Download as pdf or txt
Download as pdf or txt
You are on page 1of 37

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
March I mars
Volume 18 §3
1992

CONTENTS I TABLE PES MATIERES

The Olympiad Corner: No. 133 R.E. Woodrow 65

Problems: 1721-1730 74

Appeal for Problems 76

Solutions: 1580, 1613-1625, 1627 76

Canadian Mathematical Society

Q> Soci&6 irath&natique du Canada


Founding Editors: Leopold Sauve, Frederick G.B. Maskell
Bditors-in-Chief: G.W. Sands and R.E. Woodrow
M a n a g i n g Editor: G.P. Wright
EDITORIAL BOARD
G . W . Sands (Calgary)
R . E . Woodrow (Calgary)
G.P. Wright (Ottawa)
R. Guy (Calgary)
C. Fisher (Regina)
D. Hanson (Regina)
A. Liu (Alberta)
R. Nowakowski (Dalhousie)

GENERAL INFORMATION

Crux Mathematicorum is a problem-solving journal at the senior secondary and university under-
graduate levels for those who practice 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 publications should be sent to the Editors-
in-Chief:

G.W. Sands and R.E. Woodrow


Department of Mathematics & Statistics
University of Calgary
Calgary, Alberta, Canada, T2N 1N4

SUBSCRIPTION INFORMATION

Crux is published monthly (except July and August). The 1991 subscription rate for ten issues is
$ 20.00 for members of the Canadian Mathematical Society and $40.00, for non-members. Back
issues: $4.00 each. Bound Volumes with index: volumes 1 & 2 (combined) and each of 3, 7, 8 &
9: $10.00 (Volumes 4, 5, 6 & 10 are out-of-print). All prices 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

ACKNOWLEDGEMENTS

The support of the Department of Mathematics and Statistics of the University of Calgary and of
the Department of Mathematics of the University of Ottawa is gratefully acknowledged.

© Canadian Mathematical Society, 1992


Published by the Canadian Mathematical Society
Printed at Ottawa Laser Copy

ISSN 0705-0348 Publications Mail Registration Number 5432


65

THE OLYMPIAD CORNER


No. 133
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 lty.
Last month we gave some problems from China and from Ireland. This month we
continue each set. First we give five more problems used in the competition to determine
the Irish 1990 I.M.O. team. Thanks go to Andy Liu of The University of Alberta for
collecting and forwarding them.

M O R E S E L E C T I O N Q U E S T I O N S F O R T H E 1990 I R I S H I.M.O. T E A M

1 . Let ABC be a right-angled triangle with right angle at A. Let X be the foot of
the perpendicular from A to EC and Y the midpoint of XC. Let AB be extended to D
so that AB = ED. Prove that DX is perpendicular to AY.
2. The real number x satisfies all the inequalities

2k <xk + xh*1 < 2k+1


for k = 1, 2 , . . • ? n. What is the greatest possible value of n?
3* Three sides of a quadrilateral are given, of lengths a^b, c^ respectively. If the
area of the quadrilateral is as large as possible prove that the length x of the remaining
side satisfies the equation

x3 - (a2 + b2 + c2)x - 2abc = 0.

4 . Let ABC be a triangle and let the internal bisectors of the angles at A and B
meet the sides EG and AC at D and 25, respectively* Let CF and CG be the perpendiculars
from C to the lines BE and AD, respectively. Prove that the line FG is parallel to AB.
5* Let n = 2k — 1 where k > 6 is an integer. Let T be the set of all n-tuples
(xux2j... , xn) where Xi is 0 or 1 (t = 1,2,..., n). For x = (a?i,..., xn) and y = ( j h , . . . , yn)
in T, let d(x,y) denote the number of integers j with 1 < j < n such that Xj ^ yj. (In
particular, rf(x5x) = 0). Suppose that there exists a subset S of T with 2^ elements which
has the following property: given any element x in T, there is a unique element y in S
with d(x,y) < 3. Prove that n = 23,
*

The next problems were translated and forwarded to us by Edward T.H. Wang,
Wilfrid Laurier University. Thanks to Professor Wang!

Prated on recycled paper


Imprime sur papier recycle
@
66

1988 C H I N E S E O L Y M P I A D T R A I N I N G C A M P

Second Test

1 . Divide each side of an equilateral triangle ABC into n equal parts. Connect the
dividing points to produce line segments which are parallel to the sides of A ABC. Call
such a configuration a triangular grid of length n.
(a) How many rhombuses of side length \BC\/n are contained in a triangular grid
of length n?
(b) How many parallelograms are contained in a triangular grid of length n?
2« For a finite sequence A of O's and l f s f let f(A) denote the sequence obtained from
A by replacing each 1 by 0,1 ? and each 0 by 1,0; e.g., /((1 ? 0,0,1)) = (0? 1,1? 0,1,0,0,1).
Let fn(A) denote the nth iterate of / on A. Determine the number in / n ( ( l ) ) of two
consecutive terms which are 0,0.
3« A mathematics teacher wants her two intelligent students S and P to derive
the exact value of a 2-digit natural number n by revealing the number of positive divisors
of n to S and the sum of the digits of n to P. A brief conversation between S and P goes
as follows:
Pi I can not determine n.
Si I can't either but I know whether n is even or not.
Pi Now I know what n is.
Si So do I now.
Suppose both students are honest and have perfect logical reasoning for whatever they say.
Determine n and justify your answer.

Third Test

1 . Suppose the inscribed circle of AA1A2A3 touches the sides ^ 2 ^ 3 , A3A1 and
A1A2 at 2\, T2 and I3 respectively. Prom the midpoints Mi, M 2 and M 3 of A2A3j A3A1
and A1A2, draw lines perpendicular to 72T3, T3T1, and T1T2 respectively. Prove that these
three perpendicular lines are concurrent at a point P and determine the location of P.
2» Consider a quadrilateral ABCD inscribed in a circle. Suppose we fix A and C
and move B and D along the arcs AC and CA in the clockwise direction in such a way
that BC = CD. Let M denote the point of intersection of AC and BD. Find the locus of
the circumcenter of triangle AMB.
3* Let P = A1A2 .. - A2n be a regular 2n-gon, n > 2. If i x , i 2 , . . . , «2n denotes a per-
mutation of 1,2,..., 2n, then the polygon Q = A^ A{2... Ai2nAil (whether convex or not)
is called an accompanying 2n-gon of P. Prove that there always exists an accompanying
2n-gon Q of P such that exactly one pair of sides of Q are parallel to each other.
67

Fourth Test

1 . Suppose Xi$x2i***iXn are positive reals with sum equal to 1. Determine the
minimum value of the function
1
f{xux29 ...,*»)= T . I * , + TT T ^ i + 9 ''
1 + x2 + • • • + xn 1 + xx + x3 + h xn

1 + Xi + X2 + • • • + Xn-i
and find all n-tuples (#j ? a?2 ? ..., xn) which yield this minimum value.
2* (a) Prove that there exist positive real numbers A such that [An] and n have the
same parity for all positive integers n. ([•] denotes the greatest integral part function.)
(b) Find one such number A,
3* Suppose convex hexagon ABCDEF can be inscribed in a circle. Prove that a
necessary and sufficient condition for the three diagonals AD, BE and CF to be concurrent
is AB<CD<EF = BC<DB° FA.
$ * *

Before turning to solutions from the readers we give a comment about a recent
number of Crux. This certainly leaves me with egg on my face! Professor Francisco Bellot,
Valladolid (Spain) points out that the six problems sent in by Florentin Smarandache and
published under the title of the Turkish Mathematical Olympiad last December [1991:
289-290] had appeared three years earlier as the problems of the 29th International Math-
ematical Olympiad [1988: 196-197]. There must have been some error in translation from
the Turkish. I thought the last problem looked interesting and familiar (and did track
down its solution in the November 1991 number). Too bad 1 didn't follow the reference
given there back to the 1988 source where all six occur! 1 am relieved that some readers
have better memories than I do.
9|C 3p 3|C

Last month we discussed solutions to some problems from the October 1990 number
of Crux. No sooner had the column been finalized than a packet of solutions arrived from
Michael Selby, University of Windsor, Here is his solution to a problem not discussed last
issue. (He sent in solutions to numbers 1, 3, 5, 7, 8 as well.)
2 . [1990: 225] 25th Spanish Mathematics Olympiad.
In the square ABCD% let M and B be the midpoints of AB and CD, respectively.
Consider the distance preserving transformation T of points in the plane of the square such
that T(A) = if, T(H) = B, and T(D) = M. Is there a straight line 1 which is preserved
by T (i.e. T(l) = 1)1 Is there a point X of the plane with T(X) = XI
68

Solution by Michael Selby, University of Windsor.


(0,2)

Draw the square on an x,y coordinate system:


(0,0)

We are given T((0,2)) = (1,0), r((0,0)) = (1,2) and T((1,0)) = (2,2). Then

|T((0,1)) - T((0,2))| = 1 = |T((0,1)) - (1,0)|,

|r((0,l))-T((0,0))| = l = |T((0,l))-(l,2)|.
Hence T((0,1)) = (1,1). Next consider T((l, 1)). Then

|T((1,1)) - T((0,1))| = 1 = |T((1,1)) - (1,1)|,

|T((1,1)) - T((0,2))| = V2 = |T((1,1)) - (1,0)|,


and
|r((i,i))-r((i,o))| = i = |r((i,i))-(2,2)|,
so the only solution is T((l, 1)) = (2,1).
We claim that the line y = 1 is invariant; in fact if p = (x, 1), T(p) = (x + 1,1).
To prove this, let p = (x, 1) and write T(p) = (a,/3). Then

|(a,/9)-r((0,l))|-|*|-|(a,/9)-(1,1)1, (1)
|(a,0) - T((l, 1))| = \x - 1 j = |(o, ft) - (2,1)|. (2)
Therefore

( a - l ) 2 + ( / ? - l ) 2 = x2 and (a - 2)2 + (0 - I) 2 = (x - I) 2 .

Thus
(a~l)2-(a-2)2 = x2-(x-l)2
and so a = x + 1 and /? = 1. This gives T((a?, 1)) = (x + 1,1), proving the claim.
There is no point X such that T(X) = X. For a contradiction, suppose T(X) = X,
for some X. Then

\T(X) - T((0,1))| = |X - (0,1)| = \T(X) - (1,1)|.

Since T(X) = X, \T(X) - (0,1)| = \T(X) - (1,1)|. Applying T again,

\T(T(X)) - T((l, 1))| = \T(X) - (1,1)| = \T(X) - (0,1)|.

This gives
\T(X) - (2,1)| = \T(X) - (1,1)| = \T(X) - (0,1)|.
69

This means that T{X) lies on the perpendicular bisector of the line segment joining (0,1)
and (1,1), i.e. x = 1/2, and the perpendicular bisector of the line segment joining (1,1)
and (2,1), namely x = 3/2, This is impossible. Hence T(X) ^ X for any X,

Before turning to solutions to problems from the December 1990 number of Crux,
1 must correct the records since two solutions got misplaced, Edward T.H. Wang, Wilfrid
Laurier University^ Waterloo, Ontario? submitted a solution to # 2 of the 1990 Asian Pacific
Mathematical Olympiad similar to the solution given last number [1992: 40]. Senng-Jin
Bang, Seoul, Republic of Korea, sent a solution to # 2 of the Fourth Nordic Mathematical
Olympiad for which we gave a solution last issue [1992: 42].
4c 4s 4s

We next give solutions to four of the problems of the 20th Austrian Mathematical
Olympiad, Advanced Level [1990: 289-290].
1 • Let Sn be the set of all the 2n numbers of the type

2±^2±~v^± T7: .
Here the number 2 appears n + 1 times and there are n square roots preceded by +1 or
-1.
(a) Show that all members of Sn are real.
(b) Compute the product Pn of all members of Sn.
Solution by Edward T.H. Wang, Wilfrid Laurier University? Waterloo, Ontario.
(a) We show by induction on n that 0 < x < 4 for all x € Sn and for all n. Since
Si = {2 + ^^2 — V2} ? the assertion holds for n = 1. Suppose it holds for some n > 1.
Since S n+ i = {2 ± y/x : x € Sn} where y/x is real such that 0 < y/x < 2 by the induction
hypothesis, we immediately see that 0 < y < 4 for all y € 5^+1.
(b) We show by induction on n that Pn = 2 for all n. This is clear for n = 1.
Suppose that Pn = 2 for some n > 1. For x = 2 + y/y where y G Sn-i (if n = 1, this
means y = 2), let us write x = 2 - yjy. Since Sn+t = {2 ± V^ : * € &n} and since
(2 + V£)(2 - v ^ ) = 4 - ^ = 2 - y ^ = x , ( 2 + V^)(2 - v ^ ) = 4 - x - 2 + y ^ = x, we
have P n + i = Pn = 2, by the induction hypothesis.
2 , Determine all triples (a 9 6 ? c) of whole numbers such that a • 6 • c = 1989 and
a + 6 - c = 89.
Solutions by Stewart Metchette, Culver City, California; and by Edward T.H. Wang,
Wilfrid Laurier University, Waterloo, Ontario. We give Wang*s solution.
If whole numbers mean positive integers, then the only solutions are (39,51,1)
and (51,39,1). If they mean integers, then there are four more solutions: (39, - 1 , -51),
( 5 1 , - 1 , - 3 9 ) , ( - 1 , 3 9 , - 5 1 ) and ( - 1 , 5 1 , - 3 9 ) .
Let (a,ft,c) be a solution. Assume first that a,ft,and c are all positive. (Clearly none
of them can be zero.) From the prime power decomposition 1989 = 3 2 x 13 X 17, we see
70

that the set of positive divisors of 1989 is {1,3,9,13,17,39,51,117,153,221,663,1989}.


Due to symmetry, we may assume that 1 < a < b. Then 26 > a + b > 89 implies
b > 45. On the other hand if 6 > 117, then c = a + b - 89 > 117 - 89 = 28. But
c ~ 1989/a6 < 1989/117 = 17, a contradiction. Thus b = 51. Since a + b > 89, we must
have a = 39 which yields the solution (39,51,1). Interchanging a and 6 we obtain the
other solution (51,39,1).
Now suppose a, 6, and c are not all positive. Then clearly either (i) a > 0, b < 0
and c < 0 or (ii) a < 0, b > 0 and c < 0. In case (i) we set bf = —6 and d = —c. Then
aVd =5 1989 and a + d — b1 = 89 where &' > 0 and c/ > 0. Prom the discussion of the last
paragraph [a^d^b1) = (39,51,1) or (51,39,1) which yield two more solutions for (a^bjc):
(39, —1, —51) and (51, —1, —39). Similary, in case (ii) we set a1 = —a and d = —c, so that
a'bd = 1989 and 6 + </ - a' = 89 where a1 > 0 and d > 0. Thus (6, cW) = (39,51,1) or
(51,39,1) yielding still two more solutions (—1,39, —51) and (—1,51, —39).
5 . Determine all real solutions of the system

x2 + 2yz = x

y2 + 2xz = y
z2 + 2xy = z.

Solutions by Seung-Jin Bang, Seoul, Republic of Korea; Bob Prielipp, University


of Wisconsin-Oshkosh; and by Edward T.H. Wang, Wilfrid Laurier University, Waterloo,
Ontario. We give Bang's solution.
If x = 0 then y = 0 or z = 0. It follows that (s,y,s) = (0,0,0), (0,0,1), (0,1,0),
(1,0,0) are the solutions in case one of a?, j / , or z is zero. So we assume that xyz ^ 0.
If x}y,z are distinct then xyz = — 2xyz because

x(x — x2) = y(y — y2) = z(z — z2) = 2xyz

giving that a?, y and z are the three roots of w 3 — w2 + k = 0 where k = 2xj/z. This is
plainly a contradiction.
We may now assume, by symmetry, that x = y. Since x2+2xz = x and 2 2 -f2# 2 = z
we have that x + 2J2T = 1 (since x ^ 0) and z2 + 2(1 — 2z) 2 = z giving z = 2/3 and
a? = — l / 3 = j / o r z = l / 3 and x = 1/3 = y. The solutions are therefore

{(0,0,0), ( I , | , I ) , ( l , 0 , 0 ) , ( 0 , l , 0 ) , ( 0 , 0 , l ) , ( - i , - i , | ) , ( ? , - i , - i ) , ( - I , | - i ) } .

6 . Determine all functions / : No —• No such that for n G N 0

/ ( / ( * ) ) + / ( * ) = 2n + 6
(N 0 = {0,1,2,...}).
Solutions by Seung-Jin Bang, Seoul, Republic of Korea; and by Edward T.H. Wang,
Wilfrid Laurier University, Waterloo, Ontario. We first give Bang's solution.
71

Denote fm=fofo'--of(m times). Let fm(n) = amf(n) + bmn + c^. Since

/ m + 1 ( n ) = fm(f(n)) = a m / ( / ( n ) ) + 6 m /(n) + c*
= Om(2n + 6 - /(n)) + 6 m /(n) + ^ = (bm - am)f(n) + 2amn + (6a m + c m ),

we see that a m + i = bm — am, bm+\ = 2a m , c„+i = eaTO+Cn. Solving the resulting recurrence
relations, with a% = 1, &i = cj = 0, yields

«m = \{l - ( - 2 ) " ) , 6ro = | ( 1 - ( - 2 r + 1 ) , cm = i ( 6 m - 2 - ( - 2 r + 1 ) .

Since

^ = ( £ - (-D") /(»)+»(£ - (-iy-*)»+£ - ^ - 2 ( - D - > o


we have —/(n) + n + 2 > 0 (m even9 m —• oo) and f(n) — n — 2 > 0 (m odd9 m -» oo). It
follows that f(n) = n + 2*
[iVearf we £we Wang's solution? which is a little less "tricky".]
Clearly f(n) = n + 2 satisfies the given condition. We show that there are no others.
Suppose / : No —-> No satisfies

/(/(»»)) + f(n) = 2n + 6. (1)


If /(*) = / then
/(/) = 2k + 6 - f(k) = 2k + 6-L (2)
In particular, if /(0) = /, then 6 — I = /(/) > 0 implies / < 6. Using (2) repeatedly, we
find that
(i) /(0) = 0 =» /(0) = 6, a contradiction;
(ii) /(0) = 1 =• / ( l ) = 5 =» /(5) = 3 =» /(3) = 13 =» /(13) = - 1 , a contradiction;
(Hi) /(0) = 3 =• /(3) = 3 =4> /(3) = 9, a contradiction;
(iv) /(0) = 4 =• /(4) = 2 =• /(2) = 12 =• /(12) = - 2 , a contradiction;
(v) /(0) = 5 =*> /(5) = 1 =• / ( l ) = 15 =» /(15) = - 7 , a contradiction;
(vi) /(0) = 6 =>• /(6) = 0 =• /(0) = 18, a contradiction,
Prom this we conclude that /(0) = 2. Using (1) it is now easy to check by induction that

f(2k) = 2k + 2 for all k € N 0 . (3)

Now suppose / ( l ) = /. Then f(l) = 8 — I from (2). Furthermore, / can not be even for
otherwise I + 2 = f(l) = 8 — / =W = 5, a contradiction. Thus / ( l ) = 1,3,5 or 7. As
before, using (2) repeatedly, we find that:
(a) / ( l ) = 1 =*> / ( l ) = 7, a contradiction.
(b) / ( l ) = 5 =$- /(5) = 3, a contradiction as in (ii).
(c) / ( l ) = 7 =» /(7) = 1 =*- / ( I ) = 19, a contradiction.
72

Therefore / ( I ) = 3. Using this and (1) we deduce by induction that


f(2k + 1) = 2k + 3 for all k G N 0 . (4)
From (3) and (4) f(n) = n + 2 for all n G N0*

To complete this month's Corner we give solutions to the four problems of the 20th
Austrian Mathematical Olympiad, Final Round, Beginner's Level [1990: 290].
1* Let a, 6, c, df where a < 6 < c < d, be natural numbers such that a+b+c+d = 30.
Find the maximum value of the product P = a • 6 • c • rf.
Solution and generalization by Charlton Wang, student at Waterloo Collegiate
Institute, and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario.
Generalization: For positive integers n and k let
k
S = {(&i,&2?• • •?&&) * 0 < &i < ^2 < • • • <fljfej5 Z a ^ = n5 a« positive integers}.
i=i

Also, let M = max{nf=i a* : ( a i , . . . , a&) G 5 } . Then


M = [n/k\k-n+kWkl . ([n/fcj + l)^~HnAJ?
attained uniquely for the tuple (a, a , . . . , a, a + 1 , . . . , a + 1) where a = [n/k\ and there
are fc — n + &Lu)j a ' s ' ^ n Particulars if k\n then M = (n/k)k attained uniquely at
(n/&,... ,n/fc).
Proof. If &(n, then by the arithmetic-geometric-mean inequality we find easily
that M = (n/fc)fe, attained uniquely at (n/fe,n/&,... ,n/fc). Suppose fc / n and suppose
T = ( a i ? . . . ,afc) G 5 such that n?=i flt = ^ - Then there must be at least two distinct
values among the a f 's. On the other hand there cannot be more than two distinct values
for otherwise we must have aj > a,- + 2 for some i < j* We may suppose j — i is least
possible. Then ( a i , . . . , a,*_i, a,- + 1 , . . . , aj — 1 , . . . , a*) G 5 and since (a,- + l)(flj — 1) =
ajaj + a^ — af- — 1 > a*^- we have a contradiction. Therefore the components of T consist
of two values a and a + 1? say, with m occurrences of a + 1 and fe — m a's for some m
with 1 < m < &. Then from n = (& — m)a + m(a + 1 ) = &a + m < &a + k we get
fca < n < k(a + 1) or a < n/k < a + 1. Hence a = [n/fcj from which we immediately
obtain m = n — ka = n — k [n/k\ and k — m = fc — n +fc[ n AJ •
For the given problem, n = 30, fc = 4, [ra/fcj = 7 , ra = & — ra = 2. Hence
M = 7 x 82 = 3136 attained uniquely for the tuple (7,7,8,8).
2

[Editor's Note. The problem was also solved by Stewart Metchette, Culver City, California,
and a further solution with other generalizations was sent in by an anonymous solver.]
2« Let a and b be non-negative real numbers such that a2 + b2 = 4. Show that

a+b+2
and determine when equality occurs.
73

Solutions by Seung-Jin Bang, Seoul, Republic of Korea; George Evagelopoulos,


Athens, Greece; Bob Prielipp, University of Wisconsin-Oshkosh; Edward T.H. Wang,
Wilfrid Laurier University, Waterloo, Ontario; and an anonymous solver.
We may clearly suppose a > 0 and b > 0. Since

a6 = i [ ( a + 6 ) 2 - ( a 2 + 62)] = i[(a + 6 ) 2 - 4 ] = i ( a + 6 + 2)(a + 6 - 2 )

we have
ab a+b-2 a+ b
-1.
a+b+2 2 2
Hence it suffices to show that a + b < 2%/2 and determine when we have equality. By the
arithmetic-geometric-mean inequality

a 6 = V o 2 6 2 < ^ - ^ = 2-

Therefore (a + ft)2 = a 2 + b2 + 2ab < 8 from which a + b < 2\/2 follows. Clearly equality
occurs if and only if a = 6 = y/2,
3* Let a be a real number. Show that if the equation x2 — ax + a = 0 has two real
solutions x\ and &2? then
x\ + x\ >2(a?i + x 2 ) 9
Solutions by Seung-Jin Bang, Seoul, Republic of Korea; by Bob Prielipp, University
of Wisconsin-Oshkosh; and by Edward T.H. Wang, Wilfrid Laurier University, Waterloo,
Ontario.
If a?i, X2 are the two real roots of x2 — ax + a = 0 where a is real, then £i + X2 =
a = a?i^2. Now we have
x
i + xl "" 2(^1 + ^2) = ^i + ^2 "" 2^i^2 = (^1 — ^2) 2 > 0.

4 . Show that for any triangle each exradius is less than four times the circumradius.
Solutions by Bob Prielipp, University of Wisconsin-Oshkosh; and by Edward T.H,
Wang, Wilfrid Laurier University, Waterloo, Ontario. We give Wang's solution.
Let O denote the center of the exterior
circle tangent to BC and the lines extended
from AB and AC. Let R denote the radius
of this circle and let R denote the circumradius
of A ABC. Also let H denote the foot of the
perpendicular from O to BC. Then from

a = BH + CH

B C
= R'Ctan — + tan —
2 2)
74

we get
(tanf+ tanf)
Since it is well known that a = 2i2sin A we have
1 1 B O
# < 4 i 2 « * - s i n ( £ + C) = - s i n A < t a n — + tan —
1 . /19 „ x 5 C . 5 C . C B . B +C
& - sm{B + C) cos — cos — < sin — cos — + sin — cos — = sin — - —
/B + C\ BB C ,
4=^ cos ( — - — \ 1 cos — cos — < 1
-1 cos —«
/ 2
which is clearly true since 0 < ^f^ < TT/2.
* * *

That concludes the solutions we have on file for the 1990 volume of Crux. The
Olympiad Season is nigh; please collect and send me your national and regional contests
as well as your "nice" solutions to problems from the Corner.

PROBLEMS
Problem proposals and solutions should he sent to B. Sands, Department of
Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada T2N IN4.
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 permission.
To facilitate their consideration, your solutions, typewritten or neatly handwritten
on signed, separate sheets, should preferably be mailed to the editor before October 1,
1992, although solutions received after that date will also be considered until the time when
a solution is published.

1 7 2 1 . Proposed by Murray S. Klamkin, University of Alberta.


Verify the vector identity

[A x (B - C)] 2 + [B x (C - A)] 2 + [C x (A - B)] 2

= (B x C) 2 + (C x A) 2 + (A x B) 2 + (B x C + C x A + A x B) 2 .
75

1722. Proposed by Toshio Seimiya, Kawasaki, Japan.


ABCD is a cyclic quadrilateral with BD < AC. Let E and F be the intersections
of AB, CD and of BC, AD, respectively, and let L and M be the midpoints of AC and
BD. Prove that
Hi = I (*£. _ i ^
i5F " 2 VJW? AC J '
1723. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria.
Let n > 3 be an integer. Determine all continuous functions / : [0,1] —> R such
that
/ ( * l ) + --- + /(*n) = l
for all a?i,..., a?n € [0,1] satisfying a?i + • • • + xn = 1.
1724. Proposed by Stanley Rabinowitz, Westford, Massachusetts.
A fixed plane intersects a fixed sphere forming two spherical segments. (Each seg-
ment is a region bounded by the plane and one of the spherical caps it cuts from the
sphere.) Let $ be one of these segments and let A be the point on the sphere furthest
from S. A variable chord of the sphere through A meets the boundary of S in two points
P and Q. Let 7 be a variable sphere whose only constraint is that it passes through P
and Q. Prove that the length of the tangent from A to 7 is a constant. (Note: in view of
Crux 1070 [1987: 31], this generalizes problem 1155 of Mathematics Magazine, solution on
p. 47 of the 1984 volume.)
1725. Proposed by K.R.S. Sastry, Addis Ababa, Ethiopia.
Note that the Pythagorean triangle with sides 3?4,5 has the following property: one
leg (3) is a triangular number; the other leg (4) is a square number; and the hypotenuse
(5) is a pentagonal number, that is, a number of the form n(3n — l)/2. Find another
Pythagorean triangle with the same property.
1726. Proposed by Hidetosi Fukagawa, Aichi, Japan.
Circular arc AC lies inside rectangle ABCD, dividing it into two parts with in-
scribed circles of radii r% and f2- Show that r\ + f2 = 2r, where r is the inradius of the
right-angled triangle ADC.
1727. Proposed by Joaquin Gomez Rey, LB. Luis Bunuel, Alcorcon, Madrid,
Spain.
For n a positive integer, evaluate A/B where
n [n/2] j

A-JI(2*-1) and B - g (g _^ . ^ •
1728. Proposed by D.J. Smeenk, Zaltbommel, The Netherlands.
AD, BE and CF are the altitudes of an acute triangle ABC, and H is its ortho-
centre. Points Kj L, M lie on EF, FD, DE respectively. Show that AK, BL and CM meet
in a point of the Euler line of AABC if
(i) K, X, M are the points where the incircle of ADEF touches the sides of ADEF;
(ii) Kj Ly M are the feet of the altitudes of ADEF.
76

1729. Proposed by Susan Gyd, Nose Hill, Alberta.


The story is well known that when Hardy visited Ramanujan in hospital and sug-
gested that the number of his taxi, 1729, was not very interesting, Ramanujan immediately
observed that, on the contrary, it was the least number that was expressible as the sum of
two cubes in two different ways:
123 + I 3 = 1728 + 1 = 1729 = 1000 + 729 = 103 + 9 3 .
These sums are easy to see because 1 is one of the cubes, and because 729 is the last
segment of 1729. Are there any numbers whose squares behave in the same way? That is,
can we have x2 + 1 = y2 + z2 with x, y, z distinct positive integers and z2 a final segment
of x2 + 1?
1730. Proposed by George Tsintsifas, Thessaloniki, Greece.
Prove that

where a^b^c^s are the sides and semiperimeter of a triangle, and the sum is cyclic over the
sides.

APPEAL FOR PROBLEMS


For some issues now, Crux's ongoing shortage of good problems has become even
more acute. Readers are asked to please put their minds to inventing problems in their
favourite elementary subjects and sending them in. Especially welcome are problems
from students. Problems should be in a variety of subjects and should be interesting and
accessible to the readership. A solution is preferred, but not always essential. For further
guidelines see the message at the beginning of this issue's problem list. One suggestion:
readers might like to consider problems referring to specific problem numbers, of which one
example (1729) occurs in this issue, and even more obvious candidates are in the offing.
Perhaps some American reader will send in an appropriate problem 1776, or some French
reader a 1789!

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 5 8 0 * . [1990: 240; 1991: 308] Proposed by Ji Chen, Ningbo University, China.
For every convex ra-gon, if one circle with centre O and radius R contains it and
another circle with centre I and radius r is contained in it, prove or disprove that

£2>r2sec2~+702.
n
77

II. Partial solution by G.P. Henderson, Campbellcroft, Ontario.


Set 01 = d. The proposed Inequality is

R2>r2sec2- + d2. (1)

We will discuss the somewhat stronger inequality

R2 >rR sec - + d2. (2)


n
This seems to be true for all n but we have only been able to prove it for n = 3 and n = 4,
If (2) is true, i? > rsec(?r/n) and therefore (1) is true.
Let the given n-gon have vertices A*, i = 1,2,... ,n. Extend 7Af- to intersect the
outer circle at AJ. The polygon A^Ag... A§n is convex, is between the circles and we can
take it to be the given n-gon, since it has the same i2, r and d.
Starting at A!{ = A[ draw AJ'AJ+x, t = 1,2,...,n, tangent to the inner circle with
A^t on the outer circle. After n tangents we will have gone around the outer circle and
A^ +1 will be at or beyond A!{. We now increase r, keeping 0 , J, J? and As( fixed. A" +1 will
move back toward Ajf. There will be a unique value of r such that A{J+1 = A!{. We will
then have an n-gon with vertices on the outer circle and with sides tangent to a possibly
larger inner circle. Since R and d are unchanged and r has been increased, (2) is true in
general if and only if it is true for such n-gons. We now drop the superscripts and consider
only these polygons.
Let the positive angle of rotation from 0A,- to 0A t - +1 be 20,-. Let OB{ be the vector
obtained by giving 0A« a positive rotation of 0% and let af- be the rotation from 01 to
OBi. We have
OAi = OI + IAh

and taking components along 05,-,

R cos Oi = dcos OL{ + r, $ = 1,2,..., n, (3)

where
a,- = a,--i + Oi-i + 0«? t = 2 , 3 , . . . , n,
n

tsl

For given i?, d? and a = a i , these equations determine 0f- and r. If we can eliminate the 0{
we will have r as a function of i2, d and a.
For triangles the critical r Is given by Euler's formula

E 2 = 2ri2 + cf2.

Hence (2) Is true* for n = 3.


78

Now consider the case n = 4. Here also, we can get an explicit expression for r.
The equations (3) are

R cos $i = c?cos a + r, (4)


R cos 02 = d cos(a + 0X + 02) + r, (5)
R cos 03 = dcos(a + 0X + 202 + 0B) + r = -dcos{a + 02- 0 4 ), (6)
i?cos0 4 = dcos(a + 0i + 202 + 203 + 04) + r = ~<icos(a + 02 + 03)9 (7)

where in (6) and (7) we have used J2i=i 0* = *"• Subtracting (5) from (4),

J2(cos #i — cos $2) = d(cos a — cos(a + 0i + 02))

or
R sin I — - — J sin I — - — I = o sin I — —— I sin [a + (?i + — - —

Since 0 < 0X + 02 < *, sin \{BX + 02) ^ 0. Therefore, putting 0 = a + Ou

icsm I — - — 1 = asm I p H — I = asmpcos I ——— I + acospsm I — - —

Hence
dsin/3
tan
ffl-
so (since fl > d and |(02 - 0i)/2| < TT/2)
R — dcos (3

. (92-6i\ dsind (92-9i\ R-dcos0


Sm = CS = (8)
(~2~J ~ F " ' ° 1""2~J F
where
F2 = R?- 2dRcosf3 + d2.
Adding (4) and (5) gives

it cos I — - — I cos I — - — I = a cos I — - — cos LO + — - — + r,

so by (8)

C S U
° V 2 J (i2-<fcos/3)cos(^)+dsin/?sin(^) F'

Similarly, from (6) and (7) we get by subtracting that


79

or, putting 7 = a + $2?

i J s m l — — 1 = - d sin 7 cos 1 1 + d cos 7 sin I - ± ^ ~ - J ,

so
—dsin7
tan
( * * * ) i2 — dcos'j
and
. /0 -M -rfsin7 jd4-93\ R-dcos-f
0
4
Sm = C S
l"2~j ~ G ~' °
where
G2 = (J? - dcos7) 2 + (cisln7) 2 = R2 - 2Rdcos 7 + d2*
By adding, (6) and (7) yield

QS
i 2 1CQSI 2 J = •"" c o s li r —Y~ jcos[~2—) +r?

so

'03 + M r
cos
* / COS cfsin7
rG rG
(R-dcos 7)(i2 + dcos 7) - d2 sin2 7 2
£ - <P

Prom (9), and because ^ _ 1 0{ = ?r,

cos + 02\ r
(—j=
sin 1 — - — 1 — w«» 1
r- ^ 1— _ .

Therefore
r 2G2 f:
2
(R ~~(Pf + -S5
F2 = 1.
and
2 = F 2 (i? 2 - <py
(10)
F*G2 + (R? - d 2 ) 2
We have

G2 = F2 + 2dR(cos 0 - cos 7) = F 2 + 4dE sin (j1^1 J sin f 0 + ^ - = ^ -

Using (8),
G2 = ^ + 4<Wsin^
80

Replacing G2 by this in (10) and substituting for F , we get

R2 - d2 i?2 - rf2

This proves (2) for n = 4.


Editor's comment Henderson then points out that r is independent of a in the
case n = 4 as well as ra = 3, and that the same seems to hold for larger values of n. This
is explained by a special case of Poncelet's Porism: if two circles, one inside the other,
allow an ra-gon circumscribed around the smaller and inscribed in the larger, then they
allow infinitely many such n-gons (e.g., see the recent article "Triquetras and porisms'9,
by D.N. Mackenzie, in the College Math. Journal 23 (1992), pp. 118-131). Poncelet's
porism was also mentioned recently in these pages in Chris Fisher's solution of Crux 1425
[1991: 175]; for those nonexpert readers, like the editor, who see no connection between
the result as given then and this one, the earlier remarks of Fisher and Michal Szurek on
[1985: 329-331] may help. Can any reader settle Henderson's conjecture (2) for all n > 4?

1 6 1 3 . [1991: 43] Proposed by Murray S. Klamkin, University of Alberta.


Prove that
/sina?\ 2 p /tanz\p

for p > 0 and 0 < x < ?r/2. (The case p = 1 is problem E3306, American Math. Monthly,
solution in March 1991, pp. 264-267.)
Solution by Jeff Higham, student, University of Toronto.
The result is clearly true for p = 0, so assume p > 0. By the Power Means Inequality
[or just the A.M.-G.M. inequality],

/sinx\2p /tan#\p /sin 2 xtanaA ( .

Since 0 < x < TT/2, we show that the right side of (1) is > 2 by showing sin2 x tan x > x3}
or upon rearrangement,
/sinx\ 3 ^ /rtX
(—J >cos*. (2)
Writing the Taylor series for (sin#)/£ and cos a?, (2) is equivalent to

(3
K+iH^-M-"- >
Now
81

and
x2 , x4 ^ ^ x2 xA
2! + 4! ^ 2 +
24 ?
thus (3) certainly holds if

xu
T
2 24 - \ 6/ 2 ^ 12 216 5
or equivalently x2 < 9, which is true for 0 < x < ic/2. This completes the proof.
Also solved by RICHARD I. HESS, Rancho Pahs Verdes, California; WALTHER
JANOUS, Ursulinengymnasium, Innsbruck, Austria; MARGIN E. KUGZMA, Warszawa,
Poland; KEE-WAI LAU, Hong Kong; CHRIS WILDHAGEN, Rotterdam, The Nether-
lands; and the proposer. Partial solutions were sent in by SEUNG-JIN BANG, Seoul,
Republic of Korea; G FESTRAETS-HAMOIR, Brussels, Belgium; and VEDULA K
MURTY, Penn State University at Harrisburg (all for p > 1 only); and by GUO-GANG
GAO, Universite de Montreal (p a positive integer).
For inequality (2), the proposer referred to his problem 141 ? Pi Mu Epsilon Journal
4 (1966) 182, and also to item 3,4.18, p, 238 of D»S. Mitrinovic, Analytic Inequalities,

1 6 1 4 . [1991: 43] Proposed by Toshio Seimiya, Kawasaki, Japan.


Let D and E be points on side BC of a triangle ABC, Draw lines through D9 E
parallel to AC, AB respectively, meeting AB and AC at F and G. Let P and Q be the
intersections of line FG with the circumcircle of AABC. Prove that D ? J5, P and Q are
concyclic.
Solution by C Festraets-Hamoir, Brussels, Belgium.
Si BC et PQ sont secantes?
soit M leur point d'intersection.
L'homothetie de centre M et de rap-
port ME/MB applique F sur G (car
BF || EG) et Drnir C (car FD \\ GC),
done
ME __MG _ MC
9
MB~~ MF~~ MD
d?ou MB'MD^MC* MB. Or MC •
MB est la puissance de M par rapport
au cercle circonscrit et on a MC-MB =
MP-MQ. Done ME-MD = MP-MQ,
ce qui exprime la puissance de M par rapport a un cercle circonscrit au quadrilatere
DEPQ,
Si BC et PQ sont paralleles, alors il est immediat que DEPQ est un trapeze isocele5
done inscriptible dans un cercle.
82

Also solved (usually about the same way) by ILIYA BLUSKOV, Technical Univer-
sity, Gabrovo, Bulgaria; JORDIDOU, Barcelona, Spain; PETER HURTHIG, Columbia
College, Burnaby, B.C.; MARCIN E. KUCZMA, Warszawa, Poland; ANDY LIU,
University of Alberta; MARIA ASCENSldN L6PEZ CHAMORRO, LB. Leopoldo Cano,
Valladolid, Spain; P. PENNING, Delft, The Netherlands; D.J. SMEENK, Zaltbommel,
The Netherlands; DAN SOKOLOWSKY, Williamsburg, Virginia; and the proposer,
<jj^ ^£ %&£ <JL> *k^

1615. [1991: 44] Proposed by Clark Kimberling, University of Bvansville,


Evansville, Indiana.
Consider the following array:
2 3 4 5 6 7 8 9 10
4 5 6 7 8 9 10 11
H
2 5| 6 7 8 9 10 11 12
2 7 8 9 10 11 12 13
7 9
22 m 6 11 12 13 14
2 11 9 12 13 8 14 15
m
For example, to produce row 5 from row 4, write down, in order: the 1st number to the
right of 4, the 1st number to the left of 4, the 2nd to the right of 4, the 2nd to the left of
4, the 3rd to the right of 4, the 3rd to the left of 4, and then the 4th, 5th, . . . numbers to
the right of 4.
Notice that a number will be expelled from a row if and only if it is the diagonal
element in the previous row (the boxed numbers in the array), and once missing it of course
never reappears.
(a) Is 2 eventually expelled?
(b)* Is every positive integer eventually expelled?
I. Solution to part (a) by Iliya Bluskov, Technical University, Gabrovo, Bulgaria.
Let an be the position of a particular number in the nth row. From the statement
of the problem we obtain: if an = n, then the number is expelled from the next and all
future rows; otherwise,

{ 2(n — an)

2(aw — n) — 1 if n < an < 2n,


if an < n,

an — 1 if an > 2rc.
Computing successively for the number 2, with a\ = 2, we obtain
a\ = 2 a2 = 1 dz = 2 a 4 = 2 a 5 = 4
ae = 2 a? = 8 ag = 1 &9 = 14 aio = 9
a n = 2 ai2 = 18 aiz = 11 au = 4 a i 5 = 20
a i 6 = 9 an = 14 ais = 6 a\$ = 24 a2o = 9
=
G&21 = 2 2 CJ22 = 1 #23 42 c&24 = 37 a2$ = 25!
Thus 2 disappears after the 25th row.
83

After this I asked my computer for some further information. Here is the result.

n is the diagonal n is the diagonal n is the diagonal


element in row... element in row... element in row...
FT I his- nr \W 18
2 25 19 49595(!) 36 23
3 2 20 9 37 58
4 4 21 69 38 65
5 3 22 16 39 60
6 22 23 24 40 93167 '
7 6 24 12 41 68
8 8 25 43 42 17
9 10 26 47 43 1523
10 5 27 7598 44 39
11 32 28 15 45 75
12 83 29 133 46 20
13 44 30 109 47 99
14 14 31 13 48 34
15 7 32 198 49 117
16 66 33 19 50 123
17 169 | 34 33

Quite a hopeless affair!


II. Comment by the editor.
Upon seeing this intriguing problem, expert colleague RICHARD GUY? and ANDY
GUY, who is a computer science student at Cambridge (and also Richard's grandson!),
found by computer the row r(n) in which n is the diagonal element, for n from 1 to 1200.
Thus a (very) partial answer to part (b) is that every integer up to 1200 is eventually
expelled. The sequence of r(n)'s is wild and unpredictable. The largest value of r(n) they
found is
r(669) = 653494691,
with r(1194) = 608341970 coming close. In R. Guy's words: a My guess is that every
integer is eventually expelled, but that the next few generations of human beings won't
be able to prove this". However, he did manage to find several infinite families of expelled
integers, of which a typical example is that, for each k > 1,

9 • 2* - U - 10 is expelled on row 3 • 2k - 3.

Meanwhile this problem also attracted the attention of David Gale, editor of the
Mathematical Entertainments column of the Mathematical Intelligencer. The result is that
his column on pp. 54-56 of the January 1992 issue is largely devoted to this problem and
related problems. It is an instructive and highly entertaining read (true to billing) and is
most recommended.
84

Part (a) also solved by SEUNG-JIN BANG, Seoul, Republic of Korea;


MARGHERITA BARILE, student, Universitd degli Studi di Genova, Italy; RICHARD
A. GIBBS, Fort Lewis College, Durango, Colorado; HANS HAVERMANN, Weston,
Ontario; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; P. PENNING,
Delft, The Netherlands; CHRIS WILDHAGEN, Rotterdam, The Netherlands; and the
proposer. Several of these calculate the expulsion rows for other values ofn.
Gibbs and Guy both point out the answer to part (a) had to be yes, otherwise part
(b) would not have been starred!
%£« ^J^J *^# %Sf *yy*

1616. [1991s 44] Proposed by D.J, Smeenk, Zaltbommel, The Netherlands.


In triangle ABC} angles a and 7 are acute, D lies on AC so that BD ± AC\ and
E and F lie on BC so that AE and AF are the interior and exterior bisectors of IB AC,
Suppose that BC is the exterior bisector of lABD. Show that AE = AF,
Solution by Giannis G. Kalogerakis, Canea, Crete, Greece,
Let the line AE intersect the line BD
at if, and extend line AB to G. We have that
AF ± AE, so

LFAB = 90° - LB AH
= 90° - LHAD = LBHE. A D v
Since BD is the exterior bisector of lABD, IHBE = LEBG = lABF. Thus the triangles
FAB and BHE are similar, and lAFB = lAEB, Therefore AFE is an isosceles right-
angled triangle, and AF = AE,
Also solved by MIGUEL AMENGUAL COVAS, Gala Figuera, Mallorca, Spain;
SAM BAETHGE, Science Academy, Austin, Texas; SEUNG-JIN BANG, Seoul, Repub-
lic of Korea; FRANCISCO BELLOT ROSADO, LB. Emilio Ferrari, Valladolid, Spain;
ILIYA BLUSKOV, Technical University, Gabrovo, Bulgaria; JORDI DOU, Barcelona,
Spain; C. FBSTRABTS-HAMOIR, Brussels, Belgium; IAN GOLDBERG, student, Uni-
versity of Toronto Schools; JEFF HIGHAM, student, University of Toronto; L.J. HUT,
Groningen, The Netherlands; WALTHER JANOUS, Ursulinengymnasium, Innsbruck,
Austria; MARGIN E, KUCZMA, Warszawa, Poland; ANDY LIU, University of Alberta;
P. PENNING, Delft, The Netherlands; TOSHIO SEIMIYA, Kawasaki, Japan; and the
proposer,
Liu points out that the problem is a partial converse of problem 6 of the November
1985 Senior Tournament of the Towns.

1617. [1991: 44] Proposed by Stanley Rabinowitz, Westford, Massachusetts.


If p is a prime and a and k are positive integers such that pk\(a — 1), then prove
that
pn+*|(ap"_l)

for all positive integers n.


85

Solution by Ian Goldberg, student^ University of Toronto Schools.


This Is easily proved by induction on n. We are given that p* +0 |(a p0 — 1), which
means a = 1 modp. For some integer n > 0, suppose p* +n |(a pn — 1). Let or = a pn s Note
that o s l mod p means x = 1 mod p, so

a-p-i + X P^ + ...+a. + l s l + 1 + ... + 1 + 1 = p s o mod p.

Therefore
p\(xpml + a*""2 + • • • + x + 1);
but by assumption pk^n\(x — 1), so

By induction, p* +n |(a pn — 1) for all integers n > 0.


Also solved by B,L, ABBOTT, University of Alberta; SEUNG-JIN BANG, Seoul,
Republic of Korea; ILIYA BLUSKOV, Technical University, Gabrovo, Bulgaria; DANIEL
BROWN, North York, Ontario; CURTIS COOPER, Central Missouri State University,
Warrensburg; GUO-GANG GAO, Universite de Montreal; RICHARD I HESS, Rancho
Pahs Verdes, California; JEFF HIGHAM, student, University of Toronto; WALTHER
J ANGUS, Ursulinengymnasium, Innsbruck? Austria; MURRAY S* KIAMKIN, Univer-
sity of Alberta; MARGIN E. KUCZMA, Warszawa, Poland; KEE-WAI LAV, Hong Kong;
PAVLOS MARAGOUDAKIS, student, University of Athens, Greece; LEROY F.
MEYERS, The Ohio State University; BOB PRIELIPP, University of Wisconsin-Oshkosh;
LAWRENCE SOMER, Washington, D.C.; EDWARD T.H. WANG, Wilfrid Laurier
University, Waterloo, Ontario; CHRIS WILDHAGEN, Rotterdam, The Netherlands; and
the proposer. There was one incorrect solution sent in.
Several solvers pointed out that (as the above proof shows, actually) p need not be
prime, but can be any positive integer satisfying p*|(a — 1).
3fe $ $ * $

1 6 1 8 . [1991: 44] Proposed by Jordan Tabov, Sofia, Bulgaria. (Dedicated to Dan


Pedoe on the occasion of his 80th birthday,)
The sides of a given angle a intersect a given circle 0(r) in four points. Four circles
Ok(rk)$ k = 1,2,3,4, are inscribed in a so that Oi(ri) and 0^(r^) touch 0(r) externally,
and 02(r 2 ) and Oa(rs) touch 0(r) internally. Prove that r%r4 = r2r^ (This is more general
than problem 1.8-9, page 20, of Japanese Temple Geometry Problems by H. Fukagawa and
D. Pedoe.)
86

Solution by Jordi Don, Barcelona, Spain.

We put Oi(ri) = Q,. Let V be the vertex of a and T; the contacts of 0 S with a side
of a. Consider the inversion / of centre V and power P = VO -r2. We have /(ft) = ft,
I(ili) = ft4, J(n a ) = «3, and so 7(Ti) = T4,1(T2) = T3. Thus VTX-VT4 = P = VT2• VT3.
Since rt = VT,- • tan(a/2),

nr4 = r 2 r 3 = tan 2 (a/2) • P = tan 2 (a/2) • (FO 2 - r 2 ).

A/so solved by FRANCISCO BELLOT ROSADO, LB. Emilio Ferrari, and MARIA
ASCENSION L6PEZ CHAMORRO, LB. Leopoldo Cano, Valladolid, Spain; C.
FESTRAETS-HAMOIR, Brussels, Belgium; WALTHER JANOUS, Ursulinengymnasium,
Innsbruck, Austria; MARCIN E. KUCZMA, Warszawa, Poland; P. PENNING, Delft,
The Netherlands; D.J. SMEENK, Zaltbommel, The Netherlands; DAN S0K0L0WSKY,
Williamsburg, Virginia; J.B. WILKER, Scarborough College, West Hill, Ontario; and the
proposer. Two other readers sent in incorrect solutions.
The solutions of Festraets-Hamoir, Wilker, and the proposer were similar to the
above solution.

1 6 1 9 . [1991: 44] Proposed by Hui-Hua Wan and Ji Chen, Ningbo University,


Zhejiang, China.
Let P be an interior point of a triangle ABC and let Ri,R2,R3 be the distances
from P to the vertices A,B,C, respectively. Prove that, for 0 < k < 1,

Rj[ + R% + R% < (1 + 2T^f-k{ak + bk + ck).


87

Solution by the proposers.


Let
lAPB > lAPC > TT/2,
let S = AP (~1 BC, and let Q and R be the feet of
the perpendiculars from B and C, respectively, to
AP. Then

a + b + c = c + BS + b + CS
>AQ + BQ + AR + CR
= AP + PQ + BQ + AP + PR + CR
>x + y + x + z = 2x + y + z. (1)
Since 0 < k < 1, by the weighted power mean inequality we have

2^T(2^X) + yk + zk h^-2^x + y+
2*=T + 1 + 1 "~ ^ 2^r + 1 + 1
so
zfc + / + ** = 2 ^ ( 2 r i * s ) +y* + z*< ( 2 ^ t + 2 ) 1 _ (2x + y + z)k. (2)
Also, since 0 < k < 1 and la < a + 6 + c, etc.,
fc
2fc(afc +frfc+ c ) _ ^ / 2a \k >Tp_J®_ = 2
k
(a + b + c) ^\a + b + c) ~^a +b+c '
where the sums are cyclic, and so
21-k{a + b + c)k<ak + bk + ck. (3)

Hence by (2), (1) and (3),

a* + y* + ** < (2^r + 2) 1 "* (a + 6 + cf < (2^r + i) 1_fc (a* + 6* + c


k
).

Also solved by MARGIN E. KUCZMA, Warszawa, Poland.

1 6 2 0 . [1991: 45] Proposed by Riya Bluskov, Technical University, Gabrovo,


Bulgaria.
Find a finite set S of (at least two) points in the plane such that the perpendicular
bisector of the segment joining any two points in S passes through exactly two points of
S.
88

I. Solution by several readers (see below).

The outer vertices form a square. The


inner vertices are formed by drawing equi-
lateral triangles inwardly on the sides of the
square.
Alternatively, one can start with the
inner square and erect equilateral triangles
outwardly on the sides.

II. Comment by the editor.


This problem turns out to be an old one. It is given as problem F10? page 157? in
the book Unsolved Problems in Geometry, by H.T. Croft, K J . Falconer^ and R.K. Guy
(Springer-Verlag, 1991). The problem was first asked by Croft? and the above example was
found by L.M. Kelly. It is unknown whether there are any other examples! (My thanks to
Richard Guy for this information).
Chris Fisher and H.S.M. Coxeter have written to point out that the above config-
uration (but not the problem) appears in Fig. 8.2B in Professor Coxeter's book Regular
Polytopes (Dover, Third ed.5 1973; First ed.9 1948) as well as on pp. 30 and 47 of his
Regular Complex Polytopes (Cambridge Univ. Press, Second ed., 1990). Professor Coxeter
believes that this figure is indeed the only solution to the above problem, but that this
would be very difficult to prove.
Found by JORDI DOU, Barcelona, Spain; C FESTRAETS-HAMOIR, Brussels,
Belgium; RICHARD I. HESS, Rancho Palos Verdes, California; PETER HURTHIG,
Columbia College, Burnaby, B.C.; P. PENNING, Delft, The Netherlands; CORY PYE,
student, Memorial University of Newfoundland, St. John}s; and the proposer.

1 6 2 1 * . [1991: 78] Proposed by Murray S. Klamkin, University of Alberta.


(Dedicated to Jack Garfunkeh)
Let P be a point within or on an equilateral triangle and let c\ < C2 < c3 be the
lengths of the three concurrent cevians through P. Determine the minimum value of C2/C3
over all P.
89

Solution by Gerd Baron, Technische Universitat Wien, Austria.


We show that the required minimum value is 2\/2/3.

B U B
Let the triangle be ABC with center M and m^ ? mg, rrtc the medians through A,
B, C* Further let k be the circle through A, M, £ ; i.e., ZAQS = 120° for all Q on k
between A and B in the triangle. Let the side lengths of ABC be 2 and the height y/3.
For P , without loss of generality let CA < c^ < cc be the lengths of the three
cevians and SAI £B 5 £C the angles between the cevians and the corresponding medians. It
follows that 6A < £B < £c and therefore, for given c^, P must lie between the intersection
of CB with rriA and A [on the heavy segments in the figures]. Since CQ is increasing with £c5
the minimum of CB/CC is attained when 6c is largest, i.e. at P on k when e^ = e# (= e).
Let ec = V > £•
Now let 3 = tane and i = tan 17. Then
c 5 = 1/3sece and cc == v^secrfj so

4 = 3(l + 5 2 ), 4 = 3(l + t2). (i)


Since the cevians are concurrent, by the theorem
of Ceva

i + v/3s\2 A - V s Y = 1.
kl + V5t,
This leads to

[(1 - VSs)2 + (1 + V3«) 2 ]v^i = (1 + VSs)2 - (1 - Vis)2


or
1 (2)
l + 3s2'
Using (1) and (2), we have to determine the minimum value of
1+S2 (l+52)(l+3*2)2 (l+32)(l+3*2)2 (1+35 2 ) 2
4 1 +t2 (1 + 3s 2 ) 2 + 4s 2 1 + 10s2 + 9s 4 l + 9s 2
90

Applying calculus, this leads to the equation

0 = (1 + 9s 2 )12s(l + 3s 2 ) - 185(1 + 3a 2 ) 2 = 6s(l + 3s 2 )(9s 2 - 1).

Therefore we get the minimum value at s = 1/3, leading to

cB 2y/2
cc ~ 3

Also solved by JORDI DOU, Barcelona, Spain; PETER HURTHIG, Columbia


College, Burnaby, B.C.; MARCIN E. KUCZMA, Warszawa, Poland; TOM LEINSTER,
New College, Oxford; P. PENNING, Delft, The Netherlands; and TOSHIO SEIMIYA,
Kawasaki, Japan.

1622. [1991: 78] Proposed by Marcin E. Kuczma, Warszawa, Poland.


Let n be a positive integer.
(a) Prove the inequality

for real a, 6, and find conditions for equality.


(b) Show that the constant 2n — 1 in the right-hand expression is the best possible,
in the sense that on replacing it by a smaller one we get an inequality which fails to hold
for some a, 6.
Solution by Pavlos Maragoudakis, student, University of Athens, Greece.
(a) If n = 1 then the inequality becomes equality. Let n > 2. If a + 6 = 0 the
inequality becomes a2n < (2n — l) n a 2 n . This is true for all real numbers a. If a + h ^ 0
then the inequality is equivalent to

1
(^H^H-- ^))-
Let t = (a — b)/(a + 6); then
«
2a , 26
a+b t + 1 and a+bT = 1 - t.
So the inequality (1) is equivalent to

i[(t + i)*» + (t - 1)2"] < (i + ( 2 n _ !)*»)»,

or
(2 i)V (2)
s(3^sC) "- '
91

It is enough to prove for all a and all k < n that

We do this by induction on k. (3) is true if k = 0. If (3) holds for some k then

/ 2n \ _ (2n)! / 2 n \ (2n — 2Jb — l)(n - k)


\2k + 2 / (2k + 2)!(2n -2k- 2)! ~ \2k) (2k + l)(fc + 1)

JfcJ 1 * 1 1J
(2*+ !)(* + !) " U + 1/1 J
V 2A + 1
Thus, to prove that (3) holds when k Is replaced by k + 1 it Is enough to prove that

2n - 2k - 1
< 2n - 1.
2k+ 1
This inequality is equivalent to 2n — 2k — 1 < Akn — 2fc + 2rc — 1, i.e., 4&n > 0, which is
true.
If t = 0 then we have equality in (2). We show that equality never holds in (2)
when * ^ 0 and n > 2. It is enough to show that Q*] < u ) ( 2 n — I) 2 . This inequality is
equivalent to 4n > 0. Therefore equality holds in (1) only if n = 1 or a = 6.
(b) Let c be a real number and suppose for all real a and 6 and for some positive
integer n that

Then arguing as in (a), for all real numbers t,

static) ckt2k.

or, for all t ^ 0,


2n\ . rt
k=2 &-&}'-+&-( 2 20-

This implies that as i -4 0 we have m c — P2nJ > 0 or c > 2n — 1.


ylfeo solt;erf by H.L. ABBOTT, University of Alberta; SEUNG-JIN BANG, Seoul,
Republic of Korea; MARGUERITA BARILE, Universita degli Studi di Genova, Italy;
ILIYA BLUSKOV, Technical University, Gabrovo, Bulgaria; RICHARD L HESS,
Rancho Pahs Verdes, California; WALTHER JANOUS, Ursulinengymnasium, Innsbruck,
Austria; HURRAYS. KLAMKIN, University of Alberta; BEATRIZ MARGOLIS, Paris,
France; JEAN-MARIE MONIER, Lyon, France; and the proposer.
92

Janous notes that the inequality is true for n G [0, l/2]U[l,oo); that it is an identity
for n = 1/2 and n = 1; and that the inequality is reversed for n £ (1/2,1) and, in this
case, 2n — 1 cannot be replaced by a larger constant.

1 6 2 3 . [1991: 78] Proposed by Stanley Rabinowitz, Westford, Massachusetts.


(Dedicated to Jack GarfunkeL)
Let £ be any line through vertex A of triangle ABC that is external to the triangle.
Two circles with radii r% and r2 are each external to the triangle and each tangent to £
and to line BC, and are respectively tangent to AB and AC.
(a) If AB = AC, prove that as £ varies, r\ + r2 remains constant and equal to the
height of A above BC.
(b) If A ABC is arbitrary, find constants ki and &25 depending only on the triangle,
so that kiri + k2r2 remains constant as £ varies.
Solution by P. Penning, Delft, The Netherlands.
Points are labelled as in the figure.
The angles of the triangles are A, i?, C
and the opposite sides a, 6, a Because of
symmetry, the lengths TU and VW are
equal. Further, since LPMB = B/2 and
LQNC = C/2,

TB = BP = rt tan(B/2)

and

CU = QC = r2tan(C/2)}
so

a + n tan — -f r 2 tan — = I U = •= — — • — = s,

the semiperimeter of A ABC. Therefore

rt tan(B/2) + r2 tan(C/2) = s - a,

and
&i = tan(£/2), fc2 = tan(C/2)
solves part (b). For part (a), if B = C then clearly ri + r 2 is constant. The constant itself
can be found from the result above, but it is simpler to consider the case that £ is parallel
to BC: both radii are then h/2 and the sum is equal to h.
Also solved by BENO ARBEL, TeUAviv University, Israel; SAM BAETHGE,
Science Academy, Austin, Texas; SEUNG-JIN BANG, Seoul, Republic of Korea;
FRANCISCO BELLOT ROSADO, LB. Emilio Ferrari, Valladolid, Spain; ILIYA
93

BLUSKOV, Technical University, Gabrovo, Bulgaria; JORDI DOU, Barcelona, Spain;


JOHN G. HEUVER, Grande Prairie Composite H.S., Grande Prairie, Alberta; L.J, HUT,
Groningen, The Netherlands; WALTHER J ANGUS, Ursulinengyrnnasium, Innsbruck,
Austria; MURRAY S. KLAMKIN, University of Alberta; MARGIN E. KUCZMA,
Warszawa, Poland; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zalthommel,
The Netherlands; JOSE YUSTY PITA, Madrid, Spain; and the proposer.
Klamkin, Kuczrna, and the proposer all gave the solution in the form

n 4. l i — i
rc n
where r& and rc are exradii. This can be obtained from Penning's solution.
The problem generalizes problem 2.1.11, page 25, of H. Fukagawa and D. Pedoe,
Japanese Temple Geometry Problems (Charles Babbage Research Centre, Winnipeg, 1989J,
in which AABC is equilateral

1 6 2 4 * . [1991: 78] Proposed by Walther Janous, Ursulinengyrnnasium, Innsbruck,


Austria.
Let [n] = {1,2,... , n}. Choose Independently and at random two subsets A, B of
[n]. Find the expected size of A fl B. What if A and B must be different subsets?
Solution by Tom Leinster, New College, Oxford.
We assume that cat random* means that all possible subsets A and B are equally
likely. In order to choose a subset we proceed as follows: flip a fair coin n times, and
include or exclude the element k from the subset according as a head or a tail occurs on
the kth flip. It is easy to see that each subset occurs with probability (l/2) n , so they are
equally likely as required. Now for each element k of [n] there are four possibilities, each
with probability 1/4: k is in neither A nor J3, k is in A but not I?, k is in B but not A,
or k is in both A and B, Of these, only the last case will contribute to the size of A fl B.
Thus \A n B\ has a binomial distribution with probability 1/4, and E(\A 0 B\) = njL
For the second question, we have

E(\A fl B\ i A and B may be equal) =

E(\A n J3| : A = J3) Pr(A = J3) + E(\A nB\:A?B) Pr(A ^ B). (1)
From above, E(\A f) B\ : A and B may be equal) = n/4. By a similar argument we find
that E(\A fl B\ : A = B) = n/2. Pr(A = B) = 2~ n , since [n] has T subsets. Substituting
into (1), and rearranging, we get

E(\AnB\:A?B)=nf^~]1)) .

Also solved by H. L. ABBOTT, University of Alberta; ARNOLD ADELBERG,


Grinnell College, Grinnell, Iowa; MARGHERITA BARILE, student, Universita degli Studi
94

di Geneva, Italy; GERD BARON, Technische Universitat Wien, Austria; RICHARD I.


HESS, Rancho Pahs Verdes, California; MARGIN E. KUCZMA, Warszawa, Poland;
DAVID G. POOLE, Trent University, Peterborough, Ontario; CHRIS WILDHAGEN,
Rotterdam, The Netherlands; and the proposer. Two incomplete or incorrect solutions
were also sent in,
Baron gave a generalization to m subsets of[n], obtaining for the expected size of
the intersection

n2" if the subsets are arbitrary,

n2n-l(2n"1 - 1 ) . . . (2n~l - m + 1)
if the subsets are all different
2n(2n-l)...{2n-rn + l)

1 6 2 5 . [1991: 78] Proposed by Toshio Seimiya, Kawasaki, Japan,


Isosceles triangles A3Ai02 and AxA203 (with 02AX = 0 2 A 3 and O3A1 = 03A2)
are described on the sides A3AX and AXA2 outside the triangle AXA2A3. Point Ox outside
AAXA2A3 is such that LOxA2A3 = \lAx02A3 and LOxA3A2 = \LAi03A2. Prove that
AxOx ± 0203, and that
AxOx : 0203 = 20iT : A2A3,
where T is the foot of the perpendicular from Ox to A2A3.
Solution by the proposer.
We put LO\A2A3 = a and
ZO1A3A2 = 0; then lAi02A3 = 2a,
ZA1O3A2 = 2/3, and 0 < a,/? < TT/2.
We construct AAXA2B outwardly on
the side AXA2 of the given triangle so
that LAXA2B = a and LAXBA2 = fi.
Because 03A% = 03A2 and lAx03A2 =
2ZAiBA 2 , 03 is the circumcenter of
AAXA2B, and therefore O3A1 = 03B
and ZAi0 3 J? = 2Zi4ii42B = 2a. Let
P , Q be the intersections of BA3 with
0203, A\0\. Since AA 2 A30i and AA 2 i?Ai are similar, AA2AiOi and AA2.BA3 are similar
by SAS. Therefore
A1O1 = A 2 0i sin/9
l J
J5A3 A2A3 sin(a + 0) '
where the last equality uses the sine law in A01A2A3; furthermore, the angle between the
lines Oi Ai and A3B is
LOxQA3 = LOxA2A3 = a. (2)
95

Since 0ZAX = 03B and LAx03B = 2a, &AxBOz and AA 1 A 3 0 2 are similar, so that
AA1BA3 and AA1O3O2 are similar by SAS. Therefore,
BA3 BAi 203Aisina _ .
=2sma
0^-0^1= 03AX ' ^
where the middle equality makes use of the isosceles ABO3A1; furthermore

lA3P0% = lAsA^ = TT/2 - a. (4)


Prom (2) and (4) we obtain £AXQB + lA3P02 = ?r/29 whence we obtain At0i ± 0203 as
required.
{Editor's note. Chris Fisher, University of Regina, adds the following comment. For those
who prefer proofs by similarities, note that the dilative rotation about A2 through an
angle of a that takes A\0\ to J?A3, followed by the dilative rotation about A\ through
the angle TT/2 — a that takes BA3 to 0302^ is a dilative rotation through an angle of
a + (TT/2 — a) = TT/2 that takes A1O1 to 0302j giving the result with fewer intermediate
steps.]
For the second part, from (1) and (3) we get
A1O1 _ A1O1 BA3 _ 2 sin a sin ft _ 2 _ 2 _ 20xr
;
O 2 0 3 " 5 A 3 ' O 2 0 3 ~" sin(a + fi) " cot a + cot/? " ^ f + §^k " ^ H
i.e. we have AxOi : 0 2 0 3 = 2 0 i T : A 2 A 3 .
,4Jso so/ved % SEUNG-JIN BANG, Seoul, Republic of Korea; FRANCISCO
BELLOT ROSADO, LB. Emilio Ferrari, Valladolid, Spain; ILIYA BLUSKOV,
Technical University, Gabrovo, Bulgaria; JORDIBOU, Barcelona, Spain (first part only);
RICHARD I. BESS, Rancho Pahs Verdes, California; WALTHER JANOUS, Ursulinen-
gymnasium, Innsbruck, Austria; P. PENNING, Delft, The Netherlands; and D.J. SMEENK,
Zaltbommel, The Netherlands.
Most solvers used Cartesian coordinates and obtained solutions that were short and
simple. For example, letting A% = (0 9 2ai); A2 = (2a 2? 0), and A3 = (2a 3? 0) (with a% > 0
and a2 < a3), one calculates that
03 = (a2 — ai cot /?, ai — a2 cot /?), 0 2 = (a 3 + a\ cot a, ai + a 3 cot a),
__ /2a 2 cot/3 + 2a 3 cota 2a2 — 2a 3
s etc.
\ cot a + cot /? cot a + cot /?

1 6 2 7 . [1991: 79] Proposed by George Tsintsifas, Thessaloniki, Greece. (Dedicated


to Jack Garfunkel.)
Two perpendicular chords MN and ET partition the circle ( 0 , R) into four parts
Qu Q2, $3? QA* We denote by (0,-, rf-) the incircle of Q,-, 1 < i < 4. Prove that

n+r2 + r3 + r4< 4(V^ - l)i?.


96

Solution by Niels Bejlegaard, Stavanger, Norway.


Note that OOx = R-ru OOz = R~rS} and
0\Oz = %/2(^i + 7*3)- Thus the triangle inequality
immediately gives

R-rx + R-r3> V2(ri + r 3 ),

so
(V2 + l ) ( n + r 3 ) < 2 i 2

or
ri + r$ <
In the same way we have
r2 + r 4 < 2 ( V 2 - 1 ) J ? ,
so that
n + r 2 + r 3 + r 4 < 4(V2 - 1)12.
Equality only if the lines go through O.
Also solved by SEUNG-JIN BANG, Seoul, Republic of Korea; ILIYA BLUSKOV,
Technical University, Gabrovo, Bulgaria; JORDIDOU, Barcelona, Spain; HANS EN GEL-
HAUPT, Franz-Ludwig-Gymnasium, Bamberg, Germany; RICHARD I. HESS, Rancho
Pahs Verdes, California; JEFF HIGH AM, student, University of Toronto; WALTHER
JANOUS, Ursulinengymnasium, Innsbruck, Austria; MARGIN E. KUCZMA, Warszawa,
Poland; TOM LEINSTER, New College, Oxford University, England; PAVLOS
MARAGOUDAKIS, student, University of Athens, Greece; DANIEL REISZ, Vincelles,
Champs-sur-Yonne, France; K.R.S. SASTRY, Addis Ababa, Ethiopia; TOSHIO SEIMIYA,
Kawasaki, Japan; D.J. SMEENK, Zaltbommel, The Netherlands; and the proposer.
Engelhaupt, Kuczma, Maragoudakis and Reisz sent in solutions nearly as nice as
Bejlegaard's. Most other solutions were quite a bit more involved!
Janous obtained two generalizations: one, with a hypersphere in n-space partitioned
into 2n parts (with inradii ri ? r2 9 .. *,r2n) by n mutually perpendicular hyperplanes through
an interior point; the other involving the planar situation but with n rays starting at an
interior point and making equal angles, rather than four. The resulting inequalities for the
sum of the inradii are mostly obtainable as above, and are left to the reader to work out.
* • * * BOUND VOLUMES * • * *

J • • • VOLUMES RELIES•• • |

THE FOLLOWING BOUND VOLUMES OF/ CHACUN DES VOLUMES SUIVANTS DE

CRUX MA THEM A TICORUM


ARE AVAILABLE AT $10.00 PER VOLUME /ESTD/SPON/BLE A 10$

1 & 2 (Combined / ensemble), 3, 7, 8 and 9

PLEASE SEND CHEQUES MADE PAYABLE TO


VEUILLEZ ACHEMINER VOS CHEQUES LIBELLES AU NOM DE

CANADIAN MA THEMA TICAL SOCIETY / SOCIETE MA THEMA TIQUE DU CANADA

577 King Edward


Ottawa, Ontario
Canada K1N 6N5

Volumes : x $10.00 = $

(Canadian residents, please add GST / Residents canadiens, veuillez ajouter la TPS)

MAILING ADDRESS / ADRESSE POSTALE

Name/ Norn:

A d d r e s s / Adresse:

City, Province / Ville, Province:

Postal Code,, Country / Code postal, pays:


PUBLIC A TIONS
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 OUT OF PRINT

Book II Problems 51-200 and Solutions 51-150, 85 pages ($7.50, CMS Members: $6.00) |

Book!!! Problems 151-300 and Solutions 151-350, 95 pages OUT OF PRINT

Book IV Problems 251-400 and Solutions 251-350, 115 pages ($7.50, CMS Members: $6.00)

J BookV Problems 351-500 and Solutions 351-450, 86 pages ($7.50, CMS Members: $6.00) ||

The Canadian Mathematical Olympiads


(1969-1978) and (1979-1985)

Problems set in the first ten Olympiads (1969-1978) together with suggested
solution. Edited by E.J. Barbeau and W.O J . Moser. 89 pages.
($7.50, CMS Members: $6.00)

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


Edited by C M . Reis and S.Z. Ditor. 84 pages.
($7.50, CMS Members: $6.00)

(The Set of 2 books is available at $12.00)

(Canadian residents, please add GST)

Prices are in Canadian dollars and include handling charges.


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

You might also like