Rouen, France; Geoffrey A. Kandall, Hamden, CT, USA;

Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA; and Titu

Zvonaru, Cománeşti, Romania. We give Kandall’s write-up.

a+b+c

Let K = [ABC], s = 2

. It is well known (Heron) that

= −(a4 + b4 + c4 ) + 2(a2 b2 + b2 c2 + a2 c2 )

abc

and K = rs = rA (s − a) = rB (s − b) = rC (s − c) = .

4R

Let

2 r 2 r 2 r

X = a2 − = b2 − + c2 − .

rA rB rc rB rC rA rC rB rA

We have

2 r 2(s − b) K s−b s−c

a2 − = a2 − · ·

rA rB rC K s K K

2

a a+b+c a−b+c

= 2· ·

sK 2 2

a−b+c a+b−c

− ·

2 2

1

= (−3a4 + 3a2 b2 + 3a2 c2 + 2a2 bc).

4sK

41

Similarly,

2 r 1

b2 − = (−3b4 + 3a2 b2 + 3b2 c2 + 2ab2 c),

rB rC rA 4sK

2 r 1

c2 − = (−3c4 + 3a2 c2 + 3b2 c2 + 2abc2).

rC rB rA 4sK

Consequently,

1

X = (−3(A4 + B 4 + C 4 )

4Sk

+6(A2 B 2 + B 2 C 2 + A2 C 2 ) + 2ABC(A + B + C))

1 2 K

= 3 · 16K + 2 · 4RK · 2

4( K

r

)K r

= 4(R + 3r),

as required.

12. Consider n distinct points in the plane n ≥ 3, arranged such that the

n2

number r(n) of segments of length l is maximized. Prove that r(n) ≤ .

3

Solved by Oliver Geupel, Brühl, NRW, Germany.

We will apply Turán’s Theorem: Let G be a simple graph with n vertices

which does not contain a complete subgraph containing p vertices. Let r be the

remainder of n modulo p. Then the number of edges of G is not greater than

(p − 2)n2 − r(p − 1 − r)

f (n, p) = .

2(p − 1)

Consider the graph G whose vertices are the n given points and where two

vertices P and Q are connected by an edge if and only if P Q = l. Clearly, G

does not contain a complete subgraph with 4 vertices. By Turán’s Theorem, the

number of edges of G is not greater than

2n2 − r(3 − r) n2

f (n, 4) = ≤ .

6 3

(1978) asked for proving the inequality r(n) < 2n3/2 . This is a sharper

upper bound for each n ≥ 36. See: The William Lowell Putnam Mathematical

Competition problems and solutions: 1965-1984, ed. by G.L. Alexanderson, L.F.

Klosinski, and L.C. Larson, MAA, 1986, p. 104f.

which is positive. Prove that P (X) = X n − b1 X n−1 − · · · − bn−1 X − bn , has

a single positive root p, which is simple, and that the absolute value of each root

of P (X) is not greater than p.

42

Spain; and Oliver Geupel, Brühl, NRW, Germany. We give the solution by Dı́az-

Barrero.

We consider the continuous function A : (0, +∞) → R defined by

b1 b2 bn

A(x) = + + ... + −1

x x2 xn

b1 2b2 nbn

Since A0 (x) = − − −. . .−

< 0 for all x > 0, then A is a decreasing

x2 x3 xn+1

continuous function. Furthermore, lim A(x) = −1 and lim A(x) = +∞.

x→+∞ x→0+

P (x)

Therefore, on account of Bolzano’s theorem the equation A(x) = − = 0

xn

has only one positive root, say p, which is a zero of polynomial P . On the other

hand, from A0 (p) < 0 follows P 0 (p) > 0 and p is a simple zero.

To see that all the zeros of P have modulus less than or equal to p we

argue by contradiction. Assume that x0 is a zero of P and let |x0 | = α with

α > p. Then A(α) < A(p) = 0 and P (α) > 0. On the other hand from

n−1

xn

0 = bn + bn−1 x0 + . . . + b1 x0 we have

n−1

|xn

0 | = |bn + bn−1 x0 + . . . + b1 x0 | ≤ bn + bn−1 |x0 | + . . . + b1 |xn−1

0 |

Comment. The first part of the statement also follows immediately applying

the well-known Descartes Rule of Signs.

15. A circle is tangent to the sides AB and AC of the triangle ABC and to

its circumcircle at P , Q and R respectively. If P Q ∩ AR = {S} prove that

m(∠SBA) = m(∠SCA).

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

Let Γ be the circle through A, B, and C, Γ............................

......

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

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

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

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

and let ∆ be the circle through P , Q, and .............

.. .. ..

..

.. ... ...

... ...

.....

....

....

.. .. . .. ...

R. Let D be the second intersection of ∆ .

..

.... .

. .....

.

..

.

.

.

.

.. .

..

..

.

...

...

...

. ... .. ..

.... . . ...

and the line AR. Since AP is tangent to ∆, ...

...

..... .

.

..

.

. .

..

.

..

..

..

...

... .. . ∆ D .

. ... ..

..

we have ∠P DR = ∠BP R. Since ∆ and ..

..

.

..

.

.

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

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

.

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

.

.

.

.

...... ..

.

...

.

.

..

..

..

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

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

Γ have a common tangent t at R, it holds ...

..

... P . .

.

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

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

. S .

.

.

.

..

..... ..

.

.

.

.....

. ..

.

..

.. .

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

... ..

∠DP R = ∠(AR, t) = ∠ABR = ∠P BR. ...

...

.. ..

..

.

..... ...

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

...

....

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

. ......

.

.

Q

... ..

.

.

.

.

.

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

Hence, the triangles DP R and P BR are ... .. .......... .. .... ..

.......

... ...

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

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

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

......

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

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

.

similar. It follows that ∠ARP = ∠P RD = ...........

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

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

. ...

... ..

..

...... ..

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

...

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

∠BRP , that is, the line RP is the internal B .... ...........

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

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

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

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

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

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

.... ......

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

C

bisector of ∠R in 4ABR. ................... ........ ... ...... .........

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

R

43

Therefore,

AP AR

= .

BP BR

Similarly,

AQ AR

= .

CQ CR

By AP = AQ, we obtain

BP CQ

= .

BR CR

By

PS sin ∠P AS sin ∠BAR BR

= = = ,

QS sin ∠QAS sin ∠CAR CR

we deduce that

BP PS

= .

CQ QS

By ∠BP S = ∠CQS, it follows that 4BP S ∼ 4CQS. Consequently,

16. Prove that there are infinitely many primes p for which there exists a positive

integer n such that p divides n! + 1 and n does not divide p − 1.

A proof is by Paul Erdős, (c.f. p. 558 of G.E. Hardy and M.V. Subbarao, “A

modified Problem of Pillai and Some Related Questions”, The American Math.

Monthly, Vol 109, pp. 554–559).

44

BOOK REVIEWS

Amar Sodhi

Alex’s Adventures in Numberland

by Alex Bellos

Bloomsbury Publishing, 2010

ISBN-13: 978-0747597162, hardcover, 448 pages, $30.95

Reviewed by Bruce Shawyer, Memorial University of Newfoundland, St.

John’s, NL

readers of all abilities. It was published in March 2010 in the UK by Bloomsbury.

In the US the book appeared in June with the title Here’s Looking at Euclid,

published by Free Press. The editions have different covers. I have the UK edition

(448 pages) and have not seen the US edition (320 + xi pages). Both are available

in Canada via the usual retailers.

The US edition of the book only contains a brief bibliography. Thanks

to the wonders of the internet, the author has been able to make available on

his WEB site (alexbellos.com) the complete chapter-by-chapter bibliography, with

comments and suggested further reading. It is enhanced with links!

Here are some of the difference between the two editions, gleaned from the

WEB site:

1. The UK edition has a cartoon preceding each chapter.

2. Both the British and American publishers felt that their respective titles

worked best for their respective audiences.

3. The American one is slightly shorter (the section on the British 50p piece

is omitted, for example, since the shape means nothing in Arkansas or Wyoming)

and it has less diagrams overall.

4. The British version also has a 12-page colour plate section.

The author, Alex Bellos (with degrees in Mathematics and Philosophy) is a

writer, broadcaster, football (soccer) lover and self-proclaimed math geek, with a

colourful career including several books and short films.

The book consists of mathematics (real mathematics) for the lay person.

But yet, it is mathematics for the mathematics student and for the mathematics

teacher at all levels. It covers a very broad range of topics, which are all related

in an engaging narrative style.

For example, Bellos describes how Yorkshire shepherds count (his list agrees

with the song by Jake Thackray); why business card origami is abhorrent to the

Japanese; a mnemonic for the digits of π which is a remarkable modernist pastiche

of Poe’s The Raven; and why we most commonly x as the name of a variable. We

meet fanatics, crackpots, anthropologists and gurus as well as a few mathemati-

cians (such as Aitken, Brahmagupta, Cantor, Descartes, Euler and Fibonacci).

45

All of the mathematics (with one exception) is well explained and correct. It

is a pity that he has an error in the penultimate paragraph of the book. Here, he

describes infinite cardinals, and makes two claims: first, that the number of curves

in the plane is larger than c, the cardinality of the continuum (this is false if curves

must be continuous; there are only c of them); and second, that nobody has been

able to come up with a larger set (Cantor proved that the set of subsets of any set

is larger than the original set - perhaps Alex meant a larger “naturally-occurring”

set).

Despite this, we have a book well worth reading. As I read it, I thought that

it might well form the basis for an elementary course on the History of Mathemat-

ics. Now there is something that every student of mathematics should study. This

book puts much of mathematics into context, and so, will encourage students to

want to know more.

Editor’s note: the North American edition, Here’s Looking at Euclid: A

Surprising Excursion Through the Astonishing World of Math (ISBN 978-1-4165-

8825-2) retails at $32.99.

by us “Universe”, a part limited in time and space. He experiences himself, his

thoughts and feelings as something separated from the rest, a kind of optical

delusion of his consciousness. This delusion is a kind of prison for us, restricting

us to our personal desires and to affection for a few persons nearest to us. Our task

must be to free ourselves from this prison by widening our circle of compassion to

embrace all living creatures and the whole of nature in its beauty. Nobody is able

to achieve this completely, but the striving for such achievement is in itself a part

of the liberation and a foundation for inner security. In H. Eves “Mathematical

Circles Adieu”, Boston: Prindle, Weber and Schmidt, 1977.

46

PROBLEMS

Solutions to problems in this issue should arrive no later than 1 August 2010.

An asterisk (?) after a number indicates that a problem was proposed without a solution.

Each problem is given in English and French, the official languages of Canada. In

issues 1, 3, 5, and 7, English will precede French, and in issues 2, 4, 6, and 8, French

will precede English. In the solutions’ section, the problem will be stated in the language

of the primary featured solution.

The editor thanks Jean-Marc Terrier of the University of Montreal for translations

of the problems.

solutions past the posted due date. Solutions will be accepted until we process

them for publication. Currently we are delayed by about four months. Check the

CMS website, cms.math.ca/crux, for our status in processing problems.

Suppose that b is a positive real number such that there are exactly two

integers strictly between b and 2b, and exactly two integers strictly between 2b

and b2 . Find all possible values of b.

Vietnam.

Prove that if ai > 0 for i = 1, 2, 3, 4, then

X 1 12

≥

a2

cyclic i

+ a2i+1 + a2i+2 (a1 + a2 + a3 + a4 )2

Let ABC be a given triangle and 0 < λ < 12 . Let D and E be points

on AB such that AD = BE = λ · AB, and F , G points on AC such that

AF = CG = λ · AC. Let BF ∩ CE = H and BG ∩ CD = I. Show that

i) HI k BC and

1 − 2λ

ii) HI = BC.

λ2 −λ+1

Evaluate

R1

(x2 − x − 2)n dx

lim R 10 .

n→∞ 2 n

0 (4x − 2x − 2) dx

47

Catalunya, Barcelona, Spain.

Let A(x) = xn + an−1 xn−1 + · · · + a1 x + 1 be a real polynomial with

positive coefficients and having all its zeros real. Prove that

È

n

A(1)A(2) · · · A(n) ≥ (n + 1)!

Let ABC be a triangle with ∠A = 20◦ . Let BD be the angle bisector of

∠ABC with D on AC. If AD = DB + BC, determine ∠B.

Let c1 = 9, c2 = 15, c3 = 21, c4 = 25, . . ., where cn is the nth composite

odd integer. Evaluate

cn

lim .

n→∞ cn+1

Let

e1/x − 1

f (x) = .

e1/(x+1) − 1

(a) Show that for all x ∈ (0, ∞),

r

x+1

f (x) > .

x

x+1

f (x) <

x−1

for all x ∈ (1, ∞).

Italy.

Let r be a real number. and let D, E, and F be points on the sides BC, CA,

and AB of a triangle ABC with

BD CE AF

= = = r.

DC EA FB

The cevians AD, BE, and CF bound a triangle P QR whose area we denote by

[P QR]. Find the value of r for which the ratio of the areas, [DEF ]

[P QR]

equals 4.

48

Let S = {2, 3, 4, 6, 8, 9, 12, 16, 18, 24, 27, . . .} be the set of positive

integers whose only prime divisors are 2 or 3. Let a1 = 2, a2 = 3, . . . , be

the elements of S, with a1 < a2 < . . . .

∞

X

1

(i) Determine .

i=1

ai

(ii) ? For each positive integer n, let s(n) be the sum of all its divisors in-

s(n)

cluding 1 and n itself. Prove n

< 3 for all members of S.

Buzău, Romania.

Given x, y, and z are positive integers such that

, , and,

x−1 y−1 z−1

are positive integers. Find the smallest positive integer N such that xyz ≤ N .

Find all nonconstant polynomials P such that P ({x}) = {P (x)}, for all

x ∈ R, where {a} denotes the fractional part of a.

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

On suppose que b est un nombre réel positif tel qu’il existe exactement deux

entiers strictement compris entre b et 2b, de même qu’exactement deux entiers

strictement compris entre 2b et b2 . Trouver toutes les valeurs possibles de b.

3602. Proposé par Pham Van Thuan, Université de Science de Hanoı̈, Hanoı̈,

Vietnam.

Montrer que si ai > 0 pour i = 1, 2, 3, 4, alors

X 1 12

≥

a2

cyclique i

+ a2i+1 + a2i+2 (a1 + a2 + a3 + a4 )2

49

Soit ABC un triangle et 0 < λ < 12 . Soit D et E deux points sur

AB tels que AD = BE = λ · AB, F et G deux points sur AC tels que

AF = CG = λ · AC. Soit BF ∩ CE = H et BG ∩ CD = I. Montrer que

i) HI k BC et

1 − 2λ

ii) HI = BC.

λ2 −λ+1

Calculer

R1

(x2 − x − 2)n dx

lim R 10 .

n→∞ 2 n

0 (4x − 2x − 2) dx

Catalogne, Barcelone, Espagne.

Soit A(x) = xn +an−1 xn−1 +· · ·+a1 x+1 un polynôme réel à coefficients

positifs n’ayant que des racines réelles. Montrer que

È

n

A(1)A(2) · · · A(n) ≥ (n + 1)!

Soit ABC un triangle avec ∠A = 20◦ . Soit BD la bissectrice de l’angle

au sommet B avec D sur AC. Si AD = DB + BC, trouver ∠B.

Soit c1 = 9, c2 = 15, c3 = 21, c4 = 25, . . ., où cn désigne le ne entier

impair non premier. Calculer

cn

lim .

n→∞ cn+1

50

Soit

e1/x − 1

f (x) = .

e1/(x+1) − 1

r

x+1

f (x) > .

x

Ê

x+1

f (x) <

x−1

Noci, Italie.

Soit r un nombre réel et D, E et F des points sur les côtés BC, CA et AB

d’un triangle ABC avec

BD CE AF

= = = r.

DC EA FB

par [P QR]. Trouver la valeur de r pour laquelle le rapport [DEF ]

[P QR]

des aires est

égal à 4.

3610. Proposé par Peter Y. Woo, Université Biola, La Mirada, CA, É-U.

Soit S = {2, 3, 4, 6, 8, 9, 12, 16, 18, 24, 27, . . .} l’ensemble des entiers po-

sitifs dont les seuls diviseurs premiers sont 2 ou 3. Notons a1 = 2, a2 = 3, . . .

les éléments de S, avec a1 < a2 < . . . .

∞

X

1

(i) Trouver .

i=1

ai

(ii) ? Pour chaque entier positif n, soit s(n) la somme de tous ses diviseurs, y

s(n)

compris 1 et n lui-même. Montrer que n

< 3 pour tous les éléments de

S.

51

3611. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie.

On donne trois entiers positifs x, y et z tels que

, et

x−1 y−1 z−1

sont des entiers positifs. Trouver le plus petit entier positif N tel que xyz ≤ N .

Trouver tous les polynômes non constants P tels que P ({x}) = {P (x)}

pour tout x ∈ R, où {a} désigne la partie fractionnaire de a.

are sent to us in electronic format. Material sent in TEX or LATEX is preferred, but

we will also accept pdf, Microsoft Word as well as handwritten material (mailed

or scanned).

When sending electronic solutions, please name the files in a meaningful way

to identify yourself and the problem. For example, if I was to submit a solution

to problem 3603 from this issue, I would name it February 3603 Godin.tex.

Please place each solution on its own separate sheet(s) with the problem

number, your name and affiliation on each page. Multiple solutions on one page

means we have to do lots of photocopying and it increases the chances that

something will get overlooked or misfiled.

As always no problem is ever closed. We always accept new solutions and

generalizations to past problems. Also, in the last issue [2010 : 545, 547], Chris

Fisher published a list of unsolved problems from Crux. Below is a sample of one

of these unsolved problems:

609?. [1981 : 49; 1982 : 27-28] Proposed by Ian June L. Garces, Ateneo de

Manila University, The Philippines.

A1 B1 C1 D1 is a convex quadrilateral inscribed in a circle and M1 , N1 , P1 ,

Q1 are the mid-points of sides B1 C1 , C1 D1 , D1 A1 , A1 B1 , respectively. The

chords A1 M1 , B1 N1 , C1 P1 , D1 Q1 meet the circle again in A2 , B2 , C2 , D2 ,

respectively. Quadrilateral A3 B3 C3 D3 is formed from A2 B2 C2 D2 as the latter

was formed from A1 B1 C1 D1 , and the procedure is repeated indefinitely. Prove

that quadrilateral An Bn Cn Dn “tends to” a square as n → ∞.

What happens if A1 B1 C1 D1 is not convex?

Enjoy!

52

SOLUTIONS

No problem is ever permanently closed. The editor is always pleased to

consider for publication new solutions or new insights on past problems.

Let N be the set of positive integers, E the set of all even positive integers,

and O the set of all odd positive integers. A set S ⊆ N is closed if x + y ∈ S for

all distinct x, y ∈ S, and unclosed if x + y 6∈ S for all distinct x, y ∈ S. Prove

that if N is partitioned into A and B, where A is closed and nonempty, and B is

unclosed and infinite, then A = E and B = O.

York, USA.

We prove first that 1 ∈ B.

Assume by contradiction that 1 ∈ A. Chose m, n ∈ B then m + n ∈ A.

But since 1 ∈ A, it follows that all integers greater than m + n are in A, which

contradicts B infinite.

We prove now that 2 ∈ A. Again assume by contradiction that 2 ∈ B.

Chose n > 2, n ∈ B. Then 3 ∈ A, n + 1 ∈ A, n + 2 ∈ A and hence

n + 4, n + 5 ∈ A. Then 2n + 5, 2n + 6, 2n + 7 ∈ A, and since 3 ∈ A, it

follows that A contains all the integers greater than 2n + 5, which contradicts B

infinite.

Now we show that 3 ∈ B. Assume by contradiction 3 ∈ A. Then 5 ∈

A, 7 ∈ A, 8 ∈ A, and then, since 2 ∈ A, it follows that A contains all the

integers greater than 7. Again this contradicts B infinite.

Next we prove that neither A nor B contains two consecutive integers.

If n, n + 1 are in A, then n ≥ 4, and since 2 ∈ A, it follows that A

contains all the integers greater than n, a contradiction.

Assume now that B contains two consecutive integers n, n + 1. Since B is

infinite, there exists m ∈ B; m > n + 1. But then A contains the consecutive

integers m + n, m + n + 1, a contradiction.

Hence 1 ∈ B, 2 ∈ A and neither A nor B contains two consecutive integers.

Then it follows by induction that 2k − 1 ∈ B, 2k ∈ A for all k ≥ 1.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ŠEFKET

ARSLANAGI Ć, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; ROY BARBARA,

Lebanese University, Fanar, Lebanon; MICHEL BATAILLE, Rouen, France; CHIP CURTIS,

Missouri Southern State University, Joplin, MO, USA; JOSEPH DiMURO, Biola University,

La Mirada, CA, USA; IAN JUNE L. GARCES, Ateneo de Manila University, Quezon City,

The Philippines; OLIVER GEUPEL, Brühl, NRW, Germany; JOHAN GUNARDI, student,

SMPK 4 BPK PENABUR, Jakarta, Indonesia; MICHAEL JOSEPHY, Universidad de Costa

Rica, San Pedro, Costa Rica; KATHLEEN E. LEWIS, SUNY Oswego, Oswego, NY, USA;

MISSOURI STATE UNIVERSITY PROBLEM SOLVING GROUP, Springfield, MO, USA;

JOEL SCHLOSBERG, Bayside, NY, USA; DIGBY SMITH, Mount Royal University, Calgary,

AB; EDMUND SWYLAN, Riga, Latvia; PETER Y. WOO, Biola University, La Mirada, CA,

USA; and the proposer. There was one incorrect solution.

53

3502. [2010 : 44, 46] Proposed by José Luis Dı́az-Barrero, Universitat Politècnica

de Catalunya, Barcelona, Spain.

Find all real solutions of the following system of equations

È È

x21 + x22 + 21 = x22 + 77 ,

È È

x22 + x23 + 21 = x23 + 77 ,

··· ··· ···

È È

x2n + x21 + 21 = x21 + 77 .

Similar solutions by Michel Bataille, Rouen, France ; and Titu Zvonaru, Cománeşti,

Romania. We give Bataille’s write up.

√ √

Let ui = x2i and f (x) = x + 77 − x + 21. Then, the equations are:

1 1

0

f (x) = √ − √ 1 , hence

2 x+77 x+21

1 1

|f 0 (x)| < √ ≤ √ .

2 x + 21 2 21

Let k = 2√121 and f m denote the composition f ◦ f ◦ f ◦ ... ◦ f . Then

k < 1 and it follows by induction that

Let 1 ≤ i ≤ n. Then ui ∈ [0, ∞) and f n (ui ) = ui . We show that

ui = 4.

Assume by contradiction that ui 6= 4. Then by the Mean Value Theorem,

there exists a c between ui and 4 so that

This shows that ui = 4 for all i. Conversely u1 = u2 = .. = un = 4 is

obviously a solution for our system.

Thus, all the real solutions of the system are (±2, ±2, ..., ±2).

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ROY BARBARA,

Lebanese University, Fanar, Lebanon; BRIAN D. BEASLEY, Presbyterian College, Clinton,

SC, USA; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; OLIVER

GEUPEL, Brühl, NRW, Germany; RICHARD I. HESS, Rancho Palos Verdes, CA, USA; JOEL

SCHLOSBERG, Bayside, NY, USA; ALBERT STADLER, Herrliberg, Switzerland; EDMUND

SWYLAN, Riga, Latvia; and the proposer. One incomplete solution was submitted.

54

Newfoundland, St. John’s, NL.

Given a triangle and the midpoints of its sides, with the use of a straight

edge and only three uses of a pair of compasses, bisect all three angles of the

triangle.

Soit ABC le triangle donné ; I, J et K les milieux respectifs de BC, CA

et AB. On va utiliser une seule fois le compas.

On trace le cercle de centre K et de rayon KA = KB. Soit E le point

d’intersection de la droite KJ avec ce cercle. Comme KE = KB, alors E est sur

la bissectrice de l’angle ABC car KBE triangle isocèle et KJ parallèle à BC.

De même, soit F le point d’intersection du cercle avec la droite KI, alors

on a KF = KA, et donc F est sur la bissectrice de l’angle BAC car KAF

triangle isocèle et KI est parallèle à AC.

Ces deux bissectrices se coupent en H, centre du cercle inscrit et donc sur

la troisième bissectrice CH.

Les trois bissectrices sont donc AH, BH et CH.

Also solved by the following readers, with the number of uses of the compass

indicated in parentheses: MICHEL BATAILLE, Rouen, France (2); OLIVER GEUPEL,

Brühl, NRW, Germany (2); JOHAN GUNARDI, student, SMPK 4 BPK PENABUR, Jakarta,

Indonesia (3); GEOFFREY A. KANDALL, Hamden, CT, USA (2); V ÁCLAV KONE ČN Ý,

Big Rapids, MI, USA (2); MISSOURI STATE UNIVERSITY PROBLEM SOLVING GROUP,

Springfield, MO, USA (2); EDMUND SWYLAN, Riga, Latvia (1); PETER Y. WOO, Biola

University, La Mirada, CA, USA (2); TITU ZVONARU, Cománeşti, Romania (2); and the

proposer (3).

Swylan was the only other solver who used the compass just once. Bataille, the Missouri

State University Problem Solving Group, and Woo all noted that by the Poncelet–Steiner theorem

just one use of the compass suffices to carry out the construction.

Given triangle ABC, set Q = a cos2 A + b cos2 B + c cos2 C, and let

ABC have area S and circumradius R. Prove that

S

(a) Q ≥ , with equality if and only if ABC is equilateral.

R

√

S 2

(b) Q ≤ if ABC is not obtuse, with equality if and only if ABC is an

R

isosceles right triangle.

Solution to part (a) by Michel Bataille, Rouen, France; solution to part (b) by the

proposer, modified by the editor.

(a) Let a = BC, b = CA, c = AB and let I and H be the incentre and

orthocentre of 4ABC. Since 2sI = aA + bB + Cc, we have

55

−→ −

−→ −−→ −−→

2sHI = aHA + bHB + cHC, and so

4s2 HI 2 = a2 HA2 + b2 HB 2 + c2 HC 2

−−→ −−→ −−→ −−→ −

−→ −− →

+ 2abHA · HB + 2bcHB · HC + 2caHC · HA

+ bc(HC 2 + HB 2 − a2 ) + ca(HA2 + HC 2 − b2 )

Now, if A0 is the midpoint of BC, then HA2 = (2OA0 )2 = 4R2 cos2 A and

similar results hold for HB 2 and HC 2 . It follows that 2sHI 2 = 4R2 · Q − abc

and

abc sHI 2

Q= + .

4R2 2R2

abc S S

Since = , we see that Q ≥ , with equality if and only if H = I, that

4R2 R R

is, if and only if ABC is equilateral.

c = 2R sin C, S = 2R2 sin A sin B sin C into the inequality and canceling

2R from each side yields the equivalent inequality

√

sin A cos2 A + sin B cos2 B + sin C cos2 C ≤ 2 sin A sin B sin C .

Assume that A ≥ B ≥ C and set φ = 21 (B + C), ψ = 1

2

(B − C). Then from

π π

the hypotheses we have 0 ≤ ψ ≤ ≤ φ ≤ , and

4 3

A = π − 2φ , B = φ+ψ, C = φ−ψ.

In terms of φ and ψ the desired inequality takes the form

sin 2φ cos2 2φ + cos2 (φ + ψ) sin(φ + ψ) + cos2 (φ − ψ) sin(φ − ψ)

√

≤ 2 sin 2φ sin(φ + ψ) sin(φ − ψ) .

In this inequality make the replacements sin 2φ = 2 sin φ cos φ, cos 2φ =

cos2 φ − sin2 φ, cos(φ ± ψ) = cos φ cos ψ ∓ sin φ sin ψ, and sin(φ ± ψ) =

sin φ cos ψ ± cos φ sin ψ, then expand each side and cancel like terms, then

cancel a common term sin φ > 0 from each side, then apply the identities

sin2 x = 1 − cos2 x for x = φ, ψ. This yields the equivalent inequality

cos3 φ − (1 − cos2 φ) cos φ + cos2 φ cos3 ψ

+ (1 − cos2 φ) cos ψ(1 − cos2 ψ) − 2 cos2 φ(1 − cos2 ψ) cos ψ

√ √

≤ 2(1 − cos2 φ) cos φ cos2 ψ − 2 cos3 φ(1 − cos2 ψ)

√ √

1 2 2

Set x = cos φ, y = cos ψ, so x ∈ , and y ∈ I = ,1 .

2 2 2

Making these substitutions and simplifying yields the equivalent inequality

√ √

x3 (2 + 2) + 4x2 y 3 − 3x2 y − 2xy 2 − x − y 3 + y ≤ 0 .

56

√

Let t = x 2. Then t and y are in the interval I, and we need to prove that

√

2+ 2 3 t

f (t, y) = √ + 2t2 y 3 − t2 y − ty 2 − √ − y 3 + y ≤ 0 .

2 2 2 2

We have

1 1 1

y + = (y − 1) y 2 −

f (1, y) = y 3 − y 2 − ≤ 0, y ∈I,

2 2 2

√

2

with equality only for y = 1 and y = .

2

Next we prove that f (t, y) ≤ f (1, y).

Let g(y) = 4y 3 − 2y 2 − 3y + 1 and h(y) = 4y 3 − 3y. We then have

√ √

2[f (t, y) − f (1, y)] = (t − 1) t2 ( 2 + 1) + t( 2 + 1) + th(y) + g(y) .

√ √

t2 ( 2 + 1) + t( 2 + 1) + th(y) + g(y) > 0 . (1)

Note that (1) follows from

√

t( 2 + 1) + g(y) ≥ 1 , (2)

√

t( 2 + 1) + h(y) ≥ 1 , (3)

since the left side of (1) is t times the left side of (3) plus the

√

left side of (2).

√

0 1 ± 10 1 + 10

The quadratic equation g (y) = 0 has roots y = and <

√ 2

√ √ 2

2 2 − 2

, so g(y) is strictly increasing on I and g(y) ≥ g = . Also,

2 √ √ 2 2

2 √ 2

t≥ , so that t( 2 + 1) ≥ 1 + , and (2) follows.

2 2

Furthermore, h(y) ≥ g(y) for y ∈ I, since h(y) = g(y) + (2y 2 − 1) and

2

2y − 1 ≥ 0 on I, so (3) follows from (2).

This completes the proof of the inequality in part (b).

Equality holds in f (t, y) − f (1, y)

√

≤ 0 only for t = 1, and equality holds

2

in f (1, y) ≤ 0 only for y = 1 and y =

. These cases correspond to a triangle

2

π π π

with A = and B = C = , or to a degenerate triangle with A = B =

2 4 2

and C = 0.

Part (a) also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ŠEFKET

ARSLANAGI Ć, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; SCOTT BROWN,

Auburn University, Montgomery, AL, USA; OLIVER GEUPEL, Brühl, NRW, Germany; JOE

HOWARD, Portales, NM, USA; THANOS MAGKOS, 3rd High School of Kozani, Kozani,

Greece; and the proposer. One incomplete solution to part (a), one incorrect solution to part

(a), and three incomplete solutions to part (b) were received.

The proposer said she was influenced by Crux problem 3167, which asked to show that

a cos3 A+b cos3 B+c cos3 C ≤ abc/4R2 holds for non-obtuse triangles ABC. She indicated

that the inequality in part (a) occurs in a different form on p. 14 of V.P. Soltan and I. Majdan’s

book Identities and Inequalities in a Triangle, Kishinev, 1982 (Russian), although the proof there

is of a general nature and different from the one she constructed.

57

3505. [2010 : 45, 47, 107, 109] Proposed by Yakub N. Aliyev, Qafqaz University,

Khyrdalan, Azerbaijan.

The circles Γ1 and Γ2 have a common centre O, and Γ1 lies inside Γ2 . The

point A 6= O lies inside Γ1 ; a ray not parallel to AO that starts at A intersects

Γ1 and Γ2 at the points B and C, respectively. Let tangents to corresponding

circles at the points B and C intersect at the point D. Let E be a point on the

line BC such that DE is perpendicular to BC. Prove that AB = EC if and

only if OA is perpendicular to BC.

Let A0 be the projection of O on the line ` = BC. A and A0 lie on the

same side of B on ` (because they are interior points of Γ1 ), so we have

OA ⊥ ` ⇔ A = A0 ⇔ AB = A0 B. (1)

Since ∠OBD = ∠OCD = 90◦ , B and C lie on the circle whose diameter is

OD, which we will denote by Γ3 ; let M be its centre. Since M B = M C, the

projection of M on ` is the midpoint N of BC. The lines OA0 , M N , and DE

are parallel (they are all perpendicular to `). Since OM = M D it follows that

A0 N = N E. Hence, A0 B = A0 N − BN = N E − N C = EC. That is,

A0 B = EC. (2)

only if OA ⊥ `.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ŠEFKET

ARSLANAGI Ć, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; MICHEL

BATAILLE, Rouen, France; CHIP CURTIS, Missouri Southern State University, Joplin,

MO, USA; PRITHWIJIT DE, Homi Bhabha Centre for Science Education, Mumbai,

India; RICHARD EDEN, student, Purdue University, West Lafayette, IN, USA; OLIVER

GEUPEL, Brühl, NRW, Germany; JOHAN GUNARDI, student, SMPK 4 BPK PENABUR,

Jakarta, Indonesia; JOHN G. HEUVER, Grande Prairie, AB; JOEL SCHLOSBERG, Bayside,

NY, USA; MIHAÏ STOËNESCU, Bischwiller, France; PETER Y. WOO, Biola University,

La Mirada, CA, USA; KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh, PA,

USA; and the proposer.

3506. [2010 : 45, 47] Proposed by Pedro Henrique O. Pantoja, student, UFRN,

Brazil.

Prove that Q(n) + Q n2 + Q n3 is a perfect square for infinitely many

positive integers n that are not divisible by 10, where Q(n) is the sum of the

digits of n.

Solution by the Missouri State University Problem Solving Group, Springfield, MO,

USA.

More generally, we will show that the following result holds: If Q(n, b)

denotes the sum of the digits of n in base b, then Q(n, b) + Q(n2 , b) + Q(n3 , b)

58

is a perfect square for infinitely many positive integers n that are not divisible by

b.

Let a be the square-free part of b−1, k = a`2 with ` ∈ N, and n = bk −1.

Now n consists of k b − 1’s when written in base b and hence Q(n) = k(b − 1).

Using the binomial theorem, it is easy to see that the base b representation of n2

consists of k−1 b−1’s, one b−2, k−1 0’s, and one 1 (hence Q(n2 , b) = k(b−1))

and n3 consists of k − 1 b − 1’s, one b − 3, k − 1 0’s, one 2, and k b − 1’s

(hence Q(n3 , b) = 2k(b − 1)). Therefore Q(n, b) + Q(n2 , b) + Q(n3 , b) =

4k(b − 1) = 4a(b − 1)`2 , but since a is the square-free part of b − 1, a(b − 1)

is a perfect square and we’re done.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ROY

BARBARA, Lebanese University, Fanar, Lebanon; BRIAN D. BEASLEY, Presbyterian

College, Clinton, SC, USA; CHIP CURTIS, Missouri Southern State University, Joplin, MO,

USA; JOSEPH DiMURO, Biola University, La Mirada, CA, USA; OLIVER GEUPEL, Brühl,

NRW, Germany; JOHAN GUNARDI, student, SMPK 4 BPK PENABUR, Jakarta, Indonesia;

RICHARD I. HESS, Rancho Palos Verdes, CA, USA; PETER HURTHIG, Columbia College,

Vancouver, BC; MICHAEL JOSEPHY, Universidad de Costa Rica, Costa Rica; R. LAUMEN,

Deurne, Belgium; ALBERT STADLER, Herrliberg, Switzerland; EDMUND SWYLAN, Riga,

Latvia; LI ZHOU, Polk Community College, Winter Haven, FL, USA; and the proposer.

The other submitted solutions were similar and they can be summarized by the following

2

sequences nk = 2 × 10k + 7 (k ≥ 2), nk = 7 × 10k + 2 (k ≥ 2), nk = 10k − 1 (k ≥ 1),

nk = 10k + 17 (k ≥ 4), and nk = 18 × 10k + 18 (k ≥ 4).

3507. [2010 : 45, 47] Proposed by Pham Huu Duc, Ballajura, Australia.

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

Ê Ê Ê

a(b + c) b(c + a) c(a + b)

+ +

a2 + bc b2 + ca c2 + ab

Ê

1 1 1

≤ 2(a + b + c) + + .

a+b b+c c+a

By the Cauchy-Schwarz inequality, we have that

Ê !2 !

X a(b + c) X a

≤ 2(a + b + c)

cyc

a2 + bc cyc

a2 + bc

X a X 1

≤

cyc

a2 + bc cyc

a+b

c c 1

c2 + ab ≥ ac + bc and c2 +ab ≤ ac+bc = a+b . Therefore, it now suffices to

show that

a b 1 1

+ 2 ≤ +

a2 + bc b + ac b+c a+c

59

Since this inequality holds for a = b, we can assume that a > b ≥ c. This

simplifies to

Since a − b > 0, we obtain

c2 (a − b) ≤ a3 − b3 + c(a2 − b2 )

c2 ≤ a2 + ab + b2 + c(a + b)

The result follows.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; OLIVER

GEUPEL, Brühl, NRW, Germany; PAOLO PERFETTI, Dipartimento di Matematica,

Università degli studi di Tor Vergata Roma, Rome, Italy; and the proposer.

3508. [2010 : 45, 47] Proposed by Hung Pham Kim, student, Stanford

University, Palo Alto, CA, USA.

Let a, b, c, d be nonnegative real numbers such that a + b + c + d = 4.

Prove that

√ √ √ √ √

a bc + b cd + c da + d ab ≤ 2 1 + abcd .

Let (x, y, z,

√ t) be√a permutation

√ √of (a, b, c, d) such that x ≥ y ≥ z ≥ t.

We clearly have x ≥ y ≥ z ≥ t and

√ p √ p

xyz ≥ xyt ≥ xzt ≥ yzt,

√ √ √ p √ √ √ p

x xyz + y xyt + z xzt + t yzt

√ √ √ √ √ √ √ √

≥ a abc+ b bcd+ c cda+ d dab.

It remains to prove that

√ √ √ p √ √ √ p √

x xyz + y xyt + z xzt + t yzt ≤ 2(1 + abcd),

or √ √ p p

√

( xy + zt)( xz + yt) ≤ 2(1 + xyzt).

1

Since uv ≤ 2

u2 + v 2 , it is enough to prove that

√ √ √ p p

( xy + zt)2 + ( xz + yt)2 ≤ 4(1 + xyzt),

60

or

xy + zt + xz + yt ≤ 4,

which is equivalent to (x + t)(y + z) ≤ 4. This is clearly true by the AM–GM

Inequality, since x + y + z + t = 4, and we are done.

There were 2 incomplete solutions.

3509. [2010 : 45, 48] Proposed by Hung Pham Kim, student, Stanford

University, Palo Alto, CA, USA.

Let a, b, and c be nonnegative real numbers such that a + b + c = 3. For

each positive real number k, find the maximum value of

a2 b + k b2 c + k c2 a + k .

solutions to several other problem proposals, while all of the other submitted

solutions to this problem were either incomplete or incorrect.

The editor has therefore elected to leave this problem open until a correct

and complete “one piece” solution is received.

3510. [2010 : 45, 48] Proposed by Cosmin Pohoaţă, Tudor Vianu National

College, Bucharest, Romania.

Let d be a line exterior to a given circle Γ with centre O. Let A be the

orthogonal projection of O on the line d, M be a point on Γ, and X, Y be the

intersections of Γ, d with the circle Γ0 of diameter AM . Prove that the line XY

passes through a fixed point as M moves about Γ.

Jakarta, Indonesia, modified by the editor.

When M lies on OA the lines OA and XY coincide, so that a fixed point

would necessarily lie on OA. For any position of M on Γ off OA let Q denote the

intersection of XY and OA. We must prove that the position of Q is independent

of the choice of M . Let P be the centre of Γ0 ; we first show that OP QX is cyclic.

To that end, note that ∠OP M is both an exterior angle of ∆AOP and half the

apex angle of the isosceles triangle P XM . Therefore,

∠OAP = ∠Y M A. From (1) therefore,

that is,

∠P OA = 90◦ − ∠Y M X. (2)

61

the corresponding inscribed angle at M , ∠Y M X = 21 ∠Y P X. Note also that

90◦ − 12 ∠Y P X = ∠P XY = ∠P XQ. Consequently, equation (2) becomes

∠P OQ = ∠P XQ, and we conclude that O, P, Q, X are concyclic, as desired.

This now implies that ∠OQP = ∠OXP ; moreover, because triangles P XM and

OM X are isosceles we have ∠OXP = ∠P M O. Define R to be the point where

the line parallel to P Q through M meets OA. We have ∠ORM = ∠OQP =

∠P M O = ∠AM O. Therefore, triangles M RO and AM O are similar and we

OR

have OM = OM

OA

; hence

OM 2

OR = ,

OA

Q must be the midpoint of AR. We conclude that line XY passes through the

midpoint of the fixed segment AR as M moves about Γ.

II. Solution by Li Zhou, Polk Community College, Winter Haven, FL, USA.

We do not need the assumption that d is exterior to Γ so long as it is

neither tangent to Γ nor passing through O. Let p be the power of A with respect

to Γ. Note that p is negative if A is inside Γ. Let i be the transformation

defined by i(U ) = V if and only if A, U, V are collinear and as signed lengths,

AU · AV = p. [Editor’s comment. Zhou called this transformation an inversion.

When A is outside Γ, then i is indeed an inversion in the circle with centre A

that is orthogonal to Γ; when A is inside Γ, i is the commutative product of

a halfturn about A and inversion in the circle with centre A whose diameter is

the chord intercepted from d by Γ.] Note that d and Γ are invariant under i.

Let M 0 = i(M ), X 0 = i(X), and Y 0 = i(Y ). Then i(Γ0 ) is the line passing

through M 0 , X 0 , Y 0 and i(XY )is the circle passing through A, X 0 , Y 0 . Since

AX ⊥ M X, we have AM 0 ⊥ M 0 X 0 , whence M X 0 is a diameter of Γ. Let B

be the point symmetrical with A about O. Then the triangles OBX 0 and OAM

are congruent, which implies that ∠OBP = ∠OAM . This last angle equals

the directed angle between the lines Y 0 A and Y 0 X 0 (because corresponding sides

are perpendicular). We conclude that B is on the circle through A, X 0 , and Y 0 ,

whence i(B) is the fixed point through which XY passes.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece(2 solutions);

MICHEL BATAILLE, Rouen, France; CHIP CURTIS, Missouri Southern State University,

Joplin, MO, USA; OLIVER GEUPEL, Brühl, NRW, Germany; JOEL SCHLOSBERG, Bay-

side, NY, USA; PETER Y. WOO, Biola University, La Mirada, CA, USA; and TITU

ZVONARU, Cománeşti, Romania; and the proposer. There was one incorrect submission.

The final equation of solution I implies that R is the inverse of A with respect to Γ,

something Woo proved in his solution using projective geometry. Many of the other solutions

easily solved the problem with the help of coordinates or trigonometry.

62

3511. [2010 : 46, 48] Proposed by Pham Van Thuan, Hanoi University

of Science, Hanoi, Vietnam.

Let a, b, c, and d be nonnegative real numbers. Prove that

Y 1

a2 + b2 + c2 ≤ (a + b + c + d)8 .

cyclic

64

Q

We write f (a, b, c, d) = a2 + b2 + c2 , and without loss of generality

cyclic

we assume that a ≥ b ≥ c ≥ d.

Since

2

c+d

b2 + c2 + d2 ≤ b + ,

2

c+d 2

c2 + d2 + a2 ≤ a + ,

2

2

c+d 2 c+d

a2 + b2 + c2 ≤ a + + b+ ,

2 2

and

2 2

2 2 2 c+d c+d

a +c +d ≤ a+ + b+ ,

2 2

c+d c+d c+d

we obtain f (a, b, c, d) ≤ f a + ,b+ , 0, 0 . Let x = a +

2 2 2

c+d

and y = b + , so that x + y = a + b + c + d.

2

2 1

We now need to prove that x2 + y 2 x2 y 2 ≤ (x + y)8 . However, this

64

inequality follows from an application of the AM–GM inequality:

2

1 2 1 x2 + y 2 + 2xy 1 4

x2 + y 2 xy = x + y 2 (2xy) ≤ = (x + y) ,

2 2 2 8

Equality holds precisely when two of a, b, c, d are equal and the remaining

two are zero.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece (second solution);

ŠEFKET ARSLANAGI Ć, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; OLIVER

GEUPEL, Brühl, NRW, Germany; PAOLO PERFETTI, Dipartimento di Matematica,

Università degli studi di Tor Vergata Roma, Rome, Italy; PETER Y. WOO, Biola University,

La Mirada, CA, USA; TITU ZVONARU, Cománeşti, Romania; and the proposer.

63

3512. [2010 : 46, 48] Proposed by Ovidiu Furdui, Campia Turzii, Cluj, Romania.

Let α be a real number and let p ≥ 1. Find

Y

n

np + (α − 1)kp−1

lim .

n→∞

k=1

np − kp−1

Q

n

np + (α − 1)kp−1

Let pn = . Then

k=1 np − kp−1

p−1 Ǳ

k 1

X

n 1 + (α − 1) ·

ln pn = ln n n

p−1

k 1

k=1 1− ·

n n

p−1

X

n

k 1

= ln 1 + (α − 1) ·

k=1

n n

p−1

X

n

k 1

− ln 1 − · . (1)

k=1

n n

p−1 p−1

k 1 k 1 1

ln 1 + (α − 1) · = (α − 1) · +O , (2)

n n n n n2

p−1 p−1

k 1 k 1 1

ln 1 − · =− · +O . (3)

n n n n n2

n p−1

X

k 1 1

ln pn = α · +O . (4)

k=1

n n n2

P

n p−1

k 1

Note that the sum Sn = · is a Riemann sum for the function

n=1 n n

f (x) = xp−1 over the interval [0, 1]. Hence,

Z 1 1

lim Sn = xp−1 dx = . (5)

n→∞ 0 p

α

From (4) and (5) we have lim ln pn = , hence lim pn = eα/p .

n→∞ p n→∞

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; MICHEL

BATAILLE, Rouen, France; PAUL BRACKEN, University of Texas, Edinburg, TX,

USA; OLIVER GEUPEL, Brühl, NRW, Germany; JOE HOWARD, Portales, NM, USA;

ANASTASIOS KOTRONIS, Heraklion, Greece; and the proposer. Two incorrect solutions were

submitted.

64

Let α and β be positive real numbers, and r be a positive rational number.

Prove that there exist infinitely many integers m and n such that

bmαc

= r,

bnβc

where bxc is the greatest integer not exceeding x.

Solution by Oliver Geupel, Brühl, NRW, Germany, modified by the editor.

We show that the statement is false. It is well known (see D.O. Shklarsky,

N.N. Chentzov, and I.M. Yaglom, The USSR Olympiad Problem Book: selected

problems and theorems of elementary mathematics, Dover, New York, 1993,

1 1

Problem 108) that if α and β are positive irrational numbers such that + = 1,

α β

then all positive integers appear with no duplications in the two sequences bαc,

b2αc, b3αc, . . . and bβc, b2βc, b3βc, . . . . [Ed.: These are known in the

literature as complementary Beatty sequences.]

√ √ 1 1

If we take r = 1, α = 2, and β = 2 + 2, then clearly + = 1, so

α β

bmαc

there are no positive integers m and n such that = 1. It is also clear that

bnβc

m 6= 0, n 6= 0, and if m and n are of opposite signs, then bmαc 6= bnβc.

Finally, suppose m < 0 and n < 0. Using the trivial fact that

bxc + b−xc = −1 for all reals x which are not integers, we see that the

sequences b−αc, b−2αc, b−3αc, . . . and b−βc, b−2βc, b−3βc, . . .

contain all nonnegative integers with no duplications. Hence it is again

impossible for bmαc = bnβc to hold, and our proof is complete.

Two incorrect solutions were submitted.

Crux Mathematicorum

with Mathematical Mayhem

Former Editors / Anciens Rédacteurs: Bruce L.R. Shawyer, James E. Totten, Václav Linek

Crux Mathematicorum

Founding Editors / Rédacteurs-fondateurs: Léopold Sauvé & Frederick G.B. Maskell

Former Editors / Anciens Rédacteurs: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer

Mathematical Mayhem

Founding Editors / Rédacteurs-fondateurs: Patrick Surry & Ravi Vakil

Former Editors / Anciens Rédacteurs: Philip Jong, Jeff Higham, J.P. Grossman,

Andre Chang, Naoki Sato, Cyrus Hsia, Shawn Godin, Jeff Hooper, Ian VanderBurgh

65

Please send your solutions to problems in this Skoliad by October 15, 2011.

A copy of CRUX with Mayhem will be sent to one pre-university reader who

sends in solutions before the deadline. The decision of the editors is final.

Our contest this month is the National Bank of New Zealand Junior

Mathematics Competition, 2010. Our thanks go to Warren Palmer, Otago

University, Otago, New Zealand for providing us with this contest and for

permission to publish it.

La rédaction souhaite remercier Rolland Gaudet, de Collège universitaire de

Saint-Boniface, Winnipeg, MB, d’avoir traduit ce concours.

Nouvelle Zélande, 2010

Durée : 1 heure

1. Raymonde tient un séminaire à son lieu de travail. Elle désire

créer un anneau ininterrompu de tables identiques en formes de

polygones réguliers. (Dans un polygone régulier, les côtés sont de

longueurs égales et les angles sont de mesures égales. Les carrés et

les triangles équilatéraux sont réguliers.) Chaque table doit avoir

deux côtés complets qui coı̈ncident avec les côtés d’autres tables, tel qu’illustré à

droite dans le cas du carré ombré. Raymonde a l’intention d’étaler du matériel à

l’intérieur de l’anneau, de façon à ce que ce matériel soit visible par chacun.

(Si vous ne pouvez pas nommer une forme, simplement en fournir le nombre

de côtés. Par exemple, si vous croyez que la forme a 235 côtés, mais n’en connaissez

pas le nom, simplement l’appeler un 235-gone ; noter que ceci fait partie de la

réponse à aucune des questions ci-bas.)

Quel est le nombre minimal de tables carrées à placer les unes contre les

autres, de façon à ce qu’il y ait un espace vide au centre ?

2. Si Raymonde utilise le nombre minimal de tables carrées, quelle est la forme

de l’espace vide au centre ?

3. Raymonde considère maintenant utiliser des tables en formes d’octagones

(huit côtés chacune).

(a) Quel est le nombre minimal de tables octagonales que doit utiliser

Raymonde afin qu’il y ait un espace vide au centre de l’enclos ?

(b) Quel est le nom de la forme de l’espace vide au centre ? Si vous n’en

connaissez pas le nom, il suffit d’en donner le nombre de côtés.

4. À part les carrés et les octagones, y a-t-il d’autres formes de tables possibles ?

Si oui, les nommer. Sinon, indiquer qu’il n’y en a pas.

66

2. Une horloge analogue affiche le temps à l’aide de deux aiguilles. Chaque heure,

l’aiguille des minutes tourne par 360 degrés, tandis que l’aiguille des heures (qui

est plus courte que l’aiguille des minutes) tourne par 360 degrés sur une période

de 12 heures. Deux exemples suivent.

3h00 6h00

1. Dessiner une horloge qui affiche 9h00. Assurez-vous que l’aiguille des heures

est plus courte que celle des minutes.

2. Quel est l’angle entre les deux aiguilles à 3h00 et aussi à 9h00 ?

3. Quel temps est affiché par l’horloge qui suit, à l’heure et à la minute près ?

(a) 1h00.

(b) 2h00.

(c) 1h30.

5. À quel moment, à la minute près, entre 7h00 et 8h00, les aiguilles sont-elles

à la même position ?

3. Un entier à six chiffres “abcdef ” est formé en utilisant les chiffres 1, 2, 3, 4,

5, et 6, une et une seule fois chacun, de façon à ce que “abcdef ” est un multiple

de 6, “abcde” est un multiple de 5, “abcd” est un multiple de 4, “abc” est un

multiple de 3 et “ab” est un multiple de 2.

1. Déterminer une solution “abcdef .” Montrer votre travail.

2. La solution que vous avez obtenue est-elle unique (la seule possible) ? Si oui,

expliquer brièvement pourquoi. Sinon, donner une deuxième solution.

4. Un rectangle 3 × 2 est divisé en six carrés égaux, chacun contenant une

bibitte. Lorsqu’une cloche sonne, chaque bibitte saute soit horizontalement soit

verticalement pour atterrir dans un carré voisin ; les bibittent ne peuvent pas

sauter diagonalement ; aussi, elles doivent rester à l’intérieur du rectangle ; on ne

peut pas savoir d’avance où le bibittes sauteront ; enfin, toute bibitte doit changer

de carré, aucune ne pouvant rester au même carré qu’avant.

Comme exemple de façon de représenter ceci, le sextuplet (1, 1, 1, 1, 1, 1)

dénoteune situation où chaque carré contient une bibitte, que ce soit au départ

ou que ce soit à un moment plus tard, comme ça pourrait bien se produire. Deux

67

sextuplet (2, 2, 1, 0, 0, 1) représente la situation donnée par le

diagramme à droite ; il y a plusieurs successions de sauts pouvant

0 0 1

donner ceci. Le premier chiffre représente un coin, le second un carré au milieu

d’un côté, et ainsi de suite.

savoir où les bibittes sauteront ?

2. À partir de la situation initiale d’une bibitte par carré, est-ce possible que

trois bibittes se retrouvent dans le même carré, après un seul son de cloche ?

Si vous croyez que oui, écrire un sextuplet ordonné comme les deux ci-

haut pour indiquer comment ceci pourrait se produire. Sinon, expliquer

brièvement pourquoi.

3. Pour un rectangle 3 × 2, avec une situation initiale d’une bibitte par carré,

il n’est certainement pas possible que quatre bibittes se retrouvent dans

le même carré, après un seul son de cloche. Donner la taille du plus petit

rectangle pour lequel aurait été possible.

4. À partir d’une situation initiale avec une bibitte dans chaque carré, il est

impossible d’avoir cinq bibittes dans un carré, après un seul son de cloche,

quel que soit la taille du rectangle. Dans quelques mots, expliquer pourquoi.

5. Pour le cas 3 × 2, avec une bibitte dans chaque carré au départ, combien de

sextuplets non uniques, comme (1, 1, 1, 1, 1, 1), sont possibles après un seul

son de cloche ? Vous n’avez pas besoin d’en fournir la liste, bien que vous

pourriez le faire.

physique en courrant une fois par semaine autours

des deux enclos situés à le ferme de leur père, à

Kakanui ; ils vont de A à B à C à D, puis de

retour à A. (Voir le schéma.) En ligne directe de

A à C, la distance est de 6250 mètres. AB est D

plus court que BC. B C

déterminer les longueurs de ses côtés.

2. Si 4ABC est rectangle en ratio 3 : 4 : 5, avec l’angle rectangle à B,

déterminer la mesure de l’angle ∠CAB, à une décimale près.

3. L’angle à B est effectivement un angle rectangle, et AB et BC sont de

longueurs entières en mètres, mais, cette fois-ci, les côtés ne sont pas en

ratio 3 : 4 : 5. Déterminer les valeurs possibles de AB et BC.

4. L’angle à D n’est pas rectangle, mais égale plutôt 40◦ . CD est de longueur

600 mètres. Utiliser cette information pour déterminer la longueur de AD.

68

x y z

loi de sinus : = = ,

sin X sin Y sin Z

loi de cosinus : x2 = y 2 + z 2 − 2yz cos X,

où le côté x est opposé à l’angle X, le côté y est opposé à l’angle Y et le côté z

est opposé à l’angle Z.

Mathematics Competition, 2010

One hour allowed

1. Rebecca is holding a seminar at the place at which she works.

She wants to create an unbroken ring of tables, using a set of

identical tables shaped like regular polygons. (In a regular polygon,

all sides have the same length, and all angles are equal. Squares

and equilateral triangles are regular.) Each table must have two

sides which completely coincide with the sides of other tables, such as the shaded

square table seen to the right. Rebecca plans to put items on display inside the

ring where everyone can see them.

(If you cannot name a shape in this question, just give the number of sides.

For example, if you think the shape has 235 sides, but don’t know the name, just

call it a 235-gon—that isn’t an answer to any of the parts.)

1. Rebecca first decides to use identical square tables. What is the minimum

number of square tables placed beside each other so that there is an empty

space in the middle?

2. If Rebecca uses the minimal number of square tables, what shape is left bare

in the middle?

(a) What is the minimum number of octagonal tables which Rebecca must

have in order for there to be a bare space in the middle so that the

tables form an enclosure?

(b) What is the name given to the bare shape in the middle? If you can’t

name it, giving the number of sides will be sufficient.

4. Apart from squares and octagons, are there any other shaped tables possible?

If there are any, name one. If there isn’t, say so.

2. An analogue clock displays the time with the use of two hands. Every hour

the minute hand rotates 360 degrees, while the hour hand (which is shorter than

the minute hand) rotates 360 degrees over a 12-hour period. Two example times

are shown below:

69

3 o’clock 6 o’clock

1. Draw a clock face which shows 9 o’clock. Make sure the hour hand is shorter

than the minute hand.

2. What is the angle between the two hands at both 3 o’clock and 9 o’clock?

3. What time to the closest hour (and minute) does the following clock face

show?

4. What is the angle between the two hands at the following times?

(a) 1 o’clock.

(b) 2 o’clock.

(c) Half past one.

5. At what time (to the nearest minute) between 7 and 8 o’clock do the hands

meet?

5, and 6 once and only once so that “abcdef ” is a multiple of 6, “abcde” is a

multiple of 5, “abcd” is a multiple of 4, “abc” is a multiple of 3, and “ab” is a

multiple of 2.

2. Is the solution you found unique (the only possible one)? If it is, briefly

explain why. If it isn’t, give another solution.

When a bell rings, the bugs jump either horizontally or vertically (they cannot

jump diagonally and they stay within the rectangle) into a square adjacent to their

previous square in any direction, although you cannot know in advance which exact

square they will jump into. Every bug changes square; no bug stays put.

As an example, the ordered sextuplet (1, 1, 1, 1, 1, 1) (where this represents

the result, not the movement) represents the situation where every bug jumped

so that each square still had one bug in it (it could happen). Alternatively, two

70

the only way this could happen) of this might be represented

by (2, 2, 1, 0, 0, 1) —see the diagram to the right. The first

0 0 1

number in the sextuplet represents a corner square, the second represents a square

on the middle of a side, and so on.

matter how the bugs jump?

2. From the initial situation of one bug in every square, is it possible for three

bugs to end up in the same square if the bell rings only once? If you think

it is, write an ordered sextuplet like the two above where this could happen.

If you think it can’t happen, briefly explain why not.

3. From the initial situation of one bug in every square, it is certainly not

possible in a 3 × 2 rectangle for four bugs to end up in the same square if

the bell rings only once. Write down the dimensions of the smallest rectangle

for which it would be possible.

4. From the initial situation of one bug in every square, five bugs can never end

up in the same square if the bell rings only once, no matter the size of the

rectangle. In a few words, explain why not.

are possible from the initial situation of one bug in every square, if the bell

rings only once? You do not have to list them, although you might like to.

around two paddocks on their father’s farm near

Kakanui from A to B to C to D and then back

to A (see the diagram). In a direct line from A

to C, the distance is 6250 m. AB is shorter than D

BC. B C

right angle, find the lengths of the sides.

right angle, find the size of ∠CAB to one decimal place.

3. The angle at B is in fact a right angle, and AB and BC are whole metres in

length, but the sides are not in the ratio of 3 : 4 : 5. Find possible lengths

for AB and BC.

4. The angle at D is not a right angle but is 40◦ , and CD is 600 m. Use this

information to find the length of AD.

71

x y z

Sine Law: = = ,

sin X sin Y sin Z

Cosine Law: x2 = y 2 + z 2 − 2yz cos X,

opposite to angle Z.

given in Skoliad 125 at [2010:194–196].

1. Find all natural numbers n such that the sum of n and the digit sum of n

is 2010.

Consider the n-digit number abcd; that is, n = 1000a + 100b + 10c + d.

Then the digit sum of n is a + b + c + d, and the condition is that 1000a +

100b + 10c + d + a + b + c + d = 2010, so 1001a + 101b + 11c + 2d = 2010.

Since all variables represent digits, a = 1 or a = 2.

If a = 2, the condition is that 2002 + 101b + 11c + 2d = 2010, so

101b + 11c + 2d = 8. Since all variables represent digits, b and c must both be

zero, and thus d = 4. Hence n = 2004.

If a = 1, the condition is that 1001 + 101b + 11c + 2d = 2010, so

101b + 11c + 2d = 1009. Again, all variable represent digits, so

11c + 2d ≤ 11 · 9 + 2 · 9 = 117, so 101b ≥ 892, so b = 9. With a = 1 and

b = 9, the condition is that 1001 + 909 + 11c + 2d = 2010, so 11c + 2d = 100.

Since d is a digit, 2d ≤ 18, so 11c ≥ 82, so c ≥ 8 because c also is a digit.

If c = 9, then 2d = 100 − 11c = 1, which is impossible. If c = 8, then

2d = 100 − 11c = 12, so d = 6. Hence n = 1986.

Only two natural numbers satisfy the condition, namely 2004 and 1986.

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; GESINE GEUPEL, student, Max Ernst Gymnasium, Brühl, NRW, Germany;

and RICHARD I. HESS, Rancho Palos Verdes, CA, USA.

2. A regular 18-gon can be cut into congruent pentagons as in the figure below.

Determine the interior angles of such a pentagon.

72

Solution by Lena Choi, student, École Dr. Charles Best Secondary School,

Coquitlam, BC.

e d d

ca ca

b c

d

d

b d c e b e

b

c e a a d

e c e b e d

b c d

c

b d a e

a b a e a a b c

c a

b aaa e b a

e d b

d c e b

c b e e

d c

d a a

e c

b b d

e e c d b

d

c b

ac

ac

d d e

Label the angles of the congruent pentagons as in the figure. In the centre

of the figure, you can see that 6a = 360◦ , so a = 60◦ .

The angle sum of an 18-gon is (18 − 2) · 180◦ = 2880◦ , so each interior

1

angle in a regular 18-gon is 18 · 2880◦ = 160◦ . In the figure, the interior angles of

the 18-gon are e, a + c, and 2d. Thus e = 160◦ , a + c = 160◦ , and 2d = 160◦ ,

so c = 100◦ and d = 80◦ .

In the figure you will also find that 2b + d = 360◦ , so 2b = 280◦ , so

b = 140◦ .

The angles of the pentagon are (a, b, c, d, e) = (60◦ , 140◦ , 100◦ , 80◦ , 160◦ ).

Also solved by VINCENT CHUNG, student, Burnaby North Secondary School, Burnaby,

BC; NATALIA DESY, student, SMA Xaverius 1, Palembang, Indonesia; RICHARD I. HESS,

Rancho Palos Verdes, CA, USA; and ROWENA HO, student, École Dr. Charles Best Secondary

School, Coquitlam, BC.

C

3. In the figure on the right, 4ABE is

isosceles with base AB, ∠BAC = 30◦ , E

and ∠ACB = ∠AF C = 90◦ . Find the S

ratio of the area of 4ESC to the area

of 4ABC.

A F B

Solution by Vincent Chung, student, Burnaby North Secondary School, Burnaby,

BC.

Since ∠BAC = 30◦ and ∠ACB = 90◦ , ∠ABC = 60◦ ; that is 4ABC

is a 30◦ –60◦ –90◦ triangle. Assume without loss of generality that AB = 2,

√ √ √

AC = 3, and BC = 1. Then the area of 4ABC is 23·1 = 23 .

Since 4ABE is isosceles and ∠BAC = 30◦ , ∠ABE = 30◦ and

∠AEB = 120◦ . Thus ∠CES = 60◦ . Since ∠ABE = 30◦ and ∠AF C = 90◦ ,

73

Since ∠ABC = 60◦ and ∠ABE = 30◦ , ∠CBE = 30◦ . As

∠ACB = 90◦ , it follows that ∠BEC = 60◦ , and 4BCE is a 30◦ –60◦ –90◦

triangle. Since BC = 1, BE = √23 and CE = √13 .

1

Since 4ESC is equilateral with side √

3

, its height is

s r r

2 2

1 1 1 1 1 1

√ − √ = − = = .

3 2 3 3 12 4 2

1 1 √

√ · 2

Therefore the area of 4ESC is 3 = 123 .

2

√

The ratio of the area of 4ESC to the area of 4ABC is, then,

√

3 3 1

12

: 2

= 12 : 12 = 1 : 6.

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; NATALIA DESY, student, SMA Xaverius 1, Palembang, Indonesia; RICHARD

I. HESS, Rancho Palos Verdes, CA, USA; ROWENA HO, student, École Dr. Charles Best

Secondary School, Coquitlam, BC; and KENRICK TSE, student, Point Grey Secondary,

Vancouver, BC.

zn−1

4. Given two nonzero numbers z1 and z2 , let zn be

zn−2

for n > 2. Then

z1 , z2 , z3 , . . . form a sequence. Prove that if you multiply any 2009 consecutive

terms of the sequence, then the product is itself a member of the sequence.

Using the recursive definition of zn ,

z2 z3 z /z 1

z3 = , z4 = = 2 1 = ,

z1 z2 z2 z1

z4 1/z1 1 z5 1/z2 z1

z5 = = = , z6 = = = ,

z3 z2 /z1 z2 z4 1/z1 z2

z z1 /z2 z z1

z7 = 6 = = z1 , z8 = 7 = = z2 , . . .

z5 1/z2 z6 z1 /z2

z2

Since z7 = z1 and z8 = z2 , the sequence will repeat with period six: z1 , z2 , ,

z1

1 1 z1 z

, , , z1 , z2 , 2 , . . . . Note that the reciprocal of each of the first six terms

z1 z2 z2 z1

is itself one of the first six terms. Therefore, whenever zn is a member of the

sequence, then so is its reciprocal, z1n .

The product of the first six terms is 1. Therefore the product of any six

consecutive terms is 1. Since 2010 = 6 · 335, it follows that the product of any

2010 terms is 1. Therefore the product of any 2009 consecutive terms is the

reciprocal of the next term, but this reciprocal is itself a member of the sequence

as noted above.

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; JONATHAN FENG, student, Burnaby North Secondary School, Burnaby, BC;

RICHARD I. HESS, Rancho Palos Verdes, CA, USA; and ROWENA HO, student, École

Dr. Charles Best Secondary School, Coquitlam, BC.

74

5. Let 4ABC be an isosceles triangle such that ∠ACB = 90◦ . A circle with

centre C cuts AC at D and BC at E. Draw the line AE. The perpendicular to

AE through C cuts the line AB at F , and the perpendicular to AE through D

cuts the line AB at G. Show that the length of BF equals the length of GF .

Impose a coordinate system such that

A

C = (0, 0), A = (0, 1), B = (1, 0),

D = (0, r), and E = (r, 0), where r is G

the radius of the circle. Then the line AB D

has the equation y = 1 − x.

The slope of the line AE is − r1 . Since F

AE is perpendicular to CF , the slope of CF

must then be r. (The product of the slopes

of perpendicular lines is −1.) Therefore the C E B

equation of CF is y = rx. Now, F is the in-

tersection point of AB and CF , so it is the

intersection of y = 1 − x and y = rx.

Solving these two equations simultaneously

1 r

yields that rx = 1 − x, so x = r+1 and, thus, y = rx = . That is,

r+1

1

F = , r

r+1 r+1

.

Similarly, the slope of DG is also r, so the equation of DG is y = rx + r.

1−r

Intersection with AB, that is y = 1 − x, yields that rx + r = 1 − x, so x = r+1

r+1 1−r 2r 1−r

and, thus, y = 1 − x = r+1

− r+1

= r+1

. That is, G = , 2r

r+1 r+1

.

You can now calculate the distance from B to F using the Pythagorean

2 2 2 2 2

1 r −r r 2r

Theorem: |BF |2 = r+1 − 1 + r+1 − 0 = r+1 + r+1 = (r+1)2,

√ 2 2

r 1−r 1 2r r

so |BF | = r+1 2. Likewise, |GF |2 = r+1 − r+1 + r+1 − r+1 =

2 √

−r 2 r 2r 2 r

r+1

+ r+1 = (r+1)2 , so |GF | is also r+1 2. Thus |BF | = |GF |.

Also solved by GEOFFREY A. KANDALL, Hamden, CT, USA, who uses this geometric

approach:

Let H be the point of intersection between CF and

A

AE, and let the line through E parallel to AB meet CF

x

at I. Since |CD| = |CE| and 4ABC is isosceles, you w

can label the lengths as in the figure. The problem now is G

to show that y = z. D y

Since CH ⊥ AE, 4ACE ∼ 4AHC. Therefore

|AC| |AH|

= |AC| , so (r + w)2 = |AC|2 = |AE| · |AH|. F

|AE| r H

|CE| |HE|

Again, since 4ACE ∼ 4CHE, |AE|

= |CE|

, I z

so r2 = |CE|2 = |AE| · |HE|. Hence

(r+w)2 |AE|·|AH| |AH| C r E w B

r2

= |AE|·|HE|

= |HE|

.

2

|AF | |AH| (r+w)

Since EI k AF , 4HAF ∼ 4HEI, so |IE|

= |HE|

= r2

. Moreover,

|EI| |CE| r x+y |AF | |AF | |IE|

4CIE ∼ 4CF B, so |F B|

= |CB|

= r+w

. Therefore z

= |F B| = |IE| · |F B|

=

(r+w)2 r r+w

r2

· r+w

= r

.

75

x+y

Since DG k CF , 4ADG ∼ 4ACF , so x

= w+r

w

, so 1 + x y

= 1+ w r

, so

y r x+y w+r

x

= w , so x

y

= w

r

, so x

y

+1 = w

r

+ 1, so y

= r

. The previous paragraph shows that

x+y

z

= r+wr

. Thus x+y

y

= x+y

z

, and it follows that y = z.

We leave it to the reader to judge whether this geometric argument is preferable to the

analytic geometry of the first solution.

6. A gaming machine randomly selects a divisor of 20092010 and displays its ones

digit. Which digit should you gamble on?

Solution by Lena Choi, student, École Dr. Charles Best Secondary School,

Coquitlam, BC.

Since 2009 = 72 · 41, 20092010 = 74020 · 412010. Thus any divisor

of 20092010 has the form 7a · 41b , where a and b are integers such that

0 ≤ a ≤ 4020 and 0 ≤ b ≤ 2010.

Note that the ones digit of a product depends only on the ones digits of the

factors, and use x ≡ y to mean that x and y have the same ones digit. Then

41b ≡ 1 for all 2011 possible values of b. Moreover, 7a · 41b ≡ 7a · 1 = 7a ,

so you just have to study the powers of 7: 70 = 1, 71 = 7, 72 = 49 ≡ 9,

73 = 72 · 7 ≡ 9 · 7 = 63 ≡ 3, and 74 = 73 · 7 ≡ 3 · 7 = 21 ≡ 1.

Since the ones digit reached 1 again, the sequence of ones digits now repeats

with period four. That is,

7a ≡ 3 if a = 3, 7, 10, . . . , 4019

7a ≡ 9 if a = 2, 6, 9, . . . , 4018

7a ≡ 7 if a = 1, 5, 8, . . . , 4017

7a ≡ 1 if a = 0, 4, 7, . . . , 4016, 4020 .

Thus 7a has ones digit 1 for 1006 values of a, while 7a only has ones digit

7, 9, or 3 for 1005 values of a each.

Hence the ones digit 1 occurs slightly more often among the divisors of 20092010

than any other digit, so you should gamble on the digit 1.

Also solved by RICHARD I. HESS, Rancho Palos Verdes, CA, USA.

This issue’s prize of one copy of Crux Mathematicorum for the best

solutions goes to Kenrick Tse, student, Point Grey Secondary, Vancouver, BC.

We look forward to receiving our readers’ solutions to our featured contest.

76

MATHEMATICAL MAYHEM

Mathematical Mayhem began in 1988 as a Mathematical Journal for and by

High School and University Students. It continues, with the same emphasis,

as an integral part of Crux Mathematicorum with Mathematical Mayhem.

The interim Mayhem Editor is Shawn Godin (Cairine Wilson Secondary

School, Orleans, ON). The Assistant Mayhem Editor is Lynn Miller (Cairine

Wilson Secondary School, Orleans, ON). The other staff member is Monika Khbeis

(Our Lady of Mt. Carmel Secondary School, Mississauga, ON).

Mayhem Problems

Veuillez nous transmettre vos solutions aux problèmes du présent numéro avant le

15 septembre 2010. Les solutions reçues après cette date ne seront prises en compte

que s’il nous reste du temps avant la publication des solutions.

Chaque problème sera publié dans les deux langues officielles du Canada

(anglais et français). Dans les numéros 1, 3, 5 et 7, l’anglais précédera le français,

et dans les numéros 2, 4, 6 et 8, le français précédera l’anglais.

La rédaction souhaite remercier Jean-Marc Terrier, de l’Université de Montréal,

d’avoir traduit les problèmes.

On définit comme s(n) la somme des chiffres de l’entier positif n. Par

exemple, s(2011) = 2 + 0 + 1 + 1 = 4. Trouver le nombre d’entiers positifs de

quatre chiffres n avec s(n) = 4.

M477. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie.

Soit m un paramètre entier tel que l’équation x2 − mx + m + 8 = 0 ait

une racine entière. Trouver la valeur du paramètre m.

On considère l’ensemble des points (x, y) du plan tels que

x2 + y 2 − 22x − 4y + 100 = 0 .

y

Soit P le point de cet ensemble pour lequel est maximal. Déterminer la distance

x

de P à l’origine.

77

M479. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie.

Soit A = 1 · 2 · 3 · · · · · 2011 = 2011!.

(a) Trouver le plus grand entier positif n pour lequel 3n est un diviseur de A.

(b) Trouver le nombre de zéros en queue de la représentation de A en base 10.

Soit x, y et k trois nombres positifs tels que x2 + y 2 = k. Trouver la valeur

minimale possible de x6 + y 6 en fonction de k.

M481. Proposé par Edward T.H. Wang, Université Wilfrid Laurier, Waterloo,

ON.

On suppose que a, b et x sont des nombres réels avec ab 6= 0 et a + b 6= 0.

sin4 x cos4 x 1 sin6 x cos6 x

Si + = , trouver la valeur de 3

+ en fonction de

a b a+b a b3

a et b.

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

Define s(n) to be the sum of the digits of the positive integer n. For

example, s(2011) = 2 + 0 + 1 + 1 = 4. Determine the number of four-digit

positive integers n with s(n) = 4.

Buzău, Romania

Let m be an integer parameter such that the equation x2 −mx+m+8 = 0

has one integer root. Determine the value of the parameter m.

Consider the set of points (x, y) in the plane such that

x2 + y 2 − 22x − 4y + 100 = 0 .

y

Let P be the point in this set for which is the largest. Determine the distance

x

of P from the origin.

Buzău, Romania

Let A = 1 · 2 · 3 · · · · · 2011 = 2011!.

(a) Determine the largest positive integer n for which 3n divides exactly into

A.

(b) Determine the number of zeroes at the end of the base 10 representation of

A.

78

minimum possible value of x6 + y 6 in terms of k.

ON

sin4 x cos4 x 1 sin6 x cos6 x

+ = , determine the value of 3

+ in terms of a

a b a+b a b3

and b.

Mayhem Solutions

x2 + (y 2 − y − 2)2 = 0 .

y 2 − y − 2 = (y − 2)(y + 1) = 0 must hold, so the solutions are: (0, 2), (0, −1).

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ADAMAS AQSA

F.S., student, SMA Kharisma Bangsa, Indonesia; JACLYN CHANG, student, University

of Calgary, Calgary, AB; NATALIA DESY, student, SMA Xaverius 1, Palembang,

Indonesia; RICHARD I. HESS, Rancho Palos Verdes, CA, USA; ANTONIO LEDESMA

LÓPEZ, Instituto de Educación Secundaria No. 1, Requena-Valencia, Spain; PEDRO

HENRIQUE O. PANTOJA, student, UFRN, Brazil; RICARD PEIRÓ, IES “Abastos”,

Valencia, Spain; PAOLO PERFETTI, Dipartimento di Matematica, Università degli studi

di Tor Vergata Roma, Rome, Italy; CAO MINH QUANG, Nguyen Binh Khiem High

School, Vinh Long, Vietnam; BRUNO SALGUEIRO FANEGO, Viveiro, Spain; NECULAI

STANCIU, George Emil Palade Secondary School, Buzău, Romania; GUSNADI WIYOGA,

student, SMPN 8, Yogyakarta, Indonesia; and KONSTANTINE ZELATOR, University of

Pittsburgh, Pittsburgh, PA, USA.

1 1 1

Determine the positive integer x for which + = .

log2 x log5 x 100

79

1

Note that since loga b · logb a = 1 then loga b = logb a

. Consequently, we

1 1

have that log2 x

= logx 2 and log5 x

= logx 5. Thus,

1

logx 2 + logx 5 =

100

1

logx 10 =

100

1

x 100 = 10

x = 10100

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; BRUNO

SALGUEIRO FANEGO, Viveiro, Spain; G.C. GREUBEL, Newport News, VA, USA;

RICHARD I. HESS, Rancho Palos Verdes, CA, USA; YOUNGHUAN JUNG, The Wood-

lands School, Mississauga, ON; MAR ÍA ASCENSI ÓN L ÓPEZ CHAMORRO, I.B. Leopoldo

Cano, Valladolid, Spain; DRAGOLJUB MILO ŠEVIĆ, Gornji Milanovac, Serbia; NATALIA

DESY, student, SMA Xaverius 1, Palembang, Indonesia; PEDRO HENRIQUE O. PANTOJA,

student, UFRN, Brazil; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain; PAOLO

PERFETTI, Dipartimento di Matematica, Università degli studi di Tor Vergata Roma,

Rome, Italy; CAO MINH QUANG, Nguyen Binh Khiem High School, Vinh Long, Vietnam;

ALEJANDRO S. CONCEPCIÓN RODRÍGUEZ, student, University of Las Palmas de Gran

Canaria; NECULAI STANCIU, George Emil Palade Secondary School, Buzău, Romania;

KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh, PA, USA; and ALLEN ZHU,

Conestoga High School, Berwyn, PA, USA.

M441. Proposed by Katherine Tsuji and Edward T.H. Wang, Wilfrid Laurier

University, Waterloo, ON.

What is the maximum number of non-attacking kings that can be placed

on an n × n chessboard? (A “king” is a chess piece that can move horizontally,

vertically, or diagonally from one square to an adjacent square.)

We will denote by (i, j) the square situated on row i and column j, where

1 ≤ i, j ≤ n.

Case I: If n is even

In each of the n rows, we can place at most n 2

kings, because between any two

neighbouring kings there must exist at least one free square. Analogously, in each

of the n columns we can place at most n 2

kings too. If we place a king at the

squares (i, j), where i, j are odd numbers then we have n 2

kings in each column

and in each row that are non-attacking. Thus, in the n × n chessboard with n

2

n n n

even, we can place a maximum of 2

· 2

= 2

kings.

Case II: If n is odd

In each of the n rows we can place a maximum number of n+1

2

non-attacking kings.

Similar to above, if we want to have the maximum number of non-attacking kings,

80

then we must place a king at the squares (i, j), where i, j are odd numbers giving

n+1

2

kings in each column and in each row that are non-attacking. Thus, in the

n+1 n+1 n+1 2

n × n chessboard with n odd, we can place a maximum of 2

· 2

= 2

kings.

GESINE GEUPEL, student, Max Ernst Gymnasium, Brühl, NRW, Germany; RICHARD

I. HESS, Rancho Palos Verdes, CA, USA; ANTONIO LEDESMA LÓPEZ, Instituto

de Educación Secundaria No. 1, Requena-Valencia, Spain; GUSNADI WIYOGA, student,

SMPN 8, Yogyakarta, Indonesia; ALLEN ZHU, Conestoga High School, Berwyn, PA, USA;

and the proposers.

Buzău, Romania.

Let bxc denote the greatest integer not exceeding x. For example, b3.1c =

3 and b−1.4c = −2. Let {x} denote the fractional part of the real number x

(that is, {x} = x − bxc). For example, {3.1} = 0.1 and {−1.4} = 0.6. Find

all positive real numbers x such that

§ ª

2x + 3 2x + 1 14

+ = .

x+2 x+1 9

First, note that 2x+3

x+2

= 2 − x+21

. Similarly, 2x+1

x+1

1

= 2 − x+1 .

1 1

Since x is a positive real number, 0 < x+1 < 1 and 0 < x+2 < 1, which

1 1

implies that 1 < 2 − < 2. This means that 2 − = 1. Also, as

x+2 x+2

1 1

1 < 2− x+1

< 2, then that means that 2 − = 1. Now, assembling the

x+1

values we obtained plus the definition of {n}, we have

§ ª

2x + 3 2x + 1 14

+ =

x+2 x+1 9

2x + 3 2x + 3 2x + 1 14

− + =

x+2 x+2 x+1 9

2x + 3 14

−1+1 =

x+2 9

2x + 3 14

=

x+2 9

9(2x + 3) = 14(x + 2)

4x = 1

1

x =

4

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; BRUNO

SALGUEIRO FANEGO, Viveiro, Spain; RICHARD I. HESS, Rancho Palos Verdes, CA,

USA; MAR ÍA ASCENSI ÓN L ÓPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain;

81

SAMUEL GÓMEZ MORENO, Universidad de Jaén, Jaén, Spain; RICARD PEIRÓ, IES

“Abastos”, Valencia, Spain; PAOLO PERFETTI, Dipartimento di Matematica, Università degli

studi di Tor Vergata Roma, Rome, Italy; NECULAI STANCIU, George Emil Palade Secondary

School, Buzău, Romania; EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, ON;

GUSNADI WIYOGA, student, SMPN 8, Yogyakarta, Indonesia; KONSTANTINE ZELATOR,

University of Pittsburgh, Pittsburgh, PA, USA; and ALLEN ZHU, Conestoga High School,

Berwyn, PA, USA.

Catalunya, Barcelona, Spain.

Let a and b be real numbers. Prove that

p p √

a2 + b2 + 6a − 2b + 10 + a2 + b2 − 6a + 2b + 10 ≥ 2 10 .

Let A = (−3, 1), B = (3, −1) and P = (a, b). Then

p È

a2 + b2 + 6a − 2b + 10 = (a + 3)2 + (b − 1)2 = P A,

p È

a2 + b2 − 6a + 2b + 10 = (a − 3)2 + (b + 1)2 = P B,

È √ √

AB = (−6)2 + 22 = 40 = 2 10.

P A + P B ≥ AB

p p √

a2 + b2 + 6a − 2b + 10 + a2 + b2 − 6a + 2b + 10 ≥ 2 10.

Note that equality holds if and only if P is on the line segment, l, connecting A

and B. Since the slope of l is − 13 , its equation is y = − 13 x. Hence, equality

holds if and only if a = −3b, where −1 ≤ b ≤ 1.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; SHAMIL

ASGARLI, student, Burnaby South Secondary School, Burnaby, BC; RICHARD I. HESS,

Rancho Palos Verdes, CA, USA; BRUNO SALGUEIRO FANEGO, Viveiro, Spain; MAR ÍA

ASCENSI ÓN L ÓPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain; SAMUEL

GÓMEZ MORENO, Universidad de Jaén, Jaén, Spain; PEDRO HENRIQUE O. PANTOJA,

student, UFRN, Brazil; PAOLO PERFETTI, Dipartimento di Matematica, Università degli

studi di Tor Vergata Roma, Rome, Italy; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain;

CAO MINH QUANG, Nguyen Binh Khiem High School, Vinh Long, Vietnam; NECULAI

STANCIU, George Emil Palade Secondary School, Buzău, Romania; GUSNADI WIYOGA,

student, SMPN 8, Yogyakarta, Indonesia; ALLEN ZHU, Conestoga High School, Berwyn, PA,

USA; and the proposer.

82

Ian VanderBurgh

problems. This month, we’ll continue by looking at two more problems on this

topic.

On a calculus exam, the average of those who studied was 90% and the average

of those who did not study was 40%. If the average of the entire class was 85%,

what percentage of the class did not study?

Solution 1 to Problem 1. Let x be the number of people who studied for the

exam and let y be the number of people who did not study. We assume without

loss of generality that the exam was out of 100 marks.

Since the average of those who studied was 90%, then those who studied

obtained a total of 90x marks. Since the average of those who did not study was

40%, then those who did not study obtained a total of 40y marks. Since the overall

average was 85%, then 90x+40y

x+y

= 85. Therefore, 90x + 40y = 85x + 85y or

5x = 45y or x = 9y.

Therefore, x : y = 9 : 1 = 90 : 10. This means that 10% of the class did

not study for the exam.

This is a good solution, but doesn’t take advantage of what we looked at

last month related to weighted averages. Let’s try this approach.

Solution 2 to Problem 1. The combined average (85%) splits the two partial

averages (40% and 90%) in the ratio 45 : 5 or 9 : 1. This means that the number

of people in the two categories must be in the inverse ratio, or 1 : 9. This is the

ratio of the number of students who did not study to the number of students who

did study. This ratio is equivalent to 10 : 90.

Therefore, the percentage of the class that did not study is 10%.

That was much easier, wasn’t it? Here is a second problem for this month

involving averages.

Problem 2 (2010 Cayley Contest) Connie has a number of gold bars, all of

different weights. She gives the 24 lightest bars, which weigh 45% of the total

weight, to Brennan. She gives the 13 heaviest bars, which weigh 26% of the total

weight, to Maya. She gives the rest of the bars to Blair. How many bars did Blair

receive?

(A) 14 (B) 15 (C) 16 (D) 17 (E) 18

This problem is one of my favourites from the past couple of years. One

reason that I like this problem is that it’s not at all obvious that there is enough

information to solve the problem. In fact, a couple of people involved in the contest

83

creation process were convinced that there was something missing! However, there

is enough information to solve Problem 2. Give it a try before reading on!

Solution to Problem 2. Connie gives 24 bars that account for 45% of the

total weight to Brennan. Thus, each of these 24 bars accounts for an average of

45

24

% = 158

% = 1.875% of the total weight.

Connie gives 13 bars that account for 26% of the total weight to Maya. Thus,

each of these 13 bars accounts for an average of 26

13

% = 2% of the total weight.

Since each of the bars that she gives to Blair is heavier than each of the bars

given to Brennan (which were the 24 lightest bars) and is lighter than each of the

bars given to Maya (which were the 13 heaviest bars), then the average weight of

the bars given to Blair must be larger than 1.875% and smaller than 2%.

Note that the bars given to Blair account for 100% − 45% − 26% = 29%

of the total weight. If there were 14 bars accounting for 29% of the total weight,

29

the average weight would be 14 % ≈ 2.07%, which is too large. Thus, there must

be more than 14 bars accounting for 29% of the total weight.

If there were 15 bars accounting for 29% of the total weight, the average

weight would be 2915

% ≈ 1.93%, which is in the correct range. If there were 16

bars accounting for 29% of the total weight, the average weight would be 29

16

%≈

1.81%, which is too small. The same would be true if there were 17 or 18 bars.

Therefore, Blair must have received 15 bars.

When we read this problem for the first time, the fact that averages might

enter in is not clear. But averages are useful in a pretty natural way, and perhaps

in a “real life” way too. Often, using an average is a great way of estimating the

size of objects in a collection, and that’s exactly what we’ve done here. We don’t

know the actual sizes of the gold bars, but we can estimate and compare by using

averages.

As one final note on this problem, can you find the piece of information in

this problem that seemed important, but was never used?

84

No. 292

R.E. Woodrow

We begin the section of solutions from our readers with the file of solutions

to problems of the Thai Mathematical Olympiad Examinations 2006, Selected

problems, given at [2010 : 83–84].

Solved by Arkady Alt, San Jose, CA, USA; David E. Manes, SUNY at Oneonta,

Oneonta, NY, USA; and Titu Zvonaru, Cománeşti, Romania. We use an edited

version of the solution of Alt.

Let p (x) = x2 + x + 3, q (x) = x2 − 3x + 5. Since

p (1 − x) = (1 − x)2 + (1 − x) + 3 = x2 − 3x + 5 = q (x) ,

2

6 (1 − x) − 10 (1 − x) + 17 = 6x2 − 2x + 13

then

§

f (p (x)) + 2f (q (x)) = 6x2 − 10x + 17

2f (p (x)) + f (q (x)) = 6x2 − 2x + 13

we obtain

= 2 6x2 − 2x + 13 − 6x2 − 10x + 17

= 6x2 + 6x + 9 ⇐⇒ f (p (x)) = 2p(x) − 3.

11 11

Since p (x) = x2 + x + 3 ≥ then for any y ≥ there is x such that

4 4

11

p (x) = y and, therefore, for any y ≥ we have

4

f (y) = f (p (x)) = 2p(x) − 3 = 2y − 3.

85

2. Evaluate

X

8000

k 8084 − k

.

k=84

84 84

Solved by Arkady Alt, San Jose, CA, USA; Oliver Geupel, Brühl, NRW,

Germany; David E. Manes, SUNY at Oneonta, Oneonta, NY, USA; and

Edward T.H. Wang, Wilfrid Laurier University, Waterloo, ON. We give Wang’s

combinatorial argument.

More generally, we consider the sum

X

n−d

k n−k

S(n, d) =

k=d

d d

j k

n

where n, d ∈ N such that d ≤ . We use a 2-way counting argument to

2

n+1

show that S(n, d) = 2d+1

. Let T = {0, 1, 2, . . . , n}. Then |T | = n + 1

n+1

so 2d+1

is the number of (2d + 1)-subsets of T . On the other hand, each

(2d + 1)-subset of T is completely determined by first choosing a positive integer

k to be the middle number, d ≤ k ≤ n − d and then select any d numbers

from the k-subset {0, 1, 2, . . . , k − 1} and any d numbers from the (n − k)-

subset {k + 1, k + 2, . . . , n}. This procedure which is possible since d ≤ k and

d ≤ n − k would yield all the (2d + 1)-subsets of T . Hence

X

n−d

k n−k n+1

S(n, d) = =

k=d

d d 2d + 1

8085

follows. In particular, the value of the given summation is S(8084, 84) = 169

.

Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA; and Titu

Zvonaru, Cománeşti, Romania. We give the solution by Zonvaru.

If n2 + 59n + 881 is a perfect square, then 4(n2 + 59n + 881) is a perfect

square. We have

k2 = 4n2 + 4 · 59n + 3524

⇔ k2 = 4n2 + 4 · 59n + 592 + 3524 − 592

⇔ k2 = (2n + 59)2 + 43 ⇔ (k − 2n − 59)(k + 2n + 59) = 43.

Since k, n are integers and 43 is prime, we have the following possibilities

§

k − 2n − 59 = 1

(i) ⇔ k = 22, n = −19;

k + 2n + 59 = 43

§

k − 2n − 59 = −1

(ii) ⇔ k = −22, n = −40;

k + 2n + 59 = −43

86

§

k − 2n − 59 = 43

(iii) ⇔ k = 22, n = −40;

k + 2n + 59 = 1

§

k − 2n − 59 = −43

(iv) ⇔ k = −22, n = −19.

k + 2n + 59 = −1

It results that n2 + 59n + 881 is a perfect square for n = −19 and

n = −40.

√ n+1

3z − zn − 1 = 0

has a complex root z with |z| = 1.

Let z = cos ϕ + i sin ϕ, ϕ ∈ [0, 2π). Since

nϕ nϕ nϕ

1 + cos nϕ + i sin nϕ = 2 cos2 + 2i cos sin

2 2 2

nϕ nϕ nϕ

= 2 cos cos + i sin

2 2 2

then

√ n+1

3z = 1 + cos nϕ + i sin nϕ

√ n+1

nϕ nϕ nϕ

⇐⇒ 3z = 2 cos cos + i sin

2 2 2

yields

√ n+1

3 z = 2 cos nϕ cos nϕ + i sin nϕ

2 2 2

√ √

nϕ 3 nϕ

⇐⇒ 3 |z|n+1 = 2 cos ⇐⇒ = cos

2 2 2

√

3 nϕ 3 nϕ 1

⇐⇒ = cos ⇐⇒ = 2 cos2 ⇐⇒ cos nϕ = .

2 2 2 2 2

√ √

1 3 1 3

If cos nϕ = then sin nϕ = ± and z n = ± i.

2 √ 2 2 2

√

3 i 3 1 i 3

Since z n + 1 = ± , z n+1 = z · z n = z ± then

2 2 2 2

√ n+1 n

√ 1 √

i 3 3

√

i 3

3z = z + 1 ⇐⇒ 3z ± = ±

2 2 2 2

√ √

1 3 3 1

⇐⇒ z ± i = ± i

2 2 2 2

π π π π

⇐⇒ z cos ± + i sin ± = cos ± + i sin ±

3 3 6 6

π π

⇐⇒ z = cos ∓ + i sin ∓ .

6 6

87

π nπ

Hence, ± = ∓ + 2kπ ⇐⇒ n = ±12k − 2, k ∈ Z and, therefore,

3 6

the smallest positive integer n satisfying this equation is n = 10. So, a necessary

condition is n ≥ 10.

π π

Let z = cos + i sin and n = 10 then

6 6

√

10π 10π 5π 5π 1 i 3

z 10 = cos + i sin = cos + i sin = − ,

6 6 3 3 2 2

√

11π 11π 3 1

z 11 = cos + i sin = − i,

6 6 2 2

√ √

√ 3 3 1 3

and 3z n+1 = − i=1+ − = 1 + zn .

2 2 2 2

√

Thus, the least positive integer n such that 3z n+1 − z n − 1 = 0 has a complex

root with |z| = 1 is 10.

5. Let pk denote the kth prime number. Find the remainder when

X

2550

p4 −1

p kk

k=2

is divided by 2550.

First we factor 2550 into prime powers: 2550 = (255)·(10) = 5·51·10 =

5 · 3 · 17 · 2 · 5. So, we easily see that

First, note that p2 = 3, p3 = 5, and p7 = 17. We will first find the congruence

p4 −1 p4 −1 p4 −1

classes the three integers p22 , p33 , and p77 ; belong to modulo 2550. We

p4 −1 4

start with p22 = 33 −1

= 380 .

Clearly

2

2 2

38 ≡ 34 ≡ (81) ≡ (6) ≡ 36 ≡ 11 (mod 25) .

So that,

≡ (−4)4 · (−4) ≡ (256)(−4) ≡ 6 · (−4)

≡ −24 ≡ 1 (mod 25) (4)

88

§ ª

380 ≡ 1 (mod 2) , 380 ≡ 1 (mod 17) ,

80 (5)

3 ≡ 1 mod 52 and 380 ≡ 0 (mod 3)

380 − 1 ≡ 0 mod 2 · 52 · 17 ; 380 ≡ 1 mod 2 · 52 · 17 . Thus,

380 = 2 · 52 · 17 · k + 1, for some positive integer k. Since k = 3 · k0 + r where

r ∈ {0, 1, 2} and k0 ∈ Z+ , and since 380 ≡ 0 (mod 3) and 42 · 52 · 17 ≡

2 · 1 · 2 ≡ 4 ≡ 1 (mod 3) we see that

380 = 2 · 3 · 52 · 17 · k0 + 2 · 52 · 17 · r + 1

≡ r + 1 ≡ 0 (mod 3)

so r = 2. Thus,

380 = 2 · 3 · 52 · 17 · k0 + 2 · 52 · 17 · 2 + 1

= 2550 · k0 + 1700 + 1 = 2550 · k0 + 1701

p4 −1

p2 = 3, p2 2 ≡ 1701 (mod 2550) . (6)

p4

3 −1

4 2

−1)(52 +1)

Next, consider p3 = 55 −1

= 5624 = 5(5 = 524·26. Clearly

5624 ≡ 0 mod 52 ;

5624 ≡ 1 (mod 2) ; and

5624 = 524·26 = 58·2·3·13 = (516 )39 ≡ 139 ≡ 1 (mod 17)

by Fermat’s Theorem.

We see that 5624 ≡ 1 (mod 3), 5624 ≡ 1 (mod 2), 5624 ≡ 1 (mod 17)

which implies that 5624 ≡ 1 (mod 2 · 3 · 17) so 5624 = 2 · 3 · 17 · l + 1 for

some positive integer l. Observe that 2 · 3 · 17 ≡ 6(−8) ≡ −48 ≡ −(−2) ≡ 2

(mod 25). And so, 5624 = 2 · 3 · 17 · l + 1 ≡ 2l + 1 (mod 25). But 5624 ≡ 0

(mod 25); and so we must have 2l + 1 ≡ 0 (mod 25) ⇔ l ≡ 12 (mod 25);

l = 25 · L + 12; for some L ∈ Z+ . Thus,

| · 3 ·{z

2550

p4 −1

p3 = 5, p3 3 ≡ 1225 (mod 2550) . (7)

p4 −1 4 2

−1)(172 +1)

p7 7 = 1717 −1

= 17(17 = 1716·18·290.

89

And

1716·18·290 ≡ 1 (mod 2) and

1716·18·290 ≡ 1 (mod 3) (since 172 ≡ 1 (mod 3))

And also, 1716·18·290 ≡ 0 (mod 17)

Consider 1716·18·290 modulo 52 = 25. Observe that, 172 = 289 ≡ 275 + 14 ≡

14 (mod 25); or equivalently, 172 ≡ −11 (mod 25)

1716 ≡ (172 )8 ≡ (−11)8 ≡ [(−11)2 ]4 ≡ (121)4

≡ (−4)4 ≡ 256 ≡ 1 (mod 25) .

And thus 1716·18·290 ≡ [(17)16 ]18·290 ≡ 118·290 ≡ 1 (mod 25). We have

1716·18·290 ≡ 1 (mod 2), 17

16·18·290

≡ 1 (mod 3), and 1716·18·290

≡ 1

2 16·18·290

mod 5 , which implies that 17 ≡ 1 mod 2 · 3 · 52 ; and so

1716·18·290 = 2 · 3 · 52 · m + 1; for some m ∈ Z+ . And since 2 · 3 · 52 ≡

2 · 3 · 25 ≡ 2 · 3 · 8 ≡ 2 · 3 · 8 ≡ 48 ≡ −3 (mod 17). We see that we must

have 2 · 3 · 52 · m + 1 ≡ 0 (mod 17); −3m + 1 ≡ 0 (mod 17); 3m ≡ 1

(mod 17); m ≡ 6 (mod 17). So that m = 6 + 17 · M ; for some M ∈ Z+ .

Altogether,

1716·18·290 = 2 · 3 · 52 · (6 + 17M ) + 1

52 · 17} ·M + 2 · 3 · 52 · 6 + 1 = 2550M + 901

= |2 · 3 ·{z

2550

p4 −1

p7 = 17, p7 7 ≡ 901 (mod 2550) (8)

p4

k −1

We now come to the last part of the problem by considering pk ; where

k ≥ 2 and k 6= 2, 3, 7. In other words, pk 6= 3, 5, 17; and pk > 2.

Since pk is odd, we have p2k ≡ 1 (mod 8); from which it follows (just write

p2k = 8λ + 1; and square both sides) that

p4k ≡ (mod 16) . (9)

By Fermat’s Little Theorem, we also have (since pk 6= 5),

p4k ≡ 1 (mod 5) . (10)

From (9) and (10) it follows that

p4k − 1 ≡ 0 (mod 16 · 5) ;

p4k − 1 = 80t, for some positive integer t. (11)

p4 −1

Thus, pkk = p80t

k ; which implies

8 9

< p80t

k ≡ 1 (mod 2) , p80t k ≡ 1 (mod 3) , =

and (by Fermat’s Little Theorem since 80 is divisible by 16) (12)

: ;

p80t

k ≡ 1 (mod 17) .

90

k modulo 25. Since pk is not divisible by 5; we have

pk = 5 · q + r; where q is a positive integer and r = 1, 2, 3, or 4. Consider

p20 20

k = (5q + r) . Since both 20 and 5q are divisible by 5; it is clear that in the

binomial expansion (5q + r)20 every term, except for the last one; is divisible by

25; thus p20

k = (5q + r)

20

≡ r 20 mod 25

20

When r = 1, r ≡ 1 (mod 25)

When r = 2, r 20 ≡ 220 ≡ (26 )3 · 22 ≡ (64)3 · 22 ≡ (−11)3 · 22 ≡

(−11)2 · (−11) · 22 ≡ (121)(−11)(22 ) ≡ (−4)(−11) · 4 ≡ (−16)(−11) ≡

9(−11) ≡ −99 ≡ 1 (mod 25).

When r = 3; p20 ≡ 320 ≡ (34 )5 ≡ (81)5 ≡ (−6)5 ≡ (−6)2 · (−6)2 ·

(−6) ≡ (36)(36) · (−6) ≡ (11)(11) · (−6) ≡ (121)(−6) ≡ (−4)(−6) ≡

24 ≡ −1 (mod 25).

When r = 4; r 20 ≡ (44 )5 ≡ (256)5 ≡ 15 ≡ 1 (mod 25). We see that

in all cases; r 20 ≡ ±1 (mod 25). Therefore p20k = (5q + r)

20

≡ r 20 ≡ ±1

(mod 25). And so,

p80

k ≡ (±1)4 ≡ 1 (mod 25) (13)

p80t

k ≡ 1 (mod 25). From (13) and (12), it is clear that for k 6= 2, 3, 7

p4 −1

p80t

k = p kk ≡ 1 mod 2 · 3 · 52 · 17 (14)

≡ 1 (mod 2550)

P2550 p4 −1

Thus, in the sum k=2 pkk ; every term with k 6= 2, 3, 7; is congruent to 1

modulo 2550 by (14). There are (2550 − 2) + 1 − 3 = 2550 − 4 such terms.

We have

X

2550

p4 −1 p4 −1 p4 −1 p4 −1

p kk ≡ p2 2 + p3 2 + p7 7 + (2550 − 4) · 1

k=2

Ǳ

0 (mod 2550)

z }| {

≡ 1701 + 1225 + 901 + 2550 −4 · 1 by (6), (7), and (8)

7. A triangle has perimeter 2s, inradius r, and the distance from its incenter to

the vertices are sa , sb and sc . Prove that

3 r r r s2

+ + + ≤ .

4 sa sb sc 12r 2

91

Solved by Arkady Alt, San Jose, CA, USA; George Apostolopoulos, Messolonghi,

Greece; and Titu Zvonaru, Cománeşti, Romania. We give the write-up by Zvonaru.

r

We have sa = sin A

, sb = sinr B , sc = r

sin C

, and we have to prove

2 2 2

3 A B C s2

that + sin

4 2

+ sin 2

+ sin 2

≤ 12r 2

. The last inequality follows by known

inequalities

A B C 3

sin + sin + sin ≤ (item 2.9 in [1])

2 2 2 2

and 27r 2 ≤ s2 (Item 5.11 in [1]).

[1] 0. Bottema, Geometric Inequalities, Groningen, 1969.

p

is a perfect square.

Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA. We give

Mane’s version.

2p−1 −1

The only such primes are p = 3 and p = 7. Assume that p

= n2 for

some integer n. Then

p−1

p−1

2p−1 − 1 = 2 2 +1 2 2 −1 = pn2 .

1 p2 · · · p2e

r , where the exponent e and

r

p−1 p−1 p−1 p−1

gcd 2 2 + 1, 2 2 − 1 = 1 implies that one of the terms 2 2 +1 or 2 2 −1

is equal to pe a2 and the other is equal to b2 for some integers a and b. Assume

p−1 p−1

2 2 + 1 = b2 and let b = 2x + 1 so that 2 2 + 1 = 4x2 + 4x + 1. Therefore

p−1

2 2 = 4x(x + 1). The product x(x + 1) is a power of 2 only if x = 1, and so

2 2 = 23 . Therefore, p−1

p−1 p−1

2

= 3 or p = 7. If 2 2 − 1 = b2 = 4x2 + 4x + 1,

p−1

then 2 2 = 2(2x2 + 2x + 1). Since 2x2 + 2x + 1 is odd, it follows that it can

p−1

be a power of 2 only if 2x2 + 2x + 1 = 1 when x = 0. Therefore 2 2 − 1 = 1

so that p−1

p−1

2

= 1 or p = 3. Accordingly, the only values of p for which 2 p −1 is

a perfect square are p = 3 and p = 7.

Mathematical Olympiad 2006 given at [2010: 84–85].

B and C, in acute triangle ABC, respectively, and let OA , OB and OC be the

incenters of the triangles AB1 C1 , BC1 A1 and CA1 B1 , respectively. Let TA , TB

and TC be the points of tangency of the incircle of the triangle ABC to the sides

BC, CA and AB, respectively. Show TA OC TB OA TC OB is a regular hexagon.

92

and r be the inradius. The triangles AB1 C1

and ABC are similar, with B1

TB

AC1 AB1 B1 C 1

= = = cos A. I

AC AB BC

The points A, OA , I are collinear, and we have OA

r r A TC C1 B

AI = A

, AOA = A

cos A.

sin 2

sin 2

We deduce that

r r r · 2 sin2 A

2

OA I = A

− A

cos A = A

,

sin 2

sin 2

sin 2

hence OA I = 2r sin A2

. In 4IOA TC we have ITC = r, OA I = 2r sin A

2

and

∠OA ITC = 90◦ − A 2

. By the Law of Cosines, we obtain:

2 A A A

OA TC2 = 2

r + 4r sin 2

− 2r · 2r sin cos 90 − ◦

2 2 2

2 2 2 A 2 2 A 2

= r + 4r sin − 4r sin = r .

2 2

It results that OA TC = r and all sides of TA OC TB OA TC OB are equal to

r.

But this hexagon has no equal angles; From the isosceles triangle OA TC I

we have

A

∠IOA TC = ∠OA ITC = 90◦ − .

2

It follows that ∠TB OA TC = 180◦ −A = B+C, and similarly ∠TC OB TA =

A + C, ∠TA OC TB = A + B.

We also obtain that

A

∠OA TC I = 180◦ − 2 90◦ − = A,

2

hence ∠OA TC OB = A + B, ∠OB TA OC = B + C, ∠OC TB OA = A + C.

for IMO 2007 given at [2010: 85].

1. An airline company is planning to run two-way flights between some of the six

cities A, B, C, D, E and F . Determine the number of ways these flights can be

arranged so that it is possible to travel between any two of these six cities using

only the flights of this company.

93

For positive integers n1 , . . . , np let dn1 ,...,np denote the number of labeled

graphs consisting of p disjoint connected subgraphs with n1 , . . . , np vertices. We

are to determine d6 . The result will be d6 = 26704.

To begin with, note that d1 = d2 = 1, d3 = 4.

The total number of labeled graphs with 4 vertices is

4

64 = 2(2) = d1,1,1,1 + d2,1,1 + d2,2 + d3,1 + d4

4 1 4 2 4

= d41 + d2 d21 + d2 + d3 d1 + d4

2 2 2 3

= 1 + 6 + 3 + 16 + d4 = 26 + d4 ;

hence d4 = 38.

The total number of labeled graphs with 5 vertices is

5

1024 = 2(2) = d1,1,1,1,1 + d2,1,1,1 + d2,2,1 + d3,1,1 + d3,2 + d4,1 + d5

5 1 5 3 5 5 5

=1+ + + d3 + d3 + d4 + d5

2 2 2 2 3 3 4

= 1 + 10 + 15 + 40 + 40 + 190 + d5 = 296 + d5 ;

thus d5 = 728.

Finally, the total number of labeled graphs with 6 vertices is

6

32768 = 2(2) = d1,1,1,1,1,1 + d2,1,1,1,1 + d2,2,1,1 + d2,2,2 + d3,1,1,1 + d3,2,1

+ d3,3 + d4,1,1 + d4,2 + d5,1 + d6

6 1 6 4 1 6 4 6 6 3

=1+ + + + d3 + d3

2 2 2 2 6 2 2 3 3 2

1 6 2 6 6 6

+ d3 + d4 + d4 + d5 + d6

2 3 4 4 5

= 6064 + d6 ;

consequently d6 = 26704.

Remark. The problem is well-known. A different approach is given in Herbert

S. Wilf’s book Generating Functionology, Second edition, Academic Press, 1994,

page 87, formula (3.10.2).

1 1 1 1

+ + ≥ .

ab + 2c2 + 2c bc + 2a2 + 2a ca + 2b2 + 2b ab + bc + ca

Solved by Arkady Alt, San Jose, CA, USA; José Luis Dı́az-Barrero, Universitat

Politècnica de Catalunya, Barcelona, Spain; and Michel Bataille, Rouen, France.

We give the version of Dı́az-Barrero.

94

1 1 1 ab + bc + ca

+ + ≥

ab + 2c2 + 2c bc + 2a2 + 2a ca + 2b2 + 2b (ab + bc + ca)2

1 ab

≥

ab + 2c2 + 2c (ab + bc + ca)2

b2 c2 + c2 a2 √

≥ a2 b2 c4 = abc2

2

which holds on account of AM-GM inequality. Likewise,

1 bc

≥

bc + 2a2 + 2a (ab + bc + ca)2

and

1 ca

2

≥ .

ca + 2b + 2b (ab + bc + ca)2

Adding the preceding three inequalities the statement follows. Equality holds

when a = b = c = 1/3, and we are done.

Now we turn to solutions to the Estonian Team Selection Contest 2007 given

at [2010: 149].

2. Let D be the foot of the altitude of triangle ABC drawn from vertex A.

Let E, F be the points symmetric to D with respect to the lines AB, AC,

respectively. Let triangles BDE, CDF have inradii r1 , r2 and circumradii R1 ,

R2 , respectively. If SK denotes the area of figure K, prove that

95

Solved by Oliver Geupel, Brühl, NRW, Germany; and Titu Zvonaru, Cománeşti,

Romania. We give Geupel’s solution.

Without loss of generality let AD = 1. Let denote ϕ1 = ∠BAD and

ϕ2 = ∠CAD, where ϕ1 , ϕ2 ∈ [0, π/2). It holds BD = BE = tan ϕ1 and

1

DE = 2 sin ϕ1 . Hence SABD = tan ϕ1 and

2

DE · BD · BE sin ϕ1 tan2 ϕ1 1 1

r1 R1 = = = −1 .

2(DE + BD + BE) 2(sin ϕ1 + tan ϕ1 ) 2 cos ϕ1

1 1 1

Similarly, SACD = tan ϕ2 and r2 R2 = − 1 . We therefore have

2 2 cos ϕ2

1 1 1

to prove that |tan ϕ1 − tan ϕ2 | ≥ − . Since tan x and are

cos ϕ1 cos ϕ2 cos x

1 1

both increasing for 0 < x ≤ π/2, the terms tan ϕ1 −tan ϕ2 and −

cos ϕ1 cos ϕ2

have the same sign.

It therefore suffices to prove

1 1

tan ϕ1 − tan ϕ2 ≥ − (1)

cos ϕ1 cos ϕ2

1

under the hypothesis ϕ1 ≥ ϕ2 . But the function f (x) = − tan x is

cos x

0 sin x − 1

decreasing for 0 ≤ x < π/2 as can be seen from the derivative f (x) = .

cos2 x

This implies (1) and the proof is complete.

b−1

is a prime power for

some positive integer b then n is prime.

Assume that,

8 bn −1 9

> = pk ; bn − 1 = pk · (b − 1) >

< b−1 =

where p is a prime number, (1)

>

: n, b, k positive integers such >

;

that n ≥ 2 and b ≥ 2

First, we will treat the case in which p equals 2. This is done in Case 1. In Case

2, p ≥ 3; the proof splits into three subcases.

Case 1. p = 2.

We have, by (1),

¨ «

bn −1

b−1

= 2k ; or equivalently

n−1 n−2 (2)

b +b + · · · + b + 1 = 2k .

It follows from (2) that b must be odd and n even; since there are n terms, with

each term being an odd number; in the sum bn−1 + bn−2 + · · · + b + 1. Thus

§ ª

n = 2l

(3)

where l is a positive integer.

96

their greatest common divisor is 2:

(bl − 1, bl + 1) = 2. (5)

If l ≥ 2; then

(bl−1 + bl−2 + · + b + 1) · (bl + 1) = 2k , (7)

Each of the factors on the lefthand side of (7) is greater than 2. Thus (7)

implies that § ª

bl−1 + bl−2 + · · · + b + 1 = 2k1

(8)

and bl + 1 = 2k2 ; with k1 , k2 positive.

Also,

bl − 1 and bl + 1, contrary to (5).

Case 2. p is an odd prime, p ≥ 3.

Recall from number theory that if p is an odd prime and a an integer not

divisible by p, then the order of a modulo p is the least positive integer k such

that ak ≡ 1 (mod p). When a ≡ 1 (mod p); then obviously the order of a

modulo p equals 1. Otherwise the order of a is ≥ 2. The following Lemma is

well-known in number theory and easily provable by using the division algorithm.

Lemma 1. Let p be an odd prime; and b a positive integer not divisible by p, such

that b 6≡ 1 (mod p). If m is a positive integer, such that bm ≡ 1 (mod p),

then m is divisible by d, d | m, where d is the order of b modulo p; d ≥ 2.

Since by Fermat’s (Little) Theorem, bp−1 ≡ 1 (mod p). We have the

following corollary of Lemma 2.

Lemma 2. Let p be an odd prime; b ∈ Z+ , b 6≡ 0, 1 (mod p). Then the order

of b modulo p is a divisor d ≥ 2 of p − 1.

Back to the problem.

Subcase 2a. Assume that b ≡ 1 (mod p).

97

regardless of whether n is a prime or not.

From b ≡ 1 (mod p) and (1) we have

n

bn−1 + bn−2 + · · · + b + 1 = pk = bb−1 −1

(10)

and b = p · t + 1, for some positive integer t.

we have,

1 + 1 + · · · + 1 ≡ 0 (mod p) ;

| {z }

terms

n

Clearly, since n ≥ 3 and b = p · t + 1, we have

bn − 1 = pk · (b − 1) ⇒ (p · t + 1)n − 1 = pk · (pt).

0

z }| {

n n n−1 n

(pt) + (pt) + ··· + (pt) + 1 − 1 = pk+1 · t. (12)

1 n−1

Let pf be the highest power of p dividing n (see (11)); and pe be the highest power

of p dividing t. Then n = n1 · pf , t = t1 · pe , f ≥ 1; and e ≥ 0 and n1 · t1 6≡ 0

(mod p). The highest power p dividing the left hand side of (12); is the highest

n

power of p dividing the term n−1 p · t = n · p · t = n1 · t1 · pf +e+1 . The right

hand side is pk+1 · t = pk+1+e . Thus, we must have 1 + f + e = k + 1 + e;

and so f = k; n = n1 · pk . If we look at the term (pt)n (left hand side of (12);

k k

(pt)n = pn ·tn = pn1 ·p ·tn1 ·p > pk+1 ·t, since t ≥ 1 and n1 pk ≥ pk > k+1,

in view of p ≥ 3 and k ≥ 3.

We have a contradiction to (12).

In subcases 2b and 2c below; we argue by contradiction. We assume that

b 6≡ 1 (mod p) and n to be a composite number ≥ 4, and we show that this

leads to a contradiction. Observe that a composite number is either a prime power

with exponent at least 2; or otherwise it has two distinct prime bases in its prime

factorization.

Subcase 2b. Assume that b 6≡ 1 (mod p); and n ≥ 4 has at least two

prime bases in its prime factorization into prime powers.

This then implies that we can write,

8 9

< n = n1 · n2 ≥ 4 (actually 6) =

with 1 < n1 , n2 < n; where n1 , n2 (13)

: ;

are relatively prime positive integers; (n1 , n2 ) = 1.

98

positive divisors. We have,

bn1 n2 − 1 = pk · (b − 1);

(b − 1) · (bn2 −1 + · · · + b + 1)[(bn1 )n2 −1 + · · · + bn1 + 1] = pk · (b − 1);

(bn1 −1 + b + 1) · [(bn1 )n2 + · · · + bn1 + 1] = pk (14)

Each factor on the left hand side of (14) is greater than 2. Clearly then (14)

implies that each factor must be a power of p, so that,

§ ª

bn1 −1 + · · · + b + 1 = pk1

(15)

k1 a positive integer.

Similarly, by factoring

§ ª

bn2 −1 + · · · + b + 1 = pk2

(16)

for some positive integer k2

§ ª

bn1 − 1 = pk1 · (b − 1) and

(17)

bn2 − 1 = pk2 · (b − 1)

Thus, (17) ⇒ (bn1 ≡ 1 (mod p) and bn2 ≡ 1 (mod p)). By Lemma 1, both

n1 and n2 must be divisible by the order ρ of b modulo p; since b 6≡ 1 (mod p);

we have ρ ≥ 2. And 2 ≤ ρ | n1 and ρ | n2 , a contradiction of (n1 , n2 ) = 1 in

(13).

Subcase 2c. n = q j , j ≥ 2, where q is a prime number.

2

Since j ≥ 2, bn − 1 is divisible by bq − 1; and hence by bq − 1 as well.

Indeed,

j 2 j−2 2 j−2

bq − 1 = bq ·q − 1 = (bq )q −1

2 j−2

bn − 1 = (bq )q −1

¨ 2

«

(bq − 1), if j = 2

= 2

q 2 q j−2 −1 2 (18)

(bq − 1) · [(b ) + · · · + (bq ) + 1], if j ≥ 3.

of the factor b − 1 from both sides of the last equation; it follows that bq−1 +

bq−2 + · · · + b + 1 must be a power of p. Indeed, since

2

bq − 1 = (bq )q − 1 = (bq − 1)[(bq )q−1 + · · · + bq + 1]

= (b − 1)(bq−1 + · · · + b + 1)[(bq )q−1 + · · · + bq + 1],

99

bq − 1 = pλ · (b − 1) ⇒ bq ≡ 1 (mod p) . (19)

By Lemma 1, (19) implies that the order d of b modulo p must divide q: 2 ≤ d | q.

But q is a prime and therefore it follows that d = q. The order of b modulo p

must equal the prime q. Hence, by Lemma 2 it follows that

p − 1 ≡ 0 (mod q) . (20)

2

Recall from (18) that bq − 1 is a factor of bn − 1.

2

bn − 1 = (bq − 1) · N, N a positive integer and bn − 1 = pk (p − 1)

Thus,

2

(bq − 1) · N = pk · (b − 1);

((bq )q − 1) · N = pk · (b − 1);

(bq − 1) [(bq )q−1 + · · · + bq + 1] = pk · (b − 1);

| {z }

(b−1)(bq−1 +···+b+1)

So

(bq−1 + · · · + b + 1)[(bq )q−1 + · · · bq + 1] = pk . (21)

So each factor (on the left hand side of (21) must be a power of p: In particular

(bq )q−1 + · · · + bq + 1 = pw , w ≥ 1. (22)

By (19) and (22) we deduce that,

1 + 1 + · · · + 1 ≡ 0 (mod p) ⇒ q ≡ 0 (mod p) ; (23)

| {z }

q

and since p and q are both primes; (23) implies p = q. But (20) then implies

p − 1 ≡ 0 (mod p), an impossibility. 2

D F C

are chosen in the interior of sides BC and

CD, respectively. The line drawn from F

perpendicular to AE passes through the in-

tersection point G of AE and BD. A point K

K is chosen on F G such that |AK| = |EF |.

E

Find ∠EKF . G

A B

Solved by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; Michel Bataille,

Rouen, France; Geoffrey A. Kandall, Hamden, CT, USA; Bruce Shawyer, Memo-

rial University of Newfoundland, St. John’s, NL; Konstantine Zelator, University

of Pittsburgh, Pittsburgh, PA, USA; and Titu Zvonaru, Cománeşti, Romania. We

give the solution of Amengual Covas.

100

D F C

45◦ 45◦

E

G

A B

circumcircle,

∠GF A = ∠GDA = ∠BDA = 45◦

Thus, right-triangle AGF is isosceles with AG = GF .

Since we also have AK = EF , right triangles AGK and F GE are

congruent (side-angle-side) with KG = GE.

Hence 4KGE is isosceles right-angled. Thus, ∠EKF = 180◦ −∠GKE =

180 − 45◦ = 135◦ .

◦

f (x + f (y)) = y + f (x + 1) .

The functions x 7→ x + 1 and x 7→ −x + 1 are solutions (readily checked).

We show that there are no other solutions.

Let f : R → R be a continuous function satisfying the given functional

equation that we will denote by (E). Taking x = −1 in (E) yields f (f (y)−1) =

y + f (0) for all y ∈ R. An immediate consequence is f (y) = f (y 0 ) =⇒ y = y 0

and f is injective.

Letting y → 0 and using the continuity of f gives f (x+f (0)) = f (x+1),

hence f (0) = 1 and so f (f (y) − 1) = y + 1 for all y ∈ R. Using (E), it follows

that

equation g(x + y) = g(x) + g(y). It is known that g must be a linear function

x 7→ ax and so f must be x 7→ ax + 1. Returning to (E), we must have

a(x + ay + 1) + 1 = y + a(x + 1) + 1 for all x, y that is, a2 y = y for all y.

Hence a = 1 or a = −1 and the conclusion follows.

101

Mathematical Olympiad 2007, 10th grade, given at [2010: 150–151].

an , let m = min{a0 , a0 +a1 , . . . , a0 +a1 +· · ·+an }. Prove that P (x) ≥ mxn

for all x ≥ 1.

By the definition of m, it holds

a0 − m ≥ 0, a0 + a1 − m ≥ 0, ..., a0 + a1 + . . . + an − m ≥ 0.

Hence, we have for x ≥ 1

P (x) = a0 xn + a1 xn−1 + · · · + an

= a0 (xn − xn−1 ) + (a0 + a1 )(xn−1 − xn−2 )

+ (a0 + a1 + a2 )(xn−2 − xn−3 )

+ · · · + (a0 + a1 + · · · + an−1 )(x − 1) + (a0 + a1 + · · · + an )

= (a0 − m)(xn − xn−1 ) + (a0 + a1 − m)(xn−1 − xn−2 ) + · · ·

+ (a0 + a1 + · · · + an − m)(x − 1) + (a0 + a1 + · · · + an − m)

+ mxn

≥ mxn .

chosen on the smaller arc BC of the circumcircle, such that B1 K and AC are

perpendicular. Point L is chosen on line AC such that BL and AK are also

perpendicular. Line BB1 meets the smaller arc AC at point T . Prove that

points K, L, T are collinear.

Fixing the typo in the problem, we

substitute the hypothesis B1 K ⊥ AC

by the condition B1 K ⊥ BC. Let D A

be the point of intersection of the lines

T

BC and B1 K, and let the lines AK

and BL intersect at the point E. Since

the inscribed angles ∠CBK = ∠DBK

and ∠CAK = ∠LAE have equal size, B1

the right triangles DBK and EAL are

similar. E L

Hence, D

B C

∠BKB1 = ∠BKD = ∠ALE = ∠BLB1 ,

K

that is, the points B, B1 , K, and L are

concyclic.

102

Thus,

∠BB1 D = ∠BB1 K = ∠BLK = ∠KLE,

which implies that the right triangles BB1 D and KLE are similar. Therefore,

6. (S. Berlov) Two circles ω1 and ω2 intersect at points A and B. Let P Q and

RS be the segments of common tangents to these circles (points P and R lie on

ω1 , while points Q and S lie on ω2 ). Ray RB intersects ω2 again at point W .

If RBkP Q, find the ratio RP/BW .

sity of Pittsburgh, Pittsburgh, PA, USA; and Titu Zvonaru, Cománeşti, Romania.

We give Kandall’s solution.

P

A ω2

ω1

O1 O2

y

W

B y

N

x

x

R M

at M (N ). Since ∠O1 P Q and ∠O2 QP are right angles and RW kP Q, it follows

that ∠O1 M N and ∠O2 N M are right angles (so P QN M is a rectangle) and

M (N ) is the midpoint of RB (BW ). Let RB = 2x, BW = 2y.

Then RS = P Q = M N = x + y. Since RS 2 = RW · RB, we have

(x + y)2 = 2(x + y) · 2x. Dividing by x + y, we obtain x + y = 4x, that is

y = 3x.

RB x 1

Therefore, = = .

BW y 3

103

BOOK REVIEWS

Amar Sodhi

The Calculus of Friendship

by Steven Strogatz

Princeton University Press, 2009

ISBN 978-0-691-13493-2 hardcover, 166 + xii pp. US$19.95

Reviewed by Georg Gunther, Sir Wilfred Grenfell College (MUN), Corner

Brook, NL

New books and new cars have this in common: one cannot wait to open

them. Their lure is irresistible. If chemists could ever be bothered to synthesize

the smell of a new car, undoubtedly L’eau d’Auto would soon become the fragrance

of choice.

Cracking open the cover of a new book carries a similar appeal: there is

a promise of things to come, an anticipation of words to be read, a sense of

excitement. And the magic happens when the book delivers on this promise,

when the words capture and enthrall.

The Calculus of Friendship is such a book. It is a remarkable story, based

upon a thirty-year correspondence between the author and his former high-school

mathematics teacher. The letters, many of which are reproduced in the book, are

all about calculus. It is the passion and love for this discipline that connects these

two men as they move through their respective lives, and it is calculus, with its

unparalleled ability to provide deep and profound metaphors, that provides the

connecting context. “Some books are to be tasted ”, wrote Francis Bacon, “others

to be swallowed, and some few to be chewed and digested ”. This book certainly

falls into Bacon’s third category.

The author, Steven Strogatz, is the Jacob Gould Schurman Professor of

Applied Mathematics at Cornell University. He has contributed widely to the

study of synchronization in dynamical systems. In addition to writing numerous

mathematical papers and books, he is the author of the best selling Sync: the

emerging science of spontaneous order (2003).

The book being reviewed here is written with a deceptive simplicity and

grace of style that draws the reader effortlessly from page to page. It provides a rare

glimpse into the mind and heart of a top-notch mathematician. Always fascinating,

at times deeply profound, it escapes being maudlin by the sheer simplicity of its

prose, the elegance of its mathematics and the, at times, brutal self-appraisal of

the author.

There is a great deal of fascinating and exciting mathematics in this book.

Interesting problems are tossed back and forth between former teacher and former

student in a wonderful point and counterpoint of question and answer, leading to

yet more questions, more avenues of investigation. Reading these letters, one is

struck with the exuberance of the intellectual exercises being posed, with the thrill

of the exploration, and the sheer fun of doing mathematics.

104

chronology of two lives, from 1974 to the present; each explores one of the

mathematical topics raised in the course of the correspondence between these two

men. In addition to including the actual text of the original letters, the author

provides some of the mathematical background, as well as placing these discussions

within a personal context.

The mathematical content of this book is far from trivial. It covers diverse

topics, from chase problems, discussion of irrationality, subtleties of infinite

series, explorations of randomness, all the way to chaos theory and questions

about bifurcation. This book is well worth reading for its mathematical content

alone.

However, this book is so much more than simply a book about calculus. It

is a testimonial to the bonds of friendship and to the complex and ever evolving

connections between teacher and student. In mathematics, fixed-point theorems

play a central role. The major focus of Strogatz’s mathematical work is the science

of synchrony: how spontaneous order can arise within inherently chaotic

systems. In The Calculus of Friendship, the author explores one more facet of this

theme: “like calculus itself, this book is an exploration of change. It’s about the

transformation that takes place in a student’s heart, as he and his teacher reverse

roles, as they age, as they are buffeted by life itself. Through all these changes,

they are bound together by a love of calculus. For them it is more than a science.

It is a game they love playing together – so often the basis of friendship between

men – a constant while all around them is flux ”.

Towards the end of this book, Steven Strogatz acknowledges that his former

mentor taught him “something profoundly mathematical, about how to live ”; the

lesson is to “balance the inevitable against the unforeseeable, the two sides of change

in this world. The orderly and the chaotic. The changes that calculus can tame,

and the ones it cannot. He confronts them all, and not, like Zeno, with his mind

alone but also with his heart ”.

“The Calculus of Friendship ” is recommended for all. Practicing

mathematicians will find much that might be new to them, as well as meeting

up with many old friends. Students will get glimpses into numerous directions

that will compel their interest. Non-mathematicians can skip the hard stuff, the

mathematical equations, and find a different set of valuable truths in the vastly

more complicated rules that lie at the heart of human relationships.

105

Crux Chronology

J. Chris Fisher

1 Highlights

March 1975, Eureka.

Crux Mathematicorum with Mathematical Mayhem began life as the journal

Eureka. Early in 1975, six members of the Carleton-Ottawa Mathematics

Association (COMA) met privately and decided to launch Eureka

especially interesting problems and solutions, among the members of

the mathematical community in the Ottawa region, students and

teachers alike. [1975 : 1]

Three of the six were from Algonquin College: Léo Sauvé, who served as the first

editor (until 1986); Fred G.B. Maskell, COMA’s secretary-treasurer who became

the first managing editor (through 1984); and H.G. Dworschak. The other three

were Viktors Linis of the University of Ottawa, R. Duff Butterill of the Ottawa

Board of Education, and Richard J. Semple of Carleton University.

While Dworschak and Linis provided some support and numerous problems,

it was Léo and Fred who provided the the energy and dedication required to

turn their modest local venture into a journal that within two years developed

an international following, a following that included some of the world’s finest

mathematicians. In the words of his friend and colleague Kenneth S. Williams

[1987 : 240-242], “Léo’s dedication and hard work, his broad knowledge and love

of mathematics, his careful eye for detail, all enabled Crux to grow from a four-page

problem sheet to the international mathematical problem-solving journal that it is

today.” Many of Crux’s faithful contributors became Léo’s friends; their tributes

to him can be found in the issue dedicated to him [1986 : 163-168].

March 1978, Crux Mathematicorum.

Beginning with volume 4 number 3, the name of the journal changed to Crux

Mathematicorum. After 32 issues had been published under the name Eureka, it

was discovered that there was a journal Eureka published once a year by the

Cambridge University Mathematical Society. Sauvé chose the new name: it is an

idiomatic Latin phrase meaning a puzzle or problem for mathematicians [1978 :

89-90].

January 1979, the Olympiad Corner.

Murray S. Klamkin initiated the Olympiad Corner [1979 : 12] to “provide,

on a continuing basis, information about mathematical contests taking place in

Canada, the U.S.A., and internationally.” It would also provide “practice sets of

problems on which interested students could test and sharpen their mathematical

skills and thereby possibly qualify to participate in some Olympiad.” Klamkin

served as editor of the first 80 columns, through December, 1986.

106

Frederick G. B. Maskell (1904-1985) steps down as managing editor [1984 :

340], and died a few months later [1985 : 14, 34]. He was replaced by Kenneth S.

Williams.

October 1 1985, the CMS.

For nearly eleven years Crux was published by Algonquin College and

sponsored by the Carleton-Ottawa Mathematics Association. In 1979 The

Canadian Mathematical Olympiad Committee and the Carleton University

Mathematics Department added their support, joined later by the University of

Ottawa Mathematics Department. In March, 1985, the Canadian Mathematical

Society was asked to assume responsibility for its publication; that organization

called for suggestions concerning the future of the journal and began the search

for a new editor. [1985 : 100] On October 1, 1985, Crux became an official publi-

cation of the Canadian Mathematical Society. [1985 : 234, 236] The October 1985

issue was dedicated to Philip Kileen, President of Algonquin College, who strongly

supported Crux during its first eleven years. For all future editors, the

mathematics department of the current editor’s university lent support to the

journal.

February 1986, G.W. (Bill) Sands.

Starting with Volume 12, number 2, Bill Sands (University Calgary) became

the second editor (through 1995) . [1986 : 17, 37] He announced [1986 : 65] that

the September issue would be dedicated to Léo Sauvé.

January 1987 Robert Woodrow.

The Olympiad Corner’s Murray Klamkin was replaced by Robert E. Woodrow

[1986 : 263; 1987 : 2-3, 34], but Klamkin continued making valuable contributions

to Crux until his death in 2004.

June 19, 1987, Léo.

Léo Sauvé died (Dec. 12, 1921 - June 19, 1987). [1987 : 240-242] Not

only was Léo the consummate scholar, but he injected into each page of Crux

a spicy liveliness that will probably never be matched. When a problem failed

to attract solutions to his liking, he would write his own, assuming the guise of

his contributor persona, Gali Salvatore (Salvatore = Sauvé). If there happened

to be a gap he could not fill, he would attach an editorial comment in which he

criticized Salvatore’s proof; see [1984 : 31] where he attacked the solution (his

own!) to problem 783, complaining that the whole argument rested on a formula

that came without a proof “presumably because he (or she: is Gali a man’s name

or a woman’s?) felt it was ‘easy.’ ” Edith Orr (= editor) was another persona who

was able to make comments that an editor could not; her poetry was sometimes so

racy that it would not have been allowed to pass through the mail had the postal

inspectors thought to look closely at a math journal! See her comment on lambs

[1983 : 211-212], which Léo, wearing his editor’s hat, dismissed as “scrofulous.”

January 1988, Colour.

Starting with volume 14, each issue came with a coloured front and back

cover; although it continued to be printed on 8 21 × 11 three-holed paper and

107

January 1989, Kenneth Williams.

Kenneth Williams steps down from his position of managing editor and his

role as technical editor. [1988 : 300-301] His duties were taken over by Graham

P. Wright, the executive director of the Canadian Mathematical Society, together

with members of his staff; Wright served (except for a short period between 1999

and 2000 when the position was filled by Robert Quackenbush) until his retirement

in May, 2009.

January 1991, The Editorial Board.

After five years of performing all the editorial duties with only occasional

help from others, Sands organized the first formal editorial board. Robert Woodrow

was promoted to joint editor, and six others agreed to form the board. The board

members will be listed later.

January 1995, New format and Skoliad.

Starting with volume 21, the appearance was changed to its current 10-inch

format with purple covers. Also, the Skoliad Corner was inaugurated with its

simpler “Pre-Olympiad” problem sets, with Robert Woodrow as its first editor

[1995 : 5]. The name was suggested by Richard Guy, who had searched a map

of the Mount Olympus area, finding the mountain Scollis, and then making a

portmanteau of this with scholar and Olympiad. At about 1/3 the height of Mount

Olympus, Mount Scollis seems the appropriate metaphor for a junior Olympiad.

It was later learned that skolion, an unrelated ancient Greek word, referred to

songs sung by invited guests at banquets in ancient Greece extolling the virtues

of the gods or heroic men! It seems to have come from the word for crooked,

which seems appropriate for a problem section: short, diverse, and slightly twisted

entertainments.

January 1996, Bruce L.R. Shawyer.

Bruce Shawyer (Memorial University of Newfoundland) became the third

editor starting with volume 22 [1995 : 354; 1996 : 1]; he served until December

2002. Colin Bartholomew, also of Memorial University, served as assistant editor

for one year, after which Clayton Halfyard took over that position from 1997

through 2002. With the new editor, the journal went from 10 issues of 36 pages

(360 pages per volume) to 8 issues of 48 pages (384 pages per volume). Although

the January and June issues were dropped, there were 24 extra pages per year,

allowing increased efficiency of printing and a decrease in mailing costs. The

journal went on-line for subscribers later that year [1996 : 289].

The Academy Corner.

Shawyer produced the Academy Corner during his tenure as editor. It dealt

with problem solving at the undergraduate level. [1996 : 28] It ended with column

49 when he stepped down in 2002. [2002 : 480]

February 1997, Crux Mathematicorum with Mathematical Mayhem.

Mathematical Mayhem had been founded in 1988 by Ravi Vakil and Patrick

Surry, two Canadian IMO alumni, as a journal of high-school and college level

108

mathematics written by and for students. [1996 : 337; 1997 : 1-2, 30-31] When

it amalgamated with Crux in volume 23, it brought additional high-school level

material and gained, in return, a wider exposure. It was agreed that the new

journal would continue the volume numbering of Crux as well as maintain its

general external appearance. The number of pages per issue jumped from 48 to

64. The then current Mayhem editor Naoki Sato and assistant editor Cyrus Hsai

continued in their positions for four years, through December 2000. Other editors

and staff are listed below.

The statement of all problems would henceforth appear in both English and

French. Jean-Marc Terrier has been translating from the start; other translators,

serving for various periods, have been Hidemitsu Sayeki, Martin Goldstein, and

Rolland Gaudet.

James Edward Totten (University College of the Cariboo, renamed

Thompson Rivers University in 2005) became the fourth editor starting with

volume 29. [2002 : 287-288; 2003 : 1] His plan was to step down in June of

2008, after serving the final six months as co-editor, but he died on March 9, 2008

in his 61st year (born August 9, 1947). His Assistant editor was Bruce Crofoot.

The Mayhem portion of the journal became open to the public on the

internet starting with volume 32. Currently, all issues of Crux with Mayhem

become free to the public after five years, while the Mayhem portion is always

available for free.

Vazz Linek (University of Winnipeg) became co-editor of the journal, then

became the journal’s fifth editor starting in July. [2007 : 449; 2008 : 193-194] His

assistant editor is Jeff Hooper.

September 2009.

Johan Rudnick became the executive director of the Canadian Mathematical

Society and, thereby, the new managing editor of Crux.

2 Crux Editors

Editors-in-chief

Léo Sauvé March 1975 through January 1986

G.W. Sands February 1986 through December 1990

G.W. Sands and Robert Woodrow January 1991 through December 1995

Bruce L.R. Shawyer January 1996 through December 2002

James Totten January 2003 through May 2008

Václav (Vazz) Linek and James Totten January 2008 through May 2008

Vazz Linek June 2008 through March 2011

Shawn Godin April 2011 to present

109

Managing Editors

Frederick G.B. Maskell March 1975 through November 1984

Kenneth S. Williams December 1984 through December 1988

Graham P. Wright January 1989 through September 1999

Robert Quackenbush October 1999 through December 2000

Graham P. Wright January 2001 through August 2009

Johan Rudnick September 2009 to present

Olympiad Corner

Murray S. Klamkin 1979 through 1986

Robert E. Woodrow 1987 through 2010

J. Chris Fisher 1991 to present

Richard Guy 1991 through 2003

Denis Hanson (articles) 1991 through September 1999

Andy Liu (book reviews) 1991 through 1998

Richard Nowakowski 1991 through 1994

Edward T.H. Wang 1993 through 2010

Rod De Peiza 1994

Jim Totten 1994 through 2002 (when he became editor-in-chief)

Catherine Baker 1995 through 1999

Loki Jörgenson 1998 through 2002∗

Alan Law (book reviews) 1999 through 2001

Bruce Gilligan (articles) 2000 through 2007

Iliya Bluskov 2000 through 2009

John Grant McLoughlin 2002-2009 (book review editor until 2008)

Richard (Rick) Brewster 2003 through 2005

Bruce Shawyer (editor-at-large) 2003 to present

Maria Torres 2006 through 2009

James Currie (articles) 2008 through 2010

Amar Sodhi (book reviews) 2009 to present

Nicolae Strungaru 2009 to present

Jonatan Aronsson 2010

Dzung Minh Ha 2010

Robert Craigen 2011 to present

Robert Dawson (articles) 2011 to present

Chris Grandison 2011 to present

Cosmin Pohoata 2011 to present

∗

Although he was responsible for the on-line edition starting in 1996, he was a board member

for only five years.

Skoliad Corner

Robert Woodrow 1995 through May 2001

Shawn Godin September 2001 through 2004

Robert Bilinski 2005 through 2008

Václav Linek February 2009

Lily Yen and Mogens Hansen March 2009 to present

110

3 Mathematical Mayhem

Mayhem editors

Before joining Crux: Patrick Surry, Ravi Vakil, Philip Jong, Jeff Higham, J.P.

Grossman, Andre Chang, and Naoki Sato.

After the amalgamation:

Naoki Sato 1997 through 2000

Shawn Godin 2001 through 2006

Jeff Hooper 2007

Ian VanderBurgh 2008 through 2010

Mayhem assistant editors

Cyrus Hsai 1997 through 2000

Chris Cappadocia 2001 through 2002

John Grant McLoughlin 2003 through 2005

Jeff Hooper 2006

Ian VanderBurgh 2007

Lynn Miller 2011 to present

Mayhem editorial staff (various terms)

Richard Hoshino, Wai Ling Yee, Adrian Chan, Jimmy Chui, David Savitt, Donny

Cheung, Paul Ottaway, Larry Rice, Dan MacKinnon, Ron Lancaster, Eric Robert,

Monika Khbeis, Mark Bredin.

Special Issues

3:4 April 1977, The Gauss bicentennial Issue [1976 : 131, 161]

3:10 December 1977, Special Morley Issue; see also [1978 : 33-34, 132; 1978 :

304-305]

12:7 September 1986, Issue dedicated to Léo Sauvé [1986 : 65]

27:2 March 2001, Murray Klamkin 80th Birthday Issue [2001 : 65-85]

31:5 September 2005, Issue dedicated to Murray S. Klamkin [2004 : 361]

35:5 September 2009, Issue dedicated to James Edward Totten [2008 : 193]

Encyclopedic Articles

Léo Sauvé, The Celebrated Butterfly Problem. [1976 : 2-5]

Léo Sauvé, The Steiner-Lehmus Theorem, and

Charles W. Trigg, A Bibliography of the Steiner-Lehmus Theorem. [1976 : 19-24,

191-193]

Honours and Awards

Murray Klamkin (University of Alberta), M.A.A. Distinguished Service Award

[1988 : 33]; David Hilbert International Award [1992 : 224]

Marcin E. Kuczma (University of Warsaw), David Hilbert International Award

[1992 : 224]

111

Andy Liu (University of Alberta), David Hilbert International Award [1996 : 201];

Outstanding University Professor [1999 : 65-66]

Bruce Shawyer (Memorial University of Newfoundland), Adrien Pouliot Award

[1998 : 19]

Francisco Bellot Rosado (Institute Emilio Ferrari), Erdös Prize [2000 : 320]

Contributor Profiles

R. Robinson Rowe [1977 : 92,184,248; 1978 : 6,63,129,189]

Kestraju Satyanarayana [1981 : 294]

Jack Garfunkel [1990 : 318]

Leon Bankoff [1995 : 292]

Jordi Dou [2002 : 56; 2006 : 65]

K.R.S. Sastry [2006 : 2]

Toshio Seimiya [2000 : 114; 2006 : 129]

Christopher J. Bradley [2006 : 257]

D.J. Smeenk [2006 : 353]

Michel Bataille [2007 : 1]

Richard K. Guy [2007 : 65]

Walther Janous [2007 : 385]

Peter Y. Woo [2008 : 1]

Arkady Alt [2010 : 65]

John G. Heuver [2010 : 193]

Death Notices

Richard J. Sempel, Carleton University (1930-1977) [1977 : 188]

R. Robinson Rowe (1896-1978) [1978: 152]

Herman Nyon, Paramaribo, Surinam (? - 1982) [1982 : 268]

Viktors Linis, University of Ottawa (1916-1983) [1983 : 192]

Kestraju Satyanarayana (1897-1985) [1985 : 268]

Geoffrey James Butler, University of Alberta (1944-1986) [1986 : 203]

Samuel L. Greitzer, Rutgers University (1905-1988) [1988 : 161-162]

Charles Trigg, San Diego, CA (1898-1989) [1989 : 224]

Jakob T. Groenman, Arnhem, The Netherlands [1989 : 96]

W.J. Blundon, Memorial University of Newfoundland (1916-1990) [1990 : 160]

Jack Garfunkel (1910-1990) [1991 : 64]

Oene Bottema, Delft, The Netherlands (1901-1992) [1993 : 32]

Peter Joseph O’Halloran, University of Canberra (1931-1994) [1994 : 248-249]

Pál Erdös (1913-1996) [1996 : 339]

Leon Bankoff, Los Angeles, CA (1908-1997) [1997 : 145]

Jessie Lei, University of Toronto (1980-2000) [2000 : 1-2]

Herta Freitag, Roanoke, VA (1908-2000) [2000 : 535]

H.S.M. Coxeter, University of Toronto (1907-2003) [2003 : 232]

Murray S. Klamkin, University of Alberta (1921-2004) [2004 : 361]

Robert Barrington Leigh, University of Toronto (1986-2006) [2006 : 453]

Jordi Dou (1911-2007) [2007 : 457]

James Edward Totten, Thompson Rivers University (1947-2008) [2008 : 194]

112

PROBLEMS

Toutes solutions aux problèmes dans ce numéro doivent nous parvenir au plus tard

le 1er septembre 2011. Une étoile (?) après le numéro indique que le problème a été

soumis sans solution.

Chaque problème sera publié dans les deux langues officielles du Canada

(anglais et français). Dans les numéros 1, 3, 5 et 7, l’anglais précédera le français,

et dans les numéros 2, 4, 6 et 8, le français précédera l’anglais. Dans la section des

solutions, le problème sera publié dans la langue de la principale solution présentée.

La rédaction souhaite remercier Jean-Marc Terrier, de l’Université de Montréal,

d’avoir traduit les problèmes.

3613. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie.

Résoudre le système d’équations

= 7, = 5 , et = 12 .

x−1 y−1 z−1

où x, y et z sont des entiers positifs.

À partir des décimales consécutives de 17 on obtient l’ensemble des points

A(1, 4), B(4, 2), C(2, 8),. . . dans le plan. Montrer que tous ces points

appartiennent à une même ellipse. Calculer l’aire de cette ellipse.

3615. Proposé par Pham Van Thuan, Université de Science de Hanoı̈, Hanoı̈,

Vietnam.

Montrer que si x, y, z ≥ 0 et x + y + z = 1, alors

xy yz zx 1

√ +√ +√ ≤ .

z + xy x + yz y + zx 2

Calculer

2k arctan(nk ) arctan(nk + 1)

L = lim n − ,

n→∞ nk nk + 1

où k ∈ R.

Soit r un nombre rationnel positif. Montrer que si r r est rationnel, alors r

est un entier.

113

Soit α > 3 un nombre réel. Trouver la valeur de

X

∞ X

∞

n

.

n=1 m=1

(n + m)α

3619. Proposé par Pham Kim Hung, étudiant, Université de Stanford, Palo

Alto, CA, É-U.

Soit a, b et c trois nombres réels non négatifs tels que a + b + c = 3. Montrer

que

Soit P un point intérieur du tétraèdre ABCD et désignons par A0 ,B 0 , C 0

et D 0 les points d’intersection des droites AP , BP , CP et DP avec les faces

correspondantes opposées. On a alors

AP BP CP DP AP BP CP DP

= 3+2 + + +

P A0 P B 0 P C 0 P D 0 P A0 P B0 P C0 P D0

AP BP AP CP AP DP

+ + +

P A0 P B 0 P A0 P C 0 P A0 P D 0

BP CP BP DP CP DP

+ 0 0

+ 0 0

+ .

PB PC PB PD P C 0 P D0

Soit a, b et c trois nombres réels non négatifs avec a + b + c = 1. Montrer

que

27 4 4 4 3

[(a−b)2 +(b−c)2 +(c−a)2 ]+ + + ≤ .

128 1+a 1+b 1+c ab + bc + ca

On donne un quadrilatère ABCD.

(a) Trouver des conditions nécessaires et suffisantes sur les côtés et les angles

de ABCD pour qu’il existe un point intérieur P tel que deux droites

perpendiculaires issues de P divisent le quadrilatère ABCD en quatre

quadrilatères d’aire égale.

(b) Déterminer P .

114

Soit z1 , z2 , z3 , z4 quatre nombres complexes distincts mais de même

module, α = |(z3 − z2 )(z3 − z4 )|, β = |(z1 − z2 )(z1 − z4 )| et

α(z1 − z4 ) + β(z3 − z4 )

u() = .

α(z1 − z2 ) + β(z3 − z2 )

Calculer la somme

X

∞

(−1)n−1 1 1 (−1)n+1

1− + − ··· + .

n=1

n 2 3 n

3625. Proposé par Pham Van Thuan, Université de Science de Hanoı̈, Hanoı̈,

Vietnam.

Soit a, b et c trois nombres réels positifs. Montrer que

r Ê r Ê

a b c abc

+ + ≤ 2 1+ .

a+b b+c c+a (a + b)(b + c)(c + a)

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

Buzău, Romania.

Solve the system of equations

= 7, = 5 , and = 12 .

x−1 y−1 z−1

Where x, y, and z are positive integers.

Taking consecutive decimal digits of 17 the set of points A(1, 4), B(4, 2),

C(2, 8),. . . in the plane is obtained. Prove that all these points belong to the

same ellipse. Compute the area of the ellipse.

Vietnam.

Prove that if x, y, z ≥ 0 and x + y + z = 1, then

xy yz zx 1

√ +√ +√ ≤ .

z + xy x + yz y + zx 2

115

Compute

2k arctan(nk ) arctan(nk + 1)

L = lim n − ,

n→∞ nk nk + 1

where k ∈ R.

Let r be a positive rational number. Show that if r r is rational, then r is an

integer.

Let α > 3 be a real number. Find the value of

X

∞ X

∞

n

.

n=1 m=1

(n + m)α

3619. Proposed by Hung Pham Kim, student, Stanford University, Palo Alto,

CA, USA.

Let a, b, and c be nonnegative real numbers such that a + b + c = 3. Prove

that

Let P be an interior point in tetrahedron ABCD and let AP , BP , CP ,

and DP meet the corresponding opposite faces in A0 , B 0 , C 0 , and D 0 then

AP BP CP DP AP BP CP DP

= 3+2 + + +

P A0 P B 0 P C 0 P D 0 PA 0 PB 0 PC 0 P D0

AP BP AP CP AP DP

+ 0 0

+ 0 0

+

PA PB PA PC P A0 P D 0

BP CP BP DP CP DP

+ 0 0

+ 0 0

+ .

PB PC PB PD P C 0 P D0

Let a, b, and c be nonnegative real numbers with a + b + c = 1. Prove that

27 4 4 4 3

[(a−b)2 +(b−c)2 +(c−a)2 ]+ + + ≤ .

128 1+a 1+b 1+c ab + bc + ca

116

Let ABCD be a quadrilateral.

(a) Find sufficient and necessary condition on the sides and angles of ABCD,

so that there is an inner point P such that two perpendicular lines through

P divide the quadrilateral ABCD into four quadrilaterals of equal area.

(b) Determine P .

Let z1 , z2 , z3 , z4 be distinct complex numbers with the same modulus,

α = |(z3 − z2 )(z3 − z4 )|, β = |(z1 − z2 )(z1 − z4 )| and

α(z1 − z4 ) + β(z3 − z4 )

u() = .

α(z1 − z2 ) + β(z3 − z2 )

Prove that u(+1) or u(−1) is a real number.

Calculate the sum

X

∞

(−1)n−1 1 1 (−1)n+1

1 − + − ··· + .

n=1

n 2 3 n

Vietnam.

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

r Ê r Ê

a b c abc

+ + ≤ 2 1+ .

a+b b+c c+a (a + b)(b + c)(c + a)

117

SOLUTIONS

No problem is ever permanently closed. The editor is always pleased to con-

sider for publication new solutions or new insights on past problems.

Let m be a positive real number and a, b, c real numbers such that

We set

x = b − a, y = c − b, z = a − c. (1)

Then

x + y + z = 0,

−xy − yz − zx = a(a − b) + b(b − c) + c(c − a) ,

xyz = ab(a − b) + bc(b − c) + ca(c − a) ,

Thus, the range consists of all real numbers n for which w3 − mw − n

has three real roots (if the roots x, y, z are real, then (1) can be solved for the

corresponding real numbers a, b, c).

A cubic equation has three real roots if and only if its discriminant D is not

3 2

−m −n

positive. Here the discriminant is D = + , so D ≤ 0 holds if

3 2

3/2

m

and only if |n| ≤ 2 .

3 3/2 3/2

m m

Therefore, the range is the interval −2 ,2 .

3 3

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ROY BARBARA,

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

Joplin, MO, USA; PRITHWIJIT DE, Homi Bhabha Centre for Science Education, Mumbai,

India; OLIVER GEUPEL, Brühl, NRW, Germany; PAOLO PERFETTI, Dipartimento di

Matematica, Università degli studi di Tor Vergata Roma, Rome, Italy; PETER Y. WOO, Biola

University, La Mirada, CA, USA; and the proposer. One incomplete solution was submitted.

118

3515. [2010 : 107, 110] Proposed by José Luis Dı́az-Barrero and Josep

Rubió-Massegú, Universitat Politècnica de Catalunya, Barcelona, Spain.

Let x, y, and z be positive real numbers. Prove that

{x}2 bxc2 {y}2 byc2 {z}2 bzc2 x2 + y 2 + z 2

+ + + + + ≥ ,

y z z x x y x+y+z

where bac is the greatest integer not exceeding a, and {a} = a − bac.

Herzegovina.

Note first that for all positive reals a, b, and c we have by the Cauchy–Schwarz

Inequality that

{c}2 bcc2 2

+ (a + b) ≥ ({c} + bcc) = c2 ,

a b

{c}2 bcc2 c2 c2

+ ≥ > .

a b a+b a+b+c

Hence,

{x}2 bxc2 {y}2 byc2 {z}2 bzc2

+ + + + +

y z z x x y

2 2 2

x y z x + y2 + z2

2

> + + = .

x+y+z x+y+z x+y+z x+y+z

Also solved by ARKADY ALT, San Jose, CA, USA; GEORGE APOSTOLOPOULOS,

Messolonghi, Greece; ŠEFKET ARSLANAGI Ć, University of Sarajevo, Sarajevo, Bosnia and

Herzegovina; MICHEL BATAILLE, Rouen, France; OLEH FAYNSHTEYN, Leipzig, Germany;

OLIVER GEUPEL, Brühl, NRW, Germany; JOHAN GUNARDI, student, SMPK 4 BPK

PENABUR, Jakarta, Indonesia; PAOLO PERFETTI, Dipartimento di Matematica, Università

degli studi di Tor Vergata Roma, Rome, Italy; ALBERT STADLER, Herrliberg, Switzerland;

PETER Y. WOO, Biola University, La Mirada, CA, USA; and the proposers.

The featured solution shows that the inequality is strict.

The proofs given by most solvers are similar to the one featured above. Faynshteyn

2 2 2

+y +z

obtained the stronger result in which the right side of the inequality is replaced by 3

2

· x x+y+z

x2 y2 z2 3 x2 +y 2 +z 2

by proving that y+z

+ z+x

+ x+y

≥ 2

· x+y+z

. He also claimed the following

generalization without proof:

X {x}2n bxc2n

3 x2n + y 2n + z 2n

+ ≥ · .

y z 22n−1 x+y+z

cyclic

119

Let P and Q be interior points of triangle ABC. Let AA0 , BB 0 , and

CC be three concurrent cevians through P . The line through A0 parallel to

0

through B 0 parallel to BQ intersects the lines CQ and AQ at points M and N 0 ,

respectively. The line through C 0 parallel to CQ intersects the lines AQ and BQ

at points N and L0 , respectively.

Is it true that triangles LM N and L0 M 0 N 0 have the same area?

The point P plays no role in the result: A0 , B 0 , and C 0 can be any points

on the lines BC, CA, and AB, respectively; furthermore, Q can be any point

except a vertex in the plane of ∆ABC. But even these points play no essential

role. The result is a special case of a known theorem; for triangles LM N and

L0 M 0 N 0 to have the same area, all that is required is that the six vertices satisfy

Theorem. If A, B, C, D, E, F are six points in an affine plane such that

AB||DE, BC||EF , and CD||F A, then triangles ACE and BDF have equal

areas and the same orientation.

It is easier to find a proof for the theorem than a reference. We use vec-

tors

with an arbitrary point taken to be the origin. Since

AB||DE

we have

~−B

A ~ × D ~ −E ~ = 0. Similarly, B ~ −C ~ × E ~ −F

~ = 0 and

C~ −D ~ × F ~ −A ~ = 0. Expanding these cross products we get,

~×D

A ~ −B

~ ×D

~ −A

~×E~ +B

~ ×E

~ = 0,

~ ×E

B ~ −C

~ ×E

~ −B

~ ×F

~ +C

~ ×F

~ = 0,

~ ×F

C ~ −D

~ ×F

~ −C

~ ×A

~ +D

~ ×A

~ = 0.

Subtracting the second relation from the sum of the first and third we get,

~ ×C

A ~ +C

~ ×E

~ +E

~ ×A

~=B

~ ×D

~ +D

~ ×F

~ +F

~ × B.

~

The final equation tells us that ∆ACE and ∆BDF have equal areas and the

same orientation.

Also solved by Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece;

MICHEL BATAILLE, Rouen, France; OLIVER GEUPEL, Brühl, NRW, Germany; and

PETER Y. WOO, Biola University, La Mirada, CA, USA.

120

3517.[2010 : 108, 110] Proposed by Václav Konečný, Big Rapids, MI, USA.

Is there a scalene triangle ABC for which there exists a point P in the plane

of ABC such that, for each line ` through P , the sum of the squares of the

distances of A, B, and C to ` is constant?

University, Fanar, Lebanon; Michel Bataille, Rouen, France; Oliver Geupel, Brühl,

NRW, Germany; and the proposer.

Yes, there are scalene triangles with the required property. In Cartesian

coordinates we take P to be the origin. The line through P with normal unit

vector (cos φ, sin φ) is described by the equation x cos φ + y sin φ = 0, whence

the square of the distance from a point X = (x0 , y0 ) to that line is

The examples given by our five correspondents are listed alphabetically by last

name; for each example the vertices are given in the first three columns and their

squared distances to x cos φ + y sin φ = 0 in the next three, followed in the final

column by the constant sum of those three squared distances.

A B C dA dB dC sum

(0, 5) (3, 0) (4,√0) 25 sin2 φ 9 cos2 φ 16 cos2 φ 25

(1, 0) (2, 0) (0, 5) cos2 φ 4 cos2 φ 5 sin2 φ 5

(1, −2) (2, 2) (2, −1) (cos φ− (2 cos φ+ (2 cos φ−

√

2 sin φ)2 2 sin φ)2 sin φ)2 9

(1, 0) (0, 1

2

) (0, 2

3

) cos2 φ 1

4

sin2 φ 3

4

sin2 φ 1

(0, 5) (−3, 0) (4, 0) 25 sin2 φ 9 cos2 φ 16 cos2 φ 25

Also solved by ALBERT STADLER, Herrliberg, Switzerland. There was one incorrect

submission.

Stadler also took P to be the origin, but he used complex numbers a, b, and c to represent

the vertices A,B, and C. He proved that the sum of the squares of the distances from A,B,

and C to a line through 0 is constant if and only if a2 + b2 + c2 = 0; in other words, given

vertices corresponding to the complex numbers a and b, the third vertex can be either square

root of −(a2 + b2 ). This, of course, produces all triangles, even degenerate, with the required

property with respect to the origin. The proposer was motivated by problem 3291 [2007 : 485,

487; 2008 : 492-494], which applied to isosceles triangles.

121

Khyrdalan, Azerbaijan.

Prove that if n and i are integers with 0 ≤ i ≤ n, then

Ci ¦ p © 1

0 < 1 − 2i+1 − 10(2i+1)n · 102n − 1 < ,

2 102n

th

where {x} denotes the fractional

part of the real number x and Ci is the i

1 2i

Catalan number, Ci = , i ≥ 0.

i+1 i

The case n = 0 is immediate; so assume that n > 0. We will use the

1/2

standard formula Ci = (−1)i i+1 · 22i+1 and the inequality Ci < 22i .

Pi 2n(i+1−k)

Let s = 102n(i+1) − k=1 Ck−1 · 10 22k−1 . It is well-known that the

Catalan number Ck−1 is an integer for each integer k > 0. Moreover, for 1 ≤

2n(i+1−k)

k ≤ i ≤ n, it holds that 2k − 1 < 2n ≤ 2n(i + 1 − k); hence 10 22k−1 is an

integer. Consequently,

s ∈ Z. (1)

P∞ Ck−1

k−i−2

1 1

Let t = 102n k=i+2 22k−1 · 102n

. By the binomial series, we have

that

1 X ∞

Ck−1 1 X ∞

k 1/2

0<t< = − 2n (−1)

102n k=1 22k−1 10 k=1 k

√

1− 1−1 1

= 2n

= . (2)

10 102n

Moreover,

Ci 22i 1

0 < 2i+1 + t < 2i+1 + = 1 . (3)

2 2 2

√

Let u = 10(2i+1)n · 102n − 1. Using the Binomial series, we obtain

r

1 X

∞

1/2 1

2n(i+1)

u = 10 · 1− = 102n(i+1) (−1)k

102n k=0

k 102nk

X

i

1/2 1/2

k 2n(i+1−k) i

= (−1) 10 − (−1)

k=0

k i+1

X

∞

1/2 1

− (−1)k−1

k=i+2

k 102n(k−i−1)

Ci

=s− − t.

22i+1

Ci

By (1) and (3), we obtain {u} = 1− 22i+1 −t. The conclusion follows immediately

from (2).

Also solved by ALBERT STADLER, Herrliberg, Switzerland; and the proposer.

122

3519. [2010 : 108, 110] Proposed by Nguyen Duy Khanh, student, Hanoi

University of Science, Hanoi, Vietnam.

Two triangles ABC and A0 B 0 C 0 have areas S and S 0 , respectively. Let wa ,

wb , wc be the lengths of the internal angle bisectors of ABC to the sides BC,

AC, AB, respectively, and define wa0 , wb0 , wc0 similarly. Prove or disprove that

√

wa wa0 + wb wb0 + wc wc0 ≥ 3 3SS 0 .

We show that the inequality is not valid in general.

As usual, let a = BC, b = CA, c = AB. Let ABC be a triangle with

a = 3 and c = b + 1 > 2. Using a well-known formula for the length of an angle

bisector, we obtain

Ê È

a2 2 (b − 1)b(b + 1)(b + 2)

wa = bc 1 − = ,

(b + c)2 2b + 1

and similarly

È È

2 6(b + 1)(b + 2) 2 3b(b + 2)

wb = , wc = .

b+4 b+3

Let A0 B 0 C 0 be a triangle with B 0 C 0 = c, C 0 A0 = b, A0 B 0 = a. We have

wa0 = wc , wb0 = wb , wc0 = wa . If p denotes the common semiperimeter of both

triangles, then we obtain

È È

S = S0 = p(p − a)(p − b)(p − c) = 2(b − 1)(b + 2) .

We will prove that for sufficiently large b,

√

wa wa0 + wb wb0 + wc wc0 < 3 3SS 0 .

By substituting the relations above, we obtain

È

8b(b + 2) 3(b − 1)(b + 1) 24(b + 1)(b + 2)

+

(b + 3)(2b + 1) (b + 4)2

È

< 3 6(b − 1)(b + 2) . (1)

√

8 3 √

As b → ∞, the left side of (1) is asymptotic to · b = 4 3b, while the right

√ √ 2 2 √ 2

side of (1) is asymptotic to 3 6b. Since (4 3) = 48 < 54 = (3 6) , the

right side eventually exceeds the left side. Indeed, with a calculator we find that

this occurs when b = 13.

This completes the proof.

No other solutions were received.

Geupel remarked that a related inequality appears in CRUX Problem 2029 [1995 : 91,

129–32], namely

√

wb wc + wc wa + wa wb ≥ 3 3S ,

which is true.

123

Seville, Seville, Spain.

Construct a triangle ABC such that the line through the incentre and the

circumcentre is parallel to the side AB.

Construct ∆ABC given its circumcentre O, circumradius R, and ∠C.

Suppose that we have constructed a scalene triangle ABC such that the line

through O and the incentre I is parallel to AB. Since I is interior to the

triangle, O must be on the same side of AB as C. For C 0 the midpoint of AB it

follows that ∠C 0 OA = ∠BCA (= ∠C). Observing that OC 0 = d(O, AB) =

d(I, AB) = r (the inradius) and OA = R (the circumradius), we deduce that

r

R

= cos C; from Euler’s inequality 2r < R, so that ∠C is necessarily between

60◦ and 90◦ . Moreover, if M is the point where the ray [OC 0 ) intersects the

circumcircle Γ, M C is the angle bisector of ∠C and, from a familiar theorem,

M B = M A = M I. This analysis of the figure leads to the following three-step

construction:

• Construct an angle equal to the given ∠C with vertex at O, and the point A

on one of its legs such that OA = R. Let C 0 be the orthogonal projection

of A onto the other leg, and let M be the point where the circle with centre

O and radius OA intersects that leg. Denote the circle by Γ and the point

where Γ intersects AC 0 by B.

• Since M O = AO, when 90◦ > ∠M OA > 60◦ we have M A > M O,

so that the circle with centre M , radius M A, intersects the line through O

parallel to AC 0 in two points. Let I be one of those points.

• Let C be the second point of intersection of the line M I with Γ. Then M C

is the angle bisector of ∠BCA (since M is the midpoint of the arc AB of

Γ that does not contain C), and I is the incentre of ∆ABC (since it is the

point on the angle bisector for which M I = M A = M B). The triangle

ABC satisfies the requirement.

The construction shows that there is at most one triangle with OI||AB for the

given O, R, and ∠C, and it exists if and only if 60◦ < ∠C < 90◦ .

Grande Prairie, AB; and Ricard Peiró, IES “Abastos”, Valencia, Spain.

Construct ∆ABC given O, I, and r. Should such a triangle exist, we let

F be the orthogonal projection of I onto AB; then IF = r. Consider the

right triangle OIF ; since by Euler’s formula OI 2 = R2 − 2Rr, we deduce that

OF 2 = R2 − 2Rr + r 2 , whence OF = R − r. It follows that if the point P

is at a distance r on the line OF beyond F , OP = R. The construction then

proceeds in four steps:

• Draw the circle with centre I and radius r (which will become the incircle).

Construct the perpendicular to IO at I, and denote by F one of the points

124

call it `. (The line ` will become AB.)

• Draw the circle with centre F and radius r = F I; denote by P the point

where it meets the line OP on the other side of F from O.

• Draw the circle with centre O and radius OP . (This will be the circumcircle

because OP = R.) Call the points A and B where it meets `.

• Define E to be the second point where the circle with centre A and radius

AF intersects the incircle, and define C to be the second point where AE

intersects the circumcircle. ABC is the required triangle. (Because by

construction, AF is tangent to the incircle and AE = AF , it follows that

AE must also be tangent to the incircle. BC is also tangent to the incircle

because we constructed the length R so that OI 2 = R2 − 2Rr.)

The construction shows that the required triangle exists and is unique for any

given pair of points O and I, and positive length r.

Also solved by MOHAMMED AASSILA, Strasbourg, France(a second solution); GEORGE

APOSTOLOPOULOS, Messolonghi, Greece; ROY BARBARA, Lebanese University, Fanar,

Lebanon(2 solutions); FRANCISCO JAVIER GARCÍA CAPITÁN, IES Álvarez Cubero, Priego

de Córdoba, Spain; OLIVER GEUPEL, Brühl, NRW, Germany; V ÁCLAV KONE ČN Ý,

Big Rapids, MI, USA; SHAILESH SHIRALI, Rishi Valley School, India; D.J. SMEENK,

Zaltbommel, the Netherlands; ALBERT STADLER, Herrliberg, Switzerland; PETER Y. WOO,

Biola University, La Mirada, CA, USA; and the proposer. There was one incomplete

submission.

Because the problem called for the construction of just one example of the required triangle

without specifying what we are given, our solvers provided a wide variety of constructions, several

of which were as simple as our featured solutions. Most were based on properties of triangles

for which OI||AB; such properties have been the subject of numerous problems in CRUX with

MAYHEM such as 659 [1982 : 215-216], which was based on Problem 758 in Mathematics

Magazine 43:5 (Nov. 1970) pages 285-286. Here are three of them:

(1) Construct the triangle given ∠A and side length c; Smeenk and Woo both used the

property cos A + cos B = 1.

(2) Construct the triangle given side lengths a and b; Aassila, Apostolopoulos, Garcia

Capitán, and Shirali all obtained a formula for c in terms of a and b:

p

ab + a4 − a2 b 2 + b 4

c= .

a+b

(For example, one can apply the cosine law to the formula used in (1).) Shirali observed

parenthetically that this formula implies that there are no integer-sided triangles with the

required property (because the expression under the radical sign has an integer square root

only if a = b = 0).

(3) Construct the triangle given R and the distance dA from O to the side BC; Geupel

noted that Carnot’s theorem, dA + dB + dC = R + r [Nathan Altshiller Court, College

Geometry, page 83, Section 146] becomes simpler when dC = r—the circle whose

diameter is OC contains the projections of O onto BC and AC, which are easily

constructed using the distances dA and R − dA .

125

SANU, Belgrade, Serbia.

Let f : R → R be a continuously differentiable function. Solve the

functional equation

x+y

f (x) + f (y) − 2f = p(x − y) f 0 (x) − f 0 (y) ,

2

where p is a real parameter independent of x, y.

Partial solution by the proposer, modified by the editor.

We show that if f has a continuous second derivative, then

1

(i) f (x) = ax + b if p 6= ,

4

1

(ii) f (x) = ax3 + bx2 + cx + d if p = ,

4

where a, b, c, and d are arbitrary constants.

Differentiating the given equation with respect to x, we obtain

x+y

f 0 (x) − f 0 = p f 0 (x) − f 0 (y) + (x − y)f 00 (x) . (1)

2

Next, differentiating (1) with respect to y, we obtain

1 00 x+y

f = p f 00 (x) + f 00 (y) . (2)

2 2

1

Setting y = x in (2) we then have 2p − f 00 (x) = 0.

2

1

If p 6= , then f 00 (x) = 0, from which (i) follows.

4

1

If p = , then (2) becomes

4

x+y 1

f 00 = f 00 (x) + f 00 (y) .

2 2

That is,

1 x+y

g g(x) + g(y) ,= (3)

2 2

where g = f 00 , which is a variant of Cauchy’s equation also known as Jensen’s

equation. It is a well-known fact in the theory of functional equations that the

only solution of (3) is g(x) = αx + β for some constants α and β. From this,

(ii) follows immediately.

Also solved by MICHEL BATAILLE, Rouen, France, under the same stronger assumption

on f 00 .

Three other solutions were submitted which were either incomplete or partially incorrect.

The problem remains open if we assume only that f ∈ C 1 (−∞, ∞).

That the only solution of (3) above is g(x) = αx + β for some constants α, β when g

is continuous or monotone can be found in Funtional Equations and How to Solve Them, by

C.G. Small, Springer Problem Books in Mathematics, 2007. In case (i) both sides of the given

equation vanish, while in case (ii) both sides equal 3

4

a(x + y)(x − y)2 + 1

2

b(x − y)2 .

126

Let a1 , a2 , . . . , an+1 be positive real numbers satisfying the condition

an+1 = min{a1 , a2 , . . . , an+1 }. Prove that

an+1

1 + an+1

2 + · · · + an+1

n+1 − (n + 1)a1 a2 · · · an+1

≥ (n + 1)an+1 [(a1 − an+1 )n + (a2 − an+1 )n + · · · + (an − an+1 )n

− n(a1 − an+1 )(a2 − an+1 ) · · · (an − an+1 )] .

Let t = an+1 and let xi = ai − an+1 for i = 1, 2, . . . , n, then we have

that t > 0 and xi ≥ 0 for all i. We can now write the inequality as

X

n Y

n

(t + xi )n+1 + tn+1 − (n + 1)t (t + xi )

i=1 i=1

!

X

n Y

n

≥ (n + 1)t xn

i −n xi . (1)

i=1 i=1

degree of t, the coefficient of the LHS is not less than the coefficient of the

RHS. The coefficients

Pn of t0 and

Qn t

n+1

follow immediately. The coefficient of

1 n

t is (n + 1)( i=1 xi − n i=1 xi ) for both sides of the inequality. The

remaining coefficients, ti+1 for i = 1, 2, . . . , n − 1, are equal to 0 for the RHS so

it now suffices to show that the remaining coefficients of the LHS are non-negative,

that is

X

n

n+1 X Y

n−i

xjn−i − (n + 1) xjk ≥ 0 . (2)

j=1

i+1 1≤j1 <j2 <···<jn−i ≤n k=1

Qn−i Pn−i

By the AM-GM Inequality, we have that k=1 xjk ≤ 1

n−i k=1 xjn−i

k

. Thus,

n

X Y

n−i

n+1 n−1 X

(n + 1) xjk ≤ xjn−i . (3)

1≤j1 <j2 <···<jn−i ≤n k=1

n−i i i=1

n+1 n+1 n−1

≥ .

i+1 n−i i

From the definition of binomial coefficients, we have that

n+1 n+1 n n−1 n+1 n−1

= ≥

i+1 n−ii+1 i n−i i

and we are done.

Also solved by Michel Bataille, Rouen, France; Chip Curtis, Missouri Southern State

University, Joplin, MO, USA; and the proposer.

127

a+b √ √ √ 2

Let 0 ≤ a, b ≤ 1. Prove that ≤ a + b − ab .

1 + ab

√ √

For convenience, take α = a and β = b. Then we need to show that for

all 0 ≤ α, β ≤ 1 we have

α2 + β 2

≤ (α + β − αβ)2 ,

1 + α2 β 2

or equivalently

2

(α + β − αβ) 1 + α2 β 2 − α2 + β 2 ≥ 0 .

α2 β 2 − αβ(α + β) − αβ + 2 ≥ 0 .

But

Also solved by ARKADY ALT, San Jose, CA, USA; ŠEFKET ARSLANAGI Ć,

University of Sarajevo, Sarajevo, Bosnia and Herzegovina; ROY BARBARA, Lebanese

University, Fanar, Lebanon; MICHEL BATAILLE, Rouen, France; CHIP CURTIS, Missouri

Southern State University, Joplin, MO, USA; CHARLES R. DIMINNIE, Angelo State

University, San Angelo, TX, USA; OLIVER GEUPEL, Brühl, NRW, Germany; KEE-WAI

LAU, Hong Kong, China; PAOLO PERFETTI, Dipartimento di Matematica, Università degli

studi di Tor Vergata Roma, Rome, Italy; PRITHWIJIT DE, Homi Bhabha Centre for Science

Education, Mumbai, India; ALBERT STADLER, Herrliberg, Switzerland; DANIEL TSAI,

student, Taipei American School, Taipei, Taiwan; PETER Y. WOO, Biola University, La

Mirada, CA, USA; KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh, PA, USA;

and the proposer.

128

3526.[2010 : 109, 111] Proposed by Cao Minh Quang, Nguyen Binh Khiem High

School, Vinh Long, Vietnam.

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

X a

È ≥ 1.

cyclic a2 + 2(b + c)2

By Hölder inequality we have

2

X a X

È a(a2 + 2(b + c)2 ) ≥ (a + b + c)3 .

cyclic a2 + 2(b + c)2 cyclic

X

(a + b + c)3 ≥ a(a2 + 2(b + c)2 ) ,

cyclic

or equivalently that

But this is immediate from the AM-GM inequality. This completes the proof.

Equality holds if and only if a = b = c

Also solved by ARKADY ALT, San Jose, CA, USA; GEORGE APOSTOLOPOULOS,

Messolonghi, Greece(2 solutions); ŠEFKET ARSLANAGI Ć, University of Sarajevo, Sarajevo,

Bosnia and Herzegovina; MICHEL BATAILLE, Rouen, France; CHIP CURTIS, Missouri

Southern State University, Joplin, MO, USA; PEDRO HENRIQUE O. PANTOJA, student,

UFRN, Brazil; PAOLO PERFETTI, Dipartimento di Matematica, Università degli studi di Tor

Vergata Roma, Rome, Italy; ALBERT STADLER, Herrliberg, Switzerland; PETER Y. WOO,

Biola University, La Mirada, CA, USA; and the proposer.

Crux Mathematicorum

with Mathematical Mayhem

Former Editors / Anciens Rédacteurs: Bruce L.R. Shawyer, James E. Totten, Václav Linek

Crux Mathematicorum

Founding Editors / Rédacteurs-fondateurs: Léopold Sauvé & Frederick G.B. Maskell

Former Editors / Anciens Rédacteurs: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer

Mathematical Mayhem

Founding Editors / Rédacteurs-fondateurs: Patrick Surry & Ravi Vakil

Former Editors / Anciens Rédacteurs: Philip Jong, Jeff Higham, J.P. Grossman,

Andre Chang, Naoki Sato, Cyrus Hsia, Shawn Godin, Jeff Hooper, Ian VanderBurgh

129

EDITORIAL

Shawn Godin

Hello again CRUX with MAYHEM readers. Is it April already? I just

wanted to thank you for your patience as I ease into my role as editor of this great

publication many months later than I would have liked. You may (or may not)

have noticed some subtle changes already. For one thing, we have a new font that

is a lot more user friendly from my end. We have also changed the package that

we use to create diagrams so it is much easier to create nicer looking diagrams.

We are in the process of going through a couple of years of proposals to

decide which ones we will use in the journal. The hope is that you will hear about

the fate of your proposals soon after you have submitted them. It is the plan that

in the next couple of months we will review the problems that have already be

submitted in 2011. After that point, as we continue to work our way through the

older problems still in our banks, problem proposers will be informed of the status

of their proposal in a much more timely fashion. We will also look into classifying

problems so we can inform proposers which areas already have a long queue of

problems waiting to appear and which areas have a shortage.

We are still behind publication schedule and will be for the remainder of

2011, but we should be closer to our real schedule going into 2012 (so you should

get the February issue before July!). We will continue to use the “old” deadlines

in print, but deadlines will be extended until we are ready to work on the material.

A CRUX with MAYHEM page has been created on Facebook at

www.facebook.com/pages/

Crux-Mathematicorum-with-Mathematical-Mayhem/152157028211955.

Updates to our progress will be posted on the site. Messages informing you when

certain problems have been sent to the editors for moderation will appear to give

you a better idea of what the “real” deadline for the problems will be. We will

also give you some hints as to what will appear in future issues.

As I mentioned in an earlier editorial we are looking to make some changes.

With increases in publishing and mailing costs we are exploring alternate ways to

deliver CRUX with MAYHEM. Also, in an increasingly digital world, we are

looking into ways that we can better interact with our readers. We hope that,

sometime in the not too distant future, CRUX with MAYHEM readers can

submit problem proposals and solutions online. You would then be able to check

the status of your proposal to see if it has been accepted and when it is projected

to appear.

We are preparing a short online survey to get some feedback on various parts

of CRUX with MAYHEM and to get your input into some of the proposed

changes. Check the Facebook page for details and please take a few minutes to

give us your opinion.

130

Please send your solutions to problems in this Skoliad by December 15, 2011.

A copy of CRUX with Mayhem will be sent to one pre-university reader who

sends in solutions before the deadline. The decision of the editors is final.

Our contest this month is the Maritime Mathematics Competition, 2010.

Our thanks go to David Horrocks, University of Prince Edward Island, for

providing us with this contest and for permission to publish it.

Maritime Mathematics Competition, 2010

2 hours allowed

1. The Valhalla Winter Games are held in February, and the closing ceremonies

are on the last day of the month. The first Valhalla Winter Games were held in

the year 750, and since that year, they have been held every five years. How many

times have the closing ceremonies been held on February 29th ? Note that year Y

is a leap year if exactly one of the following conditions is true:

(a) Y is divisible by 4 but Y is not divisible by 100.

(b) Y is divisible by 400.

2. A triangle with vertices A(0, 0), B(3, 4), and C(2, c) has area 5. Find all

possible values of the number c.

3. Let f (x) = ax2 + bx + c, where a, b, and c are real numbers. Assume that

f (0), f (1), and f (2) are all integers.

(a) Prove that f (2010) is also an integer.

(b) Decide if f (2011) is an integer.

4. If x is a real number, let bxc denote the largest integer which is less than or

equal to x. For example, b7.012c = 7. If n is any positive integer, find a (simple)

formula for

2n 2(n + 1) 2(n + 2)

+ + .

3 3 3

5. (a) If a is a positive number, prove that

1

a+ ≥ 2.

a

(b) If a and b are both positive numbers, prove that

1 1 1

a+ +b+ + ≥ 4.5 .

a b ab

1

You may assume without proof that f (x) = x + x

is an increasing function

for x ≥ 1.

131

6

√. A hole in a concrete wall has the shape of a semi-circle with a radius of

2 metres. A utility company wants to place one large circular pipe or two smaller

circular pipes of equal radius through the hole to supply water to Watertown. If

they want to maximise the amount of water that could flow to Watertown, should

they use one pipe or two pipes, and what size pipes(s) should they use?

2 heures a permis

1. Les Jeux d’hiver de Valhalla de déroulent en février, les cérémonies de clôture

se tenant let dernier jour du mois. Les premier Jeux d’hiver de Valhalla se sont

déroulés en l’an 750 et les Jeux prennent place depuis à tous cinq ans. Combien

de fois le cérémonies de clôture ont-elle eu lieu le 29 février ? Rappelons qu’une

année Y est une année bissextile si l’une des conditions suivantes est satisfaite :

(a) Y est divisible par 4 mais n’est pas divisible par 100.

(b) Y est divisible par 400.

2. Si les sommets d’un triangle d’aire 5 sont A(0, 0), B(3, 4) et C(2, c), trouver

toutes les valeurs possibles de c.

3. Soit f (x) = ax2 + bx + c où a, b, et c sont des nombres réel. Supposons que

f (0), f (1), et f (2) sont des entier.

(a) Montre que f (2010) est aussi un entier.

(b) Déterminer si f (2011) est un entier.

4. Si x est une nombre réel, dénotons par bxc le plus grand entier inférieur ou

égal à x. Par exemple, b7.012c = 7. Si n est un entier positif quelconque, trouver

une formule (simple) pour

2n 2(n + 1) 2(n + 2)

+ + .

3 3 3

1

a+ ≥ 2.

a

(b) Si a et b sont des nombres réels strictement positifs, montre que

1 1 1

a+ +b+ + ≥ 4.5 .

a b ab

1

Vous pouvez supposer sans preuve que la fonction f (x) = x + x

est croissante

pour x ≥ 1.

132

√ la ville de Watertown est percé d’un trou en forme de

demi-cercle de rayon 2 mètres. La ville veut se fournir de l’eau au moyens de

tuyaux passés à travers le trou. Pour maximiser le débit de l’eau, est-il préférable

d’utiliser en seul grand tuyau circulaire ou deux tuyaux circulaires plus petits de

même rayon, et quelle est la dimension des tuyaux à utiliser ?

in Skoliad 126 at [2010 : 262–263].

angles x, y, and z. Find the sum of the

angles x, y, and z. x y z

Comment by Solomon W. Golomb, University of Southern California, Los Angeles,

CA, USA.

Problem 4. on pages 262–263 of the September 2010 issue of CRUX with

MAYHEM asks for the sum of the three angles x, y and z, in the figure below,

consisting of three adjacent unit squares, and three diagonals. Two proofs that

x + y + z = 90◦ , one trigonometric and one geometric, are given.

A B C D

x y z

E F G H

z = x + y) appeared in Martin Gardner’s “Mathematical Games” column in

Scientific American, and drew a great deal of reader response. However, the most

elegant solution, by the late Leon Bankoff, has never been previously published.

It uses no trigonometry, and needs no additional lines to be drawn. Bankoff’s

keen observation was that the triangles DEG and DGF are similar! (They

have ∠DGF is common,

√ and√the including sides are in the √

same ratio,

√ namely

|DG| : |GF | = 2 : 1 = 2, and |EG| : |GD| = 2 : 2 = 2.) Hence

angle x = ∠DEG in the larger triangle equals ∠GDF = x in the smaller

triangle.

A B C D

x

x y zz

E F G H

So we have that the three angles of triangle DGF are x, y, and 90◦ + z, and

must sum to 180◦ . Hence x + y + z = 90◦ .

133

Contest, Final Round, 2009/2010, given in Skoliad 125 at [2010 : 259–260].

1. A 2009 × 2010 grid is filled with the numbers 1 and −1. For each row,

calculate the product of the entries in that row. Do likewise for the columns.

Show that the sum of all the row products and all the column products cannot be

zero.

Solution by Rowena Ho, student, École Dr. Charles Best Secondary School,

Coquitlam, BC.

Each of the 2009 row products and each of the 2010 column products is

either 1 or −1. For the sum of a collection of 1’s and −1’s to be zero, you must

have equally many of each, but this is impossible since you have 4019 numbers.

Thus the sum of all the row products and all the column products cannot be zero.

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; RICHARD I. HESS, Rancho Palos Verdes, CA, USA; MONICA HSIEH,

student, Burnaby North Secondary School, Burnaby, BC; and JULIA PENG, student, Campbell

Collegiate, Regina, SK.

2. The square ABCD has side length 6. The point P splits side AB such

that |AP | : |P B| = 2 : 1. A point Q inside the square is chosen such that

|AQ| = |P Q| = |CQ|. Find the area of 4CP Q.

BC.

Impose a coordinate system so that A = (0, 6), P B

B = (6, 6), C = (6, 0), and D = (0, 0). Since A

|AP | : |P B| = 2 : 1, P = (4, 6). Since |AQ| = |P Q|,

Q is on the line given by x = 2. Since |AQ| = |CQ|,

Q is on the line through B and D, y = x. Thus

Q

Q = (2, 2).

6·2

Now, the area of 4CDQ is 2 = 6, the area of

4BCP is 2·6 2

= 6, the area of 4AP Q is 4·4

2

= 8, the D C

6·2 2

area of 4ADQ is 2 = 6, and the area of ABCD is 6 = 36. Therefore the

area of 4CP Q is 36 − 6 − 6 − 8 − 6 = 10.

Also solved by LENA CHOI, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; and RICHARD I. HESS, Rancho Palos Verdes, CA, USA.

3. The product of three positive integers is 140. Determine the sum of the three

integers if the second integer is seven times the first one.

Germany.

Say the three integers are a, 7a, and b. Then 7a2 b = 140. If a = 1, then

b = 20, and the sum of the three positive integers is a+7a+b = 1+7+20 = 28.

If a = 2, then b = 5, and the sum is 2 + 14 + 5 = 21. If a = 3, then b = 20 9

,

134

then b ≤ 54 , so b cannot be a positive integer. Hence, the sum is either 21 or 28.

Also solved by RICHARD I. HESS, Rancho Palos Verdes, CA, USA; and MONICA

HSIEH, student, Burnaby North Secondary School, Burnaby, BC.

Rather than going through so many cases, it may be easier to consider that if

7a2 b = 140, then a2 b = 20 = 22 · 5. The only squares that divide 22 · 5 are 12 and

22 , so a = 1 or a = 2.

4. Five points are placed at the intersections of a rectangular grid. Then the mid-

point of each pair of points is marked. Prove that at least one of these midpoints

lands on an intersection point of the grid.

Solution by Lena Choi, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; Rowena Ho, student, École Dr. Charles Best Secondary School,

Coquitlam, BC; and Monica Hsieh, student, Burnaby North Secondary School,

Burnaby, BC.

Impose a coordinate system such that (0, 0) is a grid point and the size of

the grid is 1. Then the five points all have integer coordinates and the midpoints

are grid points if and only if they, too, have integer coordinates.

The midpoint between (x1 , y1 ) and (x2 , y2 ) is 12 (x1 + x2 ), 21 (y1 + y2 ) .

Therefore a midpoint is a grid point if and only if x1 + x2 and y1 + y2 are both

even. Note that “even plus even is even” and “odd plus odd is even” while the

sum of two numbers of opposite parity is odd.

In terms of parity of their coordinates, the grid points fall into four classes:

(even, even), (even, odd), (odd, even), and (even, odd). Since you have five grid

points, two must be in the same class. Say (x1 , y1 ) and (x2 , y2 ) are in the same

class. Then x1 and x2 have the same parity, so x1 + x2 is even. Likewise, y1 + y2

is even, so the midpoint between (x1 , y1 ) and (x2 , y2 ) is a grid point.

Also solved by RICHARD I. HESS, Rancho Palos Verdes, CA, USA.

placed such that |AK| = |LM |. Place

the points B and C on one side of AM C

and point D on the other side of AM such

that |BK| = |KM |, |CM | = |KL|, and

L

|DL| = |LM |, and such that BK, CM , A M

and DL are all perpendicular to AM . Prove K

that ABCD is a square. D

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

Note that |AL| = |AK| + |KL| = |LM | + |KL| = |KM | = |BK| and

that |AK| = |LM | = |DL|. Since 4ABK and 4DAL are both right-angled,

|AB| = |AD| and 4ABK ∼ = 4DAL.

Let O be the point on BK such that KM CO is a rectangle. Let P be

the intersection of CO and the extension of DL. Then OP LK is a rectangle.

Moreover, |KL| = |CM | = |P L|, so OP LK is a square.

135

B

a

Now,

b

|DP | = |DL| + |LP | b

O P a C

= |AK| + |KL| = |AL|

and

|P C| = |OC| − |OP | a K L

A b M

= |KM | − |KL|

= |LM | = |DL| . b

a

D

Again, both 4CDP and 4DAL are right-angled, so 4CDP ∼ = 4DAL.

Finally, |OC| = |KM | = |AL| and |BO| = |BK| − |OK| = |AL| −

|KL| = |AK|, so 4BCO ∼ = 4ABK since both triangles are right-angled.

Since 4ABK ∼ = 4BCO ∼ = CDP ∼ = 4DAL, the sides of ABCD are

all equal and the angles are equal as labelled. Since the angle sum in 4BAK is

180◦ , a + b = 90◦ , so ∠DAB = 90◦ , and ABCD is a square.

Several solvers showed that ABCD is a rhombus.

6. Let N be a positive integer. Ragnhild writes down all the divisors of N other

than 1 and N . She then notes that the largest divisor is 45 times the smallest

one. Which positive integers satisfy this condition?

Solution by Lena Choi, student, École Dr. Charles Best Secondary School,

Coquitlam, BC.

The smallest possible proper divisor of N would be 2. If 2 is a divisor, then

2 · 45 = 90 would be the largest proper divisor. Since N must be the product of

its largest and smallest proper divisors, N = 2 · 90 = 180.

If 3 is the smallest proper divisor, then 3 · 45 = 135 is the largest proper

divisor, and N = 3 · 135 = 405.

If m > 3 and m is the smallest proper divisor, then 45m is the largest

proper divisor, and N = 45m2 . But 45m2 is divisible by 3 since 45 is divisible

by 3. Thus m is not the smallest proper divisor after all. This contradiction shows

that the smallest divisor is at most 3. Hence N = 180 or N = 405.

Also solved by RICHARD I. HESS, Rancho Palos Verdes, CA, USA; ROWENA HO,

student, École Dr. Charles Best Secondary School, Coquitlam, BC; and MONICA HSIEH,

student, Burnaby North Secondary School, Burnaby, BC.

This issue’s prize of one copy of Crux Mathematicorum for the best

solutions goes to Monica Hsieh, student, Burnaby North Secondary School,

Burnaby, BC.

We hope that our readers will enjoy the featured contest and share their

solutions.

136

MATHEMATICAL MAYHEM

Mathematical Mayhem began in 1988 as a Mathematical Journal for and by

High School and University Students. It continues, with the same emphasis,

as an integral part of Crux Mathematicorum with Mathematical Mayhem.

The interim Mayhem Editor is Shawn Godin (Cairine Wilson Secondary

School, Orleans, ON). The Assistant Mayhem Editor is Lynn Miller (Cairine

Wilson Secondary School, Orleans, ON). The other staff members are Ann

Arden (Osgoode Township District High School, Osgoode, ON) and Monika Khbeis

(Our Lady of Mt. Carmel Secondary School, Mississauga, ON).

Mayhem Problems

Please send your solutions to the problems in this edition by 15 November 2011.

Solutions received after this date will only be considered if there is time before publication

of the solutions.

Each problem is given in English and French, the official languages of Canada. In

issues 1, 3, 5, and 7, English will precede French, and in issues 2, 4, 6, and 8, French

will precede English.

The editor thanks Rolland Gaudet, Université de Saint-Boniface, Winnipeg, MB,

for translating the problems from English into French.

Using four sticks with lengths of 1 cm, 2 cm, 3 cm, and 5 cm, respectively,

you can measure any integral length from 1 cm to 10 cm. Note that a stick may

only be used once in a particular measurement, so the 1 cm, 2 cm, and 3 cm sticks

could be used to measure 6 cm, but not the 3 cm stick twice.

(a) Find a set of ten stick lengths that can be used to represent any integral

length from 1 cm to 100 cm.

(b) What is the fewest number of sticks that are needed to represent any integral

length from 1 cm to 100 cm?

St. John’s, NL.

Triangle ABC has ∠BAC = 90◦ . The feet of the perpendiculars from

A to the internal bisectors of ∠ABC and ∠ACB are P and Q, respectively.

Determine the measure of ∠P AQ.

137

Solve the equation

2

2 x

x +4 = 45 .

x−2

ON.

Prove that

Y

n

n 1 Y

n

kk

=

k=1

k n! k=1

(n − k)!

for all n ∈ N.

Buzău, Romania.

How many distinct numbers are in the list

12 − 1 + 4 22 − 2 + 4 32 − 3 + 4 20112 − 2011 + 4

, , , ..., ?

12 + 1 22 + 1 32 + 1 20112 + 1

Let m be a positive integer. Find all real solutions to the equation

Ê É q È √

m+ m+ m + ··· m+ m+ x = x,

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

À l’aide de baguettes de longueurs 1 cm, 2 cm, 3 cm et 5 cm, on peut

mesurer toute longueur entière de 1 cm à 10 cm. Noter qu’une baguette oeut être

utilisée une seule fois lors d’une mesure ; par exemple, les baguettes de 1 cm, 2

cm et 3 cm peuvent être utilisées pour mesurer 6 cm, tandis qu’on ne peut pas

utiliser deux baguettes de 3 cm pour mesurer 6 cm.

mesurer toute longueur entière de 1 cm à 100 cm.

(b) Quel est le plus petit nombre de baguettes permettant de mesurer toute

longueur entière de 1 cm à 100 cm ?

138

St. John’s, NL.

Le triangle ABC est tel que ∠BAC = 90◦ . Les pieds des perpendiculaires

de A jusqu’aux bissectrices internes des angles ∠ABC et ∠ACB sont P et Q

respectivement. Déterminer la mesure de ∠P AQ.

Résoudre l’équation

2

x

x2 + 4 = 45 .

x−2

M485. Proposé par Edward T.H. Wang, Université Wilfrid Laurier, Waterloo,

ON.

Démontrer que

Y

n

n 1 Y

n

kk

=

k=1

k n! k=1

(n − k)!

pour tout n ∈ N.

M486. Proposé par Neculai Stanciu, École secondaire George Emil Palade,

Buzău, Roumanie.

Combien de nombres distincts y a-t-il dans la liste

12 − 1 + 4 22 − 2 + 4 32 − 3 + 4 20112 − 2011 + 4

, , , ..., ?

12 + 1 22 + 1 32 + 1 20112 + 1

M487. Proposé par Samuel Gómez Moreno, Université de Jaén, Jaén, Espagne.

Soit m un entière positif. Déterminer toutes les solutions réelles à l’équation

Ê É q È √

m+ m+ m + ··· m+ m+ x = x,

dans laquelle l’entière m a lieu n fois.

Mayhem Solutions

M440. Proposed by the Mayhem Staff.

In trapezoid ABCD, AB is parallel to DC and AD is perpendicular to

AB. If AB = 20, BC = 5x, CD = x2 + 3x, and DA = 3x, determine the

value of x.

139

There are two cases to consider.

Case I: x2 + 3x > 20.

Let P be the foot of the perpendicular from B to DC. Then BP = 3x

and, according to Pythagoras, P C = 4x. Therefore,

A 20 B

x2 + 3x = 20 + 4x,

3x 3x 5x

x2 − x − 20 = 0,

4x

(x − 5)(x + 4) = 0,

D P C

x = 5 (since x > 0).

x2 + 3x

A 20 B

Proceeding as in Case I, we obtain

x2 + 7x − 20 = 0,

√ 4x

−7 + 129 D C P

x= .

2

x2 + 3x

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; SCOTT BROWN,

Auburn University, Montgomery, AL, USA; RICHARD I. HESS, Rancho Palos Verdes, CA,

USA; ANTONIO LEDESMA LÓPEZ, Instituto de Educación Secundaria No. 1, Requena-

Valencia, Spain; DRAGOLJUB MILO ŠEVIĆ, Gornji Milanovac, Serbia; RICARD PEIRÓ,

IES “Abastos”, Valencia, Spain; BRUNO SALGUEIRO FANEGO, Viveiro, Spain; and

NECULAI STANCIU, George Emil Palade Secondary School, Buzău, Romania. Eight

incorrect solutions were received. Most of the incorrect solutions neglected one of the cases.

Consider the square array

2 3

1 2 ··· n−1 n

6 n+1 n+2 ··· 2n − 1 2n 7

6 7

6 .. .. .. .. 7

4 . . . . 5

(n − 1)n + 1 (n − 1)n + 2 ··· n − 1 n2

2

the sum of the numbers on each diagonal. How does this sum compare to the

“magic constant” that would be obtained if the n2 entries were rearranged to

form a magic square?

140

Los elementos de la diagonal principal son: 1, n + 2, 2n + 3, . . . , (n −

1)n + n. La suma es:

D1 (n) = 1 + (n + 2) + (2n + 3) + · · · + [(n − 1)n + n]

= [1 + 2 + 3 + · · · + n] + n[1 + 2 + 3 + · · · + (n − 1)]

n(n + 1) (n − 1)n

= +n

2 2

3

n +n

= .

2

Los elementos de la diagonal secundaria son: n, 2n−1, 3n−2, . . . , n·n−(n−1).

La suma es:

D2 (n) = n + (2n − 1) + (3n − 2) + · · · + [n · n − (n − 1)]

= n[1 + 2 + 3 + · · · + n] − [1 + 2 + 3 + · · · + (n − 1)]

n(n + 1) (n − 1)n

=n −

2 2

3

n +n

= .

2

La constante mágica de un cuadrado mágico n × n es:

1

M (n) = (1 + 2 + 3 + · · · + n2 )

n

1 n2 (n2 + 1)

=

n 2

3

n +n

= = D1 (n) = D2 (n).

2

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; JACLYN

CHANG, student, University of Calgary, Calgary, AB; SAMUEL GÓMEZ MORENO,

Universidad de Jaén, Jaén, Spain; RICHARD I. HESS, Rancho Palos Verdes, CA, USA;

ANTONIO LEDESMA LÓPEZ, Instituto de Educación Secundaria No. 1, Requena-Valencia,

Spain; BRUNO SALGUEIRO FANEGO, Viveiro, Spain; NECULAI STANCIU, George

Emil Palade Secondary School, Buzău, Romania; GUSNADI WIYOGA, student, SMPN 8,

Yogyakarta, Indonesia; and ALLEN ZHU, Conestoga High School, Berwyn, PA, USA.

The lines with equations y = x + 1, y = mx − 1, and y = −4x + 2m

pass through the same point. Determine all possible values for m.

Given are three linear equations

y = x + 1, (1)

y = mx − 1, (2)

y = −4x + 2m. (3)

141

x + 1 = mx − 1

x(1 − m) = −2

−2

x= (4)

1−m

From (2) and (3) we have

mx − 1 = −4x + 2m

x(m + 4) = 2m + 1

2m + 1

x= (5)

m+4

From (4) and (5) we have

−2 2m + 1

=

1−m m+4

−2(m + 4) = (1 − m)(2m + 1)

−2m − 8 = 2m − 2m2 + 1 − m

2m2 − 3m − 9 = 0

(2m + 3)(m − 3) = 0

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ALPER CAY

and LOKMAN GOKCE, Geomania Problem Group, Kayseri, Turkey; DAVINIA CERVERA

GARCÍA, Club Mathématique de l’Instituto de Ecuación Secundaria No. 1, Requena-Valencia,

Spain; MUHAMMAD HAFIZ FARIZI, student, SMPN 8, Yogyakarta, Indonesia; G.C.

GREUBEL, Newport News, VA, USA; RICHARD I. HESS, Rancho Palos Verdes, CA,

USA; WINDA KIRANA, student, SMPN 8, Yogyakarta, Indonesia; SALLY LI, student,

Marc Garneau Collegiate Institute, Toronto, ON; DEBRA A. OHL, student, Angelo State

University, San Angelo, TX, USA; KONSTANTINOS AL. NAKOS, Agrinio, Greece; RICARD

PEIRÓ, IES “Abastos”, Valencia, Spain; NECULAI STANCIU, George Emil Palade Secondary

School, Buzău, Romania; GUSNADI WIYOGA, student, SMPN 8, Yogyakarta, Indonesia; and

KONSTANTINE ZELATOR, University of Pittsburgh, Pittsburgh, PA, USA. Two incorrect

solutions were submitted.

Let a, b, and c be positive digits. Suppose that b equals the product of a,

b, and c, and ac = a + b + c. Determine a, b, and c. (Here ab is the two-digit

positive integer with tens digit a and units digit b.)

We are given

b = a·b·c

10a + c = a + b + c

142

gives a · c = 1; which implies that a = 1 = c. From 10a + c = a + b + c,

then we have 10 · 1 + 1 = 1 + b + 1; hence b = 11 − 2 = 9. Therefore

a = 1, b = 9, c = 1.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ALPER

CAY and LOKMAN GOKCE, Geomania Problem Group, Kayseri, Turkey; RICHARD

I. HESS, Rancho Palos Verdes, CA, USA; AFIFFAH NUUR MILA HUSNIANA,

student, SMPN 8, Yogyakarta, Indonesia; YOUNGHUAN JUNG, The Woodlands

School, Mississauga, ON; WINDA KIRANA, student, SMPN 8, Yogyakarta, Indonesia;

DONGCHAN LEE, University of Toronto, Toronto, ON; SALLY LI, student, Marc

Garneau Collegiate Institute, Toronto, ON; MUHAMMAD ROIHAN MUNAJIH,

student, SMPN 8, Yogyakarta, Indonesia; ALEECE NALBANDIAN, California

State University, Fresno, CA, USA; DEBRA A. OHL, student, Angelo State

University, San Angelo, TX, USA; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain;

ANDRÉS PLANELLS CÁRCEL, Club Mathématique de l’Instituto de Ecuación Secundaria

No. 1, Requena-Valencia, Spain; NECULAI STANCIU, George Emil Palade Secondary School,

Buzău, Romania; GUSNADI WIYOGA, student, SMPN 8, Yogyakarta, Indonesia; and

INGESTI BILKIS ZULFATINAAS, student, SMPN 8, Yogyakarta, Indonesia. One incorrect

solution was submitted.

Azerbaijan.

Let ABCD be a parallelogram. The sides AB and AD are extended to

points E and F (respectively) so that E, C, and F all lie on a straight line. Prove

that BE · DF = AB · AD.

BE BC

=

The triangles BCE and F DC are similar, so . Since ABCD

DC DF

BE AD

is a parallelogram we know BC = AD and DC = AB. So = , hence

AB DF

BE · DF = AB · AD.

Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca,

Spain; ALPER CAY and LOKMAN GOKCE, Geomania Problem Group, Kay-

seri, Turkey; RICHARD I. HESS, Rancho Palos Verdes, CA, USA; AFIFFAH

NUUR MILA HUSNIANA, student, SMPN 8, Yogyakarta, Indonesia; WINDA

KIRANA, student, SMPN 8, Yogyakarta, Indonesia; DONGCHAN LEE, Uni-

versity of Toronto, Toronto, ON; DEBRA A. OHL, student, Angelo State

University, San Angelo, TX, USA; PEDRO HENRIQUE O. PANTOJA, student, UFRN,

Brazil; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain; JORGE SEVILLA LACRUZ,

Club Mathématique de l’Instituto de Ecuación Secundaria No. 1, Requena-Valencia, Spain;

NECULAI STANCIU, George Emil Palade Secondary School, Buzău, Romania; LOU VANG,

California State University, Fresno, CA, USA; KONSTANTINE ZELATOR, University of

Pittsburgh, Pittsburgh, PA, USA; and INGESTI BILKIS ZULFATINAAS, student, SMPN 8,

Yogyakarta, Indonesia. One incorrect solution was submitted.

A polyhedron with exactly m + n faces has m faces that are quadrilaterals

and n faces that are triangles. Exactly four faces meet at each vertex. Prove that

n = 8.

143

The number of faces is F = m + n. Since there are 4 vertices and 4

edges in a quadrilateral, and 3 vertices and 3 edges in a triangle, then the total

number of edges would be E = 4m+3n 2

. The total number of vertices will be

V = 4m+3n4

since it is given that exactly four faces meet at each vertex. Using

Euler’s polyhedron formula, which says that the sum of the number of faces and

the number of vertices is equal to the number of edges plus two,

F + V = E + 2,

4m + 3n 4m + 3n

m+n+ = + 2.

4 2

Solving the equation, we get n = 8.

Also solved by ALPER CAY and LOKMAN GOKCE, Geomania Problem Group,

Kayseri, Turkey; JORGE ARMERO JIMÉNEZ, Club Mathématique de l’Instituto de Ecuación

Secundaria No. 1, Requena-Valencia, Spain; RICHARD I. HESS, Rancho Palos Verdes, CA,

USA; SALLY LI, student, Marc Garneau Collegiate Institute, Toronto, ON; RICARD PEIRÓ,

IES “Abastos”, Valencia, Spain; and NECULAI STANCIU, George Emil Palade Secondary

School, Buzău, Romania.

Buzău, Romania.

4x

Let E(x) = .

4x +2

(a) Prove that E(x) + E(1 − x) = 1.

1 2 2008 2009

(b) Find the value of E +E + ··· + E +E .

2010 2010 2010 2010

4x

(a) From the given equation E(x) = 4x +2

so,

4x 41−x

E(x) + E(1 − x) = +

4x + 2 41−x + 2

x 1−x

4 (4 + 2) + 41−x (4x + 2)

=

(4x + 2)(41−x + 2)

4 + 2(4x ) + 4 + 2(41−x )

=

4 + 2(4x ) + 2(41−x ) + 4

8 + 2(4x ) + 2(41−x )

=

8 + 2(4x ) + 2(41−x )

=1

4x

41−x 2 2

[Ed. – Note that E(1−x) = × 4x

= and the conclusion follows

41−x + 2 2

2 + 4x

immediately.]

144

1 2009 1 1

E +E =E +E 1− =1

2010 2010 2010 2010

2 2008 2 2

E +E =E=1 +E 1−

2010 2010 2010 2010

and so on. Then we have

1 2 2008 2009 1005

E +E +...+E +E = 1004×1+E

2010 2010 2010 2010 2010

1

1005 42 1005 1

Since E = 1 , then E = .

2010 4 2 +2 2010 2

Therefore the sum is 1004.5.

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ALPER CAY

and LOKMAN GOKCE, Geomania Problem Group, Kayseri, Turkey; CHAO-PING CHEN,

Henan Polytechnic University, Jiaozuo City, China and Mihály Bencze, Lajos Aprily High-

school, Brasov, Romania; DIANA DOMINGUEZ, California State University, Fresno, CA,

USA; MUHAMMAD HAFIZ FARIZI, student, SMPN 8, Yogyakarta, Indonesia; PABLO

PARDAL GARCES, Club Mathématique de l’Instituto de Ecuación Secundaria No. 1, Requena-

Valencia, Spain; G.C. GREUBEL, Newport News, VA, USA; RICHARD I. HESS, Rancho Palos

Verdes, CA, USA; WINDA KIRANA, student, SMPN 8, Yogyakarta, Indonesia; DONGCHAN

LEE, University of Toronto, Toronto, ON; SALLY LI, student, Marc Garneau Collegiate

Institute, Toronto, ON; KONSTANTINOS AL. NAKOS, Agrinio, Greece; CARLOS TORRES

NINAHUANCA, Lima, Perú; PEDRO HENRIQUE O. PANTOJA, student, UFRN, Brazil;

RICARD PEIRÓ, IES “Abastos”, Valencia, Spain; PAOLO PERFETTI, Dipartimento di

Matematica, Università degli studi di Tor Vergata Roma, Rome, Italy; GUSNADI WIYOGA,

student, SMPN 8, Yogyakarta, Indonesia; KONSTANTINE ZELATOR, University of

Pittsburgh, Pittsburgh, PA, USA; and INGESTI BILKIS ZULFATINAAS, student, SMPN 8,

Yogyakarta, Indonesia;

Waterloo, ON.

Prove that if n is an odd positive integer, then nn+2 + (n + 2)n is divisible

by 2(n + 1).

From the Binomial Theorem, (n + 2)n can be written as:

n

(n + 2)n = (n + 1) + 1

n n n

= (n + 1)n + (n + 1)n−1 + · · · + (n + 1) + 1.

0 1 n−1

Since n is an odd number we can also write:

nn+2 = (n + 1 − 1)n+2

n+2 n+2

= (n + 1)n+2 − (n + 1)n+1 + · · ·

0 1

n+2

+ (n + 1) − 1.

n+1

145

If we add the two expansions together, the constant terms cancel each other.

Therefore, we have:

n n+2 n n

(n + 2) + n = (n + 1)n + (n + 1)n−1 + · · ·

0 1

n n+2

+ (n + 1) + (n + 1)n+2

n−1 0

n+2 n+1 n+2

− (n + 1) + ··· + (n + 1)

1 n+1

Since all of the terms have a factor of n + 1, then (n + 2)n + nn+2 is divisible

by n + 1 and we can rewrite the expression as:

n n+2 n n−1 n

(n + 2) + n = (n + 1) (n + 1) + (n + 1)n−2 + ...

0 1

n n+2

+ + (n + 1)n+1

n−1 0

n+2 n n+2

− (n + 1) + · · · +

1 n+1

Now we must prove that the expression in the square brackets above is an

even number. Since we know that n + 1 is an even number, all of the termswith

n

a factor of n + 1 are also even. Then we are left with only two terms, n−1 and

n+2 n n n+2

n+1

.Since n−1 = n and n+2

n+1

= n + 2, then we have n−1

+ n+1

=

2n + 2 which is an even number.

Hence (n + 2)n + nn+2 is divisible by 2(n + 1).

Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; ADAM

GREGSON, teacher, University of Toronto Schools, Toronto, ON; G.C. GREUBEL, Newport

News, VA, USA; RICHARD I. HESS, Rancho Palos Verdes, CA, USA; ANTONIO LEDESMA

LÓPEZ, Instituto de Educación Secundaria No. 1, Requena-Valencia, Spain; DONGCHAN

LEE, University of Toronto, Toronto, ON; SALLY LI, student, Marc Garneau Collegiate

Institute, Toronto, ON; RICARD PEIRÓ, IES “Abastos”, Valencia, Spain; NECULAI

STANCIU, George Emil Palade Secondary School, Buzău, Romania; and KONSTANTINE

ZELATOR, University of Pittsburgh, Pittsburgh, PA, USA.

nn+2 + 1

Note that since n is odd = nn+1 − nn + nn−1 − · · · + 1 is odd. Also

n+1

(n + 2)n − 1

= (n + 2)n−1 + (n + 2)n + · · · + 1 is also odd. Thus their sum is even.

n+1

146

Ian VanderBurgh

Problems involving probability can be very interesting and can lead to lots

of discussion and debate. (If you don’t believe me about debate, try looking up

the Monty Hall Problem.) These problems also give lots of opportunity for both

creative solutions and plausible incorrect solutions.

Here are two problems involving probability that have very different flavours.

Problem 1 (2011 Euclid Contest) Three different numbers are chosen at random

from the set {1, 2, 3, 4, 5}. The numbers are arranged in increasing order. What

is the probability that the resulting sequence is an arithmetic sequence?

This type of problem is pretty standard. We are given a set of objects and

a property. We need to determine the probability that a randomly chosen object

from the set has the desired property.

Often, the most direct approach is to count the total number of objects

in the set and to count the number of objects in the set that have the desired

property. The probability that we are after is the second number divided by the

first number. Here is an illustration of this technique.

Solution to Problem 1. We consider choosing the three numbers all at once.

We list the possible sets of three numbers that can be chosen:

We have listed each in increasing order because once the numbers are chosen, we

arrange them in increasing order.

There are 10 sets of three numbers that can be chosen. Of these 10, the

4 sequences 1, 2, 3 and 1, 3, 5 and 2, 3, 4 and 3, 4, 5 are arithmetic sequences.

Therefore, the probability that the resulting sequence is an arithmetic sequence is

4

10

or 52 .

in a more straightforward way. The problem itself isn’t any easier, but it might

be easier to attack. Separating the problem into counting these two different sets

of objects makes this easier to approach.

When this contest was marked, the most popular incorrect answer to this

4 1

problem was 60 (or its reduced form of 15 ). Can you figure out what mistake

might lead to this answer?

The next problem is also about probability, but seems very different at first.

Problem 2 (2011 Euclid Contest) A 75 year old person has a 50% chance of

living at least another 10 years. A 75 year old person has a 20% chance of living

at least another 15 years. An 80 year old person has a 25% chance of living at

least another 10 years. What is the probability that an 80 year old person will live

at least another 5 years?

147

This is a really interesting problem that has a good “real life context”. The

data given is close to the actual data for Canadian adults. One approach is to use

the given probabilities directly.

Solution 1 to Problem 2. Suppose that the probability that a 75 year old

person lives to 80 is p, the probability that an 80 year old person lives to 85 is q,

and the probability that an 85 year old person lives to 90 is r. We want to the

determine the value of q.

For a 75 year old person to live at least another 10 years, they must live

another 5 years (to age 80) and then another 5 years (to age 85). The probability

of this is equal to pq. We are told in the question that this is equal to 50% or 0.5.

Therefore, pq = 0.5.

For a 75 year old person to live at least another 15 years, they must live

another 5 years (to age 80), then another 5 years (to age 85), and then another

5 years (to age 90). The probability of this is equal to pqr. We are told in the

question that this is equal to 20% or 0.2. Therefore, pqr = 0.2.

Similarly, since the probability that an 80 year old person will live another

10 years is 25%, then qr = 0.25.

pqr 0.2

Since pqr = 0.2 and pq = 0.5, then r = = = 0.4.

pq 0.5

qr 0.25

Since qr = 0.25 and r = 0.4, then q = = = 0.625.

r 0.4

Therefore, the probability that an 80 year old person will live at least another

5 years is 0.625, or 62.5%.

A second approach is actually to use the “count the objects” method that

we discussed earlier. You might wonder what the set of objects is. Here is one

way to do this.

Solution 2 to Problem 2. Consider a population of 100 people, each of whom is

75 years old and who behave according to the probabilities given in the question.

Each of the original 100 people has a 50% chance of living at least another

10 years, so there will be 50% × 100 = 50 of these people alive at age 85. Each

of the original 100 people has a 20% chance of living at least another 15 years, so

there will be 20% × 100 = 20 of these people alive at age 90.

Since there is a 25% (or 14 ) chance that an 80 year old person will live at

least another 10 years (that is, to age 90), then there should be 4 times as many

of these people alive at age 80 than at age 90. Since there are 20 people alive at

age 90, then there are 4 × 20 = 80 of the original 100 people alive at age 80.

In summary, of the initial 100 people of age 75, there are 80 alive at age

80, 50 alive at age 85, and 20 people alive at age 90. Because 50 of the 80 peo-

ple alive at age 80 are still alive at age 85, then the probability that an 80 year

50

old person will live at least 5 more years (that is, to age 85) is 80 = 58 , or 62.5%.

mathematics becomes so connected to real life. Problem 2 is related to an area of

mathematics called actuarial science, which has lots of applications to things like

insurance and pensions. If you are interested in the idea of applying mathematics

to the financial industry, check out this field!

148

No. 293

Strungaru, from Grant MacEwan University in Edmonton, has agreed to take over

from Robert Woodrow who has been the editor of the Corner since 1987. Robert’s

dedication to CRUX with MAYHEM over the years is greatly appreciated and

he will be sorely missed. Material from Robert will continue to appear in CRUX

with MAYHEM as we wrap up the solutions to the last sets of problems he

published.

The format of the Corner is changing slightly. It will still consist of problems

from Olympiads from around the world, but, rather than printing the contests in

their entirety, each column will consist of 10 questions, in both English and French,

selected from different contests. The origin of the question will be revealed when

the solutions are published.

We will have the same time lines as we do with the CRUX problems.

Solutions will be due six months from the issue date and will appear in the

same issue number of the next volume, one year later. The first set of new

Olympiad Corner problems is below, please send your solutions to Nicolae by email

(preferred) at:

crux-olympiad@cms.math.ca

or by mail to

Nicolae Strungaru

Department of Mathematics and Statistics

Grant MacEwan University

Edmonton, AB

Canada

T5J 4S2

149

The solutions to the problems are due to the editor by 1 January 2012.

f (xy) + f (y − x) ≥ f (y + x)

for all real numbers x and y.

(a) Give a nonconstant polynomial that satisfies the condition.

(b) Prove that f (x) ≥ 0 for all real x.

∠CBD = 2∠ADB,

∠ABD = 2∠CDB

and AB = CB.

Prove that AD = CD.

OC4. Consider 70-digit numbers n, with the property that each of the digits

1, 2, 3, . . . , 7 appears in the decimal expansion of n ten times (and 8, 9 and 0 do

not appear). Show that no number of this form can divide another number of this

form.

a1 ≥ a2 ≥ · · · ≥ a100 ≥ 0,

a1 + a2 ≤ 100

and a3 + a4 + · · · + a100 ≤ 100.

Determine the maximum possible value of a21 + a22 + · · · + a2100 , and find all

possible sequences a1 , a2 , . . . , a100 which achieve this maximum.

OC6. In the diagram, ABCD is a square, with U and V interior points of the

sides AB and CD respectively. Determine all the possible ways of selecting U

and V so as to maximize the area of the quadrilateral P U QV .

A D

P

U

V

Q

B C

150

1 1 1 1 1 1 1

1+ + ··· + > + + ··· + .

n+1 3 2n − 1 n 2 4 2n

OC8. For each real number r let Tr be the transformation of the plane that

takes the point (x, y) into the point (2r x, r2r x + 2r y). Let F be the family of

all such transformations i.e. F = {Tr : r ∈ R}. Find all curves y = f (x) whose

graphs remain unchanged by every transformation in F .

OC9. A deck of 2n + 1 cards consists of a joker and, for each number between

1 and n inclusive, two cards marked with that number. The 2n + 1 cards are

placed in a row, with the joker in the middle. For each k with 1 ≤ k ≤ n, the

two cards numbered k have exactly k − 1 cards between them. Determine all the

values of n not exceeding 10 for which this arrangement is possible. For which

values of n is it impossible?

OC10. The number 1987 can be written as a three digit number xyz in some

base b. If x + y + z = 1 + 9 + 8 + 7, determine all possible values of x, y, z, b.

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

w! = x! + y! + z!.

OC2. Supposer que f est une fonction à valeurs réelles qui satisfait

f (xy) + f (y − x) ≥ f (y + x)

∠CBD = 2∠ADB,

∠ABD = 2∠CDB

et AB = CB.

OC4. Considérer les nombres n à 70 chiffres avec la propriété que chacun des

chiffres 1, 2, 3, . . . , 7 apparaı̂t dix fois dans l’expansion décimale de n (et que 8,

9 et 0 n’y apparaissent pas). Montrer qu’aucun nombre de cette forme ne peut

être divisé par un autre nombre de la même forme.

151

OC5. Supposons que les nombres réels a1, a2 , . . . , a100 satisfont aux conditions

suivantes

a1 ≥ a2 ≥ · · · ≥ a100 ≥ 0,

a1 + a2 ≤ 100

et a3 + a4 + · · · + a100 ≤ 100.

les suites possibles a1 , a2 , . . . , a100 pour lesquelles ce maximum est atteint.

OC6. Sur le diagramme ci-dessous, ABCD est un carré sur lequel on choisit des

points U et V intérieurs aux côtés AB et CD respectivement. Déterminer toutes

les façons possibles de choisir U et V de telle sorte que la surface du quadrilatère

P U QV soit maximale.

A D

P

U

V

Q

B C

1 1 1 1 1 1 1

1+ + ··· + > + + ··· + .

n+1 3 2n − 1 n 2 4 2n

le point (x, y) en le point (2r x, r2r x+2r y). Trouver toutes les courbes y = f (x)

dont le graphe est invariant pour chacune des transformations de F .

OC9. Un jeu de 2n+1 cartes contient un joker et, pour chaque nombre entier de

1 à n inclusivement, 2 cartes marquées de ce numéro. Les 2n + 1 cartes sont alors

alignées avec le joker au milieu. De plus, pour chaque nombre entier k avec 1 ≤

k ≤ n, les deux cartes numérotées k ont exactement k−1 autres cartes entre elles.

Trouver toutes les valeurs de n ne dépassant pas 10 pour lesquelles cet arrangement

soit possible. Maintenant, pour quelles valeurs de n est-ce impossible ?

OC10. Le nombre 1987 s’écrit à trois chiffres, xyz, dans une certaine base b.

Si x + y + z = 1 + 9 + 8 + 7, déterminer toutes les valeurs possible de x, y, z

et b.

152

First we look at solutions from the files to the 20th Korean Mathematical

Olympiad, given at [2010: 152–153].

Γ0 has centre O 0 , is tangent to O at A and to the side BC at D, and intersects

the lines AB and AC again at E and F , respectively. The lines OO 0 and EO 0

intersect Γ0 again at A0 and G, respectively. The lines BO and A0 G intersect at

H. Prove that DF 2 = AF · GH.

T A

Γ

0

Γ

E O 0

G

B O

H

D F

A0

C

Since O E = O G and O A = O A , the quadrilateral AEA0 G is a

0 0 0 0 0

parallelogram, hence

A0 GkAE (1)

The triangles O 0 AE and OAB are isosceles. It follows that

hence

EO 0 kBO. (2)

By (1) and (2) we deduce that BHGE is a parallelogram; thus HG = BE

and we have to prove that

DF 2 = AF · BE (3)

AD is a diameter of Γ0 , BE = CF and DF ⊥ AC. It is easy to see that, in

4ADC with DF ⊥ AC, the equation (3) is true.

We may assume that b > c, and we denote by T the intersection of the line

BC with the tangent to Γ at A.

Using the power of point T with respect to Γ, we obtain

T A2 = T B · T C ⇔ T A2 = T B(T B + a),

153

and applying the Law of Cosines in 4ABT (with ∠ABT = 180◦ − B), we have

T A2 = T B 2 + AB 2 − 2T B · AB · cos ∠ABT

⇔ T B 2 + aT B = T B 2 + c2 + 2c · T B · cos B,

hence

c2

TB = .

a − 2c cos B

Since O 0 A = O 0 D, then T D = T A and, using again the Law of Cosines,

we deduce that

c2 c2 + a2 − 2ac cos B b2

TB + a = +a = = ,

a − 2c cos B a − 2c cos B a − 2c cos B

bc c2 c(b − c)

BD = T D − T B = − = .

1 − 2c cos B a − 2c cos B a − 2c cos B

b−c a(b − c) a(b − c) a

Denoting α = , we have α = 2 = 2 = ;

a − 2c cos B a − 2ac cos B b − c2 b+c

it results that BD = cα, DC = bα.

Using the power of points B and C with respect to circle Γ0 , we get:

BE · BA = BD 2 , CF · CA = CD 2 ,

hence BE = cα2 , CF = bα2 , AF = b(1 − α2 ).

The equality (3) is equivalent to:

DF 2 = AF · BE ⇔ DC 2 + CF 2 − 2DC · CF · cos C = AF · BE

⇔ b2 α2 + b2 α4 − 2b2 α3 cos C = bcα2 (1 − α2 )

⇔ b + bα2 − 2bα cos C = c(1 − α2 )

a2 a

⇔ b − c + (b + c) · − 2b · cos c = 0

(b + c)2 b+c

⇔ b2 − c2 + a2 − 2ab cos c = 0.

which is true (by the Law of Cosines in 4ABC).

Solved by Michel Bataille, Rouen, France; Prithwijit De, Homi Bhabha Centre

for Science Education, Mumbai, India; and Konstantine Zelator, University of

Pittsburgh, Pittsburgh, PA, USA. We give the solution of De.

Observe that z is odd and is at least 3. Let z = 2m + 1 where m is a

positive integer. Then the equation reduces to

4x−1 + 4y−1 = m(m + 1) (1)

Assume that x ≥ y and rewrite (1) as

4y−1 (4x−y + 1) = m(m + 1) (2)

Observe that gcd(m, m + 1) = 1. Therefore either

154

(b) m + 1 = 4y−1 , m = 4x−y + 1.

obtain

22y−3 − 22x−2y−1 = 1 (3)

The solution of (3) is (x, y) = ( 25 , 2) which is inadmissible because x is not an

integer.

Hence the solution set in positive integers of this equation is

Suppose that (p, q) is such a pair of primes. Then,

8 9

< pp + q q + 1 = kpq, =

for some positive integer k (1)

: ;

p and q are primes.

Since equation (1) is symmetric with respect to p and q; and since p 6= q (by

inspection, p = q would imply p | 1). There is no loss of generality in assuming

p < q. (2)

We distinguish between two cases: Case 1, in which p and q are both odd primes;

and Case 2 wherein, p = 2 and q is an odd prime.

Case 1. p and q are both odd primes.

Thus, by (2) we must have

3 ≤ p < p + 2 ≤ q, 5 ≤ q. (3)

We make use of the concept of the order of a positive integer a modulo an

odd prime r. If a and r are relatively prime, then the order of a modulo r is the

smallest positive integer n such that an ≡ 1 (mod r). When a ≡ 1 (mod r),

the order of a is equal to 1. Otherwise, it is some positive integer. The order of

a exists, since by Fermat’s Little Theorem, we know that ar−1 ≡ 1 (mod r).

(Thus the set of all natural numbers n such that an ≡ 1 (mod r) is nonempty).

The following lemma is well-known in elementary number theory, we state

it without proof.

Lemma 1. Let r be an odd prime, and a a positive integer not divisible by r,

and let n be the order of a modulo r. Then, if m is a positive integer such that

am ≡ 1 (mod r), n is a divisor of m.

155

q q ≡ −1 (mod p)

⇒ q 2q ≡ (−1)2 ≡ 1 (mod p) . (4)

divisor of 2q, which means that n = 1, 2, q, or 2q.

If n = 1; then q ≡ 1 (mod p); q = 1 + pl, for some positive integer l,

and going back to (1) we have

pp + (1 + p · l)q + 1 = k · p · q (5)

2 + λp = kpq, for some positive integer λ, which is impossible since this last

equation implies p | 2; we know that p ≥ 3.

Next, consider the case in which the order n (of q modulo p) is q or 2q. We

know from Fermat’s Little Theorem that

q p−1 ≡ 1 (mod p) .

and is q odd; we see that in either case 2q must divide p: therefore

p = 2qt + 1 > q,

There remains only one possibility to consider: the order n (of q modulo p)

is equal to 2.

⇔ q ≡ ±1 (mod p) ( since p is prime). (6)

The case q ≡ 1 (mod p) has already been examined above (this was done in the

case order n = 1). So, then suppose that q ≡ −1 (mod p),

q = p · v − 1, v ∈ Z, v ≥ 2 (7)

2p. Thus, f = 1, 2, p, or 2p. Once again, by Fermat’s Little Theorem, we know

that the order f must divide q − 1 by virtue of pq−1 ≡ 1 (mod q). Hence,

q−1 =f ·u

q = f · u + 1, where f = 2, p, or 2p. (8)

156

which implies (since both p and q are positive and ≥ 3) that p > q; contrary to

(3).

If f = p or 2p, then combining (7) with (8) yields p · v − f · u = 2, which

implies (since f = p or 2p) that p divides 2; an impossibility since p ≥ 3.

Finally suppose that f = 2. Then,

p2 ≡ 1 (mod q) ⇔ (p − 1)(p + 1) ≡ 0 (mod q) ;

and since q is a prime, we must have either p = 1 + q · w or p = −1 + q · w for

some positive integer w which again contradicts the conditions in (3); for either

possibility implies p > q (note that in either case, w ≥ 2). It is now clear that

there are no odd primes p and q which satisfy (1).

Case 2. p = 2 and q is an odd prime.

From (1) we have, 22 + q q + 1 = 2kq;

5 = q · (2k − q q−1 ). (9)

Equation (9) clearly shows that q | 5; and since q is a prime; we must have q = 5

and 2k − q q−1 = 1 so 2k = 54 + 1, thus k = 626 2

= 313.

Conclusion: Taking into account symmetry, there exist exactly two pairs

with the problem’s property: (p, q) = (2, 5), (5, 2).

5. For the vertex A of 4ABC, let A0 be the point of intersection of the angle

bisector at A with side BC, and let `A be the distance between the feet of the

perpendiculars from A0 to the lines AB and C, respectively. Define `B and `C

similarly, and let ` be the perimeter of 4ABC. Prove that

`A `B `C 1

≤ .

`3 64

Solved by Arkady Alt, San Jose, CA, USA; Michel Bataille, Rouen, France;

Prithwijit De, Homi Bhabha Centre for Science Education, Mumbai, India;

Geoffrey A. Kandall, Hamden, CT, USA; and Titu Zvonaru, Cománeşti,

Romania. We give Bataille’s solution.

We adopt the standard notations for the elements of ∆ABC and denote

the orthogonal projections of A0 onto AB, AC by H, K, respectively. Since the

line segment HK is a chord subtending ∠BAC of the circle with diameter AA0 ,

we have `A = HK = AA0 sin A. As it is well-known, the length of the bisector

2bc cos(A/2)

is given by AA0 = so we obtain

b+c

2bc sin(A/2) 2 2bc sin(A/2) b2 + c2 − a2

`A = · 2 cos (A/2) = · 1+ .

b+c b+c 2bc

It quickly follows that

`A 2(s − a) sin(A/2)

= .

` b+c

157

3

= .

` (b + c)(c + a)(a + b)

È r

rs = s(s − a)(s − b)(s − c), sin(A/2) sin(B/2) sin(C/2) = ,

4R

abc = 4rRs, and ab + bc + ca = s2 + r 2 + 4rR

we first deduce

= 2s(s2 + r 2 + 4rR) − 4rRs

= 2s(s2 + r 2 + 2rR)

and then

`A `B `C r3

= (1)

`3 Rs2 + Rr 2 + 2rR2

By AM-GM, we have

s (s − a) + (s − b) + (s − c) È √

3

= ≥ 3

(s − a)(s − b)(s − c) = r2s

3 3

so that s2 ≥ 27r 2. Recalling Euler’s inequality R ≥ 2r, we obtain

`A `B `C 1

and from (1), ≤ .

`3 64

given at [2010: 153].

1. Find four prime numbers less than 100 which are factors of 332 − 232 .

Solved by Arkady Alt, San Jose, CA, USA; Geoffrey A. Kandall, Hamden, CT,

USA; Henry Ricardo, Tappan, NY, USA; Edward T.H. Wang, Wilfrid Laurier

University, Waterloo, ON; Konstantine Zelator, University of Pittsburgh, Pitts-

burgh, PA, USA; and Titu Zvonaru, Cománeşti, Romania. We give Ricardo’s

write-up.

Using the familiar ‘difference of squares’ identity repeatedly, we can write

Y

4

k k

32 32

3 −2 =5 (32 + 22 ).

k=1

158

k k

For k ≤ 2, the factors 32 + 22 are less than 100 and it is easy to pick out 5,

13 (when k = 1), and 97 (when k = 2) as prime factors. Now we observe that

332 = 916 ≡ 1 (mod 17) and 232 = 416 ≡ 1 (mod 17) by Fermat’s Little

Theorem. Thus 332 − 232 ≡ 0 (mod 17) and 17 is the fourth prime factor we

seek.

that AM = M N = N B, and points P , Q lie on the side CD such that

CP = P Q = QD. Prove that

1

Area of AM CP = Area of M N P Q = Area of ABCD .

3

A. Kandall, Hamden, CT, USA; by Konstantine Zelator, University of Pittsburgh,

Pittsburgh, PA, USA; and by Titu Zvonaru, Cománeşti, Romania. We give the

solution of Amengual Covas.

A A

D D

M M

Q Q

N N

P P

B C B C

Figure 1 Figure 2

Since AM = M N , we have

Hence,

that is,

Area of AM CP = Area of M N P Q

Now, since the areas of triangles with equal altitudes are proportional to the

bases of the triangles, we have (figure 2)

1

Area of 4AM C = (Area of 4ABC)

3

159

and

1

Area of 4CP A = (Area of 4CDA)

3

Hence,

1

Area of 4AM C+Area of 4CP A = (Area of 4ABC + Area of 4CDA)

3

that is,

1

Area of AM CP = (Area of ABCD)

3

and we are done.

each of the digits 1 to 9 exactly once. It also has the property that the digits 1

to 5 occur in their natural order, while the digits 1 to 6 do not. How many such

numbers are there?

x1 x2 x3 x4 x5 x6 x7 x8 x9

1 2 3 4 5 6 7 8 9

Note that there are exactly 9! = 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 · 9 nine-digit numbers

with distinct nonzero digits.

Let S be the set of all nine-digit numbers with distinct nonzero digits and

such that the digits 1 to 5 occur in their natural order, and m = n(S) =

cardinality of the set S.

1 2 3 4 5

1 2 3 4 5 6 7 8 9

1 2 3 4 5 6

1 2 3 4 5 6 7 8 9

Let S1 be the set of all nine-digit numbers with distinct nonzero digits and such

that the digits 1 to 6 occur in their natural order, and m1 = n(S1 ) = cardinality

of the set S1 .

Let S2 be the set of all nine-digit numbers with distinct nonzero digits and

such that the digits 1 to 5 occur in their natural order; but the (numbers) digits

1 to 6 do not occur in their natural order; and m2 = n(S2 ) = cardinality of the

set S2 .

Then S = S1 ∪ S2 and S1 ∩ S2 = ∅. Therefore,

m = m1 + m2 ;

m2 = m − m1 . (1)

160

chosen; for any such choice, the numbers 1 to 5 are placed in their natural order

on those positions. Moreover, for each such choice of five positions; there are 4!

ways to place the remaining numbers 6 to 9, on the remaining four positions.

Hence,

9 9! 9!

m = (4!) · = 4! = = 6 · 7 · 8 · 9.

5 4!5! 5!

Similarly,

9 9! 9!

m1 = (3!) = 3! = = 7 · 8 · 9.

6 3!6! 6!

Hence by (1)

m2 = m − m1 = 6·7·8·9−7·8·9

= (7 · 8 · 9)(6 − 1)

= 7 · 8 · 9 · 5 = 2520

and T at B. Let AP be a diameter of S and let the tangent from P to T touch

it at Q. Show that AP = P Q.

Kandall, Hamden, CT, USA; Konstantine Zelator, University of Pittsburgh,

Pittsburgh, PA, USA; and Titu Zvonaru, Cománeşti, Romania. We give the two

solutions of Amengual Covas.

Solution 1. 0

Let O and O be the centers of S and

T , respectively. P

Then the line OO 0 , joining the

S

centers of the touching circles, goes

through the point of contact D.

Now, P A is perpendicular to AB, Q

as is the radius O 0 B to the point of O

contact with AB. Thus P A and D T

0

O B are parallel and the alternate O0

angles P OD and BO 0 D are equal.

But triangles P OD and BO 0 D are

isosceles, and since their vertical A B

angles are equal, so are their base

angles. Therefore ∠ODP = ∠O 0 DB, and D lies on P B.

Next, diameter P A subtends a right angle at D, making AD the altitude

to the hypotenuse in right-triangle ADP . By a standard mean proportion, then,

we have

P A2 = P D · P B (1)

161

hence,

P Q2 = P D · P B (2)

By (1) and (2), P Q2 = AP 2 . It follows that P Q = AP , as desired.

Solution 2.

Let R and r be the radii of circles

S and T , respectively. Let O 0 be the P

center of T and denote by C the foot

of the perpendicular from O 0 to AP . S

By the Pythagorean theorem,

applied to right triangles P QO 0 and 2R − r

Q

P CO 0 ,

r T

P Q2 + QO 0 2 = P O0 2

= P C 2 + CO 0 2 C O0

r

that is,

A B

P Q2 + QO 0 2 = P C 2 + AB 2

√

and since AB = 2 Rr (for a proof, see e.g. Japanese Temple Geometry

Problems, by H. Fukagawa and D. Pedoe, Canada, 1989, Example 1.1 on p. 3),

P C = 2R − r, QO 0 = r, we have

2

P Q2 + r 2 = (2R − r) + 4Rr .

Hence

P Q2 = 4R2

= (2R)2

= AP 2

It follows that P Q = AP , as desired.

USA; Giulio Loddi, High School student, Cagliari, Italy; Henry Ricardo, Tappan,

NY, USA; Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA;

and Titu Zvonaru, Cománeşti, Romania. We give Loddi’s solution.

We will treat a and b like constants and the right-hand side as a function

of c:

= [(a + b)2 − c2 ] · [c2 − (a − b)2 ]

= −c4 + c2 [(a − b)2 + (a + b)2 ] − (a − b)2 (a + b)2

= −c4 + 2c2 (a2 + b2 ) − (a2 − b2 )2

162

f 0 (c) = −4c3 + 4c(a2 + b2 ) = 4c[−c2 + a2 + b2 ].

zero when c = 0 or when c2 = a2 + b2 .

The derivative of f (c) is √

2 2 2 2 2 2 2 2 2 2 2 2 2

f (0) = −(a√ − b ) and f ( a + b ) = (a + b ) − (a − b ) 2= 4a 2b , so

f (0) < f ( a + b ). We can guess that there is a maximum when c = a +b2 .

2 2

But

p

f ( a2 + b2 ) = 4a2 b2 ≥ −c4 + 2c2 (a2 + b2 ) − (a2 − b2 )2 = f (c)

when c4 −2c2 (a2 +b2 )+(a2 −b2 )2 +4a2 b2 ≥ 0. By computing the discriminant

of this quadratic (in c2 ):

4 = 4(a2 + b2 )2 − 4(a2 − b2 )2 − 16a2b2 = 16a2 b2 − 16a2 b2 = 0,

√

so f ( a2 + b2 ) − f (c) ≥ 0 for all c. Finally, (a2 + b2 )2 ≥ 4a2 b2 follows from

(a2 − b2 )2 ≥ 0 and thus

p

LHS ≥ 4a2 b2 = f ( a2 + b2 ) ≥ f (c) = RHS ∀c > 0.

2

Ed. – Note that f (c) = − c2 − (a2 + b2 ) + 4a2 b2 which yields the same

result.

√

6. Let n be an integer. Show that, if 2 + 2 1 + 12n2 is an integer, then it is a

perfect square.

Solved by Arkady Alt, San Jose, CA, USA; Michel Bataille, Rouen, France; Henry

Ricardo, Tappan, NY, USA; and Edward T.H. Wang, Wilfrid Laurier University,

Waterloo, ON. We use Bataille’s write-up.

√

Suppose that 2 + 2 1 + 12n2 is a positive integer m. Then

(m − 2)2 = 4(1 + 12n2 ) so that 1 + 12n2 must be a perfect square, say

1 + 12n2 = a2 where a is a positive integer. It follows that

a2 − 3(2n)2 = 1 (1)

and the pair (a, 2n) is a solution to the Fermat equation x2 −3y 2 = 1 with x ≥ 1

and y even. It is well-known that the solutions √ to this equation

√ k in nonnegative

integers are the√

pairs (xk , yk ) such that

√ x k +y k √3 = (2+ 3) , k = 0, 1, 2, . . ..

Since xk+1 + 3yk+1 = (xk + yk 3)(2 + 3) the sequences (xk ), (yk ) are

given by the recursion xk+1 = 2xk + 3yk , yk+1 = xk + 2yk and x0 = 1,

y0 = 0. Using induction, it √ is easy to see that

√ yk is even if and only if k is even.

Note also that 2xk = (2 + 3)k + (2 − 3)k .

Returning to (1) and assuming that n ≥ 0 without lost of generality, we

must have a√= xk and 2n = √ yk for some even k. Setting k = 2`, we first deduce

2a = (2 + 3)2` + (2 − 3)2` and then

√ √ √ √

m = 2 + 2a = (2 + 3)2` + (2 − 3)2` + 2(2 + 3)` (2 − 3)`

√ √ 2

= (2 + 3)` + (2 − 3)` = x2` ,

a perfect square.

163

Olympiad, Round 2, given at [2010: 154].

bisector of ∠BAC meets BC at D. Given that AB = CD, determine AB and

BC.

USA; Konstantine Zelator, University of Pittsburgh, Pittsburgh, PA, USA; and

Titu Zvonaru, Cománeşti, Romania. We use Kandall’s version.

B

a−c

a

D

c

c

2007 C

A

Let BC = a, AB = c, so that BD = a − c. Note that 2007 = 32 · 223

(prime factorization).

Since AD is an angle bisector, we have

a−c c

= . (1)

c 2007

a positive integer). From (1),

c2

a = + c = 223(k2 + 3k).

2007

If k = 1, then c = 3 · 223, a = 4 · 223, so c + a = 7 · 223 < 2007, which

violates the triangle inequality.

Therefore, k = 2, which means that c = 6·223 = 1338 and a = 10·223 =

2230.

164

2. Show that there are infinitely many pairs of positive integers (m, n) such that

m+1 n+1

+

n m

is a positive integer.

Solved by Prithwijit De, Homi Bhabha Centre for Science Education, Mumbai,

India.

Let f (m, n) = m+1

n

+ n+1

m

. Observe that f (1, 1) = 4.

We claim that there are infinitely many pairs of positive integers (m, n)

such that f (m, n) = 4.

f (m, n) = 4 implies m2 + (1 − 4n)m + n2 + n = 0. Viewing this as a

quadratic in m and solving we get

√

(4n − 1) ± 12n2 − 12n + 1

m = . (1)

2

Observe that t = 12n2 − 12n + 1 is odd and m is an integer if and only if t is

a perfect square. So let t = p2 , for some positive integer p, then

p2 = 3q 2 − 2 (2)

Equation (2) is satisfied by (p1 , q1 ) = (1, 1) and if the positive integral pair

(pk , qk ) satisfies (2) then so does (pk+1 , qk+1 ) where

qk+1 = pk + 2qk .

Observe that {pk } and {qk } are increasing sequences and pk > qk for k > 1.

Now define

qk + 1

nk = ,

2

È

(2nk − 1) + 12n2k − 12nk + 1 2qk + 1 + pk qk+1 + 1

mk = = =

2 2 2

for k ≥ 1. Observe that both mk and nk are positive integers as qk and qk+1

are odd positive integers.

The set S = {(mk , nk ) : k ≥ 1} is an infinite set (because {qk } is an

increasing sequence) and consists of pairs of positive integers satisfying

f (m, n) = 4.

Thus we have produced infinitely many pairs of positive integers (m, n) for which

m+1 n+1

+

n m

is a positive integer.

165

Denote the circumcentre by O and the orthocentre by H and let OH meet AB

at P and AC at Q. Prove that P O = HQ.

Note: The circumcentre of triangle ABC is the centre of the circle which passes

through the vertices A, B and C. The orthocentre is the point of intersection of

the perpendiculars from each vertex to the opposite side.

Romania. We give Bataille’s solution.

O ` Q

P H

C

A0

hence 2C > B + C = 180◦ − A = 120◦ and so C > 60◦ . Also, since

∆ABC is acute-angled, AH = 2OA0 where A0 is the midpoint of BC, that

is AH = 2R cos A = R (denoting the circumcentre by H). It follows that

∆OAH is isosceles with AO = AH. Moreover, we have ∠HAC = 90◦ − C

and ∠P AO = 21 (180◦ − ∠BOA) = 12 (180◦ − 2C) = 90◦ − C as well. It

follows that the angle bisectors of ∠BAC and ∠OAH are the same line `. Now,

if ρ` denotes the reflection in `, the image ρ` (OH) of the line OH is OH itself

(since OH ⊥ `) and the image ρ` (AB) is AC. As a result, the image of P ,

the intersection of OH and AB is Q, the intersection of OH and AC. Finally,

ρ` (P ) = Q, ρ` (O) = H and so P O = QH.

166

Next we move to the May 2010 number of the Corner and solutions from

our readers to problems of the XV Olympı́ada Matemática Rioplatense, Nivel 2,

given at [2010; 214].

1. Let ABC be a right triangle with right angle at A. Consider all the isosceles

triangles XY Z with right angle at X, where X lies on the segment BC, Y lies

on AB, and Z is on the segment AC. Determine the locus of the medians of the

hypotenuses Y Z of such triangles XY Z.

Solved by Oliver Geupel, Brühl, NRW, Germany; and Konstantine Zelator, Uni-

versity of Pittsburgh, Pittsburgh, PA, USA. We give the solution of Geupel.

Z

E M

Y F

B C

D

Let AEDF be the square where D, E, and F are on the segments BC,

AB, and AC, respectively. We prove that D = X and that the locus of the

midpoints M of segments Y Z is the line segment EF . The end point E and

accordingly F is included if and only if AB ≥ AC and AB ≤ AC, respectively.

90◦ , the quadrilateral AY XZ is cyclic. From the condition XY = XZ, we see

that ∠XAY = ∠XAZ = 45◦ ; hence X = D. By ∠DM Y = ∠DEY =

90◦ , the quadrilateral DM EY is cyclic. Thus, ∠DEM = ∠DY M = 45◦ .

Consequently, M is on EF .

if and only if AB ≥ AC and AB ≤ AC, respectively. Let us suppose that M 6=

E, F . The perpendicular to DM through M cuts AB and AC at Y and Z,

respectively. By ∠DM Y = ∠DEY = 90◦ , the quadrilateral DM EY is cyclic;

hence ∠DY M = ∠DEM = 45◦ . Similarly ∠DZM = 45◦ . Consequently,

XY Z is an isosceles right triangle, which completes the proof.

167

rightmost 1/3 of each interval is deleted, an interval of length 31 remains. If the

leftmost 1/3 of each interval is deleted, an interval of length 23 remains. Let

M and m be the maximum and minimum of the lengths of an interval in the

collection, respectively. How small can M − m be?

The solution is 24.

Consider the intervals [0, 33], [19, 28], and [25, 34]. If the rightmost 1/3

of each interval is deleted, then the union of the resulting intervals is the interval

[0, 31] with length 31. If the leftmost 1/3 of each interval is deleted, then the

union of the resulting intervals is [11, 34] with length 23. We have M = 33,

m = 9; therefore M − m = 24.

We prove that generally M − m ≥ 24.

Let [a, b] be the minimal closed interval that contains all the given intervals.

If the rightmost 1/3 of each interval is deleted, then an interval [a, r] of

length 31 remains. Thus,

r − a = 31.

At least one of the intervals with right end point in the interval [r, b] will be

reduced by a segment not greater than b − r. The length of such an interval is

not greater than 3(b − r), which implies that

m ≤ 3(b − r).

If the leftmost 1/3 of each interval is deleted, then an interval [`, b] of length

23 remains. Thus,

b − ` = 23.

The initial collection of intervals contains an interval with left bound a. Its left

bound after the deletion of the left 1/3 is not less than `. Hence, its length is not

less than 3(` − a), which implies that

M ≥ 3(` − a).

We conclude

168

BOOK REVIEWS

Amar Sodhi

Pythagoras’ Revenge: A Mathematical Mystery

by Arturo Sangalli

Princeton University Press, 2009

ISBN: 978-0-6910-4955-7, cloth, 188 + xviii pp. US$24.95

Reviewed by Mark Taylor, Halifax, N.S.

Dr. Jule (formerly Jules) Davidson teaches group theory and non-Euclidean

geometry at Indiana State University. He has reached the age of 34 and is

becoming increasingly bored with the routine of academic life. On top of that he

has “all but given up hope of becoming a famous mathematician”.

Presumably in an effort to complete the work for his Fields Medal before time

runs out, Jule spends his evenings visiting canyousolveit.com, a website devoted

to math puzzles. His attention is caught by the following problem:

A group of twelve baseball players put their caps in a bag. After the caps are

well shuffled, each player picks one at random. (1) Calculate the probability that

none of the players will pick up his own hat; (2) What is this probability if there

are infinitely many players in the group?

In the preface to the book, Sangalli states that one of his aims is “to reach

those who usually shun mathematics”. The question above certainly has the

potential to fulfill this aim. The problem can be explained in such a way that

readers who may not understand the mathematical meaning of “probability” or

“infinitely many” can come to believe they understand what the question asks.

Unfortunately the calculation of probabilities can be difficult to understand even

when the problem is simple to pose – the Monty Hall and the Birthday problems

are two good examples.

Jule’s answers are 0.3679 and 1/e for parts (1) and (2) respectively. A hint

to the solution is provided in appendix 1. It uses the Inclusion-Exclusion Principle.

This may have the effect of damaging some readers for life or at least cause them to

shun the appendices; the latter being unfortunate because the appendices also in-

clude Infinitely Many Primes, A Simple Visual Proof of the Pythagorean Theorem

and some paragraphs on Perfect, Triangular and Square Numbers all of which

should be accessible to anyone with a Grade 10 education.

Jule’s solution to the hat problem results in an invitation to compete for the

“opportunity to help solve a 2,500 - year - old enigma”. Indiana Jule is quick to

seize the opportunity and, after passing a number of tests, joins an esoteric group

seeking the reincarnation of Pythagoras. The genesis of the Pythagorean cult and

its beliefs are explained in Chapter 8 where we also learn Pythagoras prophesied

his own reincarnation, and left instructions in a secret document which was to be

guarded through the generations until he reappeared on earth. When questioned

how the custodian of the secret document would recognize the reincarnation,

169

the secrets of Number, of whom many wonderful things will be persistently related ”.

Chapter 9 introduces Norton Thorp who at the age of 9 months spoke in

complete sentences. When Norton was barely five an incredible thing happened

to him. His guardian aunt, Therese, put him to bed and then sat down with

her erstwhile lover Morris to a meal that consisted of “an assortment of dips that

included grilled eggplant and lemon puree, a spread made from feta cheese spiced

with chili pepper and garlic, and meat cooked in tomato and red wine sauce. An

entre of burghul and potato cakes with lamb and apricot filling was followed by the

main course: swordfish baked in a lemon and paprika sauce and served on a bed of

pilaf rice”.

Just before dessert was to be served, the couple heard a piano playing. The

young Norton, who had never had a piano lesson in his life, was playing the third

movement from Mozart’s piano sonata in A major, K 331, with all the skill of a

concert pianist!

Ten years after the piano incident, Thorp is subject to another supernatural

incident. This time he writes an excerpt from The Odyssey in an ancient Greek

script.

The search for the reincarnation of Pythagoras is paralleled in the book by a

search for a copy of a manuscript by Pythagoras (no such manuscript was believed

to exist). Of course, the latter search, driven by Professor Elmer Galway of Oriel

College, Oxford, results in the unearthing of Pythagoras’ secret document. The

discovery takes place in Rome and the chapter heading is the old proverb - to

think that if the manuscript had been left in Greece the heading might have read

“All Roams lead to Rhodes”.

I should mention that Jule has a twin sister, Johanna Davidson, who has

a Ph.D. in computer science. Johanna’s purpose in the book seems to be to link

Jule to Norton Thorp, and to facilitate a discourse on randomness. Unfortunately

she also affords Sangalli the opportunity to display his inadequacies as a writer. I

shudder, or perhaps cringe is a better word, to recall his description of Johanna.

I can only assume that the material on random numbers had numbed the editor’s

brain to such an extent that the paragraphs immediately following failed to

register.

You may detect a certain lack of enthusiasm for the book on my part. Such

an observation is correct. However, I think Sangalli has produced the basis for

what could be a commercially successful screenplay. His ingenious twists would

transfer to film without difficulty and the pace could make more acceptable the

irrationalities within the plot because there would be little time for reflection.

Of course, the mathematics would have to be toned down; perhaps

reduced to the standard esoteric scribblings, with plenty of subscripts, superscripts,

multiple integrals, Greek letters and tensor products.

Yes, my advice would be to wait for the film.

170

PROBLEMS

Solutions to problems in this issue should arrive no later than 1 November 2011.

An asterisk (?) after a number indicates that a problem was proposed without a solution.

Each problem is given in English and French, the official languages of Canada. In

issues 1, 3, 5, and 7, English will precede French, and in issues 2, 4, 6, and 8, French

will precede English. In the solutions’ section, the problem will be stated in the language

of the primary featured solution.

The editor thanks Jean-Marc Terrier of the University of Montreal for translations

of the problems.

3626. Proposed by Thanos Magkos, 3rd High School of Kozani, Kozani, Greece.

Let x, y, and z be positive real numbers such that x2 + y 2 + z 2 = 3. Prove

that

1 + x2 1 + y2 1 + z2

+ + ≥ 2.

z+2 x+2 y+2

Catalunya, Barcelona, Spain.

Find all quadruples a, b, c, d of positive real numbers that are solutions to

the system of equations

a +b +c +d = 4,

1 1 1 1

+ + 12 + 12 (1 + 3abcd) = 16 .

a12 b12 c d

Let a, b, c and r be the edge-lengths and the inradius of a triangle ABC.

Find the minimum value of the expression

a2 b2 b2 c2 c2 a2

E= + + r −3 .

a+b−c b+c−a c+a−b

Find the greatest positive integer m such that 2m divides

2011(2013 −1)

2016

− 1.

3630. Proposed by Hung Pham Kim, student, Stanford University, Palo Alto,

CA, USA.

Let a, b, and c be nonnegative real numbers such that a + b + c = 3.

Prove that

ab(b + c) bc(c + a) ca(a + b)

+ + ≤ 2.

2+c 2+a 2+b

171

Let {xn } be the sequence satisfying x0 = 1, x1 = 2011, and

xn+2 = 2012xn+1 − xn for all nonnegative integer n. Prove that

(2010 + x2n+1 )(2010 + x2n+2 )

is independent of n.

Italy.

Let k be a real number such that 0 ≤ k ≤ 56. Prove that the equation

below has exactly two real solutions:

Let g1 (x) = x and for natural numbers n > 1 define gn (x) = xgn−1 (x) .

Let f : (0, 1) → R be the function defined by f (x) = gn (x), where

j k 1

1 3

1 1 1 3

n = . For example, f 3

= 3

. Here bac denotes the floor of a.

x

Determine lim f (x) or prove it does not exist.

x→0+

ABC is an isosceles triangle with AB = AC. Points X, Y and Z are on

−→ −→ −→

rays AC, BA and AC respectively with AZ > AC and AX = BY = CZ.

(a) Show that the orthogonal projection of X onto BC is the midpoint of Y Z.

(b) If BZ and Y C intersect in W , show that the triangles CY A and CW Z

have the same area.

172

Let ABC be an acute-angled triangle with circumradius R, inradius r,

semiperimeter s, and with points A0 ∈ BC, B 0 ∈ CA, and C 0 ∈ AB arranged

so that

Prove that:

(a) |BC 0 ||CA0 ||AB 0| = abc;

|AA0 | |BB 0 | |CC 0 |

(b) = tan A tan B tan C;

|BC 0 | |CA0 | |AB 0 |

Area(A0 B 0 C 0 ) 4R2

(c) = 2 − 1.

Area(ABC) s − (2R + r)2

Vietnam.

Let a, b, c, and d be nonnegative real numbers such that a + b + c + d = 2.

Prove that

ab(a2 +b2 +c2 )+bc(b2 +c2 +d2 )+cd(c2 +d2 +a2 )+da(d2 +a2 +b2 ) ≤ 2 .

Let x be a real number with |x| < 1. Determine

X

∞

x2 xn

n−1

(−1) n ln(1 − x) + x + + ··· + .

n=1

2 n

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

3626. Proposé par Thanos Magkos, 3ième -Collège de Kozanie, Kozani, Grèce.

Soit x, y et z trois nombres réels positifs tels que x2 +y 2 +z 2 = 3. Montrer

que

1 + x2 1 + y2 1 + z2

+ + ≥ 2.

z+2 x+2 y+2

Catalogne, Barcelone, Espagne.

Trouver tous les quadruplets a, b, c, d de nombres réels positifs qui sont

solutions du système d’équations

a +b +c +d = 4,

1 1 1 1

+ + + (1 + 3abcd) = 16 .

a12 b12 c12 d12

173

Soit a, b, c les longueurs des côtés d’un triangle ABC et r le rayon de son

cercle inscrit. Trouver la valeur minimale de l’expression

a2 b2 b2 c2 c2 a2

E= + + r −3 .

a+b−c b+c−a c+a−b

Trouver le plus grand entier positif m tel que 2m divise

2011(2013 −1)

2016

− 1.

3630. Proposé par Pham Kim Hung, étudiant, Université de Stanford, Palo

Alto, CA, É-U.

Soit a, b et c trois nombres réels non négatifs tels que a + b + c = 3.

Montrer que

ab(b + c) bc(c + a) ca(a + b)

+ + ≤ 2.

2+c 2+a 2+b

Soit {xn } une suite satisfaisant x0 = 1, x1 = 2011 et, pour tout entier

non négatif n, xn+2 = 2012xn+1 − xn . Montrer que

(2010 + x2n+1 )(2010 + x2n+2 )

est indépendant de n.

Noci, Italie.

Soit k un nombre réel tel que 0 ≤ k ≤ 56. Montrer que l’équation ci-dessous

possède exactement deux solutions réelles :

Soit g1 (x) = x et, pour les nombres naturels n > 1, on définit

gn (x) = xgn−1 (x) . Soit f : (0, 1) → R la fonction définie par f (x) = gn (x), où

j k 1

1 3

1

n = . Par exemple, f 13 = 1

3

3

. Ici, bac dénote la partie entière de a.

x

Trouver la limite lim f (x) ou montrer qu’elle n’existe pas.

x→0+

174

Soit ABC un triangle isocèle avec AB = AC. On choisit respectivement

−→ −→ −→

trois points X, Y et Z sur les rayons AC, BA et AC avec AZ > AC et

AX = BY = CZ.

(a) Montrer que la projection orthogonale de X sur BC est le point milieu de

Y Z.

(b) Si BZ et Y C se coupent en W , montrer que les triangles CY A et CW Z

ont la même aire.

Soit ABC un triangle acutangle, r le rayon de son cercle inscrit, R le

rayon de son cercle circonscrit, s son demi-périmètre. Soit de plus les points

A0 ∈ BC, B 0 ∈ CA et C 0 ∈ AB arrangés de telle sorte que

Montrer que :

|AA0 | |BB 0 | |CC 0 |

(b) = tan A tan B tan C ;

|BC 0 | |CA0 | |AB 0 |

Aire(A0 B 0 C 0 ) 4R2

(c) = − 1.

Aire(ABC) s2 − (2R + r)2

3636. Proposé par Pham Van Thuan, Université de Science de Hanoı̈, Hanoı̈,

Vietnam.

Soit a, b, c et d des nombres réels non négatifs tels que a + b + c + d = 2.

Montrer que

ab(a2 +b2 +c2 )+bc(b2 +c2 +d2 )+cd(c2 +d2 +a2 )+da(d2 +a2 +b2 ) ≤ 2 .

Soit x un nombre réel avec |x| < 1. Déterminer

X

∞

x2 xn

n−1

(−1) n ln(1 − x) + x + + ··· + .

n=1

2 n

175

SOLUTIONS

No problem is ever permanently closed. The editor is always pleased to

consider for publication new solutions or new insights on past problems.

3521. [2010 : 108, 111] Proposed by Dorin Mărghidanu, Colegiul Naţional “A.I.

Cuza”, Corabia, Romania.

Let x1 , x2 , . . . , xn be real numbers in the interval [e, ∞) and for each

x1 + x2 + · · · + xk

index k let ek = . Prove that

xk

Solution by Richard Eden, student, Purdue University, West Lafayette, IN, USA.

It is easy to show that f (x) = x1/x is decreasing on [e, ∞). [Ed.: f (x) =

exp(ln(x)/x) and d2 /dx2 ln(x)/x = (1 − ln(x))/x2 ≤ 0 for x ∈ [e, ∞), so

ln(x)/x decreases on the given interval and so does f (x).]

For each index k,

xk ≥ (x1 + x2 + · · · + xk ) ,

so

(x1 +x2 +···+xk )/xk

xekk = xk ≥ x1 + x2 + · · · + xk .

Therefore,

X

n

xekk ≥ (x1 ) + (x1 + x2 ) + · · · + (x1 + x2 + · · · + xn )

k=1

= nx1 + (n − 1)x2 + · · · + 2xn−1 + xn .

ARKADY ALT, San Jose, CA, USA; MICHEL BATAILLE, Rouen, France; CHIP CURTIS,

Missouri Southern State University, Joplin, MO, USA; OLIVER GEUPEL, Brühl, NRW,

Germany; PAOLO PERFETTI, Dipartimento di Matematica, Università degli studi di Tor

Vergata Roma, Rome, Italy; ALBERT STADLER, Herrliberg, Switzerland; PETER Y. WOO,

Biola University, La Mirada, CA, USA; and the proposer.

Most solvers used a similar approach to that of the featured solution, except Curtis, who

used Bernoulli’s inequality.

Geupel noted that problem 3521 is equivalent to Mathematics Magazine problem 1794 by

the same proposer.

176

3522. [2010 : 108, 111] Proposed by Dorin Mărghidanu, Colegiul Naţional “A.I.

Cuza”, Corabia, Romania.

If a, b, c, and d are positive real numbers satisfying abcd = 1, prove that

da bc 16

a cd b c ab d

1+ 1+ 1+ 1+ ≥ 2 a2 + b2 + c2 + d2 .

b c d a

The inequality is equivalent to

a b c d

cd ln 1 + + da ln 1 + + ab ln 1 + + bc ln 1 +

b c d a

16

≥ · ln 2,

a2 + b2 + c2 + d2

È

or, taking abcd = 1 into account and setting f (x) = x ln(1 + x),

a 2 2 c 2 2

2 2 2 2 1 1 b 1 1 d

(a + b + c + d ) f + f + f + f

a2 b b2 c c2 d d2 a

≥ 16 ln 2.

By the Cauchy-Schwarz Inequality, the left side L satisfies L ≥ M 2 , where

M = f (a/b) + f (b/c) + f (c/d) + f (d/a),

√

so it suffices to show that M ≥ 4 ln 2.

Now, let x1 = ln(a/b), x2 = ln(b/c), x3 = ln(c/d), x4 = ln(d/a), and

for real x, let g(x) = f (ex ) = ex/2 (ln(1 + ex ))1/2 . Then

−3/2

ex/2 (ln(1 + ex ))

g 00 (x) = · ln(1 + ex )2

4

2ex ln(1 + ex ) ex x ex

+ + 2 ln(1 + e ) − .

(ex + 1)2 ex + 1 ex + 1

x x

So g 00 (x) > 0 certainly holds, since 2 ln(1 + ex ) − exe+1 > exe+1 (as it follows

u

from ln(1 + u) > 1+u for positive u).

Thus, g is convex on R, and from Jensen’s inequality we obtain

x1 + x2 + x3 + x4

g(x1 ) + g(x2 ) + g(x3 ) + g(x4 ) ≥ 4g .

4

Since x1 + x2 + x3 + x4 = 0, we have the desired result

1/2 √

M ≥ 4 · e0 · ln(1 + e0 ) = 4 ln 2.

Also solved by Albert Stadler, Herrliberg, Switzerland; and the proposer. Two incorrect

solutions were submitted.

Each of the two incorrect submissions used Bernoulli’s inequality, but overlooked the fact

that (1 + x)r < 1 + rx for x > 0 and 0 < r < 1.

177

3527. [2010 : 171, 173] Proposed by Hung Pham Kim, student, Stanford

University, Palo Alto, CA, USA.

Let a, b, and c be nonnegative real numbers such that a + b + c = 3. Prove

that

X 3

3

75

a2 b + b2 c + ≤ .

cyclic

2 2 4

First we will prove that if a, b, c are nonnegative real numbers satisfying

a + b + c = 3, then

permutation of (a, b, c) such that x ≥ y ≥ z. Since xy ≥ xz ≥ yz, we have

≤ x · xy + y · xz + z · yz

= y(x + z)2 − xyz = y(x + z)2 − abc .

It suffices to show y(x + z)2 ≤ 4, which follows from the AM-GM Inequality:

3 3

2y + (x + z) + (x + z) 2(x + y + z)

≤ = = 8;

3 3

Now

X 3

3

75

a2 b + b2 c + ≤ ⇐⇒ rA + 3B − 12 ≤ 0 ,

cyclic

2 2 4

By the AM–GM Inequality we have r ≤ 1. Also, by Schur’s inequality,

27 − 3r

from which we obtain A + B ≤ . From (1) we have A ≤ 4 − r.

4

Finally, we have

rA + 3B − 12 = (r − 3)A + 3(A + B) − 12

27 − 3r

≤ (r − 3)(4 − r) + 3 − 12

4

−4r 2 + 19r − 15 15

= = −(r − 1) r − ≤ 0,

4 4

178

15

which holds because 0 ≤ r ≤ 1 and r − < 0.

4

Equality holds if and only if (a, b, c) = (1, 1, 1) or (a, b, c) is a permutation

of (0, 1, 2).

Also solved by MARIAN DINCĂ, Bucharest, Romania; KEE-WAI LAU, Hong Kong,

China; PAOLO PERFETTI, Dipartimento di Matematica, Università degli studi di Tor

Vergata Roma, Rome, Italy; STAN WAGON, Macalester College, St. Paul, MN, USA;

PETER Y. WOO, Biola University, La Mirada, CA, USA; and the proposer.

3528. [2010 : 171, 173] Proposed by Hiroshi Kinoshita and Katsuhiro Yokota,

Tokyo, Japan.

The incircle of triangle ABC touches the sides BC, AC, AB at the points

A0 , B 0 , C 0 , respectively. Let ρ, ra , rb , rc denote the inradii of the triangles

A0 B 0 C 0 , AB 0 C 0 , BC 0 A0 , CA0 B 0 , respectively, and let r be the inradius of the

triangle ABC. Prove that

1

r = (ρ + ra + rb + rc ) .

2

and Herzegovina; Oliver Geupel, Brühl, NRW, Germany; Salem Malikić, student,

Sarajevo College, Sarajevo, Bosnia and Herzegovina; and the proposers.

The desired equation is a consequence of two known theorems. Problem

1.1.4, page 3 of H. Fukagawa and D. Pedoe, Japanese Temple Geometry Problems,

says that if the tangents at points B 0 and C 0 of a circle Γ intersect in a point A,

then the incentre of ∆AB 0 C 0 lies on Γ. Since the simple proof is not given there,

we shall prove it here. Let the line joining A to the centre of Γ meet the circle at

P ; we shall show that P is the incentre of ∆AB 0 C 0 . Since AC 0 is tangent to Γ

at C 0 , we have ∠P C 0 A = ∠P A0 C 0 for any point A0 on the arc B 0 C 0 opposite

P . But P is the midpoint of arc B 0 C 0 so that ∠P A0 C 0 = ∠P A0 B 0 = ∠P C 0 B 0 ,

whence C 0 P bisects ∠AC 0 B 0 . Since P also lies on the bisector of ∠B 0 AC 0 , it

must be the incentre of ∆AB 0 C 0 , as claimed.

To set up the second theorem, let Ia , Ib , Ic , and I be the incentres of the

triangles AB 0 C 0 , A0 BC 0 , A0 B 0 C, and ABC, respectively. We have seen that

Ia , Ib , and Ic lie on the circumcircle of ∆A0 B 0 C 0 , which we again call Γ; note that

Γ has centre I and radius r. If da , db , and dc are the distances from I to the sides

B 0 C 0 , C 0 A0 , and A0 B 0 , respectively, then r = da + ra = db + rb = dc + rc .

Carnot’s theorem applied to ∆A0 B 0 C 0 with its circumradius r and inradius ρ

says that da + db + dc = r + ρ. (See, for example, Nathan Altshiller Court,

College Geometry, page 83.) It follows that

Also solved by ARKADY ALT, San Jose, CA, USA; MIGUEL AMENGUAL

COVAS, Cala Figuera, Mallorca, Spain; GEORGE APOSTOLOPOULOS, Messolonghi, Greece;

179

Chicago, IL, USA; RICHARD EDEN, student, Purdue University, West Lafayette, IN, USA;

JOHN G. HEUVER, Grande Prairie, AB; VÁCLAV KONEČNÝ, Big Rapids, MI, USA;

KEE-WAI LAU, Hong Kong, China; PRITHWIJIT DE, Homi Bhabha Centre for Science

Education, Mumbai, India; JOEL SCHLOSBERG, Bayside, NY, USA; ALBERT STADLER,

Herrliberg, Switzerland; PETER Y. WOO, Biola University, La Mirada, CA, USA; TITU

ZVONARU, Cománeşti, Romania; and the proposers (a second solution).

Geupel referred to the result as ”well known”, and provided the reference

http://www.artofproblemsolving.com/Forum/viewtopic.php?t=275874.

The solution on that web page is essentially the same as our featured solution. Most of

the other submissions were based on formulas equivalent to Carnot’s theorem which, applied to

∆ABC, become

r A B C

= cos A + cos B + cos C − 1 = 4 sin sin sin .

R 2 2 2

Let A be a point on a circle Γ with centre O and t be the tangent to Γ at

A. Triangle P OQ is such that P is on Γ, Q is on t, and ∠P OQ = 90◦ . Find

the envelope of the perpendicular to AP through Q as 4P OQ varies.

We shall see that the envelope is the

parabola, minus its vertex, with focus

O and directrix t. Point P can be P t

any point of Γ except for A and its

diametrically opposed point. We denote

by ` the perpendicular to AP through P

Q, and by U the point of intersection of Γ

` with the line AP . Since OP = OA,

∠OP A 6= 90◦ and so lines OP and ` O A

must intersect, say at R. Lastly, let H

`

be the projection of R onto t. From the U

figure we see that ∠OP A = ∠OAP = Q

∠AQU = ∠RQH, while ∠OQR =

90◦ − ∠QRP = ∠RP U = ∠OP A; it

follows that ∠OQR = ∠RQH. Thus, H

the right triangles ROQ and RHQ are R

congruent; so RO = RH, whence, R

must lie on the parabola P with focus

O, directrix t. As the perpendicular bisector of OH, ` is the tangent to P at

R. Conversely, let ` be the tangent to P at a point R of P distinct from its

vertex, and suppose that it meets t at Q. Let the perpendicular to ` through

A intersect Γ again at P and ` at U . We show that ∠QOP = 90◦ . As

above, let H be the projection of R onto t. Since, using directed angles here,

∠HOQ = ∠QHO = ∠QAU , we have ∠OP U = ∠OP A = ∠P AO =

∠OQR, hence 180◦ = ∠U QO+∠OQR = ∠U QO+∠OP U . Thus, P, O, Q, U

are concyclic and the claim follows since ∠P U Q = 90◦ .

180

Comment. Note that this problem offers an alternative construction for the points

and tangents of a parabola using the circle centred at the focus and tangent to the

directrix.

Consider Cartesian coordinates (x, y) with O = (0, 0) and A = (1, 0),

and let P = (cos ϕ, sin ϕ). Then for ϕ 6= 0, π, we have Q = (1, − cot ϕ), and

the perpendicular to AP has slope tan ϕ2

.

Firstly, take 0 < ϕ < π. The family of perpendiculars to AP through Q

as ϕ varies between 0 and π is given implicitly by U (x, y, ϕ) = 0, where

ϕ

U (x, y, ϕ) = (x − 1) tan − y − cot ϕ.

2

According to H. v. Mangoldt and K. Knopp, Einführung in die Höhere Mathematik,

Vol. 2, 10th ed. (S. Hirzel Verlag Leipzig, 1957) Paragraph 176, for the existence

of an envelope one must check that the partial derivatives Ux , Uy , Uϕ , Uϕx , Uϕy ,

and Uϕϕ are continuous on the domain of definition (which is easily verified here);

moreover, −Uϕϕ = sin2cos(ϕ/2)

ϕ sin(ϕ/2)

6= 0 and Ux Uϕy − Uy Uϕx = 2 cos 1

2 ϕ 6= 0,

as required (where we inserted the values of x and y that we obtain below into

the second derivatives). The theorem implies that the equation of the envelope in

terms of x and y can be found by solving simultaneously the pair of equations

Uϕ = 0, U = 0.

From

1 1

0 = Uϕ = (x − 1) · + ,

2 cos2 ϕ

2 sin2 ϕ

we solve for x: ϕ

2 cos2 2

x−1= − ,

sin2 ϕ

so that

1 ϕ

x= 1 − cot2 .

2 2

Plugging this into the equation U = 0 yields

1 ϕ

y = − cot ϕ − = − cot .

sin ϕ 2

1

Combining these results, we obtain x = 2

(1 − y 2 ), or

which is the lower branch of the parabola with focus O and directrix t.

For −π < ϕ < 0, we obtain the upper branch of the same parabola. The

desired envelope is thus the parabola with the exception of its vertex.

Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain;

GEORGE APOSTOLOPOULOS, Messolonghi, Greece; CHIP CURTIS, Missouri Southern

181

State University, Joplin, MO, USA; VÁCLAV KONEČNÝ, Big Rapids, MI, USA; ALBERT

STADLER, Herrliberg, Switzerland; and PETER Y. WOO, Biola University, La Mirada, CA,

USA. Two submissions dealt instead with a related problem.

All the correct solutions used the same argument as in our solution II, except for

solution I. Geupel’s version stood out, however, because he provided the required details (and,

consequently, was the only person other than the proposer to explicitly exclude the parabola’s

vertex from the envelope). The two submissions that went astray determined the locus of the

point U (in the notation of solution I) rather than the envelope of the lines QU . It turns out

that the locus is a right strophoid—the pedal curve of a parabola with respect to the point of

intersection of its axis and directrix (that is, the locus of the points where a tangent to the

parabola meets the perpendicular dropped to it from the pedal point A). It is the curve in the

3

accompanying figure with a vertic

## Гораздо больше, чем просто документы.

Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.

Отменить можно в любой момент.