Crux

Mathematicorum
Published by the Canadian Mathematical Society.

http://crux.math.ca/

The Back Files

The CMS is pleased to offer free access to its back file of all
issues of Crux as a service for the greater mathematical
community in Canada and beyond.

Journal title history:

➢ The first 32 issues, from Vol. 1, No. 1 (March 1975) to
Vol. 4, No.2 (February 1978) were published under the
name EUREKA.
➢ Issues from Vol. 4, No. 3 (March 1978) to Vol. 22, No.
8 (December 1996) were published under the name
Crux Mathematicorum.
➢ Issues from Vol 23., No. 1 (February 1997) to Vol. 37,
No. 8 (December 2011) were published under the
name Crux Mathematicorum with Mathematical
Mayhem.
➢ Issues since Vol. 38, No. 1 (January 2012) are
published under the name Crux Mathematicorum.
 CRUX MATHEMATICORUM January I Janvier
Volume 16 #1 1990

CONTENTS I TABLE PES MATIERES

Some Sums are not Rational Functions

ofR, r, and s Stanley Rabinowitz 1

The Olympiad Corner: No. Ill R.E. Woodrow 3

Problems 1501-1510 18

Solutions 1358-1392 20

Crux Cookies 32

Editorship of Crux Mathematicorum

Canadian Mathematical Society

Q Societe mathematique du Canada

ISSN 0705-0348 Second Class Mail Registration Number 5432

Founding Editors: Leopold Sauve, Frederick G.B. Maskell

Editors 6.W. Sands
Managing Editon G.P. Wright

GENERAL INFORMATION

Crux Mathematicorum is a problem-solving journal at the senior secondary and

university undergraduate levels for those who practise or teach mathematics.
Its purpose is primarily educational, but it serves also those who read it for
professional, cultural or recreational reasons.

Problem proposals, solutions and short notes intended for publication should
be sent to the Editor:

G.W. Sands
Department of Mathematics & Statistics
University of Calgary
Calgary, Alberta
Canada, T2N 1N4

SUBSCRIPTION INFORMATION

Crux is published monthly (except July and August). The 1989 subscription rate
for ten issues is $17.50 for members of the Canadian Mathematical Society and
$35.00 for non-members. Back issues; $3.50 each. Bound volumes with index:
volumes 1 & 2 (combined) and each of volumes 3, 4, 7, 8, 9 and 10: $10.00.
(Volumes 5 & 6 are out-of-print). All prices quoted are in Canadian dollars.
Cheques and money orders, payable to the CANADIAN MATHEMATICAL SOCIETY,
should be sent to the Managing Editor:

Graham P. Wright
Canadian Mathematical Society
577 King Edward
Ottawa, Ontario
Canada KIN 6N5

ACKNOWLEDGEMENT

The support of the Departments of Mathematics and Statistics of the University

of Calgary and Carleton University, and of the Department of Mathematics of the
University of Ottawa, is gratefully acknowledged.

© Canadian Mathematical Society, 1990

Published by the Canadian Mathematical Society

Printed at Carleton University
 - 1-

SOME SUMS ARE NOT RATIONAL FUNCTIONS

OF R, r, AND s
Stanley Rabinowitz

Let R, r, and s denote the circumradius, inradius, and semiperimeter of a

triangle with angles .A} B, and C. In problem 652 of this journal [2], W.J. Blundon
pointed out the well-known formulae
sin A = -TF

cos A = * - + J :
2rs
tan 4 -
t d i l <rl —
s2 - 4R2 - 4Rr - r»
i s
^i- -H
where the sums are cyclic over the angles of the triangle. He asked if there were
similar formulae for N s*n ^ / 2 a n d / , c o s ^/2- All the solutions received were
very complicated, Murray Klamkin [3] pointed out that Anders Eager in [1] tacitly
implied that there were no known simple R - r ~ s representations for the following
symmetric triangle functions:

X s i n 1sin f' X
£ sin 4 , CSC Tf S y CSC -jy CSC -pr

(1)
cos cos cos
X 4> X 1 f ' X A
sec -K- ,
V
> sec
5
T sec
C
T

Klamkin went on to conjecture that these sums cannot be expressed as rational

functions of R, r, and s. (A rational function is the quotient of two polynomials.)
This conjecture is made plausible by the fact that compendiums of such formulae
(such as chapter 4 of [4]) do not include values for these particular sums. In this
note, we will prove Klamkin's conjecture.
sin
THEOREM. V ^/2 an(
i Z, cos
^/2 can not
^ e expressed as rational
functions of R, r, and s.
Proof. Consider a triangle ABC with sides BC = 13, CA - 14, and AB = 15.
This triangle has area 84, semiperimeter 21, inradius 4, and circumradius 65/8.
 - 2 -

From the Law of Cosines, we can easily compute the cosines of the angles, finding
cos A = g sin A = -r
cos B = gg and sin 5 = TTF-
5 12
cos C = py sin (7 = jo •
From the half-angle formulae, we find that
. A 1 . 5 4 . ( 7 2
sm
sin Y - — ~<r = — s i n •*- = •

cos o- = — cos 7>- = — - cos -K- = — -

sec y = \$ sec | - ^ 5 sec ~ = ^TS

esc 4 = S esc •2"

5 = W^5
iv^5 esc 2-j = k/13" .

We thus see that

X . TJ-
sin A = — H1 . —4H , 2

which is irrational. This shows that N sinyl/2 cannot be a rational function of fZ,

r, and 5, for if it were, then in this particular case, its numerical value would be
rational (since Z£, r, and s are rational in this case), a contradiction. Similarly,
cos ;4/2 cannot be a rational function of Ry r, and 5, because that would be
i
contradicted by this particular case, in which

cos j - = — + — +

is also irrational. •
A similar calculation and argument shows that none of the expressions in
display (1) can be expressed as rational functions of R, r, and 5. In fact, the same
argument shows further than none of these expressions can be expressed as rational
functions of i2, r, s, a, 6, c, and if, where a, 6, and c are the lengths of the sides of
the triangle and K is its area.
In many cases, similar results can be shown using simpler examples. For
example, let raa, m\» rnc denote the lengths of the medians of a triangle. Using a
3-4-5 right triangle, I showed in [5] that there is no rational function, M, of a, 6,
and c such that each of raa, rab, mc can be expressed as rational functions of a, 6,
c, and JM.
 - 3 -

As an exercise, the reader can prove that sin x/2 and cos x/2 can not be
expressed as rational functions of sin x and cos x. It is well known that tan x/2 can
be so expressed, namely
tan | = \ l i n x • .
2 1 + cos x
References:
[1] Anders Eager, A family of goniometric inequalities, Puhlikacije Elektrotehnickog
Fakulteta Univerziteia U Beogradu, Serija: Matematika i Fizika Nos. 338-352
(1971) 5-25.
[2] W.J. Blundon, Problem 652, Crux Mathematicorum 7 (1981) 179.
[3] M.S. Klamkin, Comment on Problem 652, Crux Mathematicorum 8 (1982) 189.
[4] D.S. Mitrinovic, J.E. Pecaric, and V. Volenec, Recent Advances in Geometric
Inequalities, Kluwer Academic Publishers, Boston, 1989.
[5] Stanley Rabinowitz, On the computer solution of symmetric homogeneous
triangle inequalities, Proceedings of the ACM-SIGSAM 1989 International
Symposium on Symbolic and Algebraic Computation (ISSAC J89), 272-286.
12 Vine Brook Road
Westford, MA 01886-4212
U.S.A.

THE OLYMPIAD CORNER

No. I l l
R.E. WOODROW

All communications about this column should be sent to Professor R.E.

Woodrow, Department of Mathematics and Statistics, The University of Calgary,
Calgary, Alberta, Canada, T2N 1N4.

Another year has passed and It is time to thank the readership for its
continued Interest and participation. One of the projects for this year will be to
present the results of the challenges thrown out in 1988 to fill in the gaps with
solutions to problems posed in the Corner. The response was very good and Pm
sure some of you are beginning to wonder what happened to the solutions you sent
me. But I also want to continue with solutions to more current problems so as to
ensure that our new solvers9 efforts are recognized. Beginning next month the
 - 4 -

solutions will be split in two parts, "archival" and "current" problem solutions. I
would particularly like to thank those who have sent in problem sets, solutions, and
comments. Among those whose efforts have helped this column appear during the
year are Seung-Jin Bang, Francisco Bellot, Len Bos, Curtis Cooper, Nicos Diamantis,
George Evagelopoulos, Douglass L. Grant, the late J.T. Groenman, H.N. Gupta, R.K.
Guy, Denis Hanson, Walther Janous, Gy. Karoly, Murray S. Klamkin, Andy Liu,
Robert Lyness, Stewart Metchette, Dave McDonald, David Monk, John Morvay,
Gillian Nonay, Richard Nowakowski, J. Pataki, Bob Prielipp, M.A. Selby, Zun Shan,
Bruce Shawyer, Dimitris Vathis, and Edward T.H. Wang.
* * *

The problem sets that we present this month all come to us from Andy Liu,
The University of Alberta, who also translated the Chinese I.M.O. Selection Test
questions for us.

SINGAPORE MATHEMATICAL SOCIETY

INTERSCHOOL MATHEMATICAL COMPETITION 1988
Part B (2 hours)

[Ector's note: Part A consisted of 10 multiple-choice questions, to be answered in

one hour.]
*i k
L Let f(x) be a polynomial of degree n such that f(k) = L » i for each
k = 0,1,2,...,n. Find f{n + 1).

2. Suppose A ABC and

ADEF in the figure are congruent.
Prove that the perpendicular bisectors
of AD, BE, and CF intersect at the
same point.
3. Find all positive integers n such that Pn is divisible by 5, where
n
Pn = 1 + 2 + 3n + 4n. Justify your answer.
4. Prove that for any positive integer n, any set of n + 1 distinct
integers chosen from the integers 1,2,...,2n always contains 2 distinct integers such
that one of them is a multiple of the other.
 - 5 -

5. Find ail positive integers x, y, z satisfying the equation

5(xy + yz + zx) = 4xj/z.
*

The next two sets were forwarded to Andy Liu by Professor Zun Shan from
Hefei in Anhtii province of the People's Republic of China.

FIRST SELECTION TEST OF THE CHINESE I.M.O. TEAM

May 33 1988 (4± hours)

1. What necessary and sufficient conditions must real numbers i , B, C

satisfy in order that
A(x - y){x - z) + B(y - z)(y - x) + C(z - x)(z - y)
is non-negative for all real numbers x, y and zl

2. Determine all functions / from the rational numbers to the complex

numbers such that
(i) f(xx + x2 + ••• + xms) = f{xi)f{x2)-'f{xms)
8
for all rational numbers Xi,X2,- • ,£i988? and
(ii) 7(1988)/(x) = f(1988)f(x) for all rational numbers x, where z denotes
the complex conjugate of z.

3. In triangle ABC, angle C is 30°. D is a point on AC and E is a

point on BC such that AD = BE = 4 5 . Prove that 0 1 = D # and 0 / is
perpendicular to DE, where 0 and / are respectively the circumcentre and incentre of
triangle ABC.
4. Let k be a positive integer. Let 5* - {(a,6): a, 6 = 1,2,-••,&}. Two
elements (a,6) and (c,d) of S\ are said to be indistinguishable if and only if
a - c = 0 or ±1 mod k and b - d = 0 or ±1 mod &. Let r* denote the maximum
number of pairwise indistinguishable elements in S^.
(i) Determine r^ with justification.
(ii) Determine 7*7, with justification.
(iii) Determine r* in general. (Justification is not required.)

SECOND SELECTION TEST OF THE CHINESE I.M.O. TEAM

May 4? 1988 (4j hours)

X. Define xn = 3a;n-i + 2 for all positive integers n. Prove that an

integer value can be chosen for XQ such that 1988 divides xioo-
 - 6 -

2. ABCD is a trapezoid with AB parallel to DC. M and N are fixed

points on AB with M closer to A than N is. P is a variable point on CD. Let DN
cut i4P at E and CM at F, and let CM cut BP at G. For which point P is the
area of PEFG maximized?

3. A polygon in the x?/~plane has area greater than n. Prove that it

contains points (xi,yi)} 1 < i < n + 1 such that X{ - Xj and 2/i - yj are integers
for all i and j .

4. With numbers u and t; as input, a machine generates the number

uv + v as output. In the first operation, the only numbers that can be used for
input are 1, - 1 and a fixed number c. In later operations, numbers generated by
the machine as output in preceding operations may also be used. Prove that for any
polynomial f(x) = a$xn + a\xn~l + • • • + an with integer coefficients, the machine
can generate f(c) as output after a finite number of operations.
* * *

We now continue with solutions for problems posed in the 1988 numbers of
the Corner. The first is a solution to a problem from the January number.

1 9 8 5 , 3 [1988: 3] Kurschak Competition.

Let each vertex of a triangle be reflected across the opposite side. Prove that
the area of the triangle determined by the three points of reflection is less than 5
times the area of the original triangle.

Solution by Murray S. Klamkin, Mathematics Department, University of Alberta,

Edmonton.
Let F and F' denote the respective areas of the original triangle ABC and the
new triangle A'B'C'. We extend the problem by finding both upper and lower
bounds for F' jF for the case when ABC is non-obtuse, and also for several cases
when ABC is obtuse. The five cases to be considered are
(i). ABC is non-obtuse;
(it). ABC is obtuse with A > 90°, A - C < 90°, A - B < 90°;
(Hi). ABC is obtuse with A > 90°, A - C > 90°, A - B > 90°;
(iv). ABC is obtuse with 120° > A > 90°, A - C < 90°, A - B > 90°;
(v). ABC is obtuse with 135° > A > 120°, A - C < 90°, A - B > 90°.
As indicated by the five associated figures, these cases lead to different
representations of F' /F. Actually, this difference is one of sign only, but it affects
the upper and lower bounds.
 - 7 -

Case (i) Case (ii) Case (iii)

Case (iv) Case (v)

For all cases
hb = c sin A , hc = b sin .-.J4 ? /ia = c sin B ,
££f =. 2i2 cos B , Cff = 2R cos C7 , I F = ±2R cos 4
(according to whether A is non-obtuse or obtuse). Here Zia is the altitude to side a,
etc. ? R is the circumradius, and H is the orthocenter. Also,
F = 2iE2 Y cos B COS (7 sin 1. (1)
The summations here and subsequently are symmetric over the elements of ABC.
We also write [X\ for the area of figure X
Case (t). F* = [B'HC] + [CHA'] + [A'HB']
HB'-HC sinzB'HC'

=2 \ (h\> - R cos B)(hc - J? cos C)sm A (2)

Since
V feb^csin 4 = V ^c sinM = 2F V sin2j4

and
2# V (/iccos JB + /ibcos C)sin i4 = 2R V (sin B cos C + sin C cos 5)/i a

= 1R Y /lasin A

= Y afc„ = 6F ,

we have from (1) and (2)

 - 8 -

= 4 Sin2i4
P" X "5 •
It is a known result (see for example [1]) that for non-obtuse triangles
2 < Y sin2^ < | .
The upper bound is obtained only for the equilateral triangle and the lower bound
for right triangles. Thus for the non-obtuse case,
3 < F'/F < 4 .
Case (ii). For this case we have
2hh > 2R cos B and 2hc > 2R cos C .
To see this note that
2hh = 2c sin A = AR sin C sin A
= 2R cos(A - C) - 2R cos(^ + C)
> -2R cos(A + C) = 2iE cos 5
since cos(i4 - C) > 0. Similarly 2/ic > 2R cos (7 follows from cos(i4 - B) > 0.
Proceeding as before,
= 4 sin2A 5>
^ X "
Now ^4 > 90° and since 2^4 - B - C < 180°, A is also less than 120°. Here, due
to restrictions on the angles,
0 < F' jF < 3 .
The upper bound of 3, for example, corresponds to N sin 2 ^ < 2. To see the latter
inequality, we keep A fixed and minimize
cos 2B + cos 2C = 2 cos(7r - yl)cos((7 - 5).
This gives B = 0 and C = 7r - A These bounds are best possible since we can get
arbitrarily close to them by choosing triangles with angles (120 - 2e, 30 + e,
30° + e) for the lower bound and angles (90° + e, 2e, 90° - 3e) for the upper
bound. Note that for the case where the angles are (120°,30°,30°), F' - 0 since £ ' ,
C7, and H coincide.
Case (in). Here A - C > 90°, B - C > 90° so that J4 > 120° and
F* = [C"iL4'] + [4'tf£'] - [5'tfO] .
Proceeding as before we get the same expression except for a change of sign, i.e.
p- = 5- 4 Y sinM .
Hence
0 < F'/F < 5 .
Again these bounds are best possible since we can get arbitrarily close to them by
choosing triangles with angles (120° + 2e, 30° - e, 30° - e) for the lower bound and
 - 9 -

angles (180 - 2€,e,e) for the upper bound.

Case (iv). In this case we have A - C < 90°, A - B > 90° and
F' = [A'HB'\ - [B'HC] - [C'HA').
As before this leads to the same expression and bounds for F' /F as in case {%%). To
get arbitrarily close to the bounds we use triangles of the slightly different angles
(120°, 30° - £, 30° + e) and (90° + 2e, e, 90° - 3e) (to satisfy the angle
hypothesis).
Case (v). In this case A - B > 90° > A - C, and 3.4/2 > 180° so that
F' = [B'HC] + [C'HA'] - [A'HB'].
This leads to the same expression as in case (m), i.e.
= 5
T~ " 4 X Sin2±
We can get arbitrarily close to the lower bound F'/F > 0 by choosing angles
(120° + e, 30° - 36, 30° + 2e). The upper bound here is 1 and is obtained for
angles (135° - e, £, 45°). To see this, holding A fixed we want to maximize
cos 2B + cos 2C Here we can take
C = 90° - A/2 + x and £ = 90° - A/2 - x
where
90 A/2 > x > 3A/2 - 180°.
Since
cos 2 5 + cos 2C = 2 cos(180° - A)cos 2x ,
x must be as small as possible, i.e. x = 3A/2 - 180 . (Subsequently we will add an
e to it.) Then C = A - 90° and B = 270° - 2A. Then
N sin2A = sin2A + cos2A + cos22A

is a minimum for A = 135 .

In summary, for all triangles
0 < F'/F < 5.
Another way of proceeding is to reflect the triangle
across each side as in the figure. Although the case
corresponding to 5, C < 60°, A < 120° is rather simple
and leads easily to

P" = 4 X *in*A " 5?

there are different, less simple, cases to be considered when
A > 120°.
 - 10 -

Reference:
[1] O. Bottema et al, Geometric Inequalities, Wolters-Noordhoff, Groningen, 1969,
p. 18, #2.3.

The second is a solution to a problem from the February number.

9. [1988: 37] 10th Austrian-Polish Mathematics Competition.

Consider the set M of all points in the plane whose coordinates (x,y)
are both whole numbers that satisfy 1 < x < 12, 1 < y < 13.
(i) Show that every subset of M containing at least 49 points must
contain the four vertices of a rectangle having its sides parallel to the coordinate
axes.
(ii) Construct a counterexample to (i) if the subset is allowed to consist of
only 48 elements.

Solution by Curtis Cooper, Central Missouri State University.

(i) Let S C M be such that \S\ = 49. Let yt = \{x: (x,i) e S}\ for
i = 1,2,...,13. Let
13

T = {(tht2r--,tn): ti £ I , 0 < ti < 12 for i = 1,2,...,13 and Y U = 49}.

i= 1
It follows that (2/1,2/2, • • • ,2/13) £ T. Finally, let
Pi = {{r,s}: r t s and (r,i),(s,i) G 5}
for i = 1,2,...,13. Now

I 1^1 = I (I') >- »•» X(»') • 3-(2) + 1°-(2) = 69

i=l i=l i=l
66 wa
(see the claim below). However, there are 1 2 J = ys to
P ic k integers r and s
such that 1 < r, 5 < 12 and r ± s. Hence, there exist 3' ± k such that
{a,b} e Pj and {a,6} 6 Ph .
Thus (aj), (6j), (a,k), (b,k) e S form a rectangle having its sides parallel to the
coordinate axes, establishing (i).
1 6

Claim. min £ («*) = S - g ) + 10-(£)

1= 1
 - 11 -

= and
To see the claim, notice first that f!j) = (2) ° Ps* 1 ) " (2) =
* for

k > 0. Now assume that (zh° • -,^13) achieves the minimum. In this case each
Zi < 4, for if say z% > 5 then, as the remaining twelve terms sum to at most 44,
one must have Zj < 3 for some j . Decreasing Z( by 1 and augmenting zj by 1
would decrease the sum. Next we argue that in fact z* > 3 for all i Otherwise
suppose some Zi < 2. The remaining terms must sum to at least 47, but with the
observation above this gives z% = 2, eleven 4's and one 3, or z% - 1 and twelve 4's.
The values of the corresponding expressions are 70 and 72, respectively, which are
not the minimum. Consequently 3 < z\ < 4 for % = 1,2,...,13. The only possibility
J
Is to use three 3 s and ten 4's.

(ii) 13 • • • •
12 • • •
11 0 0 ® ®
10 • • • •
9 m m ® ®
8 • • ®
7 ® ® ®
6 @ • • ®
5 • G ® m
4 © ® ® ®
3 • • ®
2 ® • • Q

1 ® ® ® ®

1 2 3 4 5 6 7 8 9 10 11 12

The next solutions that we give are to problems posed in the March 1988
number of the Corner. The first are for the three problems from the Entrance
Examination of the Polytechnic of Athens (1962).

L [1988: 65] Polytechnic of Athens Entrance Exam.

Prove that

H)M<*<(-S)hS)
for all real x > 6. Then prove that
 - 12 -

00
1 48
2 Y ^
£=6
CD

and calculate V 72 accurate to three decimal places.

i=l
Solutions by George Evagelopoulos, Law student, Athens, Greece, and by Edward
T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario.
Noting that

is equivalent to
„2 1 ^ J2 ^ J2 , X 575
4
^ 482
the first inequality is evident, and for x > 6 the second follows since
x 575 N> 1 _ 57J
575 . n
2 ?
48 " 48 :
Taking reciprocals
1 . 1 . 1
2
(x - l/2)(x + 1/2) > x > (x- 2S/48)(x + 25/48) "
Equivalently
1 1 . 1 . 1 1
2
x - 1/2 " x + 1/2 > i > x - 23/48 " x + 25/48 '
Now, summing all three quantities over all x > 6, the outside series telescope with
00

X (1^17* " r + M = lim (if ~ TTTTJI) =

if
i =6
and
CD

=
X (* - 23/48 " t + 25/48J j£™ (sM " N + S5/48J " SB5
i =6
As each inequality is strict,
00
2 1 48
^ V ->
> 2 >
IT 2*
i =6 T 565
Now we have
5
V i2 ~ 1.4636 , J J - 0.1818 and ^ - 0.1811
i=l
 - 13 -

GO

This gives 1.6447 < V y 2 < 1.6454. Therefore, to three decimal places V ^2 is

1.645.
00

The well known exact value of A 7 2 *s IT w

^c^ a
6rees with this

approximation.

2. [1988: 65] Polytechnic of Athens Entrance Exam.

Determine those odd natural numbers n such that the common roots of
f{x) = {x + I f - 2* - 1
and
h(x) = (x + l)n~l - Xth'1
contain the roots of z2 + x + 1.

Solution by George Evagelopoulos, Law student, Athens, Greece.

The roots of the polynomial x2 + x + 1 are the complex cube roots of 1,
namely
-1 + h/5 ^A ,, - 1 - i^T
^ = v- anci ^ = v__
where
a;? = ul = 1

^1 + ^2 = "I , ^1^2 = 1 -
In order to find which odd natural numbers n have the property that uj\ and
u/2 are roots of both f(x) and h(x), first observe that
fM = (wi + l) n - <A - 1 = M ) » - d\ - 1 = (-l)M11 - otf - 1
and
* M = (^1 + i)n"! - ^r1 = M) n " ! - ^r1 = (~i)nMn~2 - wr1.
Next distinguish the following cases for values of n:
(i) n = 6* + 1, *' e M. Then
f(ut) = -o;l 2fct2 - a;!**1 - 1 = -atf -• (* - 1 = -(atf + «i + 1) = 0,
& M = a>l2k- a;?* = 1 - 1 = 0.
(ii) n = 6k + 3, A e IN. Then
/(a/i) = -a;l 2fc+6 - a>!**3 - l = - l - l - l = - 3 # 0 ,
2 4 2
A M = a/1 ** - a/?** = <* --w? # 0 •
(iii) n = 6* + 5, fc e W. Then
/(a;!) = -wPfc*10 - a/?**5 - 1 = -u>x - atf - 1 = 05
 - 14 -

h(vi) = uj\2k*s - a/!*+4 = w? - ojx # 0.

We have the same results for 0^2, and consequently the roots of x? + x + 1
are also roots of both f(x) and h(x) if and only if n = 6& + 1 for some integer fc.

3. [1988: 65] Polytechnic of Athens Entrance Exam.

Prove that the polynomial
fn(x) - x sin a - £ sin(na) + sin(n - l)a
is exactly divisible by
h(x) = x2 - 2x cos a + 1
where a is a real number and n is a natural number > 2.
Solutions by George Evagelopoulos, Law student, Athens, Greece, and by Edward
T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario.
There is an obvious misprint in the statement of the problem: the function
fn(x) should be a^sin a - x sin(na) + sin(n - l)a. By the quadratic formula the
roots of h(x) are easily found to be cos a ± i sin a. Thus to show h(x) divides fn(x)
we must show that fn(cos a ± i sin a) - 0. Now
/ n (cos a + i sin a) - (cos(raa) + i sin(na))sin a - (cos a + i sin a)sin(na)
+ (sin(na)cos a - cos(na)sin a) = 0 .
A similar calculation gives
/(cos a - i sin a) = 0.
[Editor's note: E.T.H. Wang continues with more on exact division.]
To see that h(x) divides fn(x) exactly consider
fn'(x) = naP'hin a - sin(na).
Now
/^(cos a ± i sin a) = n[cos(n - l)a ± i sin(n - l)a]sin a - sin(na) ^ 0
for n > 1, whence h(x) divides fn(x) but (h(x))2 does not.
Since f\(x) = 0, all powers of h(x) divide it.

The next solutions are from the 1986 Spanish Olympiad in Valladolid, Spain.

2. [1988: 67] 23rd Spanish Mathematical Olympiad.

Find all x, y, z (real numbers) such that
3 3
xyz = x + gj / + z* .
Solutions by Seung-Jin Bang, Seoul, Korea] George Evagelopoulos, Law student,
Athens, Greece, Bob Prielipp, University of Wisconsin, Oshkosh; M.A. Selby,
Department of Mathematics and Statistics, The University of Windsor, Ontario] and
 - 15 -

by Edward T.H. Wang, Wilfrid Laurier University, Waterloo.

From
x3 + t/3 + z" - 3xyz = i?L±-%±Jl{(z - y)2 + (y - z) 2 + (z - x)2] ,
the given equality holds if and only if x + y + z = 0 or x = t/ = z.

[Editofs note: At the bottom of the page of Bang's solution was a solution to
Kiirschak Competition problem 1983.1 [1988: 2] which should have been mentioned
among the alternate solutions found in [1989: 229]. My apologies.]

4. [1988: 67] 23rd Spanish Mathematical Olympiad.

Consider the equations
x2 + bx + c = 0 and x2 + b'z + C = 0
where b, c, b' and c' are integers such that
(6 - 6') 2 + (c - c') 2 > 0.
Show that if the equations have a common root, then the second roots are distinct
integers.
Solutions by Bob Prielipp, University of Wisconsin, Oshkosh, and by M.A.
Selby, Department of Mathematics and Statistics, The University of Windsor, Ontario.
Suppose that r is a common root, and let s and s' denote the other roots of
x2 + bx + c = 0 and x2 + b'x + C = 0, respectively. It follows that
r + s =.-6, r + s' = - 6 ' , rs = c and rs' = C
If s = a' then 6 = 6" and c = C contradicting (b - b')2 + (c - c') 2 > 0. Thus
s t s'. Thus b $ b'. Because r is a common root,
r2 + £>r + c = 0 and r2 + b'r + C = 0
so
(6 - b')r = C - c 7
and
c - c
r = ^ — ^
2
is rational. As x + 6x + c is a monic polynomial r must be an integer. It follows
that s = --6 - r and s' = -&' - r are both integers as well.

6. [1988: 67] 23rd Spanish Mathematical Olympiad.

Let ai ) a 2) ... ) a n be n distinct real numbers. Calculate those points x on
n
the real line which minimize \ \x - ai\, the sum of the distances from x to the
£=i
at, in the cases n = 3 and n = 4.
 - 16 -

Comment by Seung-Jin Bang, Seoul, Korea.

This problem is a special known case of a problem I posed on p. 95, vol. 16,
1987 of the Mathematical Chronicle.

Solution by M.A. Selby, Department of Mathematics and Statistics, The

University of Windsor, Ontario.
Assume ai < aj for i < j . Define

f(x) = ]T \x- ai\.

£=1

Then f(x) is continuous on (-ao,ao) and differentiable everywhere except at x = a*.

Clearly f(x) has an absolute minimum since f(x) -» QD as x -* w or x -» -OD. This
minimum occurs on [ai,a n ]. The only critical points are the ai, and hence the
minimum must occur at one of them (or an end point).
For n = 3,
f{a\) = | #2 - &i| + | ^3 - #i| = a2 - ai + a$ - a\ i

f(a2) - ai - a\ + a3 - ai ,
f(a$) = a3 - a\ + a% - a<i .
Now
&2 ~" Q>\ + ^3 ~ &i > &2 ~* &\ + &3 — ^2
and
a% — a\ + a% - a2 > ai - ^i + ^3 — #2 •
Thus the minimum occurs at £ = a*i and /(a2) = as - ai-
For n = 4,
/(ai) = a2 - ^i + as - ai + a4 - ai ,
/(a2) = a2 - ai + a3 - a2 + a4 - a2 ,
f(az) = a3 - a\ + a$ - a2 + a± - a3 ,
/(a4) = a4 — ai + a4 — a2 + a4 — a3 .
Now
/ ( a i ) > f(a2) = a4 - ai + a3 - a2
and
/ ( a 4 ) > /(a 3 ) = a4 - ai + a3 - a2 .
Hence the minimum occurs at x = a2 or x = as. Actually f(x) - a± - a\ + a$ - ai
for a2 < x < a3.
For n = 3,
a2 — ai + a3 — x , a\ < x < ai ,
! x - ai + a 3 - a2 , a2 < x < a3 ,
and for n = 4,
 - 17 -

( 0,2 - a\ + as + a& — 2x , a\ < x < a2 ,

/(x) = < a4 - ai + as — a2 ? a2 < £ < as ,
{ 2x - ai — a2 — as + a4 ? as < x < a4 .
We can now conclude that for n = 3 the minimum occurs at x = a2, while for n = 4
It occurs on the interval a2 < x < as.

7, [1988: 68] 23rd Spanish Mathematical Olympiad.

In a triangle ABC with opposite sides a, b, c respectively, show that if
a + b = (a tan A + 6 tan B)tan(C/2)
then the triangle Is isosceles.

Solution by Bob Prielipp, The University of Wisconsin, Oshkosh.

Let ABC be a triangle such that neither angle A nor angle B Is a right angle,
and let R be the circumradius. Then we have the following equivalent equalities:

a + b = (a tan A + b tan J5)tan(C/2) ,

tan (A+J?] = _ a A _ b e t B
?
[ 2 ) a + b a + b
sin A + sin B 2R sin <A t a n i4 2iZ sin B t a n B
cos A + cos B " 2J? sin ,4 + 2R s i n 5 + 2i{ sin 4 + 2R s i n 5
sin A tan ^4 + sin B tan f? .
J
sin A + sin 5
sin2>l + 2 sin A sin £ + sin2B = sinM + cos A sin B tan 5
+ cos B sin J4 tan A + sin 2 B ,
o • >i • D cos A sin 2 J5 . c o s 5 s i n 2 i 4
2 on 4 sm B = c o s jj + — ^ 3 ,

sin2yl cos2J3 - 2 sin A cos B cos A sin JB + cos2;4 sin 2 B = 0 ,

sin2(,4 - J3) = 0 ,

and thus A = B. It follows that triangle ABC is isosceles.

[Editofs note: This problem was also solved by George Evagelopoulos, Law student,
Athens, Greece, by a somewhat different sequence of identities.]

8. [1988: 68] 23rd Spanish Mathematical Olympiad.

In a right triangle ABC, with centroid G, consider the triangles AGB,
GBC, CGA. If the sides of ABC are all natural numbers, show that the areas of
AGB, BGC, CGA are even natural numbers.
 - 18 -

Solution by Boh Prielipp, University of Wisconsin, Oshkosh.

By hypothesis triangle ABC is a Pythagorean triangle. It suffices to establish
the required result when triangle ABC is a primitive Pythagorean triangle. In the
remainder we assume this is the case. Let a and b be the lengths of the legs of
triangle ABC where b is an even positive integer, and let c be the length of the
hypotenuse* Then there are positive integers r and s, r > s, such that a = r2 - s2,
and b ~ 2rs, where r and s are of opposite parity and are relatively prime. [See for
example, Theorem 5.1, p. 139 of An Introduction to the Theory of Numbers, 4th
Edition, by Niven and Zuckerman, John Wiley k Sons, New York, 1980.]

Lemma. The area of triangle ABC is divisible by 6.

Proof Because r and s are of opposite parity, 4 divides b. If 3 divides r or
3 divides s, then 3 divides b. Otherwise r2 = 1 mod 3 and s2 = 1 mod 3, and so
r2 = s2 mod 3. Thus 3 divides r2 - s2, and so 3 divides a. It now follows that 12
divides ah, giving the lemma. n

It is known that triangles AGB, BGC, and CGA all have the same area, L,
say [see Theorem 23, p. 47 of Modern College Geometry, by Davis, Addison-Wesley,
Reading, MA, 1949]. Then 3L = K, the area of triangle ABC But as 6 divides K,
we have that L is a positive even integer.
* * *

Olympiad Season is fast approaching. Remember to send me your national

and regional contests!
* * *

PROBLEMS
Problem proposals and solutions should be sent to the editor, B. Sands,
Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta,
Canada, T2N 1N4- Proposals should, whenever possible, be accompanied by a
solution, references, and other insights which are likely to be of help to the editor.
An asterisk (*) after a number indicates a problem submitted without a solution.
Original problems are particularly sought. But other interesting problems may
also be acceptable provided they are not too well known and references are given as to
their provenance. Ordinarily, if the originator of a problem can be located, it should
not be submitted by somebody else without his or her permission.
 - 19 -

To facilitate their consideration, your solutions, typewritten or neatly

handwritten on signed, separate sheets, should preferably he mailed to the editor before
August 1, 1990, although solutions received after that date will also be considered until
the time when a solution is published.

1 5 0 1 . Proposed by J.T. Groenman, Anthem, The Netherlands.

Two circles K and K\ touch each other externally. The equilateral
triangle ABC is inscribed in K, and points Ah Bh C\ lie on K\ such that AAh BBh
CC\ are tangent to K\. Prove that one of the lengths AAh TTBi, TJU[ equals the
sum of the other two. (The case when the circles are internally tangent was a
problem of Florow in Praxis der Mathematik 13, Heft 12, page 327.)
1 5 0 2 . Proposed by D.J. Smeenk, Zaltbommel, The Netherlands.
AB is a chord, not a diameter, of a circle with centre 0. The smaller
arc AB is divided into three equal arcs AC, CD, DB. Chord AB is also divided
into three equal segments AC', C'D', D'B. Let CC' and DD' intersect in P.
Show that zAPB = ^zAOB.
1 5 0 3 . Proposed by M.S. Klamkin, University of Alberta.
Prove that
1 + 2cos(B+C)cos(C+yl)cos(^+£) > cos 2 (S+C) + cos 2 (C+^) + cos 2 (^+B),
where A, B, C are nonnegative and A + B + C < TT.
1 5 0 4 . Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria.
Let A\A2° • -An be a circumscribable n-gon with incircle of radius 1,
and let Fi,F2,...,F n be the areas of the n corner regions inside the n-gon and outside
the incircle. Show that
1
+ + 1 >
60B , n\
T~x 7£ - n tan(7r/n) - TT
Equality holds for the regular n-gon.
1 5 0 5 . Proposed by Marcin E. Kuczma, Warszawa, Poland.
Let x\ - 1 and
xnH = y j l + xl - l) .
Show that the sequence (2nxn) converges and find its limit.

1 5 0 8 . Proposed by Jordi Dou, Barcelona, Spain.

Let A and P be points on a circle I\ Let I be a fixed line through A
but not through P, and let x be a variable line through P which cuts / at Lx and T
again at Gx. Find the locus of the circumcentre of AALXGX.
 - 20 -

1507. Proposed by Nicos D. Diamantis, student, University of Patras, Greece.

Find a real root of
y5 - 10t/3 + 20y - 12 = 0.
1508. Proposed by Edward T.H. Wang, Wilfrid Laurier University, Waterloo,
Ontario.
Let a < b < c be the lengths of the sides of a right triangle. Find
the largest constant K such that
a\b + c) + b\c + a) + c\a + b) > Kabc
holds for all right triangles and determine when equality holds. It is known that the
inequality holds when K = 6 (problem 351 of the College Math. Journal; solution on
p. 259 of Volume 20, 1989).
1509. Proposed by Carl Friedrich Sutter, Viking, Alberta.
Professor Chalkdust teaches two sections of a mathematics course, with
the same material taught in both sections. Section 1 runs on Mondays, Wednesdays,
and Fridays for 1 hour each day, and Section 2 runs on Tuesdays and Thursdays for
1.5 hours each day. Normally Professor Chalkdust covers one unit of material per
hour, but if she is teaching some material for the second time she teaches twice as
fast. The course began on a Monday. In the long run (i.e. after N weeks as
N -» QD) will one section be taught more material than the other? If so, which one,
and how much more?

1510. Proposed by Jack Garfunkel, Flushing, New York.

P is any point inside a triangle ABC Lines PA, PB, PC are drawn
and angles PAC, PBA, PCB are denoted by a, (3, j respectively. Prove or disprove
that
A R C
cot a + cot 0 + cot 7 > cot y + cot -^ + cot ^ ,
with equality when P is the incenter of A ABC.
* * *

SOLUTIONS
No problem is ever permanently closed. The editor will always be pleased to
consider for publication new solutions or new insights on past problems.

1 3 8 5 . [1988: 269] Proposed by Murray S. Klamkin, University of Alberta.

Show that the sides of the pedal triangle of any interior point P of an
equilateral triangle T are proportional to the distances from P to the corresponding
 - 21 -

vertices of T.

Solution by Wilson da Costa Areias, Rio de Janeiro, Brazil.

ACiPBu BA1PC1 and CAXPBX are
inscribed quadrilaterals in the circles whose
diameters are respectively PA, PB and PC.
Using the law of sines in AABxCi, we have
Bid _ BXA _ BXA
sin 2AC\BX = PA
sin 60 sin lAPBi
so
B\C\
TT 2
Similarly
AXCX _ ^ and A1Bl
TB~ ~ 2 TV
therefore
BxCx AxCr _ AiBt
TA~ TE~ ~ TV
Also solved by SEUNG-JIN BANG, Seoul, Republic of Korea] FRANCISCO
BELLOT ROSADO, LB. Emilio Ferrari, Valladolid, Spain; JORDI DOU, Barcelona,
Spain; JACK GARFUNKEL, Flushing, N.Y.; L.J HUT, Groningen, The Netherlands]
WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MARGIN E.
KUCZMA, Warszawa, Poland; P. PENNING, Delft, The Netherlands; BOB
PRIELIPP, University of Wisconsin-Oshkosh; D.J. SMEENK, Zaltbommel, The
Netherlands; COLLN SPRINGER, student, University of Waterloo; and the proposer.
Several readers, including the proposer, pointed out that the formulae
B\C\ = PA sin A , etc.
for the sides of a pedal triangle are known {e.g. Theorem 190 p. 136 of R.A. Johnson,
Advanced Euclidean Geometry), and that the result follows easily, as above. The
proposefs original solution was more involved, though.
* * . *

1 3 8 8 . [1988: 269] Proposed by G. Tsintsifas, Thessaloniki, Greece.

Let A\Ai* • ° An be a polygon inscribed in a circle and containing the
centre of the circle. Prove that

n - 2 + ~ < — < 71* • 7T

—sin
—sin -
7T - yr
IT n
£ =1

where at is the side AiAi*\ and at is the arc AiAi+i.

 - 22 -

Solution by Walther Janous, Ursulinengymnasium, Innsbruck, Austria.

Let
2(pi = iAiOAiH , i = l,...,n (-4n+i = M) ,
where 0 is the center of the circle. Then, with R the radius,
ai = 2R sin ^£ , a$ = 2R(pi ,
whence
o_£ _ sin y?i
i = l,...,n .
<Pi
Now it is either known or not hard to check that

is concave. (Indeed,
cos
y/(p) = y V ~ s i n V3 ?
V2
f»( n) = V s ^ n <p - 2<p cos y> + 2 sin y _. d(y)

and from
^'(vO = - <P2cos tp < 0
and rf(0) = 0 w e infer / ( p ) < 0 for 0 < tp < TT/2.) Thus
n n
1 /(v?i) n/ ,2

Xr =X
i=i 1=1
- { £=i
ft/ - = —sin -

?
Since the polygon contains 0, all <^i s are < TT/2. This and the concavity of /yield
(cf. [1], p.22)
n
X /(*»*) > /(I) +/(f) +(n-2)/(0) I + n- 2
7T
i= 1
[Editor's note: Other solvers noted that
( P i i V V , P > i ) « (5>J>0>"'#>0)
and applied the majorization inequality to obtain this last inequality.]

Reference:
[1] D.S. Mitrinovic, Analytic Inequalities, Springer-Verlag, 1970.
Also solved (in the same way) by MURRAY S. KLAMKIN, University of
Alberta] MARGIN E. KUCZMA, Warszawa, Poland; COLIN SPRINGER, student,
University of Waterloo; and the proposer.
The given bounds are best possible, as is evident from the above proof The
proposer conjectures that if the inscribed polygon is not required to contain the centre,
 - 23 -

then the best bounds are

n - 1 < V 5i < ^sln I .

a
• - *
i =i
-Any comments?
* * *

1387. [1988: 269] Proposed by Ravi Vakil, student, University of Toronto.

Show that for all positive integers d, a, n such that 3 < d < 2n*\ d
on
does not divide into a + 1.
Solution by Sverrir Thorvaldsson, student, University of Iceland, Reykjavik.

Let d > 3 be a number that divides into a + 1; we show that d > 2 n+1 .
2n 2n
Since a is a perfect square, it is clear that 4 n does not divide into a + 1.
2
Therefore d has an odd prime divisor p, and a = - 1 mod p. Let ra be the
m
smallest natural number such that a = 1 mod p. Now

a = (a J E (-1) 2 = 1 mod p ,
so ro|2n+1. If m < 2 n + 1 write m = 2 k with Jfc < n. But then
1 = (a ) • • = a = - 1 mod p ,
a contradiction since p is odd. Thus m = 2 n+1 . But according to Fermat's small
theorem, ap'1 = 1 mod p, and therefore p - 1 > 2 n+1 , i.e.
d > p > 2nn + 1 ,
which is exactly what we set out to prove. As a byproduct we also get that
p = 1 mod 2 n+1 .
Also solved by ANDREW CHOW, student, Albert Campbell C.I., Scarborough,
Ontario; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MARGIN E.
KUCZMA, Warszawa, Poland; KEE-WAI LAU, Hong Kong; COLIN SPRINGER,
student, University of Waterloo; C WILDHAGEN, Breda, The Netherlands; and the
proposer.
* * *

1388. [1988: 269] Proposed by Jordi Dou, Barcelona, Spain.

Given four lines a, 6, c, d in general position in the plane? show that
there is a unique line x cutting a, b, c, d in the respective points A, B, C, D and in
that order, such that AB = BC = CD.
 - 24 -

I. Solution by J. Chris Fisher, University of Regina.

Let Pi be the midpoint
of the segment joining 6c (short for
b fl c) to ab. Let Q\ be chosen so
that be is the midpoint of the segment
joining it to ac. Then we observe: for
any P € P\Qu if B' is the point of b
where it meets the parallel to c through
P, C' is the point of c where it meets
the parallel to b through P, and
A' = B'C n a, then A'B' = B> C.
The proof of this claim comes from the
figure using the affine coordinates
shown there.
Similarly let Qi be the midpoint of the segment joining be to cd} and let P2
be chosen so that be is the midpoint of the segment joining it to bd. Then given
any P 6 P2Q2, the corresponding statement that B' C' - C'D' follows by replacing a
by d and interchanging the roles of 6 and c in the above figure.
Finally, P\Q\ meets P2Q2 in a unique point 0 since we assume the given
configuration to be in general position (i.e. no pair of either the given or constructed
lines are parallel). Let B be the point of b where it meets the parallel to c through
0, and let C be the point of c where it meets the parallel to b through O. Then
x = BC is the line we were to construct. 1 /
II. Solution by P. Penning, Delft, The Netherlands.
Introduce the inter-
sections a n c = Af, b fl d = N,
af)b=F, bf)c=G, ct)d=H.
The envelope of the lines for which
AB = BC is the parabola Pi, tangent
to a, 6, c in P, Q, R with MF = PP,
PQ = QG, MG = GR. The envelope
of the lines for which BC = CD is the
parabola P2, tangent to 6, c, d in 5,
r, 17 with NG = GS, 6 T = rff,
TVif = #£/. Pi and P2 have common
tangents b and c. The third tangent
 - 25 -

that Pi and P2 have in common is the line we are looking for.

Also solved by L.J. HUT, Groningen, The Netherlands; MARGIN E. KUCZMA,
Warszawa, Poland; and the proposer (whose solution was the same as Penning*s).
Can someone supply the editor with an accessible reference for the uenvelopeu
property of the parabola, used in Solution II?
* * *

1389. [1988: 269] Proposed by Derek Chang, California State University, Los
Angeles, and Raymond Killgrove, Indiana State University, Terre Haute.
Find
n
max 7 \i - 7r(i)| ,
ireSn ^
i =i
where Sn is the set of all permutations of {l,2,...,n}.
I. Solution by Friend H. Kierstead Jr., Cuyahoga Falb, Ohio.
Let

r(7r) = £ \i-m •
t =l
We will show that
max T(ir) = [jj-l , (1)

where [ ] is the floor function.

Consider the permutation n - (n,n - 1,...,1). Then
T[TT) = |1 - n| + |2 - (n - 1)| + ••• + \n - l\ .
By expanding the absolute value functions and collecting like terms we obtain

(2Jn+(n-l)+--- + (-f+l)-l-2 || n even ,

= {< ;
IT) =
n*) ' ' ' , (2)
| 2 { n + ( n - l ) + . . . + ^ } + *+i - 2J1+2+.-.+ ^ } - 2+i , n odd

It is clear that this is the highest sum that can be obtained with these numbers,
since all of the larger numbers appear preceded by a plus sign whereas all of the
smaller numbers are preceded by a minus sign. [And the expansion of the sum T(TT)
for any % will always produce each of the numbers l,---,n twice, with n of these 2n
numbers preceded by a plus sign and n by a minus sign.] There are, of course,
other permutations that will produce the same sum. From (2) we obtain
 - 26 -

\5[n + 5" + X
J " Zl1 +
2j =
2 ' n even
'

These two results may be replaced by a single expression by use of the floor function
as in (1).
II. Generalization by Murray S. Klamkin} University of Alberta.
Let ai,a2,...,on and &i,&2,—A be nonincreasing sequences; ci,C2,.-.,cn any
permutation of the 6 sequence; F(t) a convex function; and
n
5
i= 1
Then
max 5 = i^ai - bn) + F(a2 - 6n-i) + ••• + F(an - &i) ,
the maximum taken over all rearrangements ci,C2,...,cn.
To prove this we use the majorization inequality [1], [2], i.e. the conditions
X\ > X2 > • • • > En, t/i > T/2 > • ' ' > yn,

h h
2^ Xi > Y yi for k = l,2,...,n - 1,
i= 1 i= 1
and
n n

I** = X **
i= l i=l
are necessary and sufficient in order that for every convex Junction F,
n n

£ F(*i) > £ Ifyi) •

1=1 1 = 1

(The above conditions are denoted by

( x i , ^ , - ^ ) > (yhy2,-.,l/n)
and we say that the left hand vector majorizes the right hand vector.)
Here it is clear that
(at - bn,a2 - 6fi-ii-B',fln - &i) >* (#i - ei,a2 ~ c 2,-"->^ - cn)
after the components of the latter vector are arranged in nonincreasing order, and the
result follows.
The given problem corresponds to the special case F(x) = |x|, and a* = bi = i
for i = 1,2,...,n. Here the maximum sum is 2m2 if n = 2m and 2m2 + 2m if
 - 27 -

n = 2m + 1.
References:
[1] D.S. Mitrinovic, Analytic Inequalities, Springer-Verlag, Heidelberg, 1970.
[2] A.W. Marshall, I. Olkin, Inequalities: Theory of Majorization and its
Applications, Academic Press, N.Y., 1979.
Also solved by WALTHER J ANGUS, Ursulinengymnasium, Innsbruck, Austria;
MARGIN E. KUCZMA, Warszawa, Poland] KEE-WAI LAU, Hong Kong] COLIN
SPRINGER,, student, University of Waterloo] EDWARD T.H. WANG, Wilfrid Laurier
University] PAUL YIU, University of Hong Kong] and the proposers. Two farther
readers sent in the correct value for the maximum, one with no proof and one with
an incorrect proof
Janous, Lau, and Wang note that the proposal has appeared earlier in stronger
form as problem E2424 of the American Mathematical Monthly (co-proposed by
Wang, in fact), with solution in Vol. 81 (1974) 668-670. More recently a related
problem also appeared in the Monthly (E3175, solution in Vol 96 (1989) 59-60).

1 3 9 0 . [1988: 269] Proposed by H. Fukagawa, Aichi, Japan.

A, B, C are points on a circle T such that CM is the perpendicular
bisector of AB. P is a point on CM and AP meets F again at D. As P varies
over segment CM, find the largest radius of the inscribed circle tangent to segments
PD, PB, and arc DB of V, in terms of the length of CM.

Solution by Marcin E. Kuczma, Warszawa, Poland.

The looked-for maximum equals TJM/A, attained when P is the midpoint of
CM. This is amazing! In fact, for any P on CM, the diameter of the incircle under
consideration is the harmonic mean of the lengths of CP and PM\ the claim hence
follows immediately.
There must exist some nice geometric proof of this fact, but I was not lucky
enough to find it. The only solution I was
able to work out was a lengthy one, via
unpleasant calculations. Here is an outline.
We let {0,r) be the center and radius
of r , (Q,y) the center and radius of the
incircle of the curved triangle PBD (BD the
arc of the circle), and put CM = h, MP = x.
Then our claim is
 - 28 -

X 1
y = j-(h - x) [- 2-harmonic mean of x and h - a?].
First, P<? bisects zBPD, hence P # | | , 4 J B and so
\PQ-x = Area(PQM) = Area(P<?J9) = j P 5 - y . (1)
Other relevant identities are
PQ 2 - 0<?2 - OP 2 = (r - t/)2 - (r + x - /i)2 (2)
and
PB2 = MP 2 + <9£2 - OM2 = x2 + r2 - (A - r) 2 . (3)
Putting (2) and (3) into the square of (1) leads, after some manipulation, to the
quadratic equation
ay2 + Py + 7 = 0
where
a = h(2r - h) , /? = 2 r ^ , 7 = z 2 ^ - x)(h - x - 2r) .
The positive root is
y=^jl
(verification by substitution); the other root is negative since a > 0 and 7 < 0. (Of
course, these roots can be found without guessing, by the standard algorithm.)

[Editor's note: The editor apologizes for an unclear statement of the problem
(not in the proposer's original formulation), which misled two other readers into not
completely answering the problem. Namely, it should have been said that point M
lies on AB. L.J. HUT, Groningen, The Netherlands, and P. PENNING, Delft, The
Netherlands, both took M to be the centre of the circle I\ Hut ended up
maximizing the radius for points P on the diameter through C but outside CM, and
his result also appears to be interesting. Penning's solution (when properly
interpreted) is correct, and may be shorter than the one above. The editor invites
Hut and Penning, and other Crux readers, to find a "nice" algebraic, or perhaps
purely geometric, solution.]

Also solved by the proposer, whose proof {algebraic) also contained the lovely
"harmonic mean" relation given by Kuczma.
The problem was taken from the 1840 Japanese mathematics book Sanpo
Senmonsho.

1 3 9 1 . [1988: 301] Proposed by G. Tsintsifas, Thessaloniki, Greece.

Let ABC be a triangle and D the point on BC so that the incircle of
AABD and the excircle (to side DC) of A ADC have the same radius p\. Define p2,
 - 29 -

pg analogously. Prove that

Pi + P2 + Pz > %r ,

where r is the inradius of A ABC.

L Solution by Emilia A. Velikova and Svetoslav J. Bilchev, Technical
University^ Russe, Bulgaria.
Let the points J, I\ and I\ be the
centres of the incircles and the excircle in the
triangles ABC and ABD as shown, and let E,
Ei be on .45 and F, P he on AC so that
IE±AB, ItEi±AB} IF±AC, and l[p±AC Then
from the similar triangles I\CP and ICF we
have
P \ _ J i P _ CF _ s - c
(1)
VP ~ VP~ = TF ~ —~
where a, 63 c, 5 are the sides and
semiperimeter of A ABC. It is well-known A
that
CP-AP-AC- b + AD+CD _ b = AD+CD-b (2)

and from (1) and (2) we get

rpi = AD + CD - b (3)
s- c 2
Further, from the similarity of triangles BIE and BI\E\ we obtain
Pi _ hEi _ 5#i _ c + BD - AD
r ~ ~TE~ ~ B~E~ ~ 2(8 - b) '
i.e.
(s - b)px _ c + BD - AD
r ' 2" (4)

Adding (3) and (4) we get

r . s - b\ a+ c- b
s - c
+ —=-\Pi = 9
from where
r(s- b)(s- c)
Pi = (5)
r2 + (s - b)(s - c)
From (5) and
r2 _ (s - a)(s - b)(s - c)

we obtain the beautiful expression

 - 30 -

where F is the area of A ABC. Analogously we get

pi = Pz a + b *
c + a
Then the inequality to be proved is equivalent to
F
[TT-c + TTi + i-Jri) >- \r >
i.e.
2(fl + b + c ) ^ + _ 1 _+ _1^) >9 .
But this Is equivalent to the well-known inequality

so the proof is complete.

II. Solution by Walther Janous, Ursulinengymnasium, Innsbruck, Austria.
We show more generally: if t $ 0 is a real number, then
( > 3(3/4)V* if t > 0 or t < -1 ,
p\ + p\ + p\\ = 4/r if t = -1 ,
( < 3(3/4)*r* if -1 < t < 0 .
[Editor's note: At this point Janous derived and used the expression
Pl =
1 + tan(B/2)tan(C/2) '
but in view of solution I it is easier to borrow the formula
Pl =
rs
27^
proved there. So adapted, Janous's argument continues ... ]
Thus
pji + pii + ,3-1 = 28-a + 2s-b + 2s-c __ | ?

which is the case t = - 1 . The rest follows by the general means inequality.
t > - 1 , t $ 0 we have M t > Af-i, i.e.
i/t
1 l 3r
3" 5 Pi 1
t =i t =i r
whence
> 3(3/4)V if * > 0 ,
3(3/4) V* if -1 < t< 0
i=l
For i < - 1 , Mt < M-i and so
 - 31 -

.i-t
3r
JS>» r
i=l
thus

] £ pf > 3(3/4)V .
i=l
Done!
i4fao solved by L.J. HUT, Groningen, The Netherlands; T. SEIMIYA, Kawasaki,
Japan; and the proposer.

1 3 9 2 . [1988: 301] Proposed by Angel Dorito, Geld, Ontario.

An immense spherical balloon is being inflated so that it constantly
touches the ground at a fixed point A. A boy standing at a point at unit distance
from A fires an arrow at the balloon. The arrow strikes the balloon at its nearest
point (to the boy) but does not penetrate it, the balloon absorbing the shock and the
arrow falling vertically to the ground. What is the longest distance through which
the arrow can fall, and how far from A will it land in this case?

Solution by Friend H. Kierstead Jr., Cuyahoga Falb, Ohio.

From the diagram we easily obtain
r = tan 0
^r r+- xz = sin 0
D = z sin 0
from which we derive
D = tan 0(1 - sin 6) . (1)
In order to find the maximum D, we set dD/d0 - 0:
(1 - sin 0)sec20 - tan 0 cos 0 = 0
Multiplying by 1 + sin 0 gives
1 - sin 0{1 + sin 0) = 0
or
sin20 + sin 0 - 1 = 0 .
Solving for sin 0 gives

sin 0 = VL^l
(The other root is ignored because it would make sin 0 greater than 1.) Substituting
this result into (1) gives
 - 32 -

l 3
Dm ft 3 • ^ ^ = 0.300283106-
It can easily be shown that d2D/d02 is negative throughout the region 0 < 0 < TT/2,
so this is indeed a maximum, not a minimum, [Editofs note: or just look at the
endpoints r -* 0, r -» ©.] The distance from A to the point where the arrow lands
is

x= 1- i i | ^ = sin 0 = ^ ^ = 0.618033988- - •
Also solved by HAYO AHLBURG, Benidorm, Spain; HANS ENGELHAUPT,
Gundelsheim, Federal Republic of Germany; RICHARD I. HESS, Rancho Pahs
Verdes, California; M.S. KLAMKIN, University of Alberta; J.A. MCCALLUM,
Medicine Hat, Alberta; P. PENNING, Delft, The Netherlands; and the proposer.
Not all solvers found simple expressions for the required distances. Ahlburg, on
the other hand} noted that) with

^ = 2 '
by2
the maximum falling distance is (f)' , the radius of the balloon when this occurs is
<p'l'2} and the distance from A to where the arrow hits the ground is </>~l: in his
words, "a delightful ubiquity of the Golden Section in such a simple figure".

CRUX COOKIES

Crux
Ttutter Speculating
nach altem Hausrezept
Zutaten: Weizenmehl, Zucker, Butterreinfett,
Vollei, Backtriebmittel, Farbstoff Beta-Carotin,
naturliches Aroma, Vanille
Mindestens haltbar bis Ende: siehe Clip
200 g
mmmmMm
A package bearing the above label arrived here over the Christmas holidays
from J. SUCK of Essen, and was speedily transformed into the empty set by the
editor and others. Sadly, all that remains for the Crux readership is the label. The
editor thanks Mr. Suck for his thoughtful and appropriate gift.
Hmmm. Does there perhaps exist an automobile manufacturer with a line of
Crux sports cars?
 \ Canadian Mathematical Society 577 KING EDWARD, OTTAWA, ONT. CANADA KIN 6NS

o
TEL (613) 564 2223
^ Societe Mathe manque du Canada "
EDITORSHIP OF CRUX MATHEMATICORUM
CALL FOR NOMINATIONS

The Publication Committee of the Canadian Mathematical Society invites applications for
the editorship of Crux Mathematiconim for a five-year period commencing January 1, 1991.

An application should consist of a letter from the candidate which includes an expression
of views of Crux and indicates whether any changes in direction of policy are contemplated, a
curriculum Vltae and a letter of support from the head of the potential host department.

Applications should be sent to:

Prof. Michael Barr, Chairman

CMS Publications Committee
Department of Mathematics
McGill University
805 Sherbrooke St. W.
Montreal, Quebec
H3A 2K6

The absolute deadline for receipt of applications is August 31,1990.

CRUX MATHEMATICORUM
APPELS AU1£ NOMINATIONS

Le Comite des publications de la Societe mathematique du Canada invite les mises en

candidatures pour le poste de redacteur de Crux Mathematiconim pour un mandat de cinq ans debutant
le 1 Janvier 1991.

Une mise en candidature devrait inclure une lettre du candidat (de la candidate) decrivant
sa vision de Crux et qui indique si des changements de direction ou de politiques sont envisages, un
curriculum vitae et une lettre d'appui du departement du candidat (de la candidate).

Les mises en candidature devraient etre acheminees a:

Prof. Michael Barr, President

Comite des publications de la SMC
Department of Mathematics
McGill University
805 Sherbrooke St. W.
Montreal, Quebec
H3A 2K6

La date lirnite pour la reception des mises en candidature est le 30 aout 1990.
 !!!!! BOUND VOLUMES !!!!!

THE FOLLOWING BOUND VOLUMES OF CRUX MATHEMATICORUM

ARE AVAILABLE AT $ 10.00 PER VOLUME

1 & 2 (combined), 3, 7, 8, 9 and 10

PLEASE SEND CHEQUES MADE PAYABLE TO

THE CANADIAN MATHEMATICAL SOCIETY

The Canadian Mathematical Society

577 King Edward
Ottawa, Ontario
Canada KIN 6N5

Volume Numbers Mailing:

Address
volumes X $10.00 = $

!!!!! VOLUMES RELIES !!!!!

CHACUN DES VOLUMES RELIES SUIVANTS A 10$:

1 & 2 (ensemble), 3, 7, 8, 9 et 10

S.V.P. COMPLETER ET RETOURNER, AVEC VOTRE REMISE LIBELLEE

AU NOM DE LA SOCIETE MATHEMATIQUE DU CANADA, A L'ADRESSE
SUIVANTE:

Societe mathematique du Canada

577 King Edward
Ottawa, Ontario
Canada KIN 6N5

Volumes: Adresse:

volumes X 10$ = $
 PUBLICATIONS

The Canadian Mathematical Society

577 King Edward, Ottawa, Ontario KIN 6N5
is pleased to announce the availability of the following publications:

1001 Problems in High School Mathematics

Collected and edited by E.J. Barbeau, M.S. Klamkin and W.O.J. Moser.

Book I: Problems 1-100 and Solutions 1-50 58 pages ($5.00)

Book II: Problems 51-200 and Solutions 51-150 85 pages ($5.00)
Book III: Problems 151-300 and Solutions 151-350 95 pages ($5.00)
Book IV: Problems 251-400 and Solutions 251-350 115 pages ($5.00)
Book V: Problems 351-500 and Solutions 351-450 86 pages ($5.00)

The Canadian Mathematics Olympiads (1968-1978)

Problems set in the first ten Olympiads (1969-1978) together with suggested
solutions. Edited by E.J. Barbeau and W.O.J. Moser. 89 pages ($5.00)

The Canadian Mathematics Olympiads (1979-1985)

Problems set in the Olympiads (1979-1985) together with suggested solutions.

Edited by C M . Reis and S.Z. Ditor. 84 pages ($5.00)

Prices are in Canadian dollars and include handling charges.

Information on other CMS publications can be obtained by writing
to the Executive Director at the address given above.