1
THE ACADEMY CORNER
No. 16 Bruce Shawyer
All communications about this column should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7 This month, we present some more of the solutions to a university entrance scholarship examination paper from the 1940's, which appeared in the April 1997 issue of CRUX with MAYHEM 1997: 129 . 4. a Suppose that a 6= 0 and c 6= 0, and that ax3 + bx + c has a factor of the form x2 + px + 1. Show that a2 , c2 = ab. b In this case, prove that ax3 + bx + c and cx3 + bx2 + a have a common quadratic factor. Solution. a The given conditions on the coe cients show that x = 0 is not a solution. Also, the other factor must be linear, and, on examining coe cients, must be of the form ax + c. Therefore
ax3 + bx + c = x2 + px +1ax + c = ax3 +c + apx2 +a + cpx + c:
Comparing coe cients gives
c + ap = 0; a + cp = b: c Thus, p = , a , so that a , ca2 = b, giving the required result. 1 b Since x 6= 0, we write y = x , and examine the corresponding equations in y to get this part.
5. Prove that all the circles in the family de ned by the equation
x2 + y2 , at2 + 2x , 2aty , 3a2 = 0
a xed, t variable touch a xed straight line.
2 Solution. We rewrite the equation as
t2
2 + y , at2 = a 2 + t2
2 : x,a 1+ 2 2
This shows that the line x = ,a is a tangent line. The y coordinate is 2at. 6. Find the equation of the locus of a point P which moves so that the tangents from P to the circle x2 + y 2 = r2 cut o a line segment of length 2r on the line x = r. Solution. From the circle, x2 + y 2 = a2 , we nd the points of intersection with the line y , q = mx , p. Substituting y = mx , mp + q , we get x2 + mx , mp + q 2 = a2 , leading to the two solutions:
p
2 2 1 2 2 x = mmp , q a mm+ + 1, m p + q2mp , q : 2 For the line to be a tangent, the discriminant, = a2 m2 + 1 , m2p2 + q2mp , q, must be zero. So we solve = 0, to get p pq a p2 + q2 , a2 : m = p2 , a2 p2 , a2 So, we have that the equations of the tangent lines from p; q to the circle x2 + y 2 = a2 are: ! p pq a a2 , p2 , q2 x y = , a2 , p2 p2 , a2 ! p a2 , p2 , q2 p + q: , ,pq=a2 , p2 + p2 , a2 Setting x = a in these, and subtracting the two values leads to a + p2 = p2 + q 2 , a2: Thus the locus of P is given by y2 = 2aa + x: This is a parabola opening to the right with central axis, the x axis and nose at x = ,a. There are, of course, three points that should be excluded: ,a; 0, a; 2a, and a; ,2a.
3 7. If the tangents at A, B , C , to the circumcircle of triangle 4ABC meet the opposite sides at D, E , F , respectively, prove that D, E , F , are collinear. Solution.
E
F B
P
C
These lead to
D Q R A For notational ease, we write PB = PC = , QC = QA = and RB = RA = . Apply Menelaus' Theorem for triangle 4PQR with transversals BDC , EAC and BAF QC QC ,1 = PB RD CP = BR RD ; BR DQ DQ PE RA QC = PE RA ; ,1 = ER AQ CP ER CP QF ,1 = PB RA FP = PB QF : BR AQ AQ FP
+ + QD = , BR = , ; QD QC + + ER = , CP = , ; ER RA
1 2 3
Now consider the product of ratios to prove D, E and F collinear by the converse of Menelaus: This is equal to
FQ + +
QF
AQ = , PB = , :
PE RD QF : ER DQ FP QD
FQ + + : Using 1, 2 and 3 above, we get a value of ,1, and the result is ER
proved.
+ + ER + + QD
QF
4
THE OLYMPIAD CORNER
No. 187 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. Another year has passed  How the time seems to y. I would particularly like to thank Joanne Longworth for her excellent help in pulling together A the information for the column and her work in producing a good LTEX copy for the EditorinChief usually under enormous time pressure because I am behind schedule. The Corner could not exist without its readers who supply me with good Olympiad materials and send in their interesting solutions and comments to problems in the Corner. I hope we have missed no one in the following list of contributors.
Miguel Amengual Covas S fket Arslanagi
e c Mansur Boase Christopher Bradley Sabin Cautis Adrian Chan ByungKuy Chun Mihaela Enachescu George Evagelopoulos Shawn Godin Joanne Juszunska
Joel Kamnitzer Deepee Khosla Derek Kisman Murray Klamkin Marcin Kuczma Andy Liu Beatriz Margolis Vedula Murty Richard Nowakowski Colin Percival Bob Prielipp Toshio Seimiya Michael Selby Zun Shan D.J. Smeenk Daryl Tingley Panos Tsaoussoglou Ravi Vakil Dan Velleman Stan Wagon Edward Wang
Thank you all, and all the best for 1998! The rst Olympiad we give for the new year is the 17th AustrianPolish Mathematics Competition, written in Poland, June 29 July 1, 1994. My thanks go to Richard Nowakowski, Canadian Team Leader to the 35th IMO in Hong Kong and to Walther Janous, Ursulinengymnasium, Innsbruck, Austria for sending a copy to me.
17th AUSTRIAN
1. The function f : R ! R satis es for all x 2 R the conditions
f x + 19 f x + 19 and f x + 94 f x + 94:
POLISH MATHEMATICS COMPETITION Poland, June 29 July 1, 1994
5 Show that f x + 1 = f x + 1 for all x 2 R. 2. The sequence an is de ned by the formulas and the sequence cn is de ned by the formulas Prove that
n
a0 = 1 and an+1 = 1 2an 2 for n 0; 2 +a
c0 = 4 and cn+1 = c2 , 2cn + 2 for n 0: n : an = 2c0c1 c: : cn,1 for all n 1:
n
3. A rectangular building consists of two rows of 15 square rooms situated like the cells in two neighbouring rows of a chessboard. Each room has three doors which lead to one, two or all the three adjacent rooms. Doors leading outside the building are ignored. The doors are distributed in such a way that one can pass from any room to any other one without leaving the building. How many distributions of the doors in the walls between the 30 rooms can be found so as to satisfy the given conditions? 4. Let n 2 be a xed natural number and let P0 be a xed vertex of the regular n + 1gon. The remaining vertices are labelled P1 ; P2 ; : : : ; Pn , in any order. To each side of the n + 1gon assign a natural number as follows: if the endpoints of the side are labelled Pi and Pj , then i , j is the number assigned. Let S be the sum of all the n + 1 numbers thus assigned. Obviously, S depends on the order in which the vertices have been labelled. a What is the least value of S available for xed n? b How many di erent labellings yield this minimum value of S ? 5. Solve the equation
in integers. 6. Let n
1 x + y y + zz + x + x + y + z 3 = 1 , xyz 2
x1 ; x2; : : : ; xn 0 satisfy the system of equations x2 , x1 2 + 2x2 + x1 + 1 = n2 x3 , x2 2 + 2x3 + x2 + 1 = n2 ...................................... x1 , xn 2 + 2x1 + xn + 1 = n2 : Show that either x1 = xn or there exists j with 1 j n , 1 such that xj = xj+1 .
1 be an odd positive integer. Assume that the integers
6
n = ab10 a 10a + b a 1 with the property that for every integer x the = di erence x , xb is divisible by n. 8. Consider the functional equation f x;y = a f x;z + b f y;z with real constants a, b. For every pair of real numbers a, b give the general form of functions f : R2 ! R satisfying the given equation for all x; y; z 2 R. 9. On the plane there are given four distinct points A, B, C , D lying in this order on a line g , at distances AB = a, BC = b, CD = c. a Construct, whenever possible, a point P , not on g , such that the angles APB , BPC , CPD are equal. b Prove that a point P with the property as above exists if and only if the following inequality holds: a + bb + c 4ac.
As a second source of problem pleasure for winter evenings that is if you are having winter this January!, we give the Second Round of the Iranian National Mathematical Olympiad. My thanks go to Richard Nowakowski, Canadian Team Leader to the 35th IMO in Hong Kong for collecting the problems and forwarding them to me.
7.
Determine all twodigit in decimal notation natural numbers
IRANIAN NATIONAL MATHEMATICAL OLYMPIAD February 6, 1994
Second Round
2. Let ABC be an acute angled triangle with sides and area equal to a, b, c and S respectively. Show that the necessary and su cient condition for existence of a point P inside the triangle ABC such that the distance between P and the vertices of ABC be equal to x, y and z respectively is that there be a triangle with sides a, y , z and area S1 , a triangle with sides b, z , x and area S2 and a triangle with sides c, x, y and area S3 where S1 + S2 + S3 = S. 3. Let n and r be natural numbers. Find the smallest natural number m satisfying this condition: For each partition of the set f1; 2; : : : ; mg into r subsets A1 ; A2 ; : : : ; Ar there exist two numbers a and b in some Ai 1 1 i r such that 1 a 1 + n . b 4. G is a graph with n vertices A1; A2; : : : ; An such that for each pair of nonadjacent vertices Ai and Aj there exists another vertex Ak that is adjacent to both Ai and Aj .
a Find the minimum number of edges of such a graph.
7p , 6p , 1 is divisible by 43.
1. Suppose that p is a prime number and is greater than 3. Prove that
7 b If n = 6 and A1 ; A2 ; A3 ; A4 ; A5 ; A6 form a cycle of length 6, nd the number of edges that must be added to this cycle such that the above condition holds. 5. Show that if D1 and D2 are two skew lines, then there are in nitely many straight lines such that their points have equal distance from D1 and D2. 6. f x and gx are polynomials with real coe cients suchthat for x x in nitely many rational values x, f x is rational. Prove that f x can be g g written as the ratio of two polynomials with rational coe cients. Now we turn to readers' solutions and comments for problems posed in the Corner. First solutions to a problem of the 25th United States of America Mathematical Olympiad 1996: 203 204 . 1. Prove that the average of the numbers
n sin n ; n = 2; 4; 6; : : : ; 180
is cot 1 . Solution by D.J. Smeenk, Zaltbommel, the Netherlands. Let
x = 2 sin 2 + 4 sin4 + + 90 sin90 + + 178 sin178 = 2 + 178 sin 2 + 4 + 176 sin 4 + = 180sin2 + sin 4 + + sin 88 + 90 sin 90 :
Then Now
x = 90 = 2 sin 2 + 2 sin4 + + 2 sin88 + 1 x x sin 1 = 2 sin 2 sin 1 + 2 sin4 sin 1 + + 2 sin88 sin 1 + sin 1 :
2 sin 2 sin1 = cos 1 , cos 3 2 sin 4 sin1 = cos 3 , cos 5 2 sin88 sin 1 = cos 87 , cos 89 :
Hence
Thus x = cot 1 , as required.
x sin 1 = cos 1 , cos 89 + sin 1 = cos 1 :
8 Next we give a comment on a problem, and one solution to another from the 1994 Italian Mathematical Olympiad. 2. 1996:204 Italian Mathematical Olympiad 1994. Find all integer solutions of the equation
y2 = x3 + 16:
Comment by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain. The equation has no solutions in integers di erent from x = 0, y = 4. For a proof see Theorem 20, page 102 of W. Sierpinski's Elementary Theory of Numbers. 4. 1996: 204 Italian Mathematical Olympiad 1994 Let r be a line in the plane and let ABC be a triangle contained in one of the halfplanes determined by r. Let A0 , B 0 , C 0 be the points symmetric to A, B , C with respect to r; draw the line through A0 parallel to BC , the line through B 0 parallel to AC and the line through C 0 parallel to AB . Show that these three lines have a common point. Solution by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain.
A0
q
C r
q
q
A
q
q
B B0 C0
q
There is no need that 4ABC be contained in one of the half planes determined by r. If the coordinates of A are x1 ; y1 , the coordinates of B are x2 ; y2 and those of C are x3 ; y3 in a Cartesian coordinate system with r as xaxis, then A0 has coordinates x1 ; ,y1 , B0 has coordinates x2; ,y2 and C 0 has coordinates x3 ; ,y3.
We have:
The equation of the line through A0 parallel to BC is The equation of the line through B 0 parallel to CA is The equation of the line through C 0 parallel to AB is
y2 , y3 x , x2 , x3y , x1y2 , y3 , y1x2 , x3 = 0: y3 , y1 x , x3 , x1y , x2y3 , y1 , y2x2 , x1 = 0: y1 , y2 x , x1 , x2y , x3y1 , y2 , y3x1 , x2 = 0:
Since the three equations when added together vanish identically, the lines represented by them meet in a point.
9 Its coordinates are found, by solving between any two, to be
x2 + x2 x3 y2 , y3 + x2 + x3 x1 y3 , y1 + x2 + x1 x2y1 , y2 ; 1 2 3 x1y2 , y3 + x2 y3 , y1 + x3 y1 , y2 2 2 2 y1 + y2y3x2 , x3 + y2 + y3y1x3 , x1 + y3 + y1y2 x1 , x2
: x1y2 , y3 + x2y3 , y1 + x3y1 , y2
Now we jump to the October 1996 number of the Corner, and two solutions to problems of the 10th IBEROAMERICAN Mathematical Olympiad 1996: 251 252 . 1. Brazil. Determine all the possible values of the sum of the digits of the perfect squares. Solution by Mansur Boase, student, St. Paul's School, London, England. The squares can only be 0, 1, 4 or 7 mod 9. Thus the sum of the digits of a perfect square cannot be 2, 3, 5, 6 or 8 mod 9, since the number itself would then be 2, 3, 5, 6 or 8 mod 9. We shall show that the sum of the digits of a perfect square can take every value of the form 0, 1, 4 or 7 mod 9.
10m , 12 = 102m , 2 10m + 1 = 9 9 z 9 8 0 z 0 1; m 1:  
m,1 m,1
The sum of the digits is 9m, giving all the values greater than or equal to 9 congruent to 0 mod 9
10m , 22 = 102m , 4 10m + 4 = 9 9 z 9 6 0 z 0 4; m 1:  
m,1 m,1
The sum of the digits is 9m + 1, which gives all values greater than or equal to 10 congruent to 1 mod 9.
10m , 32 = 102m , 6 10m + 9 = 9 9 z 9 4 0 z 0 9; m 1:  
m,1 m,1
The sum of the digits is 9m +4, which takes every value greater than or equal to 13 which is congruent to 4 mod 9
10m , 52 = 102m , 10m+1 + 25 = 9 z 9 0 0 z 0 2 5:  
m,1 m,1
The sum of the digits is 9m , 1 + 7 = 9m , 2, from which we get every value greater than or equal to 7 congruent to 7 mod 9. We have taken care of all the integers apart from 0, 1, 4, which are the sums of the digits of 02 , 12 , and 22 respectively.
10
5. Spain. The inscribed circumference in the triangle ABC is tangent to BC , CA and AB at D, E and F , respectively. Suppose that this circumference meets AD again at its midpoint X ; that is, AX = XD. The lines XB and XC meet the inscribed circumference again at Y and Z , respectively. Show that EY = FZ . Solution by Toshio Seimiya, Kawasaki, Japan. A Since BFY = BXF and FBY = XBF we have 4BFY and 4BXF are similar, so that FY : FX = BF : BX: 1 X Similarly we get E DY : DX = BD : BX: 2 F As BF = BD, we have from 1 and Z Y 2 that B C FY : FX = DY : DX: D
Since AX = DX we get
Since X , F , Y , D are concyclic we have
FY : FX = DY : AX:
3
FY D = AXF: 4 Thus we get from 3 and 4 that 4FY D is similar to 4FXA. Hence Y FD = XFA = XDF so that FY kXD. Similarly we have EZ kXD. Thus FY kEZ . Therefore FY ZE is an isosceles trapezoid and then EY = FZ .
We now turn to two solutions to the Maxi Finale 1994 of the Olympiade Math
matique Belge 1996: 253 254 . e 1. Un pentagone plan convexe a deux angles droits non adjacents. Les deux c^ t
s adjacents au premier angle droit out des longueurs egales. Les oe
deux c^ t
s adjacents au second angle droit ont des longueurs egales. En remoe
pla ant par leur point milieu les deux sommets du pentagone situ
s sur un c e seul c^ t
de ces angles droits, nous formons un quadrilat re. Ce quadrilat re oe e e admetil n
cessairement un angle droit? e
11 land. Solution by Mansur Boase, student, St. Paul's School, London, Eng
Let the pentagon be ABCDE and let ABC = AED = 90 . Let the midpoint of CD be M . Then the problem is equivalent to proving that given any triangle ACD with isosceles right triangles constructed on AC and AD, giving points B and E, BM = 90 and so quadrilateral BMEA must have a right angle. This result is well known. One very nice proof is that if L and N are the midpoints of AC and AD respectively then 4BLM 4MNE BL = LC = MN , LM = ND = EN and
BLM = 90 + CLM = 90 + CAD = 90 + MND = MNE: And since MLkND and ND ? EN , ML ? EN and similarly BL ? NM . Therefore BM ? EM since the other two pairs of sides are perpendicular.
4. Le plan contientil 1994 points distincts non tous align
es tels que la distance entre deux quelconques d'entre eux soit un nombre entier? Solution by Mansur Boase, student, St. Paul's School, London, England. There are 1994 pairs of solutions m;n with m n satisfying
2mn = 23988 20; 23987; : : : ; 21993; 21994 : So there are 1994 Pythagorean triples of the form m2 , n2 ; 2mn; m2 + n2 with one side of length 23988. Hence we can form a line of 1993 collinear points and place a 1994th point a distance 23998 above this line so that, with the correct arrangement
of the points, if it is joined to any other point, the resulting line is the hypotenuse corresponding to a Pythagorean triple and is therefore integral. Also the distance between any two points on the line will be integral. a1994
23988
a1993
:::
a3
a2
a1
That completes the Corner for this issue. Olympiad season is fast upon us. Send me your contest materials and your nice solutions for use in the Corner.
12
BOOK REVIEWS
Edited by ANDY LIU
MiniReviews Update
Andy Liu
This is an update of earlier MiniReviews see 1 to 9 , and is an abbreviated form of what appeared in the journals Delta K and AGATE. Written permission for reprint has been granted by the Alberta Teachers' Association which publishes the above two journals.
A. Mir Publishers' Little Mathematics Library Series see
ation of America see 2
1 This excellent series has become an unfortunate casualty of the demise of the former Soviet Union. Lost also are Mathematics Can Be Fun and Fun with Maths and Physics featured in Section I.
B. New Mathematical Library of the Mathematical Associ
In additional to the new titles listed below, there is also a revised edition of an earlier volume, Graphs and Their Uses. It was 10 in the series, but is now 34. Two further volumes have also been published recently. USA Mathematical Olympiads: 19721986, by Murray Klamkin, 1988. ISBN 0883856344. This book collects the problems of the rst fteen USA Mathematical Olympiads. While they are presented chronologically, the solutions are grouped according to subject matters, which facilitates using this book for training sessions. There is a very useful 10page glossary of mathematical terms and results, and a most extensive bibliography. Exploring Mathematics with your Computer, by Arthur Engel, 1993. see 16 Game Theory and Strategy, by Philip Stra n, 1993. see 17
C. Martin Gardner's Scienti c American Series see
3 Two more volumes have appeared, and there will be a fteenth and nal volume, about to be released by SpringerVerlag. Several earlier volumes have also changed publishers. The Mathematical Association of America has acquired Martin Gardner's New Mathematical Diversions from Scienti c American, Martin Gardner's 6th Book of Mathematical Diversions from Scienti c American, Mathematical Carnival, Mathematical Magic Show
13 and Mathematical Circus. The University of Chicago Press has acquired The Scienti c American Book of Mathematical Puzzles and Diversions, The 2nd Scienti c American Book of Mathematical Puzzles and Diversions and The Unexpected Hanging and Other Mathematical Diversions. Penrose Tiles to Trapdoor Ciphers, 1989, W. H. Freeman, 1997, Mathematical Association of America. ISBN 0883855216. Topics covered are Penrose tilings, Mandebrot's fractals, Conway's surreal numbers, mathematical wordplay, Wytho 's version of the game Nim", mathematical induction, negative numbers, dissection puzzles, trapdoor ciphers, hyperbolas, the new version of the game Eleusis", Ramsey theory, the mathematics of Berrocal's sculptures, curiosities in probability, Raymond Smullyan's logic puzzles, as well as two collections of short problems. The book contains a surprise ending, the resurrection of Dr. Matrix! There is also an update chapter. Unfortunately, Figures 3 to 6 are inadvertently left out. They are reproduced below.
Figure 3.
Figure 4.
Figure 5. Figure 6. Fractal Music, Hypercards and More, 1992, W.H. Freeman. ISBN 0716721899. Topics covered are fractal music, the Bell numbers, mathematical zoo, Charles Sanders Peirce, twisted prismatic rings, coloured cubes, Egyptian fractions, minimal sculptures, tangent circles, time, generalized ticktacktoe, psychic wonders and probability, mathematical chess problems, Hofstadter's Godel, Escher, Bach, imaginary numbers, some accidental patterns, packing squares, Chaitin's irrational number , as well as one collection of short problems.
14
D. Books from W.H. Freeman & Company, Publishers see
4 Note that the most recent book in Martin Gardner's Scienti c American Series is a Freeman publication. The rst of the books listed below has actually gone out of print, but fortunately Dover Publications Inc. has decided to pick it up. The Puzzling Adventures of Dr. Ecco, by Dennis Shasha, 1988. The title character calls himself an omniheurist, solver of all problems mathematical. The narrative is by a Watsonesque companion, Prof. Scarlet. Ecco's clients range from government o cials, industrialists, eccentric millionaires to no less than the President of a Latin American country. They brought him important, instructive and interesting problems in discrete mathematics, all of which Ecco solves to their satisfaction. The book concludes with the mysterious disappearance of Dr. Ecco. Codes, Puzzles, and Conspiracy, by Dennis Shasha, 1992. see 11 New Book of Puzzles, by Jerry Slocum and Jack Botermans, 1992. see 12 Another Fine Math You've Got Me Into : : : , by Ian Stewart, 1992. see 13
E. Oxford University Press Series on Recreations in MathematThe Mathematics of Games, by John D. Beasley, 1989. ISBN 0198532067. This book analyses mathematically some card and dice games, nimtype games, a version of John Conway's Hackenbush", as well as providing a mathematical model for the study of some sports games. The principal techniques are counting, probability and game theory. Some mathematical puzzles are also considered. The Puzzling World of Polyhedral Dissections, by Stewart T. Co n, 1990. ISBN 0198532075. This is a labour of love from an expert craftsman. Starting with two chapters of twodimensional geometric puzzles, the author eases the readers gently into the third dimension and soon launches into his specialty, the burrs, which are assemblies of interlocking notched sticks. The book is profusely illustrated with blackandwhite line drawings and photographs. It concludes with a chapter on woodworking techniques. More Mathematical Byways, by Hugh ApSimon, 1990. ISBN 019217777X. This book contains fourteen chapters. The rst three form a sequence but the others are independent of each other. Unlike the earlier volume by the same author, the problems are of uneven level of di culty, ranging from the relatively simple Alphametics to others which require a considerable
ics see 5
15 amount of what the author calls slog". One of the chapters, titled Potential Pay, is not really a problem but a commentary on a classic paradox.
F. Raymond Smullyan's Logic Series see
6 Satan, Cantor, and In nity, 1992, Alfred A. Knopf. ISBN 0679406883. In this book, a remarkable character known as the Sorcerer makes his debut. He escorts the readers on a wonderful guided tour, visiting familiar grounds such as the domains of the knights and the knaves, and those bordering on the land of Godel. There are also ventures into new territories, including an island where intelligent robots create others which can continue this process ad in nitum. This eventually leads to the pioneering discoveries on in nity of the great mathematician, Georg Cantor. The readers may be amused to discover how Satan got into the picture.
G. Dolciani Mathematical Expositions Series of the Mathematical Association of America see 7
In addition to the new titles listed below, three further volumes have been published recently. More Mathematical Morsels, by Ross Honsberger, 1991. see 10 ISBN 0883853132. This is a collection of 57 problems, almost all of which are taken from the Canadian Mathematical Society's journal Crux Mathematicorum, plus further gleanings" from its famed Olympiad Corner. Old and New Unsolved Problems in Plane Geometry and Number Theory, by Victor Klee and Stan Wagon, 1991. ISBN 0883853159. This book is divided into two halves, as suggested by the title, though the second half also covers problems about some interesting real numbers. Each half consists of two parts. In the rst, twelve problems are presented, giving the statement, known results and background information. In the second, the same twelve problems are reexamined for further results and extensions. Each half concludes with a comprehensive bibliography. Although the problems are unsolved, and therefore hard, it is not impossible for them to yield to an inspired attack. Even if this does not happen, gifted students who are willing to attempt them will nd their mathematical talent enhanced. Problems for Mathematicians Young and Old, by Paul Halmos, 1992. ISBN 0883853205. The fourteen chapters of this book are titled Combinatorics, Calculus, Puzzles, Numbers, Geometry, Tilings, Probability, Analysis, Matrices, Algebra, Sets, Spaces, Mappings and Measures. The author, a ranking mathematician and master expositor, wrote this book for fun, and hoped that it will be read the same way.
16
Excursions in Calculus, by Robert Young, 1992. ISBN 0883853175. The subtitle of this book is An Interplay of the Continuous and the Discrete. Using calculus as a unifying theme, the author branches into number theory, algebra, combinatorics and probability. The book contains a large collection of exercises and problems. The Wohascum County Problem Book, by George Gilbert, Mark Krusemeyer and Loren Larson, 1993. see 14 Lion Hunting & Other Mathematical Pursuits, edited by Gerald Alexanderson and Dale Muggler, 1995. see 18 The Linear Algebra Problem Book, by Paul Halmos, 1995. ISBN 0883853221. The whole book is a sequence of structured problems. Like the preceding volume, most of this book is beyond high school level. However, the introductory problems are certainly not intimidating, and inquisitive students may be lured into a most rewarding exploration, laying a good foundation for their undergraduate studies.
H. Books from Dover Publications, Inc. see
8 Excursions in Number Theory, by Stanley Ogilvy and John Anderson, 1988. ISBN 0486257789. This book covers the basics of classical number theory. Topics include prime numbers, congruences, Diophantine equations and Fibonacci numbers. The narrative style is very soothing. It concludes with 20 pages of elaborations and commentary on some ner points raised in the text. Excursions in Geometry, by Stanley Ogilvy, 1990. ISBN 0486265307. The rst half of this book is on inversive geometry, and the second half on projective geometry. These two topics are linked by the concept of crossratio and the study of the conic sections. It is in the same style as the preceding volume. Excursions in Mathematics, by Stanley Ogilvy, 1994. ISBN 048628283X. The original title of this volume was Through the Mathescope. The opening chapter is titled What Do Mathematicians Do? It is followed by lively tours of number theory, algebra, geometry and analysis. The last chapter is titled Topology and Apology.
I. Books from Various Publishers see
9 Selected Problems and Theorems in Elementary Mathematics has been acquired by Dover and renamed The USSR Olympiad Problem Book. Dover has also picked up The Moscow Puzzles. The Mathematical Association of America has published Five Hundred Mathematical Challenges, comprising
17 the rst ve booklets of the project, 1001 Problems in High School Mathematics. see 19 The Canadian Mathematical Society has published The Canadian Mathematical Olympiad, 1969 1993, edited by Michael Doob and Claude La amme. This book combines two earlier volumes, The First Ten Canadian Mathematics Olympiads, 1969 1978 and The Canadian Mathematics Olympiads, 19791985, and adds the contests from 1986 to 1993. see 15 1. A. Liu, Minireviews, MIR Publishers' Little Mathematics Library Series, 1989, 142147. 2. A. Liu, Minireviews, MAA's New Mathematical Library Series, 1989, 171176. 3. A. Liu, Minireviews, Martin Gardner's Mathematical Games Series, 1989, 202206 4. A. Liu, Minireviews, Books on popular mathematics from W. H. Freeman, 295297. 5. A. Liu, Minireviews, Oxford University Press' Recreation in Mathematics Series, 1990, 4243. 6. A. Liu, Minireviews, Raymond Smullyan's books on logic puzzles, 1990, 106108. 7. A. Liu, Minireviews, MAA's Dolciani Mathematical Exposition Series, 1990, 237238. 8. A. Liu, Minireviews, Books on popular mathematics from Dover, 1991, 1113. 9. A. Liu, Minireviews, Books on popular mathematics from various publishers, 1991, 7477. 10. A. Liu, Review of More Mathematical Morsels, 1991, 235236. 11. A. Liu, Review of Codes, Puzzles and Conspiracy, 1992, 204205. 12. A. Liu, Review of New Books of Puzzles, 1993, 1314. 13. R.K. Guy, Review of Another Fine Math You've Got Me Into, 1993, 4647. 14. A. Liu, Review of The Wohascum County Problem Book, 1993, 202. 15. R. Geretschlager and G. Perz, Review of The Canadian Mathematical Olympiad Book, 19691993, 1994, 1516. 16. G.C. Denham, Review of Exploring Mathematics with Your Computer, 1994, 134135. 17. M. Larsen, Review of Game Theory and Strategy, 1995, 88. 18. A. Liu, Review of Lion Hunting and Other Mathematical Pursuits, 1995, 304305. 19. M. Kuczma, Review of Five Hundred Mathematical Challenges, 1996, 313316.
Cross References to other entries in Crux Mathematicorum:
18
CANADIAN MATHEMATICAL SOCIETY AWARD FOR CONTRIBUTIONS TO MATHEMATICAL EDUCATION
The Adrien Pouliot Award
Eric Muller, Chair, CMS Education Committee
In 1995, in celebration of its 50th Anniversary, the Canadian Mathematical Society CMS instituted an Award for Contributions to Mathematics Education. This award is to recognize individuals or teams of individuals who have made signi cant and sustained contributions to mathematics education within Canada. Such contributions are to be interpreted in the broadest possible sense and might include: community outreach programmes, the development of a new programme in either an academic or industrial setting, publicizing mathematics so as to make it accessible to the general public, developing mathematics displays, establishing and supporting mathematics conferences and competitions for students, etc.. The prize is named after Adrien Pouliot. But who was Adrien Pouliot? Pouliot is certainly one of the main scienti c gures of the twentieth century in the Province of Quebec. Trained as a civil engineer at the Ecole Polytechnique de Montreal he started teaching mathematics at the Universite Laval in the early 1920s. He completed a `licence' in sciences mathematics at the Sorbonne in Paris 1928. In the words of Professor Hodgson of Laval 1 Pouliot was the main person behind the growth of science and mathematics in the Province of Quebec. To understand this one has to recall that the rst part of the century, the education system in Quebec was preparing more for traditional jobs priests, physicians than for science. Pouliot led with success a public movement to have science and math become a major part of the education system. Because of that he was accused to be antihumanities. His response was to learn Greek; when he died in 1980 he was working on a GreekFrench dictionary". Pouliot's in uence extended to other parts of Canada and beyond through his work as a governor and o cer of RadioCanada and through his participation on many missions to other countries. Pouliot was awarded a number of honorary doctorates in Canada, France and Italy. Recipients of the rst three awards are known to many Crux readers. The rst award was made in 1995, and was awarded to Professor Edward Barbeau of the University of Toronto, Ed as he is known to students, mathematics teachers, mathematicians and university colleagues, the press, local school board trustees, and all who come into contact with
19 him. Ed was recognized for his numerous contributions to mathematics education, sustained over so many years and in so many di erent forums. He has always been willing to spend the time to communicate with interested mathematics teachers, students and the general public. He continues to be involved with the International Mathematical Olympiad, an involvement that started some 16 years ago. Many have bene ted from his innovative and challenging mathematics publications accessible to secondary and post secondary students. The 1996 recipient was Professor Bruce Shawyer of Memorial University of Newfoundland. Bruce is of course the present EditorinChief of Crux. He was recognized for his substantial contributions to many di erent mathematical competitions and challenges. He proposed and was the foundingmember of the Newfoundland and Labrador Mathematical Association and was involved in the development of the Newfoundland Teachers' Association Senior Math League and the Junior High Math Challenge. Bruce was most visible as the Chief Operating O cer of the 36th International Mathematical Olympiad that was held in Canada in 1995. The third award, in 1997, went to a team from the University of Waterloo, a team of university and school mathematics educators, Ed Anderson, Don Attridge, Ron Dunkley and Ron Scoins. They were instrumental in starting the Canadian Mathematics Competition that grew out of their local competition activities in 1962. They have continued to nurture and sustain these competitions that have grown and, in the past year, involved over 200,000 students. These contests not only involve students but also support a network of mathematics teachers throughout Canada. They provide an environment that stimulates professional development and provide opportunities for mathematics teachers from schools and universities to meet and exchange ideas. 1
Reference Hodgson, Bernard, CMS Notes, JanFeb 1996
20
Cyclic ratio sums and products
Branko Grunbaum
The well known classical theorems of Menelaus and Ceva deal with certain properties of triangles by relating them to the products of three ratios of directed lengths of collinear segments. Less well known is a theorem of Euler 2 which states, in the notation of Figure 1, that
3 X
for every triangle T = A1 A2 A3 .
j =1
jjQBj =Aj Bj jj = 1
A1
t
B3
t t
B2
t
Q
t
t
t
A2
A theorem of Euler states that if Bj is the intersection of the line Aj Q with the side of the triangle A1 A2 A3 opposite to Aj , then P3=1 jjQBj =Aj Bj jj = 1. Here, and j throughout this note, jjMN=RS jj means the ratio of signed lengths of the collinear segments MN and RS .
B1 Figure 1.
A3
However, while the theorems of Menelaus and Ceva have been generalized to arbitrary polygons, and in many other ways  see, for example, 4 5 6  until very recently there have been no analogous generalizations of Euler's result. One explanation for this situation may be that attempts at straightforward generalizations lead to invalid statements. An example of such a failed theorem" is given by the question whether, in the notation of Figure 2, n
X
equals 1 or some other constant independent of Q and the polygon.
Research supported in part by NFS grant DMS9300657. Copyright c 1998 Canadian Mathematical Society
j =1
jjQBj =Aj Bj jj
21
A1
t
B4 A2
t
t
t
B3
t
A5
Q
t t
B5 A3
t
B2
t
Attempts to generalize Euler's theorem in the form Pn=1 jjQBj =Aj Bj jj = const j necessarily fail for n 3 here n = 5. However, as shown by Shephard 9 , it is possible to nd weights wj which P depend on the polygon but not on the position of Q, such that n w jjQB =A B jj = 1.
j =1 j j j j
B1 Figure 2.
t
t
A4
Recently, Shephard 9 had the idea, apparently not considered previously, of attaching to the ratios rj = jjQBj =AjBj jj certain weights wj , which dependnon the polygon P = A1 ; A2 ; : : : ; An but not on the point Q, such P that j =1 wj rj = 1. In fact, Shephard established a much more general result in this spirit; its complete formulation would lead us too far from the present aims. By sheer chance, the same day I received from Shephard a preprint of 9 , I happened to read 7 , in which two di erent sums of ratios appear, one in Bradley's solution, the other in Konen
's comments. This coincic y dence led me to consider whether these results could be generalized along Shephard's idea. As it turns out, the answer is a rmative, and leads to a number of other results. Let P = A1 ; A2 ; : : : ; An be an arbitrary n gon, and Q an arbitrary point, subject only to the condition that all the points Bj mentioned below are well determined. On each side Aj Aj +1 of P understood as the unbounded line the point Bj is the intersection with the line through Q parallel to Aj +1 Aj +2 . Here, and throughout the present note, subscripts are understood mod n. This is illustrated in Figure 3 by an example with n = 5. We are interested in the ratios
rj = jjBj Aj+1=Aj Aj+1jj:
We denote by 4UV W the signed area of the triangle 4UV W with respect to an arbitrary orientation of the plane and, more generally, by 4P the signed area of any polygon P, calculated with appropriate multiplicities for the di erent parts if P has selfintersections.
22
s
A1
B1
s
A2 B2
s s
B5
s
s
Q
s
A5
A3
s
s
B3
s
s
B4
A4
The point Bj is the intersection of the line Aj Aj +1 with the parallel through Q to the line Aj +1 Aj +2 . The ratios rj = jjBj Aj +1 =Aj Aj +1 jj of directed segments are considered in Theorem 1.
Figure 3.
Theorem 1. For each polygon P we have n=1 wj rj = 4P for all Q, where j wj = 4Aj Aj+1Aj+2 are weights that depend on the polygon P but are independent of the point Q. For a proof it is su cient to note that i by straightforward calculations or by easy geometric arguments it can be shown that rj = 4QAj +1 Aj +2 =4Aj Aj +1 Aj +2 ; and P P ii therefore the sum n=1 wj rj is equal to n=1 4QAj +1 Aj +2 = 4P, j j since the triangles with vertex Q triangulate the polygon P.
P P As a corollary we deduce at once that n=1 wj sj = n=1 wj ,4P, j j where sj = jjAj Bj =Aj Aj +1 jj = 1 , rj . In the special case that P is a regular n=d gon, all the weights wj are equal to the value w = 4 sin3 d=n cosd=n. The regular n=d gon has n vertices and surrounds its centre d times. Successive vertices are obtained by rotation through 2d=n, see 1 . It is usually assumed
P
23 that n and d are coprime, but this is a restriction that is unnecessary here and in most other contexts, and downright harmful in some cases  see, for example, 3 . Hence, in this case one can divide throughout by w, and the result becomes
P n rj = 4w = 4 sin2d=n : 1 j =1 Since the ratios rj involve only collinear lengths, the sum is invariant under a nities, and so the result 1 remains valid for all a neregular n=d gons P. An n=d gon is a neregular if it is the image of a regular n=d gon under a nonsingular a nity. Thus in this special case we
actually achieve the analogue of the generally invalid statement mentioned above. Since all triangles are a neregular, this establishes the condition for concurrency found by Konen
, mentioned in 7 . We note that Shephard c y obtains in 9 the analogous generalization of Euler's result to a neregular n gons. In the a ne case, the above corollary can be simpli ed in the same way. For n = 3 this yields the condition for concurrency obtained by Bradley in 7 . From the above it follows that in the case of a neregular polygons but not for general polygons we have
n X
d=n 2 jjBj Cj =Aj Aj+1 jj = , n cos2d=n ; 2 2 sin j =1 where the Cj is the intersection of the line Aj Aj +1 with the parallel through Q to the line Aj,1Aj see Figure 4. For n = 3, the righthand side of 2
n X
equals 1, and the result coincides with Problem 16 in 8 . It may be observed that for n = 3 and d = 1, the righthand side of condition 1 equals 1, and the equality to 1 of the ratio sum is necessary and su cient for the three parallels to the sides of the triangle to be concurrent, just as the equality to 1 of the product in Ceva's theorem for triangles is necessary and su cient for the concurrence of the Cevians. However, for n 3 it is not obvious that the weights given above are the only ones which yield the righthand constants for all Q, although one may conjecture that this is the case. Naturally, for particular choices of P and Q other weights may be used. The expression for rj obtained in i, together with the analogous formula for the ratio tj = jjAj Cj =Aj Aj +1 jj in the notation of Figure 4 leads at once to the following:
24
s
A1 C1
s s s
B1 C5 B5
s s
s
A2 B2
s
Q
s s
A5
C4
A3 C2
s
s
B3
s
s
B4
C3
s
s
A4
The point Bj is obtained as in Figure 3, while the point Cj is the intersection of the line Aj Aj +1 with the parallel through Q to the line Aj ,1 Aj . The ratios rj = jjBj Aj +1 =Aj Cj jj of directed segments are considered in Theorem 2.
Figure 4.
Theorem 2. For each polygon P with we have
for all Q. Finally, since QBj Aj +1 Cj +1 is a parallelogram for every j , we also Q have n=1 jjBj Q=QCj +2jj = 1. j This last is a Cevatype result which seems not to have been noticed previously. A referee's suggestions for improved presentation are acknowledged with thanks.
rj = j =1 tj
n Y
n Y
j =1
jjBj Aj+1=Aj Cj jj = 1
25
References
1 H.S.M. Coxeter, Introduction to Geometry, Wiley, New York 1969. 2 L. Euler, Geometrica et sphaerica quedam, M
moires de l'Acad
mie e e des Sciences de St. Petersbourg 5 1812, pp. 96 114. This was submitted to the Academy on 1 May 1789; the actual publication date is 1815. Reprinted in: L. Euleri Opera Omnia, Ser. 1, vol. 26, pp. 344 358; Fussli, Basel 1953. 3 B. Grunbaum, Metamorphoses of polygons. The Lighter Side of Mathematics, Proc. Eug ne Strens Memorial Conference, R.K. Guy and e R.E. Woodrow, eds., Math. Assoc. of America, Washington, D.C. 1994, pp. 35 48. 4 B. Grunbaum and G.C. Shephard, Ceva, Menelaus, and the area principle, Math. Magazine 68 1995, 254 268. 5 B. Grunbaum and G.C. Shephard, A new Cevatype theorem, Math. Gazette 80 1996, 492 500. 6 B. Grunbaum and G.C. Shephard, Ceva, Menelaus and selftransversality, Geometriae Dedicata 65 1997, 179 192. 7 H. Gulicher, Problem 1987. Crux Math. 20 1994, p. 250. Solution, ibid. 21 1995, pp. 283 285. 8 J.D.E. Konhauser, D. Velleman and S. Wagon, Which Way Did The Bicycle Go? Dolciani Math. Expositions No. 18. Math. Assoc. of America, Washington, DC, 1996. 9 G.C. Shephard, Cyclic sums for polygons, Preprint, August 1997. University of Washington, Box 354350, Seattle, WA 981954350 email: grunbaum@math.washington.edu
26
THE SKOLIAD CORNER
No. 27 R.E. Woodrow
To start the new year we give the problems of the Preliminary round of the British Columbia Colleges Junior High School Mathematics Contest for 1997. This round of the contest is written in the schools. The top students are invited to a mathematics day at which the nal round is written, and they get to participate in talks, tours, and a presentation ceremony. My thanks go to John Grant McLoughlin, now of the Faculty of Education, Memorial University of Newfoundland, who participated in the writing process while he was at the Okanagan University College.
BRITISH COLUMBIA COLLEGES Junior High School Mathematics Contest
Preliminary Round 1997
1. A number is prime if it is greater than one and is divisible only by itself and one. The number 1997 is prime but the sum of the digits of 1997 is not. The sum of the prime divisors of the sum of the digits of 1997 is: a 6 b 8 c 10 d 13 e 15 2. If x = 3, which expression has a value di erent from the other four? a 2x2 b x2 + 9x c 12x d x2 x , 12 e 2x2 x , 1 3. Suppose that you lace your shoes using the pattern shown below and that the horizontal spacing between the eyelets is 4 centimetres and the vertical spacing is 3 centimetres. If there is a total of ten eyelets and there are 10 centimetres of lace left at each of the upper eyelets, then the total length, in centimetres, of the lace used for one shoe is:
a 44
b 54
c 63
d 64
e 73
27
4. Consider two unequal numbers. If we subtract half the smaller number from both numbers, the result with the larger number is three times as large as the result with the smaller number. How many times is the larger number greater than the smaller number? a 2 b 3 c 6 d 8 e 9 5. Suppose your hockey card collection grew by 20 last year and last year it was 30 larger than the year before. By what percentage has your collection grown during the last two years? a 25 b 50 c 56 d 60 e 66 6. Antonino, the chocolate freak, eats x bars of chocolate every y days. In a week he eats how many bars of chocolate? 1 a 7yx b 7xy c 7xy d 7xy e 7xy 7. Six mattresses, each of which was originally 20 centimetres thick, are piled in a stack in a warehouse. Each mattress is compressed by one tenth each time an additional mattress is added to the stack. The height h of the stack, in centimetres, satis es: a h 70 b 70 h 86 c 86 h 92 d 92 h 110 e h 110 8. If a man walks to work and rides back home it takes him an hour and a half. When he rides both ways, it takes 30 minutes. How long would it take him to make the round trip by walking? 1 b 1 1 hrs c 1 1 hrs d 3 2 hrs e 2 3 hrs a 2 1 hrs 2 4 2 4 9. Tanya was asked to add 14 to a certain number and then divide the result by 4. Instead she rst added 4 and then divided the answer by 14. Her result was 5. If Tanya had followed the instructions correctly, her result would have been: a 5 b 20 c 25 d 66 e 70 10. In the diagram, a circle with centre at point C and radius 15 overlaps a circle with centre at point D and radius 20. How many square units larger is the shaded region on the right than that on the left?
C
a 400
15
20
D
d 175 d 225 , 400
b 400 , 400 c 175 , 400
28
11. ABCD is a rectangle in which AB is twice as long as BC . E is a point such that ABE is an equilateral triangle. M is the midpoint of BE . The measure, in degrees, of CMB is: A B
M
D E
C
a 30
are: a 2800; 3600 ; 5400; 6200 b 6200; 2800; 3600; 5400 c 6200; 2800 ; 5400; 3600 d 2800 ; 5400; 3600 ; 6200 e 3600; 6200 ; 2800; 5400
b 60 c 75 d 90 e 150 12. The numbers 2800, 3600, 5400 and 6200 listed in increasing order
13. Think about a stopped clock. It shows the correct time twice a day. Now consider a working clock that gains ve seconds per day. If it is never adjusted, it will rst show the correct time after: a 48 days b 360 days c 720 days d 8640 days e 17280 days 14. The diagrams below show ve pieces of rope. Imagine grasping the two loose ends of each piece rmly, then imagine pulling them until you have a straight piece of rope  either with a knot or without one. Which of the ve will give you a knot? a b c
d
e
29
15. In how many ways can 75 be expressed as the sum of at least two positive integers, all of which are consecutive? a 1 b 2 c 3 d 4 e 5
Last issue we gave the thirty problems of the 1996 Math
matiques 1997: 473 478 . Here are solutions. e 1. B 2. B 3. E 4. D 6. C 7. B 8. E 9. B 11. B 12. C 13. B 14. B 16. E 17. A 18. B 19. D 21. A 22. D 23. B 24. D 26. A 27. D 28. B 29. B Kangourou des
5. 10. 15. 20. 25. 30.
E E C C D C
Note: A number of typos were missed in proofreading the problems. In the Figure for Number 2, apparently four, not three, of the triangles should have been shaded  but didn't look to be the case on the photocopy I had! In the Figure for Number 16 the six angles not sharing a common vertex should be marked. Number 18 should start Les c^ t
s". oe More seriously in Number 26: AD = DC = CB .
That completes the Skoliad Corner for this number. I need good contest materials at this level from around the world. Please send me materials for use in the Corner as well as suggestions for future directions for the column.
30
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Shreds and Slices
The K Method In this brief article, we will present a useful geometrical tool, which we coin The K Method". As is convention, let P denote the area of polygon P . We will further stipulate that if P is labelled counterclockwise, then P is positive, and negative otherwise. This sign convention will matter. Let ABC be a triangle, labelled counterclockwise, and let P be a point in the plane. Let K = ABC , KA = PBC , KB = PCA , and KC = PAB see Figure 1. Because of the signed areas, we have that for all points P , inside or outside the triangle ABC ,
K = KA + KB + KC : A KC B KB P KA C0 C B A
P
Figure 2
B0 C
Figure 1
A0
31 Now, extend AP to A0 on BC , and de ne B 0 and C 0 similarly see Figure 2. Recall that triangles with the same height have areas in proportion to their bases. Then we have
AP = PAB = PCA = PAB + PCA = KB + KC ; and PA0 PBA0 PA0 C PBA0 + PA0C KA BA0 = ABA0 = PBA0 = ABA0 , PBA0 = KC : A0C AA0C PA0C AA0 C , PA0 C KB
We can similarly derive that
BP = KA + KC ; CP = KA + KB ; CB0 = KA ; and AC 0 = KB : PB0 KB PC 0 KC B0A KC C 0B KA Again, these hold regardless of whether P is inside or outside triangle ABC , but only with directed line segments that is, if PQ and PR are in di erent
directions, then their ratio will be negative. These expressions can be very useful in problems which involve these ratios. In fact, one half of Ceva's theorem is now trivial and with some consideration, so is the other half. Problem 1. Consider triangle P1 P2 P3 and a point P within the triangle. Lines P1 P , P2P , P3 P intersect the opposite sides in points Q1 , Q2 , Q3 respectively. Prove that, of the numbers
at least one is less than or equal to 2 and at least one is greater than or equal to 2. 1961 IMO, Problem 4 Solution. Let us use the same understood notation as above. Without loss of generality, we can assume that KA KB KC relabel the triangle if necessary. Then
P1P ; P2P ; P3P ; PQ1 PQ2 PQ3
P3P = KA + KB KC + KC = 2; and PQ3 KC KC P1P = KB + KC KA + KA = 2: PQ1 KA KA Problem 2. In triangle ABC , A0 , B 0 , and C 0 are on sides BC , AC , and AB , respectively. Given that AA0 , BB 0 , and CC 0 are concurrent at the AO BO CO point O, and that 0 + 0 + 0 = 92, nd the value of OA OB OC AO BO CO . 1992 AIME OA0 OB0 OC 0
32
Solution. The given implies
and further that Then
KA + KB + KA + KC + KB + KC = 92; KC KB KA
2 2 2 2 2 2 KA KB + KAKB + KA KC + KA KC + KB KC + KB KC = 92KAKB KC :
AO BO CO = KB + KC
KA + KC
KA + KB
OA0 OB0 OC 0 KA KB KC 2 KB + KA K 2 + K 2 KC + KA K 2 + K 2 KC + KB K 2 + 2KA KB KC B A C B C = KA KA KB KC 2 = 92KAKB KC + KKAKB KC = 94: K K
A B C
Problems
1. Prove Ceva's Theorem, which states that AA0 , BB 0 , and CC 0 as in Figure 2 are collinear if and only if
AC 0 BA0 CB0 = 1: C 0B A0C B0 A 2. Let P be a point inside the triangle ABC . Let AP meet BC at D, BP meet CA at E , and CP meet AB at F . Prove that PA PB + PB PC + PC PA 12: PD PE PE PF PF PD
The Red Book of Mathematical Problems, Williams and Hardy
33
Powers of Two
student, University of Waterloo We will solve several problems involving the number 2 and, in the process, survey various interesting mathematical results. Problem 1. Prove that for all positive integers n, pn x = xn , 2 is irreducible in Q x . Remark. Q x is the set or more precisely, ring of polynomials in x with rational coe cients. Hence, a polynomial is irreducible in Q x if and only if it cannot be factored nontrivially as a product of polynomials, also with rational coe cients. Solution. This follows directly from Eisenstein's Criterion, but we will take a more1basic approach. By DeMoivre's Theorem, the n roots of pn x in C are 2 n cis 2k , k = 0; 1; 2; : : : ; n , 1, so that n
Wai Ling Yee
pnx =
1 x , 2 n cis 0
!
1 x , 2 n cis
2
x , 2 n cis 2n , 1
: 1 n n
!
!
Suppose pn x = f xg x for some f x, g x 2 Q x , with deg f x = m and n m 0. Note: we only need to consider n 2. Let the roots of f x, respectively g x, be !1 , !2 , : : : , !m , respectively !m+1 , !m+2 , : : : , !1=nso !1, !2, : : : , !nPma permutation ofmthemroots of pnx. Let is n, !i = 2 cis i, and let = k=1 i. Note ,1 Qk=1 !k is the constant term of f x, so
,1m
m Y
But cis = cos + i sin 2 Q implies that sin =m0 and further, that cos = 1. Therefore, the constant term of f x is 2 n . It is easy to show n that 2 m 62 Q, a contradiction. Hence, xn , 2 is irreducible in Q x . Problem 2. If the sum of the proper divisors of n that is, the divisors of n that are less than n is n, then n is called a perfect number. Equivalently, the sum of the divisors of n, n, is 2n. For example, 2 6 = 6 = 1 + 2 + 3 + 6. Classify all even perfect numbers. Solution. By the Fundamental Theorem of Arithmetic, we can express n uniquely, up to permutations, in the form pa1 pa2 pak , where the pi are 1 2 k
Copyright c 1998 Canadian Mathematical Society
k=1
n !k = ,1m2 m
m Y
k=1
n cis k = ,1m2 m cis 2 Q:
34 distinct primes and the ai are positive integers. Then, all divisors of n are of the form p1 1 p2 2 pk k , where 0 i ai. Hence,
n =
k Y
1
ak +1 = pkp ,, 1 : p1 , 1 p2 , 1 k 1 It follows that if gcdm; n = 1, then mn = m n. We claim that if 2n , 1 is prime, then 2n,1 2n , 1 is a perfect number.
2
j =1 ! ! pa1+1 , 1 pa2+1 , 1
1 + pj + p2 + + paj j j
!
In such a case, we have
n 2,1 n, 2n,12n , 1 = 22 , 11 2 n , 1 , 1 2 , 1 n , 12n , 1 + 1 = 2 2n,1 2n , 1: = 2
Now consider an arbitrary even perfect number n. Then n = 2k m, where k 1 and gcd2k ; m = 1. Also, 2k+1 m = 2k m = 2k m = 2k+1 , 1 m. Now, 2k+1 , 1 is odd, so 2k+1 , 1 j m. Let m = d2k+1 , 1, so d is a divisor of m. By the above,
k+1 m m m = 22+1 , 1 = m + 2k+1 , 1 = m + d; k
but m is de ned as the sum of all of the divisors of m. Therefore, m only has the two divisors m and d, so m is prime, d is 1, and so 2k+1 , 1 is prime. We have shown that all even perfect numbers are of the form 2n,1 2n , 1, where 2n , 1 is prime. Primes of this form are called Mersenne primes. Problem 3. Perform the following transformation T on the vector x1 ; x2 ; x3; : : : ; x2n , where the xi are nonnegative integers:
T x1; x2 ; x3; : : : ; x2n = jx1 , x2j; jx2 , x3 j; : : : ; jx2n , x1j: Prove that a nite number of applications of T will transform any such vector
to the zero vector. Solution 1. Notice that mod 2 has the following special property: 1 ,1 mod 2 a ,a mod 2 for all integers a. This implies jaj a mod 2 for all integers a, and that subtraction is the same as addition in mod 2. Therefore, the values jx1 , x2 j, jx2 , x3 j, : : : , jx2n , x1 j mod 2, are congruent to x1 + x2, x2 + x3 , : : : , x2n + x1 mod 2, which will be easier to work with.
35 De ne S x1 ; x2 ; : : : ; x2n = x1 + x2 ; x2 + x3 ; : : : ; x2n + x1 . By the previous remarks, S and T give the same results mod 2. Notice that and
S2x1; x2; x3 ; : : : ; x2n = x1 + 2x2 + x3; x2 + 2x3 + x4 ; : : : ; x2n + 2x1 + x2 ; S3x1; x2; : : : ; x2n = x1 + 3x2 + 3x3 + x4 ; x2 + 3x3 + 3x4 + x5; : : : ; x2n + 3x1 + 3x2 + x3 :
The coe cients seem to be entries in Pascal's triangle. Indeed, we can prove by induction that the ith entry of S k x1 ; x2 ; x3 ; : : : ; x2n is k X k
xj+i, where xr = xs if r s mod 2n.
j =0
What does S 2n x1 ;nx2 ; x3 ; : : : ; x2n look like? First, we claim that if , 1 j 2n , 1, then 2j is even. We prove the more general result which , k states that if p is prime and 1 j pk , 1, then p divides pj . Step 1: For p prime, we determine the highest power of p dividing m!. There are b m c natural numbers less than or equal to m that are divism p ible by p, b p2 c natural numbers less than or equal to m that are divisible by p2, and so on. For each number divisible by at most pa, we have counted it exactly a times in the sums. Therefore, the highest power of p dividing m! is
j
m + m + m + : p p2 p3
Step 2: We calculate the highest power of p dividing pj , where 1 j pk , 1 and p is prime, which is simply the highest power of p dividing the numerator subtracted by the highest power of p dividing the denominator. Applying thek result from the previous step, the highest power of p di, k ! viding pj = j !ppk,j ! is
,
k
pk , j , pk , j : pi pi pi i=1 Each term in the sum is nonnegative, since bx + y c bxc + by c for all reals x, y , and when i = k, $ $ pk , j , pk , j = 1; pk pk pk
1 X
$
1 1 k pk , X j , X pk , j = X pi i=1 pi i=1 pi i=1
$
$
$
!
36 since 1 j pk , 1. Hence, there is at least one factor of p in pj . Now we relate this result to the problem. The parity of the ith entrynof T 2n x1 ; x2 ; x3 ; : : : ; x2n is the same as the parity of the ith entry of S2 x1; x2; x3; : : : ; x2n , which is
,
k
2 X
n
Therefore, the ith entry of T 2n x1 ; x2 ; x3 ; : : : ; x2n is even. We can then pull a factor of 2 out of each of the resulting elements and apply T 2n times to that vector since T is linear, and again, obtain a vector with all entries divisible by 2, and so on. Hence, after t 2n applications of T , each entry in the vector must be divisible by 2t . If yk is used to denote the largest of the 2n entries of then fyk g is a nonincreasing sequence. Furthermore, there exists t such that y1 2t. Then after t 2n applications of T , the resulting vector has entries all divisible by 2t . Since the entries are nonnegative and less than 2t , they must all be zero. Therefore, we have transformed the original vector to the zero vector with a nite number of applications of T . Solution 2. We shall prove by induction that for all natural n, there exists k such that every entry of T k x1 ; x2 ; x3 ; : : : ; x2n and, equivalently, every entry of S k x1; x2 ; x3 ; : : : ; x2n is even for all nonnegative integers x1 ; x2; : : : ; x2n . When n = 0, T x1 = x1 , x1 = 0 and our hypothesis holds. Assume the result holds for some n = r, with kr iterations of T always producing a result whose entries are all divisible by 2. Let n = r +1. Then
j =0
2n
x
j
j +i
n
n
20 xi + 2n xi+2n 2xi 0 mod 2: 2
T kx1; x2; x3 ; : : : ; x2n ;
If we take every other element of the last vector, starting with the rst, we obtain x1 + x3 ; x3 + x5 ; : : : ; x2r+1 ,1 + x1 , which is also what we obtain when we apply T to x1 ; x3 ; x5 ; : : : ; x2r+1 ,1 , and similarly with the even indexed elements. Therefore, by the induction hypothesis, after 2kr applications of T , we have a vector whose entries are all divisible by 2. By mathematical induction, for all natural n, a nite number of applications of T will transform any vector into a vector divisible by 2. As we concluded in the previous solution, a nite number of applications of T will transform any vector into the zero vector.
Sx1; x2 ; x3; : : : ; x2r+1 = x1 + x2; x2 + x3 ; x3 + x4; : : : ; x2r+1 + x1 ; 2 x1 ; x2 ; x3 ; : : : ; x r+1 S 2 = x1 + 2x2 + x3; x2 + 2x3 + x4; x3 + 2x4 + x5 ; : : : x1 + x3; x2 + x4; x3 + x5; : : : ; x2r+1 + x2 mod 2 :
37
The Cantor Set and Cantor Function
Naoki Sato
We begin by considering a problem that was given out at IMO training this past summer. Problem. Let f : 0; 1 ! 0; 1 be a function satisfying the following three properties: i f is nondecreasing that is; x y f x f y , ii f x = 1 , f 1 , x for all x 2 0; 1 , and 1 iii f 3x = 2f x for all x 2 0; 3 . Evaluate f 1=7 and f 1=13. The reader at this point should stop and try to work out the values the problem asks for; this is a good way to get a feel for the dynamics of this function, which turns out to be a very special function, and wellknown in analysis and chaos theory. What is the rst of many remarkable facts is that the three properties given above are enough to determine the value of f x for any point x 2 0; 1 , even though they do not seem to be what if x is irrational? We will come back to these questions later. We will rst describe a seemingly unrelated concept, the Cantor set. There are several possible de nitions, and this is perhaps the most straightforward. Let A0 be the interval 0; 1 . From this, we remove the middlethird" ,1 2 ; 3 ; and we obtain A1 = 0; 1 2 ; 1 , which is the union of two in3 3 3 tervals. We remove the middlethird from these two intervals, and obtain 1 2 A3 = 0; 9 2 ; 1 3 ; 7 8 ; 1 , and so forth see Figure 1. We call C , 9 3 9 9 the set obtained in this limiting process, the Cantor set. Rigorously speaking,
n 3k,1 3j , 2 3j , 1
An = 0; 1 n 3k ; 3k ; k=1 j =1
1 3k,1 3j , 2 3j , 1
C = 0; 1 n 3k ; 3k : k=1 j =1
Note that An is the union of closed intervals, and that the sum of the , lengths of these intervals is 2 n since we remove one third at each step, 3 which goes to 0 as n approaches in nity. In this sense, we have removed most" of the interval 0,1 in obtaining C .
Copyright c 1998 Canadian Mathematical Society
38 We mentioned that C has alternate de nitions. Each x 2 0; 1 has a base 3 expansion of the form
x=
where ai 2 f0; 1; 2g. Then C is the set of all x 2 0; 1 such that ai 6= 1 for all i. In other words, each digit is a 0 or 2. Caution. There are some cases where this is not exactly true  some numbers have two possible base 3 expansions, for example 1 = 0:13 = 0:0222 : : :3. Which one should we use? 3 ,1 The set of numbers which have a1 = 1 is precisely the interval 3 ; 2 , 3 which is what we threw away from ,A0 to get A1 . Then, the set of numbers ,7 8 which have a1 6= 1 and a2 = 1 is 1 ; 2 9 9 9 ; 9 , which is what we threw away from A1 to get A2 , and so on. Hence, the two constructions are the same. The Cantor set has many uses. For example, although most of 0; 1 has been removed to obtain C , it can be shown that there is an onto map from C to 0; 1 ; that is, every element in 0; 1 is equal to the image of x for some x 2 C . Not even Q, the rationals, can make this claim. In the language of set theory, Q is a countable set, and C and 0; 1 are uncountable sets, which are bigger" at this level, the size, or cardinality of sets are measured by the existence of 1 1, onto, or bijective maps between them. Also, as the reader might have suspected, the Cantor set is also one of the most basic examples of a fractal, an object, roughly speaking, with selfsimilarity. In fact, the Cantor set is one of the most important fractals, despite its almost bland simplicity and lack of interesting detail. For example, certain Julia sets are modi ed Cantor sets. But how does all this relate to our original problem? We give a formulation for f x. If x 2 C , then ai 2 f0; 2g for all i in the base 3 expansion of x, and Note we have base 3 on the left and base 2 on the right. If x 62 C , then ai = 1 for some i. Choose the smallest such i. Then
, f 0:a1a2 a3 : : : ai : : :3 = 0: a21 a22 a23 : : : ai2 1 ai 2:
i=1 3
1 X
ai ;
i
f 0:a1a2 a3 : : :3 = 0: a21 a2 a3 : : :2 : 2 2
For example,
4
= f 0:022022 : : : = 0:011011 : : : = 3 ; f 13 3 2 7 38
= f 0:02102 = 0:011 = 3 : f 243 3 2 8
39 We leave it to the reader to verify that these are the correct expressions, but we will show that the properties above are su cient to determine f . First, assume that x 2 C , so ai 2 f0; 2g for all i. We proceed by induction on the number of digits of x. Let 0:b1b2 b3 : : :2 be the binary expansion of , . Using the properties, we see that f 0 = 0, f 1 = 1, and fx f , 1 = f 2 = 1 , or f 0:0222 : : :3 = 0:0111 : : :2, and f 0:23 = 0:12. 3 3 2 1 Hence, by expressing 2 in these two forms, and since f is nondecreasing, we see that if a1 = 0, then b1 = 0, and if a1 = 2, then b1 = 1, so b1 = a1 =2. Now assume bi = ai =2 for all i from 1 up to some n. We must show that bn+1 = an+1 =2. First, consider the case a1 = 0. Then b1 = 0, as shown above, and f 3x = 2f x, or
f 0:a2a3 : : : an+1 : : :3 = 0:b2b3 : : : bn+1 : : :2 : By the induction hypothesis, bn+1 = an+1 =2. The case where a1 = 2 is left to the reader Hint: This is where we must use f x = 1 , f 1 , x . Therefore, by induction, bi = ai =2 for all i. Also, it can be seen that if x and y are the endpoints of one of the intervals removed to obtain C , then f x = f y . Thus, again since f is nondecreasing, f on each missing interval is the common value at the end
points. We also get another remarkable property as a bonus. Note that f is onto. Pick y 2 0; 1 , express y in base 2, and it should be obvious which x 2 0; 1 satis es f x = y. We can even choose x 2 C , so that f , when restricted to C , is an example in fact, the standard example of a map from C to 0; 1 that is onto. Moreover, f is nondecreasing, so the only possible discontinuities of f are jump discontinuities. But f is onto, so it cannot have any jump discontinuities, so f is continuous.
Figure 1. Figure 2. When f is graphed, the features alluded, to above become immediately ap 1 parent see Figure 2. On the interval 1 ; 2 , f is indeed the constant 2 , 3 3
40
1 3 on 1 ; 2 , the constant 4 , and on 7 ; 8 , the constant 4 , and so on. As 9 9 9 9 mentioned, this function f , sometimes called the Devil's Staircase, has some exceptional properties. By the observation just made, f 0 x = 0 almost everywhere almost everywhere" does have a technical meaning, and it is related to the fact that most of 0; 1 is missing; in fact, the derivative is not de ned precisely at points in the Cantor set. But f 0 = 0 and f 1 = 1, so f must somehow climb up abruptly at points in the Cantor set, since it is at almost everywhere. So, a fairly innocuous problem on a functional equation turns out to have some large rami cations. We end with a few more miscellaneous facts. Let T1 x = x=3 and T2 x = x + 2=3. Pick a random point x 2 0; 1 , and recursively apply Ti to x where i is randomly chosen at each step. Plot each x; you can do this on your computer. What do you get? De ne T x on 0; 1 as follows: 3x if 0 x 1=2 T x = : 3 , 3x if 1=2 x 1 This function is sometimes called the tent function, for obvious reasons. Some points in the range of T are outside of 0; 1 ; which ones? These are bad points we wish to remove from the domain of T , because we wish to apply T an arbitrarily number of times to values in 0; 1 . For which x is it true that T 2 x 2 0; 1 ? What about T 3 ? In fact, what is the set of all x such that T k x 2 0; 1 for all k? You guessed it, it is the Cantor set. In terms of the base 3 expansion, how does T x relate to x?
, ,
Problems
1. Show that for all z 2 0; 2 , there exist x, y 2 C such that x + y = z . Symbolically, C + C = 0; 2 . 2. Show that C is totally disconnected; that is, show that for all x, z 2 C , there is a y 62 C such that x y z .
41
J.I.R. McKnight Problems Contest 1981
1. If x, y , and z are all between 0 and inclusive, solve for x, y and z if 2
3 log2 sin x sin y sin z = , 2 ! sin2 x sin3 y = , 5 log2 sin z 2 ! x 4 log2 sinsinsin y = ,3 2z
2. A variable chord of an ellipse subtends a right angle at the centre. Show that the chord always touches a xed circle. 3. a Obtain the equation of the tangent with slope m to the parabola whose equation is y 2 = 4px. Assume p 0. b Obtain the equation of the tangent perpendicular to the tangent in a. c Find the equation of the locus of the points of an intersection of pairs of perpendicular tangents to the parabola in a. 4. Find the rst four terms in ascending powers of x in the expansion as an in nite series of 1 , 2x and state the restrictions on x. 5. A tent has the shape of a cone and has a capacity of 1000 m3 . Find the radius of the base if the amount of canvas used is to be a minimum. No canvas is used on the oor. 6. Find the length of the longest ladder that can be carried horizontally around the corner of the corridor shown in the diagram.
4m
1 + x , 2x2
6m 3 ?
42
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only problems  the next issue will feature only solutions. We warmly welcome proposals for problems and solutions. With the new schedule of eight issues per year, we request that solutions from this issue be submitted by 1 April 1998, for publication in the issue 5 months ahead; that is, issue 6 of 1998. We also request that only students submit solutions see editorial 1997: 30 , but we will consider particularly elegant or insightful solutions from others.
Mayhem Problems
High School Problems
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca H223. In each of the following alphametics, each letter in the addition represents a unique digit: 1997 and 1998
+
OLD
+
OLD
YEAR
YEAR
exist.
For each alphametic, nd a solution, or prove that a solution does not
H224. Let ABCD be a square. Construct equilateral triangles APB, BQC , CRD, and DSA, where P , Q, R, and S are points outside of the
square. a Prove that PQRS is a square. b Determine the ratio P Q . See how many ways you can solve this! AB
H225. Consider a row of ve chairs, numbered 1, 2, 3, 4, and 5. You are originally sitting on 1. On each move, you must stand up and sit down on an adjacent chair. Make 19 moves, then take away chairs 1 and 5. Then make another 97 moves, with the three remaining chairs. No matter how the moves are made, you will always end up on chair 3. Why is this the case?
43
H226. The smallest multiple of 1998 that only consists of the digits 0 and 9 is 9990. a What is the smallest multiple of 1998 that only consists of the digits 0 and 3? b What is the smallest multiple of 1998 that only consists of the digits 0 and 1?
Advanced Problems
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu A209. Are there an in nite number of squares among the triangular numbers? Triangular numbers are numbers of the form Tn = nn + 1=2. A210. Let P be a point inside circle C . Find the locus of the centres of all circles ! which pass through P and are tangent to C . A211. Does there exist a convex polyhedron and a plane, not passing 2 through any of its vertices, and intersecting more than 3 of all of the edges of the polyhedron? Polish Mathematical Olympiad, rst round A212. Let A and B be real n n matrices such that A2 + B2 = AB. Prove that if AB , BA is an invertible matrix, then n is divisible by 3. International Competition for University Students in Mathematics
Challenge Board Problems
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu In this issue, we introduce David Savitt as the new Challenge Board editor. David is currently a graduate student at Harvard, and comes to us with much problem solving experience under his belt, including high ranking on the Putnam competition. It looks like he may be asking some very tough problems! Ravi Vakil has recently graduated from Harvard University, and will be going on to Princeton and the IAS for postdoctoral work, and we wish him the best of luck.
44
C75. a Let n be an integer, and suppose a1, a2, a3 , and a4 are integers such that a1 a4 , a2 a3 1 mod n. Show that there exist integers Ai , 1 i 4, such that each Ai ai mod n and A1 A4 , A2A3 = 1. b Let SL2; Z denote the group of 2 2 matrices with integer entries and determinant 1, and let ,n denote the subgroup of SL2; Z of matrices which are congruent to the identity matrix modulo n. By this we mean that all pairs of corresponding entries are congruent modulo n. What is the index of ,n in SL2; Z ? C76. Let X be any topological space. The nth symmetric power of X , denoted X n , is de ned to be the quotient of the ordinary nfold product X n by the action of the symmetric group on n letters that is, it is the space of unordered ntuples of points of X . Show that the symmetric power C n is actually homeomorphic to the ordinary product C n .
From the les
For the bene t of those readers who are new to MATHEMATICAL MAYHEM, we present a few problems from back issues: J14. MAYHEM 1988: Vol 1, 1, 19 In trapezoid ABCD, we have AB kCD and jAB j = 2jCDj. Suppose that AC meets BD at X . Find the ratio BX : XD. S12. MAYHEM 1988: Vol 1, 1, 20 Find all functions with domain 0; 1 such that
f x =
U10. MAYHEM 1988: Vol 1, 1, 21 Prove that
n X r=0
Z
x
0
f t dt :
2n
2n , 2r
= 4n : r n,r
45
PROBLEMS
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 September 1998. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX. UK.
2301.
Proposed by Christopher J. Bradley, Clifton College, Bristol,
Suppose that ABC is a triangle with sides a, b, c, that P is a point in the interior of 4ABC , and that AP meets the circle BPC again at A0 . De ne B 0 and C 0 similarly. Prove that the perimeter P of the hexagon AB 0 CA0 BC 0 satis es
p P 2 pab + bc + pca :
Proposed by Toshio Seimiya, Kawasaki, Japan. Suppose that the bisector of angle A of triangle ABC intersects BC at D. Suppose that AB + AD = CD and AC + AD = BC . Determine the angles B and C . 2303. Proposed by Toshio Seimiya, Kawasaki, Japan. Suppose that ABC is a triangle with angles B and C satisfying C = 90 + 1 B, that the exterior bisector of angle A intersects BC at D, and 2 that the side AB touches the incircle of 4ABC at E . Prove that CD = 2AE .
2302.
46 Proposed by Toshio Seimiya, Kawasaki, Japan. An acute angled triangle ABC is given, and equilateral triangles ABD and ACE are drawn outwardly on the sides AB and AC . Suppose that CD and BE meet AB and AC at F and G respectively, and that CD and BE intersect at P . Suppose that the area of the quadrilateral AFPG is equal to the area of the triangle PBC . Determine angle BAC . USA.
2304.
2305.
Proposed by Richard I. Hess, Rancho Palos Verdes, California,
An integer sided triangle has angles p and q, where p and q are relatively prime integers. Prove that cos is rational.
2306. Proposed by Vedula N. Murty, Visakhapatnam, India. a Give an elementary proof of the inequality:
sin x
2 2
8
2x2 ; 1 + x2
x1,x ; 1,2x x1,x ; 1,2x
,
0 x 1: 0 x
2 1 ; 2
1
b Hence or otherwise show that
tan x :
,1
x 1 :
2
c Find the maximum value of f x =
sin x x1 , x on the interval 0; 1.
2307. Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel, Alcorc
n on, Madrid, Spain. , It is known that every regular 2n gon can be dissected into n rhom2 buses with the same side length. a How many di erent classes of rhombuses are there? b How many rhombuses are there in each class? 2308. Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel, Alcorc
n on, Madrid, Spain. A sequence fvn g has initial value v0 = 1 and, for n 0, satis es the recurrence relation
vn+1 = 2n+1 ,
Find a formula for vn in terms of n.
n X k=0
vk vn,k; :
47 UK.
2309.
Proposed by Christopher J. Bradley, Clifton College, Bristol,
Suppose that ABC is a triangle and that P is a point of the circumcircle, distinct from A, B and C . Denote by SA the circle with centre A and radius AP . De ne SB and SC similarly. Suppose that SA and SB intersect at P and PC . De ne PB and PA similarly. Prove that PA , PB and PC are collinear. Proposed by K.R.S. Sastry, Dodballapur, India. Let n 2 N. I call a positive integral divisor of n, say d, a unitary divisor if gcdd; n=d = 1. Let n denote the sum of the unitary divisors of n. Find a characterization of n so that n 2mod 4. Proposed by K.R.S. Sastry, Dodballapur, India. Let e n denote the sum of the even unitary divisors, and o n, the sum of the odd unitary divisors, of n. Assume that e n , o n = n. a If n is composed of powers of exactly two distinct primes, show that n must be the product of two consecutive integers, one of which is a Mersenne prime. b Give an example of a natural number n that is composed of powers of more than two distinct primes. Proposed by K.R.S. Sastry, Dodballapur, India. r The r n gonal number is given by P n;r = n , 2 r22 , n , 4 2 , where n 3, r = 1, 2, : : : . Prove that, in the interval P n;r; P n; r + 1 , there is an n , 1 gonal number. Proposed by HeinzJurgen Sei ert, Berlin, Germany. Let N be a nonnegative integer and let a and b be complex numbers with a; b 62 f0; ,1; ,2; : : : ; ,n , 1g. Find a closed form expression for
2310.
2311.
2312. th 2313.
where ak denotes the Pochhammer symbol, de ned by a0 = 1, ak = aa + 1 : : : a + k , 1, k 2 N.
,1k ; k=0 ak bn,k
n X
48
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems.
Erratum
On page 510 of issue 8, 1997, for Cautis", read Howard". However, the solution printed for 2181 was incorrect  see below. The name of RICHARD I. HESS, Rancho Palos Verdes, California, USA was inadvertantly omitted from the list of solvers of problem 2153. He also sumbitted a late solution to problem 2136.
2145. 1996: 170, 1997: 302 Proposed by Robert Geretschlager, Bundesrealgymnasium, Graz, Austria.n n Y, Y Prove that ak + bk,1 ,ak + bn,k for all a, b 1.
Comment by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. The given inequality is a special case of the following more general rearrangement inequality due to Oppenheim 1 : If 0 a1 a2 : : : an , 0 b1 b2 : : : bn , and the sequence fcn g is any permutation of the sequence fbng, then
n Y k=1 k=1 k=1
ak + bk
n Y
k=1
ak + ck
Reference
n Y
k=1
ak + bn,k+1 :
1. A. Oppenheim, Inequalities connected with de nite Hermition forms II, Amer. Math. Monthly 61 1954, 263 266. 2181. 1996: 318, 1997: 509 Proposed by Sefket Arslanagi c
, Berlin, Germany. Prove that the product of eight consecutive positive integers cannot be the fourth power of any positive integer. Corrected solution  several readers of the online version pointed out that the printed solution in the previous issue was incorrect  thank you! It has been pointed out by many readers that this problem has appeared before. Most readers referred to the American Mathematical Monthly, 1936, p.310 for the solution to 3703 posed by Victor Th
bault in 1934, e
49 p.522. Another reference was made to Honsberger's monograph Mathematical Morsels, where it appears on p.156 as A Perfect 4th Power". Several readers also made reference to the general problem of proving that the product of two or more consecutive integers is never a square, which was established in 1975 by Erdos and Selfridges Illinois Journal of Math 19 1975, 292301. Because the solution appears elsewhere, we will simply refer the interested reader to these other sources for a solution. Comments and or references and or solutions were submitted by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADAM BROWN, Scarborough, Ontario; GORAN CONAR, student, Gymnasium Vara din, z Vara din, Croatia; THEODORE CHRONIS, student, Aristotle University of z Thessaloniki, Greece; F.J. FLANIGAN, San Jose State University, San Jose, California, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D. KIPP JOHNSON, Beaverton, Oregon, USA; MURRAY S. KLAMKIN, University of Alberta, Ed
monton, Alberta; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; MICHAEL PARMENTER, Memorial University of Newfoundland, St. John's, Newfoundland; HARRY SEDINGER, St. Bonaventure University, St. Bonaventure, NY, USA; HEINZJURGEN SEIFFERT, Berlin, Germany; D.J. SMEENK, Zaltbommel, the Netherlands; DIGBY SMITH, Mount Royal College, Calgary, Alberta; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; KENNETH M. WILKE, Topeka, Kansas, USA; PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer. There were two incorrect solutions submitted. Sei ert remarks that A. Guibert proved the result in 1862 according to L.E. Dickson in History of the Theory of Numbers, Vol. II, 1952, pp. 679680. Prove that, if a, b, c are the lengths of the sides of a triangle,
2198. Proposed by Vedula N. Murty, Visakhapatnam, India.
2 , 1
+ c , a2 2 , 1
+ a , b2 2 , 1
0; bc a2 ca b2 ab c2 with equality if and only if a = b = c. Editor's comment: Several solvers indicated by a y beside their name in the list of solvers below pointed out that a proof of this inequality had already appeared in the comment given by the proposer on problem 1940 b , c2
1994: 321 .
50
yWALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greecetwo proofs; yHEINZJURGEN SEIFFERT, Berlin, Germany; and yPANOS E. TSAOUSSOGLOU,
Athens, Greece.
Solved by RICHARD I. HESS, Rancho Palos Verdes, California, USA;
2199. Proposed by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA. Find the maximum value of c for which x + y + z 2 cxz for all 0 x y z. Solution by HeinzJurgen Sei ert, Berlin, Germany. With x = 1, y = 1 + u; and z = 2, where 0 u 1, the inequality gives 4 + u2 2c. Letting u approach zero, we nd c 8. If 0 x y z , then
Hence, 8 is the maximum value of c. Also solved by MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; PAULOLIVIER DEHAYE, Bruxelles, Belgium; F.J. FLANIGAN, San Jose State University, San Jose, California, USA; ROBERT GERETSCHLAGER, Bundesrealgymnasium, Graz, Austria; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MURRAY S. KLAMKIN, Univer
sity of Alberta, Edmonton, Alberta; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer. There were two incomplete solutions. Klamkin used a similar proof to verify the more general result: The maximum value of c, for which
x + y + z2 2x + z 2 = 2x , z 2 + 8xz 8xz:
x1 + x2 + : : : + xn 2 cx1 : : : xn for all 0 x1 x2 : : : xn , is c = 4n , 1. Janous provided a more general result still note the inequalities here are not strict: Let n 2; 1 k l n be natural numbers. Then the maximum
constant ck;l such that
x1 + x2 + : : : + xn2 ck;l xk xl
51 is valid for all 0 x1 x2 : : : xn is given by: ,k ck;l = 4l, k n , l + 1 if 2ll , k n + 1 n + 12 if 2 , k n: In our case, this allows us to conclude c c1;3 = 8:
2200. Proposed by Jeremy T. Bradley, Bristol, UK and Christopher J. Bradley, Clifton College, Bristol, UK. Find distinct positive integers a, b, c, d, w, x, y , z , such that
z2 , y2 = x2 , c2 = w2 , b2 = d2 , a2 c2 , a2 = y2 , w2:
and
I. Solution by Richard I. Hess, Rancho Palos Verdes, California, USA. Let d1 ; d2 ; d3 ; d4 and n=d1; n=d2 ; n=d3 ; n=d4 all be distinct divisors of some positive integer n. Then
satisfy
1 d + n
; y = 1 d , n
; etc. z=2 1 d 2 1 d1 1
n = z2 , y2 = x2 , c2 = w2 , b2 = d2 , a2:
1
The other condition in the problem becomes
or
d2 , n
2 , d4 , n
2 = d1 , n
2 , d3 + n
2 ; 2 2d2 2 2d4 2 2d1 2 2d3
2
n2 d12 + d12 , d12 , d12 + 4n + d2 + d2 , d2 , d2 = 0; 2 3 1 4 2 3 1 4 or An2 + 4n + C = 0 where
A computer search exposed the following solutions. Editorial note. Hess searched for integer solutions of 2, which yield rational values for z; y etc. He then multiplied all of z; y;x; c; w; b; d; a by an appropriate positive integer , and n by 2 , to clear all denominators. He gave several solutions, some very large. We give just two of the smaller ones.
1 + 1 , 1 , 1
; A = d2 d2 d2 d2 2 3 1 4
C = d2 + d2 , d2 , d2: 2 3 1 4
52 i If A = 0, then a solution to 2 is which with = 90 gives the solution
d1 = 5; d2 = 6; d3 = 9; d4 = 90; n = 2002;
z = 18243; y = 17793; x = 15285; c = 14745; w = 10415; b = 9605; d = 5051; a = 3049: ii A solution to 2 with A = 0 is 6 d1 = 24; d2 = 20; d3 = 15; d4 = 12; n = 24 which leads = 10 to the solution z = 125; y = 115; x = 106; c = 94; w = 83; b = 67; d = 70; a = 50:
II. More solutions, and editorial comments. There were only two other contributions to this problem. The proposers give two solutions:
z = 109; y = 89; x = 81; c = 51; w = 73; b = 37; d = 63; a = 3; z = 97; y = 83; x = 79; c = 61; w = 57; b = 27; d = 51; a = 9;
and
found by computer. They wonder if there are in nitely many solutions. Of course, any solution can be multiplied through by a positive integer to create another solution, but we are only interested in primitive solutions, that is, ones in which the eight numbers z; y;: : : have no common factor. Hess claims to generate in nitely many primitive solutions, but this editor was unable to untangle his argument. Does anyone have a simple proof?
VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA, uses an idea similar to Hess's, namely he nds parametric expressions for the eight variables of the problem which automatically satisfy the rst given condition, and then searches for values of the parameters so that the second condition is satis ed too. In particular he notes that
z = fgh + e; y = fgh , e; x = egh + f; c = egh , f; w = efh + g; b = efh , g; d = efg + h; a = efg , h; satisfy 1 whenever e; f;g; h are positive integers yielding positive distinct
values for these expressions. The second condition then becomes
egh , f 2 , efg , h2 = fgh , e2 , efh + g 2;
and Kone n
gives the two solutions c y e = 4; f = 5; g = 6; h = 11 and e = 4; f = 6; g = 5; h = 11;
53 which result in the respective solutions and
z = 334; y = 326; x = 269; c = 259; w = 226; b = 214; d = 131; a = 109; z = 334; y = 326; x = 226; c = 214; w = 269; b = 259; d = 131; a = 109:
Note that these solutions di er only in that b; c have been switched and w; x have been switched. In fact this pair of switches is always possible. Condition 1 obviously remains true whenever these switches are made; moreover, in Hess's solution, the other condition 2 is symmetric in d2 and d3 , which shows that these switches do not a ect the truth of 2 either. And Hess's solution is general: given any solution of 1, put z + y = d1 and z , y = n=d1, and similarly for the other variables, and Hess's expressions for z; y etc. follow. As a consequence, we need only look for solutions to the problem which satisfy b c and w x, which hold for all solutions above except for Kone n
's second. Similarly but more easily, given any solution to the c y problem, switching z with d and y with a will always result in another solution, which means we can also assume that d z and a y , which hold for all the above solutions. At the moment this problem is still in an unsatisfactory state of disarray, and needs someone to bring some order to the chaos! Readers are encouraged to try.
ABCD is a convex quadrilateral, and O is the intersection of its diagonals. Let L; M; N be the midpoints of DB; BC; CA respectively. Suppose that AL; OM; DN are concurrent. Show that either AD k BC or ABCD = 2 OBC ; where F denotes the area of gure F .
2201.
1997: 45 Proposed by Toshio Seimiya, Kawasaki, Japan.
I. Solution by C. FestraetsHamoir, Brussels, Belgium. Let O be the origin of a coordinate system where A; B; C; D are represented by a; 0; 0;b; c; 0; 0; d with a; b
positive and c; d negative. Thus
c b L is the point 0; b+d ; M is , 2 ; 2 ; N is a+c ; 0 and 2 2
These lines are concurrent if and only if
AL : b + dx + 2ay , ab + d = 0 OM : bx , cy = 0 DN : 2dx + a + cy , da + c = 0.
b ,c 0 b + d 2a ,ab + d = 0: 2d a + c ,da + c
54 This equation reduces after some manipulation to
ab , cd a , cb , d + 2bc = 0:
Consequently, either a ab = cd, in which case ADjjBC , or , b 1 a , cb , d sin = 2 , 1 bc sin 2 2 case ABCD = 2 OBC .
where
= AOB , in which
II. Comment based on a solutionby Michael Lambrou, University of Crete, Crete, Greece. Using barycentric coordinates one can obtain a geometric characterization of those quadrilaterals ABCD for which AL; OM , and DN are concurrent at J : Either AB jjBC in which case J always exists, or JCOB is a parallelogram whose area equals ABCD. Thus to draw an accurate gure of the latter case, begin with a parallelogram JCOB and let L be any point on BO. For ABCD to be convex, L should lie between O and the midpoint of BO. Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; FLORIAN HERZIG, student, Perchtoldsdorf,Austria; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece, 3 solutions; GERRY LEVERSHA, St Paul's School,
London, England; ISTVAN REIMAN, Budapest, Hungary; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
2202. 1997: 45 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Suppose than n 3. Let A1 ; : : : ; An be a convex ngon as usual with interior angles A1 ; : : : ; An . Determine the greatest constant Cn such that
n X k=1 n 1 C X 1 : Ak n , Ak k=1
Determine when equality occurs.
55 Solution by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We assume that the intention of the problem is to determine the largest constant Cn so that the given inequality holds for all convex ngons. If n 4 then clearly Cn = 0, since we could let one of the interior angles tend to without forcing any other interior angle to tend to zero. Then n n X 1 ! +1 while X 1 is bounded: Thus the inequality cannot hold unless Cn 0. Hence Cn = 0 and equality can never occur. If n = 3, then by the arithmetic harmonic mean inequality,
1 1
A1 + A2 A + A 4 1 2
k=1 , Ak
k=1 Ak
and thus Similarly,
A1 + A2 A1 + A2 = , A3 :
1 + 1 4 A2 A3 , A1
and
1
1
4
4
1 + 1 4 : A3 A1 , A2
k
Adding, we get
3 X
It is easily seen that equality holds if and only if all the Ak 's are equal. Hence C3 = 2 and equality holds if and only if A1A2A3 is equilateral. Also solved by FLORIAN HERZIG, student, Perchtoldsdorf, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece three solutions; GERRY LEVERSHA, St Paul's School, London, England; HEINZJURGEN SEIFFERT, Berlin, Germany; and the proposer. The case n = 3 only was solved by GORAN CONAR, student, Gymnasium Vara din, Vara din, Croaz z tia. Two incorrect solutions were also sent in. Sei ert uses the known inequality
k=1 1 , ak k=1 n X
k=1 k
3 1 2X 1 : A ,A
k=1
1
n X
1 , ak
n X
where 0 ak 1=2 see pages 25 27 of Mitrinovi
, Pe ari
and Fink, Clasc c c sical and New Inequalities in Analysis, Kluwer, 1993, to prove the following
k=1 k k=1
n 1 Xa ; k a
56 related inequality: if n 4 and A1 A2 : : : An is a convex ngon such that Ak =2 for all k, then
n 1 2 X 1 k=1 Ak n , 2 k=1 , Ak n X
put ak = 1 , Ak = . Furthermore, equality holds if and only if the ngon is regular, since equality holds in the earlier inequality if and only if the ak 's are equal. One reader pointed out that the proposer had also published this problem, in German, in the journal Wissenschaftliche Nachrichten in January 1996, and that a solution was published on page 36 of the January 1997 issue. Readers are reminded that a problem submitted to CRUX with MAYHEM should not be submitted for publication elsewhere, unless and until the problem has either been rejected by CRUX with MAYHEM or withdrawn by the proposer  EditorinChief.
2203. 1997: 46 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Let ABCD be a quadrilateral with incircle I . Denote by P , Q, R and S , the points of tangency of sides AB , BC , CD and DA, respectively with I . Determine all possible values of PR;QS such that ABCD is cyclic. Comments. A quadrilateral that is simultaneously inscribed in one circle and circumscribed about another is call bicentric. Konen
and Reiman c y both report that in 2, section 39 you can nd the solution to our problem and lots more, including its converse and the fact Brianchon's Theorem that AC; BD; PR, and ST all pass through the same point see also 1, p. 79 . Bellot Rosado reminds us that our problem was part of a problem proposed by India but not used at the 1989 IMO. We present two of the many possible solutions. I. Solution by _ FestraetsHamoir, Brussels, Belgium. C. We denote by XY the angle subtended at the centre of I by the arc XY . Note that since I is an incircle, ABCD is convex and the diagonals PR and QS intersect in the interior. _ _ _ _ A = 1 PQ + QR + RS , SP , and 2
so that
1 C = 2 RS + SP + PQ , QR,
_
_
_
_
A + C =PQ + RS .
_
_
57
II. Solution by Francisco Bellot Rosado, I.B. Emilio Ferrari, Valladolid, Spain. Consider inversion in I . The images A0 ; B 0 ; C 0 ; D0 of A; B; C; D are the respective midpoints of SP; PQ; QR, and RS as in 1, Figure 5.3A , so that A0 B 0 C 0 D0 is a parallelogram. Since circles are preserved by inversion, A00 B00C 00D00 is cyclic if and only if ABCD is, in which case the parallelogram A B C D would be a rectangle. Because each side of a midpoint quadrangle is parallel to a diagonal PR or QS we conclude that PR ? QS if and only if ABCD is cyclic. References. 1 H.S.M. Coxeter and S.L. Greitzer, Geometry Revisited. MAA, 1967. 2 Heinrich Dorrie, Triumph der Mathematik, Wurzburg, 1958. English translation: 100 Great Problems of Elementary Mathematics, Dover, 1965.
PR; QS = . 2
ABCD is cyclic if and only if A + C = , and so, if and only if
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; JORDI DOU, Barcelona, Spain; YEO KENG HEE, Hua Chong Junior College, Singapore; FLORIAN HERZIG, student, Perchtoldsdorf, Austria;
VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY
LEVERSHA, St Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlandsand the proposer. Proposed by Sefket Arslanagi
, University of Sarajevo, Sarac jevo, Bosnia and Herzegovina. p For triangle ABC such that Ra + b = c ab, prove that Here, a, b, c, R, and r are the three sides, the circumradius and the inradius of 4ABC . Solution by KeeWai Lau, Hong Kong. Since c = 2R sin C ,
2204.
3 r 10 a:
p Ra + b = 2R sin C ab:
+ sin C = ap b 1: 2 ab
Using the AM GM inequality,
58 Hence, sin C = 1, C = 900, a = b and c = 2a. Thus
p
as required. Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, and
MARIA ASCENSION LOPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MIGUEL
ANGEL CABEZON OCHOA, Logro~ o, Spain; ADRIAN CHAN, student, Upper n Canada College, Toronto, Ontario; GORAN CONAR, student, Gymnasium Vara din, Vara din, Croatia; C. FESTRAETSHAMOIR, Brussels, Belgium; z z FLORIAN HERZIG, student, Perchtoldsdorf, Austria; WALTHER JANOUS,
Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA,
St Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; HEINZJURGEN SEIFFERT, Berlin, Germany; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; PANOS E. TSAOUSSOGLOU, Athens, Greecetwo solutions; and the proposer. Most of the submitted solutions are similar to the one given above.
a2 Area of 4 r = Semiperimeter ABCABC = 2a + p2a = 2 +ap2 of 4
3 a; 10
2205. 1997: 46 Proposed by V
aclav Konecn
y, Ferris State University, Big Rapids, Michigan, USA. Find the least positive integer n such that the expression
has a maximum which is a rational number where A, B , C are the angles of a variable triangle. Many of the solvers cited the relationship between this problem and 1984: 19 proposed by M.S. Klamkin which asks for the maximum value of P = sin A sin B sin C , where A; B; C are the angles of a triangle and ; ; are given. The published solution by Walther Janous 1985: 908 gives this maximum as:
sinn+2 A sinn+1 B sinn C
Pmax =
+ + + +
=2
+ + + +
=2
+ + + +
=2
:
Substitution and simpli cation yields
The rst term is rational and the smallest integer for which the second part is rational is n = 12. See also problem 2183 1996: 319; 1997: 514 , 2116 1996: 75; 1997: 116 .
n=2 n + n=2 3n+1=2 n 1n+1 n + 23n+1 Pmax = 2n+1+ n + 3n+1 2n + 1n n2n+ 1122 2n + 31=2 : = 2
59
Solved by SEFKET ARSLANAGIC, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; CHRISTOPHER J. BRADLEY, Clifton College, Bris
tol, UK; MIGUEL ANGEL CABEZON OCHOA, Logro~ o, Spain; FLORIAN n HERZIG, student, Perchtoldsdorf, Austria; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St Paul's School, London, England; HEINZJURGEN SEIFFERT, Berlin, Germany; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer. However, some of our solvers were not content to leave it here. The rst word goes to Walther Janous who provided the solution in 1985: 908 . Generalization, Part I, by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. We now show that there are in nitely many integers n such that Pmax n is rational. Let n be even. Then n=2 is an integer and
Pmax n 2 Q
s
is equivalent to
3 2n + 12n + 3 2 Q:
Because gcd 2n +1; 2n +3 = 1, one of the following two possibilities must occur: i 2n + 1 = x2 and 2n + 3 = 3y 2, or ii 2n + 1 = 3x2 and 2n + 3 = y 2, where x; y are integers. We show that i leads to the claimed in nitely many n's. Indeed, i implies x2 + 2 = 3y2; that is,
x2 , 3y2 = ,2; X 2 , 3Y 2 = 1
which has x; y = 1; 1 as one solution. The associated pure" Pellian equation has X; Y = 2; 1 as its fundamental solution. Hence, we get in nitely many solutions xm ; ym of :
p p p xm + ym 3 = 1 + 32 + 3m; for m = 0; 1; 2; : : :
2 n = xm2, 1 ; m = 1; 2; 3; : : :
Simple congruence considerations, mod 4, show that all solutions x; y of consist of odd numbers. Then
as claimed.
60 If m = 1, x1 = 5 and n = 12. In general, via conjugation, xm , ymp3 = 1 , p32 , p3m. Thus, p p p p xm = 1 1 + 32 + 3m + 1 , 32 , 3m 2 so n = 12; 180; 2520; 35112; 489060; 6811740; 94875312; etc. Generalization, Part II, by Michael Lambrou, University of Crete, Crete, Greece. We show that there are no solutions with n odd. Let n be odd and write n = 2N , 1; N 1. Then the second part of Pmax is
2
and
p p n = xm22, 1 = 1 2 + 32m+1 + 2 , 32m+1 , 4 4
N , 1 2N2,1 2N + 1 2N2,1 3N 4N , 11=2 4N + 11=2
=
p
2
We now show that the integer inside the square root sign is never a perfect square. We have
N , 12N + 14N , 14N + 1
a rational number
2N , 12N + 14N , 14N + 1 = 64N 4 , 20N 2 + 1
which is easily seen to be strictly between the perfect squares 8N 2 , 22 = 64N 4 , 32N 2 +4 and 8N 2 , 12 = 64N 4 , 16N 2 +1; so is not itself a perfect square.
2206. 1997: 46 Proposed by HeinzJurgen Sei ert, Berlin, Germany. Let a and b denote distinct positive real numbers. a Show that if 0 p 1, p 6= 1 , then 2
1 ,ap b1,p + a1,pbp 4p1 , ppab + 1 , 4p1 , p a + b
: 2 2
b Use a to deduce Polya's Inequality:
a,b log a , log b
1 2pab + a + b
: 3 2
Note: log" is, of course, the natural logarithm.
61
I. Solution to part a by Christopher J. Bradley, Clifton College, Bristol, UK. The symmetry of the inequality to be proved, between a and b, and about p = 1 , allows us, without loss of generality, to suppose that b a 2 and 0 p 1 . To save writing, set k = 4p1 , p and note that 0 k 1. 2 We are then required to prove:
This is equivalent to each of the following three inequalities:
1 ,ap b1,p + a1,pbp 2
p k ab + 1 , k a + b : 2
1 ,apb1,p + a1,p bp 2
p b1,p + a1,p bp , 2pab a
2 2 ap bp b1,2p , 2a 1 ,pb 1 ,p + a1,2p
1 , k
a + b , 2pab + pab; 2
p 1 , k a + b , 2 ab ;
2 2 1 , 2p2 a , 2a 1 b 1 + b :
It is su cient to prove the inequality obtained from taking the positive square root of this, namely:
1 1 2 2 2 2 2 2 b 2 , p a p , a 1 , p b p 1 , 2p b 2 , a 1 : Setting a = c2 and b = d2 with d c 0, this is the same as 2pd , c d , c + c1,pdp , d1,pcp
or
We now prove this inequality when p is rational. Let p = m , with m n and n coprime integers and 2m n since p 1 . Also, we set c = xn , 2 d = yn with y x 0. With these restrictions and substitutions, it is now su cient to prove that
2pd , c
dp , cp d1,p , c1,p :
,
2m yn , xn
n ym , xm ,yn,m + xn,m
,
or
is su cient to prove that
n , 2m yn , xn ny mxm y n,2m , xn,2m : On division by y , x, which is valid since it is a positive quantity, it
, n , 2m y n,1 + xyn,2 + : : : + xn,1 nxmym ,yn,2m,1 + xyn,2m,2 + : : : + xn,2m,1 :
This is true by repeated application of the Power Means Inequality. This result is then extended to irrational values of p by the familiar continuity arguments.
62
II. Solution by KeeWai Lau, Hong Kong. a For t
0, let
p
f t = 1 tp + y1,p , 4p1 , p t , 1 , 4p1 , p t + 1 :
2
2
Di erentiating, we have: and
p,1 ,p 1 2 f 0t = pt + 1 , pt , 2p1 , pt, 1 , 1 , 4p2 , p ; 2
1
or
2 4p1 , pt 1 + 2p , 12 t + 1 : 2 By integrating this inequality with respect to p from p = 1 to p = 1, we 2 obtain
1 t,1 1 pt + t + 1 : 2 log t 3 12 Polya's Inequality follows by substituting t = a .
b Part a was also solved by PAUL BRACKEN, CRM, Universit
de e
1 ,tp + t1,p 2
2 f 00t = 1 p1 , pt,2 2t 1 , tp , t1,p : 2 2 It is easy to check that f 01 = f 00 1 = 0. Since p 6= 1 , we have, for t 6= 1, 2 p 1,p p 1 1 p , t1,p 2t 2 , t 2t 2 , 2 t t = 0 ; and hence that f 00 t 0. Thus f t f 1 = 0 and f t = 0 if and only if t = 1. Now putting t = a , we easily obtain inequality a. b b From part a, we see that, for t = 1 and p = 1 , we have f t 0, 6 6 2
Montr
al, Montr
al, Qu
bec; FLORIAN HERZIG, student, Perchtoldsdorf, e e e Austria; JOE HOWARD, New Mexico Highlands University, Las Vegas, NM,
USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; VACLAV
KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece two solutions; VEDULA N. MURTY, Visakhapatnam, India; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; and the proposer. Partb was also solved by PAUL BRACKEN, CRM, Universit
de e Montr
al, Montr
al, Qu
bec; WALTHER JANOUS, Ursulinengymnasium, e e e
Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece two solutions; VEDULA N. MURTY, Visakhapatnam, India; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; and the proposer. All solvers, except Bradley, used Calculus in some way or other.
63
2207. 1997: 46 Proposed by Bill Sands, University of Calgary, Calgary, Alberta. Let p be a prime. Find all solutions in positive integers of the equation:
a b
2 + 3 = 5:
p
Solution by Florian Herzig, student, Perchtoldsdorf, Austria. Equivalently, we have p3a + 2b = 5ab; hence there are three cases to consider. First case: p = 5. Then we get 3a + 2b = ab or a , 2b , 3 = 6 = 1 6 = 2 3: Hence all pairs of positive integers a; b are 3; 9; 4; 6, 5; 5 and 8; 4. Second case: p divides a. Let a = a1 p. Hence
3a1 p = b5a1 , 2: Then p divides either b or 5a1 , 2. If b = b1 p, then 3a1 = b15a1 , 2 or 5a1 , 25b1 , 3 = 6 which has only one solution: a1 ; b1 = 1; 1, whence a; b = p; p. Otherwise, 5a1 , 2 = a2 p. Therefore, 3 a2p5+ 2 = ba2 or 3a2p + 6 = 5ba2: Hence a2 divides 6.
+2 +2 If a2 = 1, then a; b = pp5 ; 3p5 , but only if p 3 mod 5.
+1 If a2 = 2, then a; b = 2pp+1 ; 3p5 , but only if p 4 mod 5. 5
+2 If a2 = 3, then a; b = p3p+2 ; 3p5 , but only if p 1 mod 5. 5
p If a2 = 6, then a; b = 2p35 +1 ; 3p+1 , but only if p 3 mod 5. 5 Third case: p divides b. Let b = b1 p. Hence 2b1 p = a5b1 , 3: Then p divides either a or 5b1 , 3. If p divides a, then we have the same case p divides a and b as already considered above. Again p; p is the only solution. Otherwise, 5b1 , 3 = b2 p. Therefore, 2 b2p5+ 3 = ab2 or 2b2 p + 6 = 5ab2 :
64 Hence b2 divides 6.
+3 +3 If b2 = 1, then a; b = 2p5 ; pp5 , but only if p 2 mod 5.
p +1 If b2 = 6, then a; b = 2p5 ; 3p25 +1 , but only if p 2 mod 5. Note that if p = 2 or 3, this generates only two distinct solutions: 2; 2, 1; 6 for p = 2 and 3; 3, 12; 2 for p = 3. If p 5, then the three solutions are all distinct. Also solved by GERALD ALLEN, CHARLES DIMINNIE, TREY SMITH and ROGER ZARNOWSKI jointly, Angelo State University, San Angelo,
TX, USA; SEFKET ARSLANAGIC, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; MICHEL BATAILLE, Rouen, France; CHRISTOPHER J.
BRADLEY, Clifton College, Bristol, UK; MIGUEL ANGEL CABEZON OCHOA, Logro~ o, Spain; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Conn necticut, USA; KEITH EKBLAW, Walla Walla, Washington, USA; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; HEINZJURGEN SEIFFERT, Berlin, Germany; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; and the proposer. There were six incorrect and two incomplete solutions.
+3 If b2 = 2, then a; b = 2p5 ; p2p+3 , but only if p 1 mod 5. 5
+1 If b2 = 3, then a; b = 2p5 ; 3pp+1 , but only if p 4 mod 5. 5
2153. 2167.
1996:217; 1997:313 The conjectured inequality should read jxn p1=xj 2n,1 . 1996:274; 1997:381 The equation in the last line should read 2n+4=2n + n = 2.
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
65
THE ACADEMY CORNER
No. 17 Bruce Shawyer
All communications about this column should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7 This month, we present another solution to a university entrance scholarship examination paper from the 1940's, which appeared in the April 1997 issue of CRUX with MAYHEM 1997: 129 . 10. Two unequal circles of radii R and r touch externally, and P and Q are the points of contact of a common tangent to the circles, respectively. Find the volume of the frustrum of a cone generated by rotating PQ about the line joining the centres of the circles. Solution.
P T
R R
Q
0
,r
First, we note that triangle 4TOP OQ has a right angle at T , with OP OQ = R + r and OP T = R , r. Hence OQT = 2pRr. Because of parallel and perpendicular lines, all of angles PSOP , TOQOP , OP PP 0 and OQQQ0 are equal we denote the common value by . From triangle 4OP OQ T , we note that
OP P
r
OQ Q
0
S
sin = R , r R+r
and
Rr cos = 2 + r : R
p
66 Using various right triangles, we obtain:
OP P 0 = R sin = RrR+,rr ; p 0 = R cos = 2R Rr ; PP R+r P 0S = PP 0 cot p p 2R Rr 2 Rr = R+r R,r 2 = R4rR r2 ; 2 , R, OQQ0 = r sin = rr + rr ; p 0 = r cos = 2r Rr ; QQ R+r Q0S = QQ0 cot p p 2r Rr 2 Rr = R+r R,r 2 = R4r,R 2 : 2 r
Hence
h i V = PP 02 P 0S , QQ0 2 Q0S 3 4 3r 2 2 4 3 = R R r2 R4rR r2 , R Rrr2 R4r,R 2 2 , 2 3 + + r 16R2r2 3 3 = 3R + r2R 2 , rr2 R , 2 2 R2 2 = 16R r3R ++ Rr + r ; r3
which is the required volume.
67 11. Prove that
sin2 + + sin2 + , 2 cos , sin + sin + = sin2 , : 1
Solution.
2 ,sin2 + + sin2 + = 2 , cos2 + 2 + cos2 + 2 = 2 , 2 cos2 + + cos , ; ,4 cos , sin + sin + = ,2 cos , cos , , cos2 + + ;
so that
2 LHS of 1 , = 2 1 , cos2 + + cos , , cos2 , +cos , cos2 + + , = 2 1 , cos2 , = 2 sin2 , :
What is the equation of this curve?
68
THE OLYMPIAD CORNER
No. 188 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. This number we begin with the problems of The Final Round of the Japan Mathematical Olympiad. My thanks go to Richard Nowakowski for collecting these when he was Canadian Team Leader at the 35th IMO in Hong Kong.
JAPAN MATHEMATICAL OLYMPIAD
Final Round February 1994  4 hours
and let bn = n + an . Dropping all bn n = 1; 2; : : : from the set of all positive integers N , we get a sequence of positive integers in ascending order fcn g. Represent cn by n. 2. Five points are located on the plane. Any three of those points are not collinear. Let l1 ; : : : ; l10 be the length of the ten segments obtained by joining every two points of the ve points. Assume that l2 ; : : : ; l2 are 1 9 rational numbers. Prove that l2 is also a rational number. 10 3. There is a triangle A0A1A2 and seven points P0; : : : ; P6 on the plane. Assume that any Pi and Pi+1 are symmetric with center Ak , where k is the remainder of i divided by 3. a Prove that P0 = P6 . b Describe the possible position of P0 under the additional assumption that every segment connecting Pi and Pi+1 does not intersect with the interior of the triangle A0 A1 A2 . 4. We consider a triangle ABC such that MAC = 15 , where M is the midpoint of BC . Determine the possible maximum value of B . 5. There are N persons and N pieces of lot cards. Each number 1 through N is written on a card. When the N persons draw these cards, their order is determined by the numbers on their cards. After repeating this draw two times, we give gifts by the following rule. Rule: A person X gets the gift, if there is no person Y such that Y is prior to X both times. Otherwise X cannot get the gift.
1. For positive integer n, let an be the nearest positive integer to pn,
69 For example, if X is at the top in the rst lot, X always gets the gift whatever he draws in the second lot. Then determine the expected number of persons who get gifts. As a second Olympiad for your puzzling pleasure we give problems of the 30th Spanish Mathematical Olympiad, First Round, November 26 27, 1993. My thanks go to Richard Nowakowski, Canadian Team Leader to the 35th IMO in Hong Kong and to Francisco Bellot Rosado, I.B. Emilio Ferrari, Valladolid, Spain, for copies of the contest.
30th SPANISH MATHEMATICAL OLYMPIAD
First Round  November 26 27, 1993 Proposed by the Royal Spanish Mathematical Society
Time allowed: 4 hours each day. Each problem carries 10 points.
1. Show that, for all n 2 N, the fractions
First Day
are irreducible. 2. A sphere of radius R, and a right cone with base a meridian of the sphere and vertex external to the sphere, are given. Find the radius of the circle of intersection of the sphere and the cone, given that the volume of the cone is half of the volume of the sphere. 3. Solve the following system of equations: in which jtj and t represent the absolute value and the integer part of the real number t. 4. Let AD the internal bisector of the triangle ABC D 2 BC , E the point symmetric to D with respect to the midpoint of BC , and F the point of BC such that BAF = EAC . Show that BF = c . FC b
3 3
n , 1 ; n ; 2n + 1 n 2n + 1 2n2 + 2n
x jxj + y jyj = 1;
x + y = 1;
Second Day
5. Find all the natural numbers n such that the number
nn + 1n + 2n + 3
has exactly three prime divisors.
70
6. An ellipse is drawn taking as major axis the biggest of the sides of a given rectangle, such that the ellipse passes through the intersection point of the diagonals of the rectangle. Show that, if a point of the ellipse, external to the rectangle, is joined to the extreme points of the opposite side, then three segments in geometric progression are determined on the major axis. 7. Let a 2 R given. Find the real numbers x1; : : : ; xn which satisfy the system of equations
x2 + ax1 + , a,1 2 1 2 x2 + ax2 + , a,1 2 2 2 .......................... x2 ,1 + axn,1 + , a,1 2 n 2 x2 + axn + , a,1 2 n 2
= x2 = x3 = xn = x1 :
...
8. The Sisyphus's myth There are 1001 steps going up, with rocks on some of them no more than 1 rock on each step. Sisyphus may pick any rock and raise it one or more steps up to the nearest empty step. Then his opponent Hades rolls a rock with an empty step directly below it down one step. There are 500 rocks, originally located on the rst 500 steps. Sisyphus and Hades move rocks in turn, Sisyphus making the rst move. His goal is to place a rock on the top step. Can Hades stop him?
We now turn to readers' solutions received before January 1st, to problems for consideration by the International Jury at the 36th IMO in Canada 1996: 299 301 . 4. Let A, B and C be noncollinear points. Prove that there is a unique point X in the plane of ABC such that XA2 + XB 2 + AB 2 = XB2 + XC 2 + BC 2 = XC 2 + XA2 + CA2. Solutions by D.J. Smeenk, Zaltbommel, the Netherlands; and by Vasiliou Meletis, Elefsis, Greece. We give Meletis' solutions. First Solution. B
B1 A A1 X C1 C
71 From the hypothesis we have
AX 2 + AB2 = CX 2 + CB2: 1 If B1 is the midpoint of BX , applying the rst theorem of the median in the triangles 4ABX , 4CBX we get 2 2 2 2 2AB1 + 2BB1 = 2CB1 + 2BB1 or AB1 = CB1: 2 This indicates that the perpendicular bisector of the side AC passes through the point B1. Let A1 , C1 be the midpoints of AX and CX , respectively. Similarly, we obtain that the perpendicular bisectors of BC and AB pass through the midpoints A1 and C1 , respectively. 3 Furthermore we obtain AB kA1 B1 , AC kA1 C1 and BC kB1 C1. 4 From 3 and 4 we get that the circumcentre O of ABC is the orthocentre H1 of A1 B1 C1 . 5 Also from 4 the triangles ABC and A1 B1 C1 are similar with X the centre of similarity and ratio 1 . 6 2 So, their orthocentres H and H1 lie in the same straight line with the point X and HH1 = H1 X . 7 Combining 5 and 7 we get HO = OX ; that is the point X is known constant, because X is symmetric to H with respect to the orthocentre O of ABC .
Second Solution. The conditions of the problem are equivalent to the system of equations
A B
E1
E2
A1
X M A2
q
C
XB2 , XC 2 = AC 2 , AB2 1 2 2 2 2 XC , XA = BA , BC : 2 Now, taking equation 1 gives a locus of points X satisfying the condition. The relation reminds us of the second theorem of the median in a triangle. Let AA1 , XA2 be the altitudes of the triangles ABC and XBC respectively on side BC extended. Let M be the midpoint of the side BC .
If we suppose
AB AC BC; XC XB XA;
3
for illustration, we get
72 and furthermore the point M lies between the points A1 and A2 But XB2 , XC 2 = 2BC MA2 ; and 4
AC 2 , AB2 = 2BC MA1 : Hence MA1 = MA2 and A2 is a constant point on BC because it is symmetric to A1 with respect to the midpoint M . Consequently, if 1 holds, the point X lies on the line E1 perpendicular to BC at A2 . Similarly, if 2 holds, the point X lies on the line E2 perpendicular to AC at B2 where BB1 ? AC and AB1 = CB2 . Hence, the required point X lies at the intersection of E1 and E2 . 5. The incircle of ABC touches BC , CA and AB at D, E and F respectively. X is a point inside ABC such that the incircle of XBC touches BC at D also, and touches CX and XB at Y and Z , respectively. Prove that EFZY is a cyclic quadrilateral.
A
Solutionsby Toshio Seimiya, Kawasaki, Japan; and by Vasiliou Meletis, Elefsis, Greece. We give Seimiya's argument.
F Z XY B D E C P
BP CE AF = 1: PC EA FB Since CE = CD, EA = AF , and FB = BD, we get BP CD = 1 PC BD
so that
Let P be the intersection of EF with BC . Then by Menelaus' Theorem we have 1
BP = BD : PC CD
2
73 Since XZ = XY , BZ = BD and CY = CD, we have from 2
BP CY XZ = BD CD XY = 1: PC Y X ZB CD Y X BD Hence by Menelaus' Theorem P , Z and Y are collinear. Since PF PE = PD2 and PZ PY = PD2 we have PF PE = PZ PY . Hence EFZY is a cyclic quadrialteral. Comment. If AB = AC then BD = DC and then it can easily be proved that AD is the perpendicular bisector of EF and Y Z so that EFZY
is an isosceles trapezoid, and is a cyclic trapezoid. 6. An acute triangle ABC is given. Points A1 and A2 are taken on the side BC with A2 between A1 and C , B1 and B2 on the side AC with B2 between B1 and A and C1 and C2 on the side AB with C2 between C1 and B so that
AA1A2 = AA2A1 = BB1B2 = BB2B1 = CC1C2 = CC2C1: The lines AA1 , BB1 , and CC1 bound a triangle, and the lines AA2 , BB2 and CC2 bound a second triangle. Prove that all six vertices of these two
A
triangles lie on a single circle. Solutions by Toshio Seimiya, Kawasaki, Japan; by D.J. Smeenk, Zaltbommel, the Netherlands; and by Vasiliou Meletis, Elefsis, Greece. We give Meletis' solution.
x x
C2
x
H
I
x
B1
B
A1
A2
C
Let AA1 , BB1 meet at the point E ; AA1 , CC2 meet at the point F ; and BB1, CC1 meet at the point I . Also A1AA2 = B1BB2 = C1CC2 = 2x: 1 The bisectors of the angles at A1 , B1 and C in triangles 4A1 AA2 , 4B1 BB2 and 4C1CC2 respectively are perpendicular to their respective
74 bases. Hence they are the altitudes of 4ABC . Let H be the orthocentre of 4ABC . Since A1 AH = B1 BH = x and A1 AH = C1 CH = x each one of the quadrilaterals AHEB , AHDC is inscribable in a circle. These two circles have a common chord, the segment AH and since ABH = ACH = 90 , BAC , then the circles have equal radii. Thus, since the inscribed angles EAH , DAH are equal, the corresponding chords HE and HD are equal. Therefore HE = HD. Similarly, we prove that HD = HI , and so on for all six vertices of these two triangles of the problem. Thus, all six vertices lie at the same distance from the point H , and the points are concyclic. Next we give a rather novel solution by Meletis to the second problem of the 37th IMO itself 1996: 303 . 2. Let P be a point inside triangle ABC such that
APB , ACB = APC , ABC: Let D, E be the incentres of triangles APB , APC respectively. Show that AP , BD and CE meet at a point.
Solution by Meletis Vasiliou, Elefsis, Greece.
A
1
x
2
D0
C
3
y
P E
4
D
B
Figure 1. We will rst prove the converse of the proposition and then apply it to prove the stated problem. Stage I. The equality APB , APC = ACB , ABC is equivalent to
A1 , A2 = C3 , B4
1
see gure 1. Stage II. Assume the conclusion of the problem; that is. assume that the bisectors of the angles PCA and PBA concur at a point I on AP . Then we have or
AC AI AB PC = PI = PB AC = PC : AB PB
2
75 This ratio indicates that if AD is the bisector of the angle BAC , then PD is the bisector of the angle BPC , or equivalently the points A; P belong to the circle of Apollonius" whose 3 diameter DD0 lies on the line CB , with D; D0 harmonic conjugates to the points C; B " Stage III. If 3 holds then 1 holds. Since AD bisects CAB , then Since PD bisects CPB then
PAD = x = A1 , A2 : 2
4
PDB , PDC = C3 , B4 and if we draw the bisector DE of the straight angle BDC we get PDE = y = C3 , B4 :
2
5
Because of 3, and since DE ? CD we obtain that DE is tangent at D to the circumcircle of the triangle 4APD. Hence x = y . 6 Combining 4, 5 and 6 we get that
A1 , A2 = C3 , B4
so the converse of the proposition is true. Stage IV. If 1 holds then 3 holds. A
1 6 2 7
P1 P D
5 4
C
3
B
Figure 2. Consider the circle of Apollonius with respect to the angle A of the triangle ABC with DD0 as diameter. Let P1 be the point at which the circle intersects line CP . We want to show that P P1 to complete the proof. Suppose P 6 P1 . We distinguish two cases: Case a. P1 is external to the segment CP see gure 2. Denote
A1 = CAP; A6 = CAP1;
A2 = BAP; A7 = BAP1;
76
Then we have
B4 = CBP and A1 A2 B4 C3 , B4 =
=
B5 = CBP1: A6 A7 B5: A1 , A6 , C3 , C3 , A2 A7 B5 B4:
7 8
So from the hypothesis and the conclusion of Stage III
or a contradiction.
C3 , B4
C3 , B4 ;
Case b. P1 is internal to the segment CD. The proof is similar. This completes the proof.
To complete this number of the corner and our le of solutions submitted by the readers to problems from 1996 numbers of the Corner, we give one solution to a problem proposed to the jury but not used at the 36th IMO. 5. 1996: 348 36th IMO Problems proposed to the jury and not used. Let ABC be a triangle. A circle passing through B and C intersects the sides AB and AC again at C 0 and B 0 , respectively. Prove that BB 0 , CC 0 and HH 0 are concurrent, where H and H 0 are the orthocentres of triangles ABC and AB0C 0 respectively. Solution by Toshio Seimiya, Kawasaki, Japan. See diagram on the next page. Let BH , CH meet AC , AB at D, E respectively. Since BDC = BEC = 90 , B , C , D, E are concyclic and thus ADE = ABC . Now, since B , C , B 0 , C 0 are concyclic we have
AB0 C 0 = ABC: Thus ADE = AB 0 C 0 , so that DE kB 0 C 0 . Let Q be the intersection of DC 0 with EB0 . 0As0HE ? AB, B0H 0 ? AB, we get HEkH 0B0. Similarly we have HDkH C .
77
A
H C
0
0
B
0
Q P D
E B
H
C
Since DE kC 0 B 0 , HDkH 0 C 0 and HE kH 0 B 0 , DC 0 , EB 0 and HH 0 are concurrent, so that H , Q, H 0 are collinear. By Pappus' Theorem H , P , Q are collinear. Hence H , P , Q, H 0 are collinear. So BB 0 , CC 0 , and HH 0 are concurrent. That is the Corner for this issue! Send me your National and Regional Olympiad Contest materials for use in the Corner.
78
BOOK REVIEWS
Edited by ANDY LIU
From Erdos to Kiev: Problems of Olympiad Caliber, by Ross Honsberger, published by the Mathematical Association of America, 1996, ISBN 0883853248, softcover, 257+ pages, $31.00. Reviewed by Bill Sands, University of Calgary, Calgary, Alberta. This book is number 17 in the Dolciani Mathematical Expositions series of the MAA, and is the seventh in this series by Ross Honsberger. His previous book in this series, More Mathematical Morsels reviewed in Crux on 1991: 235 236 by Andy Liu, contained mostly problems from Crux. The same is true of From Erdos to Kiev  and this time, all problems taken from Crux are accompanied by references year and page number to where in Crux the problem and sometimes solution appeared. Most of the Crux problems used here are from Rob Woodrow's Olympiad Corner columns from 1987 and 1988; thus they are mostly contest problems, some easy to nd elsewhere, like the AIME and IMO contests, others quite obscure. The author has sometimes used solutions taken from Crux, duly crediting the original solvers, and other times has come up with his own, but in all cases he has written or rewritten the problems and solutions as he liked: a privilege not usually available to Crux editors! The result is a readable and enjoyable book, that every Crux fan will want to own. Altogether the book has 89 problems collected into 46 unnumbered chapters", with no discernible reason for the order they are put in. Instead of a traditional index, at the end the problems are classi ed under three subjects roughly: combinatorics, algebra number theory, and geometry with a one line description of each problem. This reviewer can't single out many of the problems in this book for special mention  as the general editor of Crux, I was not as close to the material in the Olympiad Corners as I was to the regular Problems and Solutions columns. But Honsberger includes a few items from here too, and I recognized with pleasure one of Hidetosi Fukagawa's Sangaku problems on page 223  one of several he submitted to Crux and not the most striking one in my opinion before publication of his book Japanese Temple Geometry Problems with Dan Pedoe in 1989. Of course, none of you need to be sold on the virtues of Crux! One hopes, though, that other readers of this book would thus be drawn to subscribe to Crux, and maybe some of them will, but wouldn't it have been appropriate, given the book's debt to Crux, if the words Crux Mathematicorum" had been more visible? Yes, Crux is acknowledged, and praised, in the Preface, and the book is even dedicated to Crux's late founders L
o Sauv
e e and Fred Maskell, which is a thoughtful touch and is appreciated. But why couldn't Crux get some mention in the title or at least somewhere on the
79 cover? And the MAA's advertising for this book, as far as I have seen, does not mention Crux either. Come on, MAA  fair's fair! As for other criticisms  well, we could start with the accent on the o" of Erdos, which is wrong I've used the incorrect umlaut till now in this review!, in the title and elsewhere in the book too. As an aside, I might note that the title, though catchy, isn't especially appropriate, in that the book neither begins nor ends with a problem involving Erdos or Kiev; however, there are Erdos problems, and a problem from the Kiev Olympiad, in the book. Picky, picky : : : . I didn't notice many misprints or weaknesses of exposition liable to slow readers down. On the bottom of page 242 and the top of page 243, there are two displayed equations in which most of the terms have been left out, but the reader can probably reconstruct these with little trouble. On pages 116 117, to show that cn is not a multiple root of the polynomial Pnx = 0, it is much simpler just to di erentiate the polynomial Qnx, which is a factor of Pn x; we get
n + 1xn + nxn,1 + + 1 which is obviously positive for x = cn 0. Thus cn is a single root of Qn x and so also of Pn x.
On page 107, Daniel Ropp's university should be Washington University, not Washington State University the correct name is given in Crux. In oddly similar typos, the zeds in Bruce Reznick's last name and Zvi Margaliot's rst name are replaced by s's, on pages 179 and 187 respectively. In the second half of the book there is a rash of minor errors involving Crux references; for the bene t of readers rather than as criticism, I'll list them here. On page 153, the reference given 1987: 120 is to the original publication of the problem; the solution actually appeared on 1988: 182 . On page 159, the year for this reference should be 1988, not 1987. Also, the solution for this problem appeared on 1994: 191 193 with a further comment on 1995: 82 both probably too recent to be picked up by Honsberger's searches. On page 167, the solution for this problem appeared on 1988: 199 . On page 177, again the year of publication of the original Crux problem should be 1988 not 1987, and the solution appeared on page 267 of the 1989 volume, not page 269. And by the way, these references all refer to problem 1 of this chapter, which contains two problems. On page 239, the page reference given as 491 should be just 49. Having a whole chapter on Olympiad Corner solutions by George Evagelopoulos, the rst chapter in fact, was, I think, a mistake. Some of these solutions were equally due to other readers, as reported in Crux at the time. In fact, problem 1 of this chapter is listed as coming from the 1983 Australian Olympiad, which indeed it does, but the Crux reference given 1985: 71 is to the same problem as proposed by Brazil but not used at some IMO. The correct reference for the Australian problem is 1983: 173 , with
80 the o cial solution as supplied by Peter O'Halloran published on 1986: 22 . This is the same solution that Honsberger gives in his book and attributes to Evagelopoulos. Only on 1987: 43 is the Brazil version of the problem wrapped up, and here no solution is published, only the solvers are listed. And, as Honsberger mentions, these include two others as well as Evagelopoulos. Strangely, in two other cases Honsberger fails to mention solvers given equal credit in Crux for solutions he uses and ascribes to Evagelopoulos alone. The solution to problem 4 was credited on 1989: 230 also to Zun Shan and Ed Wang; what Honsberger calls a brilliant observation" could just as well be attributed to them. And for problem 7, in the published solution in Crux 1990: 105 Duane Broline is also listed as a solver. There are also remarkable similarities between some of Evagelopoulos's solutions and earlier solutions published elsewhere. For example, for problem 3, a problem from the Russian journal Kvant and rst published in Crux in 1988, Evagelopoulos's original solution in Crux 1990: 104 contained the same square diagram, complete with the same cells labelled A" and B ", as is present in the Kvant solution see page 24 of issue 12 of the 1987 Kvant; no doubt the Kvant solution is a beautiful gem", as Honsberger calls Evagelopoulos's solution on page 5. And for problem 6, also from Kvant, Evagelopoulos's solution in Crux 1990: 102 contained the same terminology representative", translated from the Russian and the same notation " as in the solution published on page 25 of issue 9 of the 1987 Kvant, to mention only two of the amazing resemblances between these two solutions. The last chapter contains an exposition with proof of the power mean inequality, and so is of a di erent character from the rest of the book. While this is a useful inequality to know, it's a bit jarring to have it suddenly appear here, especially when there is only one small, brief, unattributed example given of a problem that can be solved with it. There must be lots of problems from Crux that could have been used. However, let none of the above reservations prevent anyone from buying this book.
Did you know?
Members of the Canadian Mathematical Society are entitled to receive CRUX with MAYHEM at half price! With membership, the cost is only $30.00 for hard copy and online access including handling charges and delivery by surface mail. Contact
memberships@cms.math.ca
81
ONE PROBLEM SIX SOLUTIONS
Georg Gunther
Sir Wilfred Grenfell College, Corner Brook, Newfoundland The three ingredients necessary to solve a problem are insight, persistence and technique. In the following, all three of these play a role as six di erent solutions to the following problem are presented.
The Problem: On the three sides of triangle ABC , construct squares facing outwards. Let A0 , B 0 , C 0 be the centres of the squares constructed on sides BC , CA and AB, respectively. Prove that distA; A0 = distB0; C 0 and AA0 ? B0C 0 .
Solution 1: Analytic Geometry
6
y
B000
C 00 C0 C 000 B
A
2b; 2c
B0
B00
C A0 A00
2a; 0

x
It is important to make the coordinates work for you. Place the origin at B , and the x axis along BC . Let C = 2a; 0 and A = 2b; 2c. Then label the vertices of the three squares as shown. It is easy to show that B 00 =2a+2c; 2a,2b, A00 = 0; ,2a and C 000 = ,2c; 2b. Now we quickly obtain A0 = a; ,a, B 0 = a + b + c; a , b + c and C 0 = b , c; b + c.
Copyright c 1998 Canadian Mathematical Society
A000
82 Now, we have that distA;A0 = 2b , a2 + 2c + a2 = distB 0 ; C 0 , 2 , 2c+a slope AA0 = 2b,a and slope B 0 C 0 = , 2b+a , showing that the lines are c a perpendicular. Solution 2: Geometry
p
A C0 B c T P Q b a M A0
B0
C
Let M be the midpoint of BC . Consider triangles ABA0 and C 0 BM . We c a b b b have ABA0 = B + 45 = C 0 BM , AB = c, C 0 B = p2 , BA0 = p2 and a. BM = 2 So these triangles are similar, and 4C 0 BM is obtained from 4ABA0 by 1 rotating through 45 about B , and scaling by a factor of p2 .
p b p2 This means that C 0 M = AA0 = p2 , and that APC 0 = 45 . Likewise, trian0 and B 0 CM are similar, and 4B 0 CM is obtained from 4ACA0 gles ACA 1 by rotating through an angle of ,45 about C and dilating by p2 . p b Hence B 0 M = p2 and M QP = 45 .
2 2 C 0B02 = C 0M 2 + B0M 2 = p2 + p2 = p2; hence C 0 B 0 = p = AA0 . Also, since 4C 0 MB 0 is a rightangled isosceles triangle, we note that c M C 0 B0 = 45 .
c But this now tells us that B 0 MC 0 = 90 , and so
83
b c b In 4TC 0 P , we have P = C 0 = 45 and so T = 90 , telling us that C 0 B0 ? AA0 . Solution 3: Trigonometry
A C 0 z0 T y0 c
q q
q
xy
B0
b a
x
q q
B
q
0
C
z
A0
q
= 45
Consider the hexagon AB 0 CA0 BC 0 and all six of its diagonals. Now we use the cosine law:
b B0C 0 2 = AC 02 + AB02 , 2AC 0 AB0 cosA + 90 2 2 b = c + b + bc sinA:
2 2
Also
b AA0 2 = AC 2 + CA02 , 2AC CA0 cosC + 45 2
b b = b2 + a , ab cosC , sinC 2 b2 + c2 + ab sinC: b =
2 2
b b bc sinA = ac sinC = 2 area4ABC : Hence B 0 C 0 = AA0 . Likewise, C 0 A0 = BB 0 and A0 B 0 = CC 0 . Now, consider triangles B 0 C 0 B and A0 AC 0 . These triangles are congruent since B 0 C 0 = AA0 , C 0 B = AC 0 and BB 0 = C 0 A0 . Hence x = x0 . Likewise,
But recall that
84
y = y0 and z = z0. But then 2x +20y +2z = 180 and0 hence x + y + z = 90 . It follows, looking at triangle A0 B T , that A0 A ? B C 0 .
Solution 4: Vectors
C 00 i P
PP = PP P
! A , a
B0
7
B00
C0
! B=, b
@ @ @ @ R @
! C=, c
A0
Here, we let the vertices of the triangle be represented by vectors a , b and ,. ! c , ! Let n be the unit vector which is perpendicular to the plane containing the diagram. , , ! ! , , ! ! Recall that for any two vectors u , v , the vector u v is perpendicular , ! , ! to both u and v , and points in the direction given by the righthand rule. It follows that
A00
?
, , !!
We next compute:
,,! = , ,! = ! BA AC 00 n ,,! = , ,! = ! CB BA00 n ,,! = , ,! = ! AC CB00 n
, , ,, ; ! ! ! n a b
, , , !, ! ; ! b c n ,
, , , : ! ! ! n c a
,!0 = 1 h, , , + ,
, , , i ; , AC 2 ! ! ! ! ! b a n a b h, , ,
, , i ,!0 = 1 ! , ! + ! ! , ! ; , AB 2 c a n c a ,!0 = 1 h, , , + ,
, , , i : , BA 2 ! ! ! ! ! c b n b c
and
85 Let u be the vector from C 0 to B 0 ; let v be the vector AA0 . Now: and
, !
, , ! !
, !
,! ,
, = ,!0 , ,!0 = 1 h, , , + ,
, + , , 2, i ; ! AB AC , , ! ! ! ! ! ! u c b n c b a 2 , = ! v
h, , ,
, , i
, , !, !+ ! !, ! , !, ! c b n b c a b h, , , ,
, , i = 1 ! + ! , 2! + ! ! , ! : a n b c 2 c b
1 2
, , = 1 h,
, + , , 2, +
, , , i = , : ! ! ! ! ! ! ! ! ! n v c b a c b u 2 n , ! , , ! ! , , ! ! And since n = 1, we conclude that v = u and that v ? u .
Solution 5: Complex Numbers
But now:
A C0 C 000 B A0 A00
B0
B00
C
Place the diagram in the complex plane, with the origin at B . Let A and C represent the complex numbers at vertices A and C . Recall that for any complex number z , the product z 0 = zei is the complex number which is obtained by rotating z about the origin through the angle . If instead, we wish to rotate w about z through an angle , then w0 = z + w , zei.
86
w0 z w
B00 = C + A , C ,i = C 1 + i , Ai; C 000 = Ai; A00 = ,Ci: 1 1 Also, A0 = 1 C 1 , i, B 0 = 2 A1 , i + C 1 + i and C 0 = 2 A1 + i. 2 , ! , ! Let u be the vector from C 0 to B 0 ; let v be the vector from A0 to A. , ! , ! Now u = 1 C 1 + i , Ai and v = A , 1 C 1 , i. 2 2 , i = A , 1 C 1 , i = , . ! ! But now note that u v 2 0 A is obtained by rotating C 0 B 0 through 90 , which proves both parts So A
of the result that we seek.
Solution 6: Transformations
B000
Using this, we obtain the following:
A C0 B A0
q
B0
q
C
q
= 45
Rotate triangle ACA0 about C through ,45 and dilate by 2 to obtain triangle B 000 CB . This tells us that B 000 B is inclined at 45 with respect to AA0 .
p
87
1 Now rotate triangle AB 000 B about A through ,45 and dilate by p2 to obtain triangle AB 0 C 0 . 1 This tells us that B 0 C 0 = p2 B 000 B , and that B 0 C 0 is inclined at 45 with respect to B 000 B . But this proves our result.
So, there it is  one problem, six di erent approaches. Which solution is the best? The late Professor Paul Erdos used to talk about God's Little Black Book", in which could be found the perfect solution to every problem. Of the solutions presented here, perhaps the last one is the closest candidate for inclusion in this Little Black Book": it is economical, to the point and carries with it a wonderful element of surprise. Those are the properties of beautiful mathematics.
Have you heard about ATOM?
ATOM is A Taste Of Mathematics" Aime T On les Math
matiques. e The booklets in the series, A Taste of Mathematics, are published by the Canadian Mathematical Society CMS. They are designed as enrichment materials for high school students with an interest in and aptitude for mathematics. Some booklets in the series will also cover the materials useful for mathematical competitions at national and international levels. Publi
s par la Soci
t
math
matique du Canada SMC, les livrets de la e ee e collection Aimeton les math
matiques ATOM sont destin
s au perfectione e nement des etudiants du cycle secondaire qui manifestent un int
r^ t et des
ee aptitudes pour les math
matiques. Certains livrets de la collection ATOM e servent egalement de mat
riel de pr
paration aux concours de math
mati
e e e ques sur l'
chiquier national et international. e This volume contains the problems and solutions from the 1995 1996 Mathematical Olympiads' Correspondence Program. This program has several purposes. It provides students with practice at solving and writing up solutions to Olympiadlevel problems, it helps to prepare student for the Canadian Mathematical Olympiad and it is a partial criterion for the selection of the Canadian IMO team. For more information, contact the Canadian Mathematical Society.
The ATOM series
La collection ATOM
88
THE SKOLIAD CORNER
No. 28 R.E. Woodrow
Last number we gave the Preliminary Round of the British Columbia Colleges Junior High School Mathematics Contest for 1997. To keep in the ow of the contest we give Part A and Part B of the Final Round. Students whose performance in class on problems of the Preliminary Round was exemplary were invited to nearby colleges to attempt the Final Round as part of a larger daylong mathematical event. Again my thanks go to John Grant McLoughlin, now of the Faculty of Education, Memorial University of Newfoundland, who participated in formulating the exams while he was at Okanagan College.
BRITISH COLUMBIA COLLEGES Junior High School Mathematics Contest
Final Round 1997  Part A
1. The buttons of a phone are arranged as shown below. If the buttons are one centimeter apart, centretocentre, when you dial the number 5927018 the distance, in centimetres, traveled by your nger is: p53 p2 p2 p p p a p + p + 2 p b 2 5 + 23 + p5 p5 + p22 + 3 5 c 2 51 +p 2 + 2 2 d p p 1 2 3 e 51 + 2 + 4 10
4 7 5 8 0 6 9
2. What is the total number of ones digits needed in order to write the integers from 1 to 100? a 11 b 18 c 20 d 21 e 100 3. The number of solutions x; y; z in positive integers for the equation 3x + y + z = 23 is: a 86 b 50 c 60 d 70 e 92
89
4. In the diagram below the upper scale AB has ten 1 centimetre divisions. The lower scale CD also has ten divisions, but it is only 9 centimetres long. If the right hand end of the fourth division of scale CD coincides exactly with the right hand end of the seventh division of scale AB , what is the distance, in centimetres, from A to C ?
A C B D
a 3:2
b 3:25 c 3:3 d 3:35 e 3:4 5. Triangle ABCpis equilateral with sides tangent to the circle with center at O and radius 3. The area of the quadrilateral AOCB , in square units is:
A
O
B
a 3 3 b 3 d 3 e 2 3 6. Times such as 1:01, 1:11,: : : are called palindromic times because their digits read the same forwards and backwards. The number of palindromic times on a digital clock between 1:00 a.m. and 11:59 a.m. is: a 47 b 48 c 55 d 56 e 66 7. Ted's television has channels 2 through 42. If Ted starts on channel 15 and surfs, pushing the channel up button 518 times, when he stops he will be on channel: a 11 b 13 c 15 d 38 e 41 8. Consider a threedigit number with the following properties: 1. If its tens and ones digits are switched, its value would increase by 36. 2. Instead, if its hundreds and ones digits are switched its value would decrease by 198. Suppose that only the hundreds and tens digits are switched. Its value would: a increase by 6 b decrease by 540 c increase by 540 d decrease by 6 e increase by 90
p
C p c 6 3
p
90
9. Speedy Sammy Seamstress sews seventyseven stitches in sixtysix seconds. The time, in seconds, it takes Sammy to stitch fty ve stiches is: 1 b 44 1 c 45 5 d 46 1 e 47 1 a 43 1 3 4 6 7 10. How many positive integers less than or equal to 60 are divisible by 3, 4, or 5? a 25 b 35 c 36 d 37 e 44
Final Round 1997  Part B
1. a The pages of a thick telephone directory are numbered from 1 to N . A total of 522 digits are required to print the pages. Find N . b There are 26 pages in the local newspaper. Suppose that you pull a sheet out and drop it on the oor. One of the pages facing you is numbered 19. What are the other page numbers on the sheet? 2. Using each of the digits 1, 9, 9, and 7 create expressions for the numbers 1; 2; 3; : : : ; 10. Note that the digits must appear separately; that is, numbers like 17 are not allowed. Only the basic operations +, ,, , and brackets if necessary may be used. Other mathematical symbols such as p are not allowed. Every expression must include one 1, two 9's, and one 7, in any order. 3. a Decide which is greater: p6 + p8 or p5 + p9. +1 b Show that x x 2 for any real number x 0. 4. In the plane gure shown on the right, ABCD is a square with AB = 12. If A0 , B0, C 0, and D0 are the midpoints of AO, BO, CO, and DO, respectively, then:
2
A A
0
B B O D
0 0
C
0
D
a Find the area of the square A0 B 0 C 0 D0 . b Find the area of the shaded region. c Find the area of the trapezoid AA0 B 0 B .
C
91
5. The gure below shows the rst three in a sequence of square arrays of dots. The number of dots in the three arrays is 1, 5, and 13.
r r r r r r r r r r r r r r r r r r r
a Find the number of dots in the next array in the sequence. b Find the number of dots in the sixth array in the sequence. c Find an expression for an , the number of dots in the nth array in the sequence, in terms of n alone. d Find a relation between an+1 , an , and n. Last issue we gave the problems of the Preliminary Round of the British Columbia Colleges Junior High School Mathematics Contest 1997. Here are the answers:
1. E 6. A 11. C 2. A 7. D 12. B 3. D 8. E 13. D 4. A 9. B 14. C 5. C 10. D 15. E
That completes this number of the Corner. Send me your comments, suggestions, and suitable materials for use in the Skoliad Corner.
Did you know?
Members of the Canadian Mathematical Society are entitled to receive CRUX with MAYHEM at half price! With membership, the cost is only $30.00 for hard copy and online access including handling charges and delivery by surface mail. Contact
memberships@cms.math.ca
92
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Shreds and Slices
The Fibonacci Triangle We present an intriguing con guration of numbers, which might be p called the Fibonacci triangle. Recall that the golden ratio is = 1+2 5 , and that bxc is the greatest integer less than or equal to x. Starting with a 2, for any entry n, write b nc and b 2nc under it. The rst few rows we obtain are as follows:
2 3 4 6 9 15 16 10 26 17 11 28 29 7 18 47 19 12 31 32 8 20 52 33 21 54 55 5 13 34 89
A general portion of the triangle is the following:
b b ncc
b nc
n
b b ncc
2
b b ncc
2
b 2 nc
b 2b 2ncc
93 Some patterns are immediately apparent in the Fibonacci triangle; we present some now. Proposition 1. n + b nc = b 2 nc. Proof. b 2 nc = b + 1nc = b n + nc = b nc + n. Proposition 2. b nc + b 2 nc = b b 2ncc. Proof. We rst make a general observation: bxc = a if and only if a is an integer and a x a + 1. Most problems of this form will use this fact. Let a = b nc, so a n a + 1. Note b nc + b 2 nc = 2a + n, and b b 2ncc = b a + nc = b a + nc. Therefore, we must prove 2a + n a + n 2a + n + 1. For the left inequality, we have if and only if if and only if
2a + n a + n
2 , a , 1n , a 2, 1 n = n;
which is true; and for right inequality, if and only if if and only if
which is also true, and the statement is proved. These results, for one, verify that the last two entries of each row do indeed form the Fibonacci sequence. We nally show the one conspicuous result in the triangle. Proposition 3. Each positive integer greater than 1 appears exactly once in the table. Proof. If a and b are distinct positive integers greater than 1, then so are b ac and b bc. This is because 1, so a and b must di er by at least 1, and hence round down to di erent integers. The same argument works for 2. Hence, the same integer cannot appear as a left leg or a right leg. Now we invoke a classic result: Beatty's Theorem. If and are positive, irrational numbers such that
a + n 2a + n + 1 2 , a , 1n , 1 , 1 a 2, 1 n , 2, = n , , 1;
1 + 1 = 1;
then the sequences fb c; b2 c; b3 c; : : : g, fb c; b2 c; b3 c; : : : g contain each positive integer exactly once.
94 We can take = and = 2 . By Beatty's theorem, no integer can appear both in the form b ac and b 2bc. Hence, no integer appears more than once. But Beatty's also tells us that each integer greater than 1 must appear at least once. Hence, each integer greater than 1 appears exactly once. Note that in this proof, we used only the fact that the sum of the reciprocals of and 2 is 1; they can be replaced with any numbers satisfying the conditions of Beatty's Theorem the fact that each is greater than 1 also follows from this relationship. Problems 1. a Prove that b 2 nc , b b ncc = 1. b Prove that b b 2ncc , b 2 b ncc = 1. What do these say about the entries in the triangle? 2. An Fsequence is a sequence in which, except for the rst and second terms, every term is the sum of the two previous terms. a Show that the positive integers cannot be partitioned into a nite number of Fsequences. b Show that the positive integers can be partitioned into an in nite number of Fsequences.
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only solutions  the next issue will feature only problems. We warmly welcome proposals for problems and solutions. With the new schedule of eight issues per year, we request that solutions from the previous issue be submitted by 1 April 1998, for publication in the issue 5 months ahead; that is, issue 6 of 1998. We also request that only students submit solutions see editorial 1997: 30 , but we will consider particularly elegant or insightful solutions from others.
95
High School Solutions
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca H217. Let a1 , a2 , a3 , a4 , a5 be a veterm geometric sequence satisfying the inequality 0 a1 a2 a3 a4 a5 100, where each term is an integer. How many of these veterm geometric sequences are there? For example, the sequence 3, 6, 12, 24, 48 is a sequence of this type. Solution by Joel Schlosberg, student, Hunter College High School, New York NY, USA. n Let m be the common ratio of the geometric sequence, where n and n m are relatively prime integers, with n m. Now a5 = a1 m , so let 4 , where k is a positive integer. Thus, our geometric series becomes a1 = km 3 km4, km n, km2n2, kmn3, kn4, and kn4 100. If n 4, then kn4 n4 256 100. So n 3. Hence, there are three cases to consider: If n = 3 and m = 2, then 81k 100, so k = 1. The only solution is 16; 24; 36; 54; 81. If n = 3 and m = 1, then 81k 100, so k = 1. The only solution is 1; 3; 9; 27; 81. If n = 2 and m = 1, then 16k 100, so k = 1, 2, : : : , 6. There are six solutions, namely 1; 2; 4; 8; 16, 2; 4; 8; 16; 32, 3; 6; 12; 24; 48, 4; 8; 16; 32; 64, 5; 10; 20; 40; 80 and 6; 12; 24; 48; 96. In total, there are eight sequences. There was one incorrect submission received, where the solver incorrectly assumed that the common ratio had to be an integer. Hence, the solution 16; 24; 36; 54; 81 was missed. H218. A Star Trek logo is inscribed inside a circle with centre O and radius 1, as shown. Points A, B , and C are selected on the circle so that AB = AC and arc BC is minor that is, ABOC is not a convex quadrilateral. The area of gure ABOC is equal to sin m , where 0 m 90 and m is an integer. Furthermore, the length of arc AB shaded as shown is equal to a=b, where a and b are relatively prime integers. Let p = a + b + m. A
4 4
O B
B B B B B B
S B
S
SB
SB
S B
C
96 i If p = 360, and m is composite, determine all possible values for m. ii If m and p are both prime, determine the value of p. Solution. i Let BOC = 2x , where 0 x 90. Otherwise, BOC would exceed 180 and ABOC would be a convex quadrilateral and not the gure of a Star Trek logo. Draw OA. Since AB = AC , OAB and OAC are congruent triangles, and so AOB = AOC . Since AOB + AOC = 360 , 2x , we have AOB = 180 , x . Thus, the area of ABOC is twice that of triangle OAB , so sin m = OA OB sin180 , x = sin x . Since 0 m; x 90, sin m = sin x implies that m = x. Hence AOB = 180 , m and the length of arc AB is
2180 , m = 180 , m : 360 180
Thus,
a and b are relatively prime integers. Hence, a 180 , m and b 180 with equality occurring i 180,m is irreducible. 180 In this case, a + b + m = 180 , m + 180 + m = 360, and so if p = 360, we must have gcdm; 180 = 1. But m is composite and cannot have any common divisors with 180 = 22 32 5. So m must be a product of primes greater than 5. But m 90, so we only have two possible choices for m: 7 7 and 7 11. Hence, if p = 360 and m is composite, then m = 49 or 77. ii If m is a prime larger than 5, we will have p = 360, since gcdm; 180 = 1. Since 360 is not prime, we must restrict our choices of m to 2, 3 and 5. If m =a 3 or178 = 589we will nd that p is composite. However, if m = 2, m , we have b = 180 = 90 , so p = a + b + m = 89 + 90 + 2 = 181, which is prime. Hence, m = 2 and p = 181. H219. Consider the in nite sum S = a0 + a1 + a2 + a3 + ;
180 180
,m = a , where b
100
102
104
106
where the sequence fan g is de ned by a0 = a1 = 1, and the recurrence p relation an = 20an,1 + 12an,2 for all positive integers n 2. If S can a be expressed in the form pb , where a and b are relatively prime positive integers, determine the ordered pair a; b. Solution. We have
97
a
a a a S , 20S , 12S = 1000 + 1012 + 1024 + 1036 + 102 104
, 20a20 + 20a41 + 20a62 + 20a83 + 10 10 10 10 12a0 12a1 12a2 12a3
, + + + +
104 106 108 1010 a a a = 1000 + 1012 , 20a20 + a2 , 201014, 12a0 10 a3 , 20a2 , 12a1 + a4 , 20a3 , 12a2 + + 106 108
Since an , 20an,1 , 12an,2 = 0 for all positive integers n 2, we have
a a S , 20S , 12S = 1000 + 1012 , 20a20 ; 2 4 10 10 10
81 7988S so S 2025 and a = 10000 1997 pS substituting inso0the a1 = 1, we have pair is=a;100,= 45=1997.. Hence, 45 = p , and desired ordered b ; 1997
H220. Let S be the sum of the elements of the set f1; 2; 3; : : : ; 2pn , 1g. Let T be the sum of the elements of this set whose representation in base 2p consists only of digits from 0 to p , 1. Prove that 2n T = p , 1=2p , 1. S
Solution. n n We have S = 2p ,1 2p . Let R denote the set of numbers that have 2 at most n digits in base 2p and contain only the digits from 0 to p , 1. Thus T is the sum of the elements of R. For example, when p = 2 and n = 3, we have R = f0; 1; 10; 11; 100; 101; 110; 111g and T = 11104 = 84. Note R will have pn elements, because each of the n digits including leading zeros can be any one of p di erent numbers. Since each number in f0; 1; : : : ; p , 1g appears the same number of times as a digit in R, there will be exactly pn,1 elements in R that have k as their tth digit, for k = 0; 1; : : : ; p , 1 and t = 1; 2; : : : ; n. Thus,
T =
=
n,1 X i=0 n,1 X
2pi pn,1 0 + pn,1 1 + + pn,1 p , 1
n,1 n X , 2pi pn,1 pp 2 1 = p p2, 1 2pi i=0 i=0 pnp , 1 2pn , 1 : = 2 2p , 1
98 Hence,
n p p,1 2p ,1 p n T = 2 2n 2n ,1 = p , 1 ; 2 S 2p ,1 2p 2p , 1 2
n n
as required. This can also be solved by induction that is an exercise left for the reader.
Advanced Solutions
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu
A193. If f x;y is a convex function in x for each xed y, and a convex function in y for each xed x, is f x;y necessarily a convex function in x and y ? Solution. No; for example, take
xy f x; y = e , 1 if x; y 0; 0 otherwise,
Then f satis es the given conditions. However, f 0; 2 = f 2; 0 = 0, and f 1; 1 = e, so f is not convex. A194. Let H be the orthocentre point where the altitudes meet of a triangle ABC . Show that if AH : BH : CH = BC : CA : AB then the triangle is equilateral. Solution by Deepee Khosla, Ottawa, ON. Let a = BC , b = AC , c = AB , as usual, and set A0 = AH BC , 0 = BH AC , C 0 = CH AB . Since AA0 B = BB 0 A, it follows that B quadrilateral ABA0 B 0 is cyclic with diameter AB . Thus AH HA0 = BH HB0. Together with the givens, we have
HB0 = AH = a ; HA0 BH b
so that
a BH = b AH; a HA0 = b HB0 :
1 2
99 Also, 2K = a AA0 = aAH + HA0 , where K is the area of triangle ABC . This implies a HA0 = 2K , a AH , and similarly for b HB0 . Putting these into equation 2 gives a AH = b BH: 3 Multiplying 1 and 3 and cancelling yields a2 = b2 or a = b since a; b 0. A similar argument shows that a = c, so triangle ABC is equilateral. A195. Compute tan 20 tan 40 tan 60 tan 80 . Solution I by D.J. Smeenk, Zaltbommel, the Netherlands. Let A = tan 20 tan 40 tan 60 tan 80 . Then A = B , where B = C 8 sin 20 sin40 sin60 sin 80 and C = 8 cos 20 cos 40 cos 60 cos 80 . Then
1 + 2 sin 10 cos 20 + sin10 cos 20 + sin10 + 2 sin 10 cos 20 + 2 sin2 10 cos 20 + sin10 + sin 30 , sin 10 + 1 , cos 20 = 3 ; 2 8 cos 20 cos 40 cos 60 cos 80 2cos 60 + cos 100 cos 20 + cos 100 1 , 2 sin 10 cos 20 , sin10 cos 20 , sin10 , 2 sin 10 cos 20 + 2 sin2 10 cos 20 , sin10 , sin 30 + sin 10 + 1 , cos 20 = 1 : 2 Therefore, A = B = 3. C
B = 8 sin 20 sin40 sin 60 sin80 = 2cos 60 , cos 100 cos 20 , cos 100
= = = C = = = = =
,9 tan , ,9 tan3 + ,9 tan5 , ,9 tan7 + ,9 tan9 tan 9 = 1 ,9 ,9 3 2 ,9 5 4 ,9 7 6 ,9 9 8 : 0 , 2 tan + 4 tan , 6 tan + 8 tan The LHS is zero for the nine values = 0 , 20 , 40 , : : : , 160 , so the roots of 9
9
9
9
9
5 7 9 1 x, 3 x+ 5 x , 7 x + 9 x =0 are the nine distinct values tan 0 = 0, tan 20 , tan 40 , : : : , tan 160 . Note that tan 100 = , tan 80 , : : : , tan 160 = , tan 20 . By taking out a factor of x, we see the product of the roots is 9
2 2 2 2 tan 20 tan 40 tan 60 tan 80 = 1 = 9;
Solution II. By a wellknown identity,
100 so that tan 20 tan 40 tan 60 tan 80 = 3. Also solved by Deepee Khosla, Ottawa, ON, and Bob Prielipp, University of Wisconsin Oshkosh, Wisconsin, USA. 2 2 2 A196. Show that r2 + ra + rb + rc 4K , where r, ra, rb, rc, and K are respectively the inradius, exradii and area of a triangle ABC . Solution by Deepee Khosla, Ottawa, ON, and Bob Prielipp, University of Wisconsin Oshkosh, Wisconsin, USA. We have Thus, by the AMGM inequality,
r = K ; ra = s K a ; rb = s K b ; and rc = s K c : s , , ,
2 2 2 2 2 2 2 r2 + ra + rb + rc = K2 + s K a2 + s K b2 + s K c2 ss , , , 2 8 K 4 s2s , a2s , b2s , c2 = 4 K = 4K: K
4
Challenge Board Solutions
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu C73. Proposed by Matt Szczesny, University of Toronto. The sequence fan g consists of positive reals such that the sum of the an diverges. Show that the sum of an =sn diverges, where sn is the nth partial sum. Solution by the proposer. Since the an are positive, the sequence fsn g is increasing. Putting s0 = 0, for any nonnegative integer N and positive integer k, we therefore have
N +k X
X an N +k an = 1 s , s = 1 , sN : N +k N sN +k n=N +1 sn n=N +1 sN +k sN +k For every nonnegative integer N there exists, by the divergence of fsn g, s some k such that sNN k 1 . It follows that for each such N we can obtain 2 aN +1 + + aN +k 1 ; and hence the sum a1 + a2 + a k so that sN +1 sN +k 2 s1 s2
+
must diverge.
101
The Order of a Zero
graduate student, Yale University Recall that a root r of a polynomial px has multiplicity m if x , rm divides px but x , rm+1 does not. In other words, px = x , rmq x for some polynomial q x, and q r 6= 0. This means that px behaves much like q rx , rm around x = r, for example in graphing px. In fact, in many applications, the values q r and m are all the data we have to know. We extend these ideas analytically to calculus, where they can play a useful and powerful role. Keep in mind that the notion of multiplicity of a root will be our main motivation. First, we give the analogue of multiplicity in calculus. De nition. We say f x has order r at x = a if for some nonzero constant c. Let us call c the rvalue of f x at x = a. First, it should be clear that the values of r and c are unique. To see this, assume that f x has order r1 and r2 at a, so
Naoki Sato
lim x!a x , ar = c
f x
lim lim x!a x , ar1 = c1 ; and x!a x , ar2 = c2 :
f x
f x
c2 c1 : Since c1 and c2 are nonzero, we must have r1 = r2 and of course, c1 = c2 . Thus, we can now speak of the order of f x at a. However, the order at a need not exist see Example 4 below. From a numerical viewpoint, f x behaves like cx , ar for x close to a, so the order gives a measure of how quickly f x approaches or recedes from 0 as x approaches a. The greater r is, the faster f x decreases. Note that if r is negative, then r must be an integer why?, and f x goes to 1 as x approaches a.
lim x!a x , a
r1,r2 =
Then
Examples. 1. If a root r of polynomial px has multiplicity m, and px = x , rm qx, then px has order m and rvalue q r at r. In particular, the order is a nonnegative integer which is at most the degree of px. For example, the polynomial x , 12xx + 3 has order 2 and rvalue 4 at x = 1.
Copyright c 1998 Canadian Mathematical Society
102 2. If f is continuous at a and f a = 0, then f x has order 0 and 6 rvalue f a at a. It neither approaches nor recedes from 0 at x = a. 3. It is well known that x!0 lim sin x = 1: Hence, sin x has order 1 and x rvalue 1 at x = 0. 4. As mentioned, the order need not exist. For example, the somewhat trivial f x = 0 has no order at any point, and f x = jxj has no order at x = 0. Also, f x = xjxj is di erentiable at x = 0, but has no order at x = 0. This shows that even regular behaviour such as di erentiability
cannot guarantee the existence of order! We do have the following su cient conditions for continuity and differentiability. Proposition. If f x , f a has order greater than 0 at a, then f is continuous at a. If f x , f a has order at least 1 at a, then f is di erentiable at a. Proof. Let r be the order of f x,f a at a. Assume that r 0. Recall that f is continuous at a if and only if f a = limx!a f x, or equivalently limx!af x , f a = 0. But we see
lim lim x!a f x , f a = x!a
f x , f a lim x , ar = 0: x , ar x!a
In this product, the rst limit exists by de nition of order, and the second limit exists and is 0 since r 0. Hence, f is continuous at a. Similarly, recall that f is di erentiable at a if and only if
lim x!a
exists, and the derivative f 0a is the value of this limit. Assume that r 1. Then
f x , f a x,a
f x , f a = lim f x , f a lim x , ar,1; x!a x , ar x!a x,a and in this product, both limits exist, so f is di erentiable at a.
lim x!a
Remarks. Example 4 shows that the converse does not hold. Also, the function f x, which is 0 at x = 0 and 1 everywhere else, has order 1 at x = 0 but is not continuous at x = 0; hence, we do require that r 0 for continuity. So order can give some useful information, but this seems academic, since continuity and di erentiability are easy to determine anyway. We now present a result which uses orders and rvalues to compute limits. Proposition. Let f1x, f2x have orders r1, r2 and rvalue c1 , c2 at a respectively. Then f1xf2x has order r1 + r2 and rvalue c1 c2 at a, and
103
f1x=f2x has order r1 , r2 and rvalue c1=c2 at a. Also, lim f1x = c1 0 2 if r1 = r2 : x!a f2x =c if r1 r2 If r1 r2 , then lim f1x = sgnc1 =c2 1; x!a f2x and if r1 , r2 is an integer, then lim, f1x = sgnc1=c2,1r ,r 1: x!a f2x Else, if r2 , r1 is not an integer, then this last limit does not exist.
+ 2 1
A proof follows almost directly from the de nition, and we will not include one here. Note that the evaluation of the limit only requires knowledge of r1 , r2, c1 , and c2 , justifying the claim that these are the only values one usually needs to know. Actually nding these values may vary from trivial to di cult, which we will address in a moment. The philosophy behind this concept is, as pointed out before, that if f x has order r and rvalue c at a, then f x behaves like cx , ar at a. So, in evaluating the limit in the former proposition, the method that should work, and the one we should always pursue is to substitute cx , ar for f x since they behave similarly at x = a, cancel all factors of x , a, and see what is left, because what is left is what precisely determines the value of the limit. For example, in taking the limit of a rational function as x approaches a, what we do is precisely as described above; that is, factor and cancel powers of x , a. Other limits are not so straightforward. For example, a recent nal exam in a rst year calculus course at the University of Toronto asked the following: Evaluate Analogously, what we would like to do is factor" out the powers of x, or more precisely, nd the orders and rvalues of the numerator and denominator at 0. For other similar limits, we must also calculate the orders and rvalues for fairly general functions. As of now, there does not seem to be a clear way of doing this. We give a systematic method for a reasonable class of functions, and we return to the polynomial case for motivation of a di erent description of order. Lemma. Let px = an xn + an,1 xn,1 + + a1 x + a0 . Then there exist unique constants cn , cn,1 , : : : , c1 , c0 such that px = cn x , an + cn,1 x , an,1 + + c1 x , a + c0. Proof. The ci are determined by the following translation:
, arctan lim sin xln1 + x x : 2 x!0 x
px + a = cnxn + cn,1 xn,1 + + c1x + c0 px = cnx , an + cn,1 x , an,1 + c1x , a + c0:
104
ck = pka=k!.
In particular, we see that by di erentiating both sides k times,
Corollary. x , ar divides px if and only if pk a = 0 for k = 0; 1; : : : ; r , 1. Proof. Both conditions are equivalent to c0 = c1 = = cr,1 = 0. Proposition. Let px be a nonzero polynomial. Then the order of prx at a is the value of r such that pka = 0 for k = 0; 1; : : : ; r , 1, and p a 6= 0, and the rvalue at a is p r a=r!. Proof. Left as an exercise for the reader. We now state the generalization we seek. Theorem. Let f x be a function which has a Taylor expansion around x = a, say
f x = c0 + c1 x , a + c2x , a2 + ; such that not all the coe cients ci are zero, or in other words, f x is not identical to zero around x = a. Then the order of f x at a exists, and is the unique nonnegative integer r such that f ka = 0 for k = 0; 1; : : : ; r , 1, and f ra = 0, and the rvalue at a is f ra=r!. 6 Proof. For all k 0, f ka = ck . Let cr be the rst coe cient which is nonzero. Then r is the order of f x at a, as described, and lim xf,xar = cr = f ra=r! 6= 0: x!a Let us now apply this to the example above. Let f1x = sin x , arctan x, f2x = ln1 + x. Then f10 x = cos x , x2 1 1 ; f10 0 = 0; + f100 x = , sin x + x2 2x 12 ; f1000 = 0; + 2x2 + 12 , 8x2x2 + 1 ; f 000 0 = 1: f1000 x = , cos x + 1 x2 + 14
1 f20 x = 1 + x ; f20 0 = 1; lim x!0 sin x , arctan x = 1 : x2 ln1 + x 6
Hence, f1 x has order 3 and rvalue 1=6 at 0, and
so f2x has order 1 and rvalue 1 at 0. By a previous proposition,
The orders cancel, and we obtain a nonzero number.
105 Remark. An approach using L'H^ pital's Rule does work, but one has o to break the limit up as follows:
lim x!0
and then, the calculations actually amount to the same thing we have done and the reader, if not sure, should check this. Lastly, we present a method for determining partial fraction expansions, using the ideas so far. Looking at one root, we wish to nd constants A1 , A2 , : : : , An such that We examine two cases. Case I. n = 1 If the exponent is 1, then there is a very simple way of determining the coe cient. In fact, it is so easy one can usually do it in one's head. We illustrate with an example, where the exponent is 1 for all the roots. We know for some constants A, B , and C . Multiplying by x , 2, we obtain
sin x , arctan x = lim sin x , arctan x lim x 2 ln1 + x 3 x!0 x!0 ln1 + x ; x x
px = A1 + A2 + + An + : 1 x , anqx x , a x , a2 x , an
x,1 A B C x , 2xx + 1 = x , 2 + x + x + 1
x,1 Bx , 2 C x , 2 xx + 1 = A + x + x + 1 : Substituting x = 2, we automatically get A = 1=6. Similarly, we also obtain B = 1=2, and C = ,2=3, and so x,1 1 1 2 x , 2xx + 1 = 6x , 2 + 2x , 3x + 1 :
x+1 = A + B + C + D + E x x , 13 x x2 x , 1 x , 12 x , 13 for some constants A, B , C , D, and E . The values of B and E can be found
2
One must agree this is much easier and faster than the usual method by solving a system of linear equations. Case II. n 1 When there are multiple roots, the situation becomes much more complicated, and the method in Case I no longer works. For example, we know
by the method in Case I, but there is no way to directly isolate the other
106 constants in the same way. The key here, as in a previous proposition, is to make a smart substitution, to make these terms where the function goes to 1 into nicer terms. Going back to 1, if we substitute x = 1=t + a, we obtain Amazing! We get a garden variety polynomial terms that blow up at a have become powers of t, with this substitution; why?. So, we make the substitution, and take the polynomial part. Furthermore, the Ai are simply the coe cients of the polynomial. But how do we know these are the right coefcients? Could there be more terms to the right, in that last equation, that contain polynomial terms? No, not if we make sure that the expression remaining on the right stays bounded as t approaches 1. Then it can't contain any powers of t. We illustrate with an example. Let
p1=t + atn = A t + A t2 + + A tn + : 1 2 n q1=t + a
The polynomial part of this expression is 2 + t; we can be sure of this since the rest of the expression goes to 0 as t approaches 1. Similarly,
C 1 rx = xx , 12 = A + x B 1 + x , 12 x , for some constants A, B , and C . Then 3 t3 r 1 = 1 1 1 12 = 1 , t2 = ,yy+ 1 sub. y = 1 , t 2 t t t, 3 1 3 = ,y + 3 , y + y12 = 2 + t , 1 , t + 1 , t2 : t3 1 r1 + 1 = 1 + 1 1 2 = t + 1 = t2 , t + 1 , t + 1 : 1 t
t t
Hence, the polynomial part is 1 , t + t2 . Therefore, A = 1 the coe cient of t in 2 + t, and B = ,1 and C = 1 the coe cients of t, t2 in 1 , t + t2 . Hence,
1 1 1 = x , x , 1 + x , 12 : xx , 1 1
2
We will leave it the reader to consider what happens in the case that the denominator of the fractions has nonreal roots.
107
PROBLEMS
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 October 1998. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX. Proposed by Toshio Seimiya, Kawasaki, Japan. Given triangle ABC with AB AC . The bisectors of angles B and C meet AC and AB at D and E respectively, and DE intersects BC at F . 1 Suppose that DFC = 2 DBC , ECB . Determine angle A.
2314.
2315. Proposed by V
aclav Konecn
y, Ferris State University, Big Rapids, Michigan, USA. s
maximum value of
1
n,1 , where F n is the Prove or disprove that F n = n 1 , n
f x1; x2; : : : ; xn = sin x1 cos x2 : : : cos xn + cos x1 sin x2 : : : cos xn + : : : + cos x1 cos x2 : : : sin xn ; xk 2 0; =2 , k = 1; 2; : : : ; n, and n 1 is a natural number.
108 Proposed by Toshio Seimiya, Kawasaki, Japan. Given triangle ABC with angles B and C satisfying C = 90 + 1 B . 2 Suppose that M is the midpoint of BC , and that the circle with centre A and radius AM meets BC again at D. Prove that MD = AB . USA.
2316.
2317. Proposed by Richard I. Hess, Rancho Palos Verdes, California,
4
b
c
2 3
a
d
e
The quadrilateral shown at the left has integer elements a through e. The angles as shown are integer multiples of the smallest. a What is the smallest possible value of c? b What is the smallest possible value of c if must be obtuse?
2318. Proposed by V
aclav Konecn
y, Ferris State University, Big Rapids, Michigan, USA. Suppose that ABC is a triangle with circumcentre O and circumradius R. Consider the bisector ` of any side say AC , and let P the pedal point" be any point on ` inside the circumcircle. Let K , L, M denote the feet of the perpendiculars from P to the lines AB, BC , CA respectively. Show that KLM the area of the pedal triangle KLM is a decreasing function of = OP , 2 0; R.
Proposed by Florian Herzig, student, Perchtoldsdorf, Austria. Suppose that UV is a diameter of a semicircle, and that P , Q are two points on the semicircle with UP UQ. The tangents to the semicircle at P and Q meet at R. Suppose that S is the point of intersection of UP and V Q. Prove that RS is perpendicular to UV . Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Two circles on the same side of the line ` are tangent to it at D. The tangents to the smaller circle from a variable point A on the larger circle intersect ` at B and C . If b and c are the radii of the incircles of triangles ABD and ACD, prove that b + c is independent of the choice of A.
2319.
2320.
109
2321. Proposed by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA. Suppose that n 2. Prove that
Here, as usual, bxc means the greatest integer less than or equal to x. Proposed by K.R.S. Sastry, Dodballapur, India. x2 y2 Suppose that the ellipse E has equation 2 + 2 = 1. Suppose that a b , is any circle concentric with E . Suppose that A is a point on E and B is a point of , such that AB is tangent to both E and ,. Find the maximum length of AB .
n n X n2 X n2 = : k=n+1 k k=2 k
2
2322.
2323.
uv2 , v2 , uv , u = c has exactly four solutions in positive integers u and v .
n=1 un
Proposed by K.R.S. Sastry, Dodballapur, India. Determine a positive constant c so that the Diophantine equation
2324. Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel, Alcorc
n on, Madrid, Spain. 1 X 1 Find the exact value of , where un is given by the recurrence
n , 1
un = n! n un,1 ; with the initial condition u1 = 2.
Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Suppose that q is a prime and n is a positive integer. Suppose that fak g 0 k n is given by
n X k=0
2325?.
akxk = q1n
Prove that each ak is an integer.
n X qn
k qk qx , 1 : k=0
110
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems.
2208. 1997: 47 Proposed by Christopher J. Bradley, Clifton College, Bristol, UK. 1. Find a set of positive integers fx; y;z; a; b; c; kg such that
y2z2 = a2 + k2 z2x2 = b2 + k2 x2y2 = c2 + k2
2. Show how to obtain an in nite number of distinct sets of positive integers satisfying these equations.
I. Solution by C. FestraetsHamoir, Brussels, Belgium. 1. Partons de triples pythagoriciens el
mentaires
e
52 12 = 32 + 42 12 52 = 42 + 32 52 52 = 242 + 72 Multiplions ces egalit
s respectivement par 32 72 , 72 42 et 42 32 ; on obtient
e 152 72 = 632 + 842 72 202 = 1122 + 842 202 152 = 2882 + 842 Donc fx; y;z; a; b;c; kg = f20; 15; 7; 63; 112; 288; 84g.
2. Etant donn
e une solution: e
il su t de multiplier ces egalit
s respectivement par b2c2 , c2 a2 , a2 b2 pour
e en obtenir une novelle:
y2z2 = a2 + k2 z2x2 = b2 + k2 x2y2 = c2 + k2
b2c2y2z2 = a2b2c2 + k2b2c2 c2 a2z2x2 = a2b2c2 + k2c2a2 a2b2x2 y2 = a2b2c2 + k2a2b2
111 On pose ce qui donne
2 2 b2y2 = y1 2 2 c z = z1 a2x2 = x2 1 2 a2b2c2 = k1 2 k2b2 c22 = a2 1 2 k2c 2a2 = b2 1 k a b = c2 1
2 2 2 y1 z1 = k1 + a2 1 2 2 z1 x2 = k1 + b2 1 1 2 x2y1 = k1 + c2 1 1
II. Solution by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We solve part 2 rst. From our construction we can easily obtain particular solutions. Let q be any even natural number with the property that q=2 can be written as the product of two distinct factors in at least three di erent ways so that q = 2u1 v1 = 2u2 v2 = 2u3 v3 where ui 6= uj , vi 6= vj , and ui 6= vj for i; j = 1; 2; 3, i 6= j . For example, we could let q = 2p1 p2 p3 where p1; p2 , and p3 are three distinct odd primes and let 2 2 2 ui = pi, vi = q=2ui. Let a1 = ju2 , v1 j, b1 = ju2 , v2 j, c1 = ju2 , v3 j. 1 2 3 Then 2 2 a2 + q2 = u2 + v1 2 = l2 2 where l = u2 2 v1 2 1 1 1 + 2 + q 2 = u2 + v 2 2 = m a2 2 where m = u2 + v2 2 2 2 2 a2 + q = u2 + v3 2 = n2 where n = u2 + v3 3 3 3 Now, let d = lmn, a = a1 d, b = b1d, c = c1 d, k = qd, x = mn, y = nl, and z = lm. Then
a22 + k22 = d2a2 + q2 = d2 l22 = l42m24n22 = y22z22 1 b 2 + k 2 = d22b22 + q22 = d22m2 = l2 m2 n4 = z2 x2 1 c + k = d c1 + q = d n = l m n = x y To get a particular solution, we could take q = 24 = 2 1 12 = 2 2 6 = 2 3 4 so u1 = 1, v1 = 12, u2 = 2, v2 = 6, u3 = 3, v3 = 4. Then a1 = 143, b1 = 32, 3c1 = 7, l = 145, m = 40, n 4= 25, d = 23 454 29, which leads to a = 2 54 111329, b = 28 5 29, c = 23 5 729, x = 23 53, y = 53 29, z = 23 52 29 and k = 26 3 54 29.
Also solved by GERALD ALLEN and CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; ROBERT B. ISRAEL, University of British Columbia, Vancouver, BC; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; DIGBY SMITH, Mount Royal College, Calgary, Alberta; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer. There was one incomplete solution.
112
2209. 1997: 47 Proposed by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain. Let ABCD be a cyclic quadrilateral having perpendicular diagonals crossing at P . Project P onto the sides of the quadrilateral. 1. Prove that the quadrilateral obtained by joining these four projections is inscribable and circumscribable. 2. Prove that the circle which passes through these four projections also passes through the midpoints of the sides of the given quadrilateral.
Comments. The recent solution to Crux with Mayhem 2194 1997: 530532 discussed the ambiguity of the word circumscribable. It is perhaps preferable to use the terminology cyclic and circumscribing for our bicentric quadrilateral; cf. Crux with Mayhem 2203. Solution by Florian Herzig, student, Perchtoldsdorf, Austria. Let the projections of P be K , L, M , N with K 2 AB , etc. Furthermore, let the midpoints of AB , BC , CD, DA be Q, R, S and T . From one of Brahmagupta's theorems 1 theorem 3.23, P is collinear with each projection such as K and the midpoint opposite that projection namely S. Furthermore QRST is a rectangle since its sides are parallel to the perpendicular diagonals AC and BD of the given quadrilateral. The diagonals QS and RT of the rectangle are diameters of its circumcircle, and this circle contains the projection points such as K since they form right triangles namely QKS that have a diameter as hypotenuse. In other words, K , L, M , N , Q, R, S, T are concyclic. Next we have four cyclic quadrilaterals each having an opposite pair of right angles: KBLP , LCMP , MDNP , and NAKP . Thus
PKN = PAN = CAD = CBD = LBP = LKP; and hence, PK is an interior angle bisector of KLMN . Likewise PL, PM , PN are interior angle bisectors, whence KLMN has an inscribed circle with incentre at P . Note that the existence of an incircle does not require AC ? BD.
Reference.
1 H.S.M. Coxeter and S.L. Greitzer, Geometry Revisited. MAA, 1967. Comments. The many references supplied by our solvers indicate that each part of problem 2209 can easily be found elsewhere; for example, see Crux 1836 1993: 113; 1994: 8485 , Crux 1866 1993: 203; 1994: 176 , and problem 3 on the 1990 Canadian Mathematical Olympiad 1990: 198199 . Seimiya found the problem itself as th
or me 159, p. 319, in F.G.M., e e Exercices de g
om
trie. For the information of newcomers to Crux with e e Mayhem, F.G.M. was a favourite reference of founding editor L
o Sauv
. e e
113 The fth edition of 1912 Cours de mathematiques No. 267, Tours: Maison A. Mame et Fils, etc. contains 2000 theorems with their proofs  nearly every result of elementary geometry known at that time. Because of his religious a liation, the author had to remain anonymous. Lambrou mentions a converse: given a bicentric quadrilateral KLMN there exists an appropriate cyclic quadrilateral ABCD with orthogonal diagonals. The sides of ABCD are the perpendiculars at K; L; M; N to the lines from the incentre P . Also solved by CLAUDIO ARCONCHER, Jundia
, Brazil; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; JORDI DOU, Barcelona, Spain; C. FESTRAETSHAMOIR, Brussels, Belgium; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Aus
tria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; GOTTFRIED PERZ, Pesta
lozzigymnasium, Graz, Austria; ISTVAN REIMAN, Budapest, Hungary; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; YEO KENG HEE, Hwa Chong Junior College, Singapore; and the proposer.
2211. 1997: 47 Proposed by Bill Sands, University of Calgary, Calgary, Alberta. Several people go to a pizza restaurant. Each person who is hungry" wants to eat either 6 or 7 slices of pizza. Everyone else wants to eat only 2 or 3 slices of pizza each. Each pizza in the restaurant has 12 slices. It turns out that four pizzas are not su cient to satisfy everyone, but that with ve pizzas, there would be some pizza left over. How many people went to the restaurant, and how many of these were hungry"? Solution by Kathleen E. Lewis, SUNY Oswego. Oswego, NY, USA. I am interpreting the problem to mean, for example, that the hungry" people would each be satis ed with 6 pieces, but would eat 7 if they were available. If x represents the number of hungry" people, and y the number of non hungry" people, then we know that 6x + 2y 48, but 7x + 3y 60. Because x and y must be integers, the quantities 6x + 2y and 7x + 3y are also integers and the rst of these must be even. Therefore we can rewrite the inequalities as 6x + 2y 50 and 7x + 3y 59:
114 Taking the di erence, we see that x + y , the total number of people, is at most 9. If x is seven or less, then 6x + 2y cannot be greater than 48, but if x = 9, then 7x 60, so x must be 8. If y were zero, then four pizzas would su ce, so y must be 1. Thus there are eight hungry" people and one non hungry" person. Also solved by MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; KEITH EKBLAW, Walla Walla, Washington, USA; ROBERT GERETSCHLAGER, Bundesrealgymnasium, Graz, Austria; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; ROBERT B. ISRAEL, University of British Columbia, Vancouver, BC; EDWARD J. KOSLOWSKA and ROSE MARIE SAENZ, students, Angelo State University, San Angelo, Texas, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; SEAN MCILROY, student, University of British Columbia, Vancouver, BC; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; JOEL SCHLOSBERG, student, Hunter College High School, New York NY, USA; HEINZJURGEN SEIFFERT, Berlin, Germany; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; and the proposer. Also, one correct but anonymous solution was sent in. Two other readers misinterpreted the problem.
2212. 1997: 48 Proposed by Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. Let S = f1; 2; : : : ; ng where n 3. a In how many ways can three integers x, y , z not necessarily distinct be chosen from S such that x + y = z ? Note that x + y = z and y + x = z are considered to be the same solution. b What is the answer to a if x, y , z must be distinct?
Solution by William Moser, McGill University, Montreal, Quebec. The answers to a and b are the cardinalities, s and t, of the sets: S = fx; y j 1 x y; x + y ng and where x and y denote integers. Note rst that the set
2
T = fx; y j 1 x y; x + y ng;
U = fx; y j 1 x; 1 y; x + y ng , has cardinality u = n . Since the set V = fx; y j 1 x = y; x + y ng
115
has cardinality v = n , it follows that the set 2
W = fx; y j 1 x; 1 y; x 6= y; x + y ng = U , V , has cardinality w = u , v = n , n . 2 2 ,, From t = w and s = t + v , we then obtain that s = 1 n + n and 2 2 2 2 ,,n , n . t=1
2 2 2
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol,
UK; MIGUEL ANGEL CABEZON OCHOA, Logro~ o, Spain; GORAN CONAR, n student, Gymnasium Vara din, Vara din, Croatia; KEITH EKBLAW, Walla z z Walla, Washington, USA; IAN JUNE L. GARCES, Manila, The Philippines, and GIOVANNI MAZZARELLO, Firenze, Italy; ROBERT GERETSCHLAGER, Bundesrealgymnasium, Graz, Austria; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; WALTHER
JANOUS, Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; WILLIAM MOSER, McGill University, Montreal, Quebec a second solution; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; HEINZJURGEN SEIFFERT, Berlin, Germany; ZUN SHAN, Nanjing Normal University, Nanjing, China; DIGBY SMITH, Mount Royal College, Calgary, Alberta; and the proposer. One incorrect solution was sent in. As expected, the answers can be given in many di erent forms. If we let f n and g n denote the numbers of ways for a and b, respectively, then the most common answers are:
f 2n = n2; f 2n +1 = nn +1; g2n = nn , 1 and g2n +1 = n2;
which, as pointed out by many solvers, are equivalent to the following formulae:
f n =
n2 if n is even n24 1 if n is odd ; , 4
gn =
nn,2 if n is even 4 n,12 if n is odd 4
:
Janous and Sei ert observed that g n = f n, 1 for n 4 and Lambrou , 1 remarked that f n = 1 n2 , 2 1 + ,1n . 4 Besides the solution given above which was also obtained by Shan, there is a variety of other interesting single expression" formulae involving the oor and or the ceiling functions. These include: 1. f n =
n2
Garces and Mazzarello; Geretschlager; and the proposer,
2 ; gn = n , 1 = nn4, 2 4 4
116 2. f n = 3. f n =
n n + 1
2 2 2
Perz; and Sei ert,
; gn = n , 1 n 2 2
n n
n n
n n
2 4. f n = 2 n , 2 ; gn = n , 1 2 , 2
Cabezon.
Hsia,
n n n 2 ; gn = 2 2 , 2
2213. 1997: 48 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. A generalization of problem 2095 1995: 344, 1996: 373 . Suppose that the function f u has a second derivative in the interval a; b, and that f u 0 for all u 2 a; b. Prove that
1. y , z f x + z , xf y + x , y f z
0 for all u 2 a; b; 2. y , z f x + z , xf y + x , y f z = 0 for all x; y;z 2 a; b,
if and only if f 00 u
z y x z y x
0 for all x; y;z 2 a; b,
if and only if f u is a linear function on a; b. Solution by Robert B. Israel, University of British Columbia, Vancouver, BC. Part 1 is incorrect. The rst statement is actually equivalent to f is strictly convex", not f 00 u 0 for all u 2 a; b". Strictly convex means that f tx + 1 , ty tf x + 1 , tf y for x; y 2 a; b and 0 t 1. If f is twice di erentiable this is equivalent to f 00 u 0 everywhere with f 00 u 0 on a dense set. For example, f x = x4 is strictly convex everywhere but f 00 0 = 0, and in this case
y , zx4 + z , xy 4 + x , yz 4 2 2 2
= x , zy , zx , y x + y + y + z + x + z 2 0
if z
y x.
117 More generally, let f be any continuous function on a; b, and n 3 an integer. Given x1 ; x2 ; : : : ; xn let
Lf =
n X
where we take x0 = xn and xn+1 = x1 . Then the following are equivalent: i Lf 0 whenever a x1 x2 xn b. ii f is strictly convex on a; b. Similarly, for part 2, we will show that f is linear on a; b if and only if Lf = 0 whenever a x1 x2 xn b. Proof of i = ii: If f is not strictly convex, then either it is not convex at all or its graph includes a straight line segment. In the rst case, there exist x, z , and t with a z x b and 0 t 1 such that
i=1
xi,1 , xi+1 f xi;
f tz + 1 , tx tf z + 1 , tf x: Taking y = tz + 1 , tx and noting that t = x , y =x , z we can rewrite this as f y x , y f z + y , z f x x,z x,z or y , z f x + z , xf y + x , y f z 0. Let x1 = z , x2 = y , and x3 = = xn = x, and we have Lf 0 if n 3, note that the terms in f xj for 3 j n are 0 and the terms in f x3 and f xn add to y , xf x. If n 3 we must move x3 ; : : : ; xn slightly so that x3 xn, but by continuity the inequality will still be true if the changes are small enough. In the second case, we can take any x1 xn in the interval over which the graph is a straight line f x = cx + d, and then X ! X n n n n X X ! Lf = c xi,1xi , xixi+1 + d xi,1 , xi+1 = 0
i=1 i=1 i=1 i=1
Proof of ii = i: As noted above, if g is linear on a; b then Lg = 0. Thus we can add a linear function g to f without changing either i or ii. With an appropriate choice of g we can make f x1 = f xn = 0, and by strict convexity we then have f xj 0 for 1 j n. Now write
Lf =
=
n X
i=1 n X i=1
xi,1 , xif xi +
n X
xi,1 , xif xi + f xi,1
i=1
xi , xi+1 f xi
118 Note that the i = 1 term is 0, and each of the other terms is strictly positive, so the sum is strictly positive. As noted above, if f is linear on a; b then Lf = 0. Conversely, suppose Lf = 0 whenever a x1 x2 xn b. By part 1, Lh 0 where hx = x2, and so since L is a linear operator Lf + ch = cLh 0 for any c 0. Again using part 1, f + ch is strictly convex, and so f = clim f + ch is convex on a; b. By the same argument, ,f is convex, 7!0 so f is concave on a; b. Any function that is both convex and concave on a; b is linear on a; b. Also solved by FLORIAN HERZIG, student, Perchtoldsdorf, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece, part 1 only. There were seven incorrect and two incomplete solutions. All of the incorrect solutions failed to notice the aw in the statement of the problem, and then proceeded to treat as equivalent the two statements: f is strictly convex on a; b" and f 00 u 0 for all u 2 a; b". Several solvers pointed out that the condition f u 0 for all u 2 a; b is unnecessary. Herzig remarks: If the interval considered is the set of real numbers instead, then it is su cient to assume y = x+z and f is a continuousfunction 2 for part 2. The given equation reduces to Jensen's equation
+
x + y
f x + f y = f
2 2
that can be solved using elementary methods tricky substitutionand Cauchy's method for solving functional equations. Lambrou lists a series of six equivalent statements for part 1, where he assumes the existence of derivatives only when needed: 1. y , z f x + z , xf y + x , y f z 0 for all x; y;z 2 a; b with z y x. 2. The rst di erence quotient f uu,f v u 6= v is strictly increasing in ,v each variable separately. 3. The second di erence quotient
f u , f w f v , f w
, u , v u,w , v,w u 6= w, v 6= w, u = v is strictly positive for all u; v;w 2 a; b. 6 4. f 0 is strictly increasing on a; b. 5. f is strictly convex on a; b.
119 6. f 00 x 0 for all x 2 a; b and there is no interval I a; b such that f 00 x = 0 for all x 2 I . In the language of topology, this says that f 00 x 0 on a dense subset of a; b.
2214. 1997: 109 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Let n 2 be a natural number. Show that there exists a constant C = C n such that for all real x1; : : : ; xn 0 we have
k=1
vn uY n Xp u xk t xk + C :
k=1
Determine the minimum C n for some values of n. For example, C 2 = 1. Solution by Michael Lambrou, University of Crete, Crete, Greece. We show that the inequality is valid for an aggregate of values of C of which the least is Let us rst do the easier task of proving the existence of C 's which make the inequality valid. Of course this part will be redundant as soon as we improve the technique to nd the least C . Setting xi = yi2 where yi 0 i = 1; : : : ; n, we are to show, equivalently, that for some C we have
n C n = n,p,n1,2 ; n 2: n
1
X !2 Y n n yi yi2 + C :
i=1 i=1
1
Treating the right hand side of 1 as a polynomial in C , we observe that all P coe cients are nonnegative and that the coe cient of C n,1 is yi2. Thus
n Y i=1
X ! n yi + C yi2 C n,1:
2
i=1
But by the Cauchy Schwarz inequality we have
i=1 i=1
X !2 X ! n n yi n yi2 ;
so inequality 1 will be valid if we choose C = n1=n,1 or larger. This completes the easier task.
120 It turns out that n1=n,1 is only a slight overestimate of the minimum C , which we now seek. For any C for which 1 is valid, set wi = yipn , 1=pC , so that 1 becomes
X !2 n n n,1 Y wi n C 1n,1 wi2 + n , 1 ,
i=1 i=1
or equivalently
X !2 n n C n,1 n Y wi2 , 1 + 1
: wi n , 1n ,1 n n
i=1 i=1
2
To nd the minimum C we shall rst show that the following inequality is valid:
X !2 n n Y wi2 , 1
wi n2 n +1 :
i=1 i=1 m Y i=1
3
We shall use the Weierstrass inequality
1 + ai 1 +
m X i=1
ai;
which holds if all ai 0 or if ,1 ai 0 for all i for example, see item 3.2.37, page 210 of D. S. Mitrinovi
, Analytic Inequalities . Without c loss of generality let w1; : : : ; wt 1 and 0 wt+1 ; : : : ; wn 1, where t 2 f0; 1; : : : ; ng. Then
n t n Y wi2 , 1
Y wi2 , 1
Y wi2 , 1
+1 = +1 n n n +1 i=1 i=1 i=t+1 X 2 !0 X 2 1 t n w ,1 , i A 1 + wi n 1 @1 + n i=1 i=t+1 !0 X 1 t n X 1 = n2 n , t + wi2 @t + wi2A i=1 0t 1 0 it=t+1 n 1 n X 2A @X 2 X 2 A 1 X = n2 @ wi2 + 1 1 + wi i=1 i=t+1 i=1 i=t+1 n ! 1 Xw 2 n2 i i=1
the last inequality by the Cauchy Schwarz inequality, which proves 3. Note that equality occurs for w1 = = wn = 1. We conclude that 2
121 is valid for any C with C n,1 nn =n , 1n,1 n2 , i.e., with
1
n C n,p,n1,2 ; n 2: n The minimum value C n we seek is then as stated at the beginning, since for wi
2 C xi = yi2 = C pn , 1 = n , 1 1
the original inequality reduces to equality. Remark. The above shows C 2 = 1,
3
2 C 3 = p3 1:1547; C 4 = p3 2 1:1905; C 5 = p4 3 1:1963;
and generally C n n1=n,1 which approaches 1 in the limit. The minimum C n for all n 2 was also found by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; KEEWAI LAU, Hong Kong; and HEINZJURGEN SEIFFERT, Berlin, Germany. The existence of C n was proved by FLORIAN HERZIG, student, Perchtoldsdorf, Austria; JOE
HOWARD, New Mexico Highlands University, Las Vegas, NM, USA; VACLAV Y, Ferris State University, Big Rapids, Michigan, USA; YEO KENG
KONECN HEE, student, Hwa Chong Junior College, Singapore; and the proposer. Of those readers who found the minimum C n, only Lambrou managed to do it without multivariable calculus. He also sent in a second solution, using calculus. Herzig found the minimum C n for n = 2 and 3, and Howard found it for n = 2. Kone n
conjectured the correct minimum value of C n for c y all n 2, but proved it only for n = 2 and 3. The proposer showed more generally that, for all real 0 and integers n 2, there is a constant C = C n such that
n X i=1
4
4
5
xi
Y n
for all real x1 ; : : : ; xn 0.
i=1
xi + C
!
1997: 109 Proposed by Theodore Chronis, student, Aristotle University of Thessaloniki, Greece. Let P be a point inside a triangle ABC . It is known how to determine P such that PA + PB + PC is a minimum known as Fermat's Problem for Torricelli. Determine P such that PA + PB + PC is a maximum.
2215?.
122 We received two types of solutions to this problem: I. P is considered to be an interior point of triangle ABC this is the problem as it was posed; II. P is considered to be inside or on the boundary of triangle ABC . Solution by Michael Lambrou, University of Crete, Greece, slightly modi ed by the editor . I Let P be an internal point. We show the existence of another internal point Q such that QA + QB + QC PA + PB + PC . Consider the ellipse E through P with foci B and C and l the tangent to this ellipse at P . Then the circle centred at A with radius AP meets l at P . Therefore, the points on l to at least one side of P are outside the circle. Take Q to be a point on l that is outside the circle hence QA PA but close to P so that it is inside triangle ABC . Since Q is on l, Q is outside E and QB + QC PB + PC . Adding we get the required result. Therefore, PA + PB + PC has no maximum. II Assume that AC AB BC and P is an interior or boundary point of triangle ABC; P 6= A. Let the line parallel to BC through P meet AB at D and AC at E. By similarity AE AD DE and AE AP . Hence
PA + PB + PC
=
=
=
PA + BD + DP + PE + EC PA + DE + BD + EC PA + AD + BD + EC PA + AB + EC AE + AB + EC AC + AB
Seimiya and Tsaoussoglou both noted that this inequality, for an interior point P , is due to Visschers 1902 see F.G.M. Exercises de G
om
trie, e e p. 228, Th. 25 IV . However, if P = A then PA + PB + PC = AB + AC . Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK I and II; JORDI DOU, Barcelona, Spain I and II; RICHARD I. HESS, Rancho Palos Verdes, California, USA I and II; PETER HURTHIG, Columbia College, Burnaby, BC II; ROBERT B. ISRAEL, University of British Columbia,
Vancouver, BC II; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA II; GERRY LEVERSHA, St Paul's School, London, England
I; ISTVAN REIMAN, Budapest, Hungary I; TOSHIO SEIMIYA, Kawasaki, Japan II; PANOS E. TSAOUSSOGLOU, Athens, Greece II.
123
2216. 1997: 110 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Suppose that 1 is a natural number. 1. Determine the set of all 's such that the diophantine equation x + y2 = z2 has in nitely many solutions. 2.? For any such , determine all solutions of this equation.
Solution by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We show that x + y 2 = z 2 has in nitely many solutions for all natural numbers by nding all solutions for any . Suppose rst that x + y 2 = z 2. Let x = p1 p2 : : : pk k 1 where p1 p2 pk are primes and i 2 N for i = 1; 2; : : : ; k. Then k k Y Y x + y2 = z2 is equivalent to z + yz , y = pi i and so z + y = pi i
1 2
and z , y =
k Y
i=1
pi i where i, i are nonnegative integers satisfying
i+ i= i k Y i=1
i=1
i=1
for i = 1; 2; : : : ; k and
2 3
pi
i
k Y i=1
pi i :
k Y
It follows that
y =
pi i ; =1 i=1 ! iY k k Y z = 1 pi i + pi i : 2
1 2
Y k
pi i ,
!
4 5
Conversely, if p1 6= 2 and x, y , z are given by 1, 4, 5 where i 2 N and i , i are nonnegative integers satisfying 2 and 3 for i = 1; 2; : : : ; k, then it is easy to see that x + y 2 = z 2 holds. If p1 = 2, then 1 and 1 must both be positive integers, since x even implies that z + y and z , y are both even. In this case, 1 1 and 1 1 and so 1 2 which is true for all 2, while if = 1 we must assume 1 2. Note that condition 3 could be replaced by the condition that 6 i 6= i
i=1
i=1
124 for at least one i if we replace 4 by
k Y k Y i=1
y=1 2
In summary, if p1 6= 2 then all solutions to x + y 2 = z 2 are given by 1, 5, and 7 where i 2 N and i , i are nonnegative integers satisfying 2 and 6. If p1 = 2 we must impose the condition that 1 1, 1 1. Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; YEO KENG HEE, Hwa Chong Junior College, Singapore; and the proposer. Four solvers solved part 1 correctly: FLORIAN HERZIG, student,
Perchtoldsdorf, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; and GERRY LEVERSHA, St Paul's School, London, England. There was one incorrect solution, and one incomplete solution.
i=1
pi i ,
pi i :
7
2217. 1997: 110 Proposed by Bill Sands, University of Calgary, Calgary, Alberta. a Prove that for every su ciently large positive integer n, there are arithmetic progressions a1 ; a2 ; a3 and b1; b2; b3 of positive integers such that n = a1 b1 + a2 b2 + a3 b3. b What happens if we require a1 = b1 = 1? This is a variation of problem 3 of the 1995 96 Alberta High School Mathematics Competition, Part II, which may appear in a future Skoliad Corner. I. Solution to part a by Joel Schlosberg, student, Bayside, New York, USA. Let the progressions be 2; 3; 4 and a; a + b; a + 2b. We have to show that
n = 2a + 3a + b + 4a + 2b = 9a + 11b for su ciently large n. There is a theorem of Euclid that any su ciently large number can be represented in the form ka + lb for some positive integers a and b if k and l are relatively prime. Since gcd9; 11 = 1, the answer
follows immediately. Editorial note. Coincidentally, in a recent issue of Mathematics and Informatics Quarterly Vol. 7, No. 4, 1997, p. 185, Crux with Mayhem regular K. R. S. Sastry gives a proof of the above theorem of Euclid", in fact showing more generally the known result that, if k and l are relatively prime, every integer greater than kl can be expressed as ka + lb for some positive integers a and b. Thus Schlosberg's proof shows that the statement of the problem holds for every integer n 100. Actually, we can allow b = 0 in this problem, which it can easily be checked lowers the bound to n 89. For more, see the editorial remarks below.
125
II. Solution to part b by Christopher J. Bradley, Clifton College, Bristol, UK. Bradley rst solved part a.  Ed. If one requires a1 = 1 and b1 = 1 then only those n are representable for which nonnegative x; y exist such that 1 + 1 + x1 + y + 1 + 2x1 + 2y = n; 1 that is, n = 3 + 3x + 3y + 5xy or
5n , 6 = 5x + 35y + 3: Since neither 5x + 3 nor 5y + 3 can equal 1 this implies 5n , 6 has to be composite as a necessary condition. When 5n , 6 is prime, n = 5 or n = 17 for example, no x and y can possibly be found. Now, by Dirichlet's Theorem, since 5 and 6 are coprime there are inde nitely large primes of the form 5n , 6. It follows that there are in nitely many n which cannot be
expressed in the form 1. III. Editorial remarks. Here are extensions of this problem observed by various solvers. For the complete list of solvers, see below. Shan and Wang calculated that the largest integer n not expressible as in part a is n = 17. Some other solvers gave larger impossible" values for n because they did not allow constant arithmetic progressions. Israel proved part a under the stronger condition that a1 is any xed positive integer. He also strengthened part b by showing that a fails whenever a1 and b1 are xed positive integers. Lambrou generalized a in another direction, namely he found all positive integers m so that every su ciently large integer n can be written in the form
n = a1b1 + a2b2 + + am bm where a1 ; : : : ; am and b1; : : : ; bm are mterm arithmetic progressions of positive integers. His answer: m = 1; 2; 3 or 6.
Most solvers noted that one of the arithmetic progressions in part a, say a1 ; a2 ; a3 , can be xed while the other varies, and the result is still true; for example see solution I, which uses a1 ; a2 ; a3 = 2; 3; 4. Lambrou also investigated exactly which xed a1 ; a2 ; a3 allow this. Letting a1 ; a2 ; a3 = a , h; a; a + h where a h 0 are integers, he proved that every su ciently large positive integer n can be expressed as n = a1 b1 + a2 b2 + a3 b3 for some positive integer arithmetic progression b1 ; b2; b3 if and only if 3a and 2h are relatively prime. As a variation on part b, Leversha noted that if we impose a1 = 1 = b3, then no integer of the form p , 6, where p is prime, can be written as in a.
126 Readers may like to verify the above results for themselves! And there are still many related problems that await investigation. For example, some readers' solutions to a use arithmetic progressions in which the common di erences are bounded in absolute value. Bradley's solution seems to be the best in this respect, with his common di erences always at most 7. Can this number be lowered? That is, can every su ciently large positive integer be written as a1 b1 +a2 b2 +a3 b3, where a1 ; a2 ; a3 and b1; b2; b3 are arithmetic progressions of positive integers with ja1 , a2 j and jb1 , b2j both at most 6, say? Or what if the common di erences of the arithmetic progressions in a are required to be bounded below by some constant? Finally, in b, what if we only require a1 = b1? Both parts also solved by FLORIAN HERZIG, student, Perchtoldsdorf, Austria; ROBERT B. ISRAEL, University of British Columbia, Vancouver, BC; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; and the proposer. Part a only solved by DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut,
USA; RICHARD I. HESS, Rancho Palos Verdes, California, USA; VACLAV
KONECNY, Ferris State University, Big Rapids, Michigan, USA; and YEO KENG HEE, student, Hwa Chong Junior College, Singapore. All proofs of part b were similar, though Lambrou observed that the full statement of Dirichlet's Theorem is not needed, only that there are innitely many primes of the form 10k , 1 as can be seen from Solution II above, and that this special case is a little easier to prove: for example, see T. Nagell's Introduction to Number Theory, Chapter 5, x50. Is there a solution to b that does not need Dirichlet's Theorem at all? The original Alberta High School Mathematics Competition problem was to show that every integer n can be written in the form a1 b1 + a2 b2 + a3 b3, where a1; a2; a3 and b1; b2; b3 are arbitrary arithmetic progressions of integers.
2218. 1997: 110 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Suppose that a, b, c are positive real numbers and that
abc = a + b , cb + c , ac + a , b:
Clearly a = b = c is a solution. Determine all others. I Solution by Goran Conar, student, Gymnasium Vara din, Vara din, z z Croatia. If any of the three quantities, a + b , c, b + c , a and c + a , b is negative, say a + b , c 0, then we have c a + b, so that b + c , a 0,
127 and c + a , b 0. This implies that abc 0, a contradiction. Hence a + b , c 0, b + c , a 0 and c + a , b 0. Note that
a + b , ca , b + c = a2 , b , c2 a2; b + c , ab , c + a = b2 , c , a2 b2; c + a , bc , a + b = c2 , a , b2 c2: a + b , cb + c , ac + a , b abc:
1 2 3
Multiply the three inequalities and take nonnegative square roots. We get 4 Since equality holds in 4, it must also hold in 1, 2 and 3. Therefore, a = b = c is the only solution. II Solution by Robert B. Israel, University of British Columbia, Vancouver, BC. Since the given equation is invariant under any permutation of a, b, c, we may assume that 0 a b c. The equation can be rewritten as
b + c , ab , c2 + ab , cc , a = 0:
Since both terms on the left are nonnegative, the only way to obtain equality is with a = b = c. Hence there are no other solutions in positive reals.
Also solved by SEFKET ARSLANAGIC, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; JOE HOWARD, New Mexico Highlands University, Las Vegas, NM, USA; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; WALTHER JANOUS, Ursulinengymnasium,
Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St Paul's School, London, England; CAN ANH MINH, University of California, Berkeley, CA, USA;
BOB PRIELIPP, University of Wisconsin Oshkosh, Wisconsin, USA; ISTVAN
REIMAN, Budapest, Hungary; JUANBOSCO ROMERO MARQUEZ, Universidad de Valladolid, Valladolid, Spain; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; DIGBY SMITH, Mount Royal College, Calgary, Alberta; YEO KENG HEE, Hwa Chong Junior College, Singapore; and the proposer. There was also one incorrect solution.
128 A few solvers remarked that if a, b and c are nonnegative, then we have the additional solutions: a; b; c = 0; t; t; t; 0; t; t; t; 0 where t 0 is arbitrary. Howard pointed out that a + b , cb + c , ac + a , b abc is problem 3 of the 1981 British Mathematical Olympiad, and referred to Rabinowitz's Index to Mathematical Problems, 1980 1984, p. 49. Ed. Actually, it is much older" than that and dates back to at least 1925. See, for example, x 1.3 on p. 12 of Geometric Inequalities by O. Bottema et al. Many solvers, after establishing the fact that a, b, c are the sides of a triangle, say 4, showed that the given equality is equivalent to the fact that 2r = R, where r and R denote the inradius and circumradius of 4 respectively. Then the conclusion follows from a celebrated theorem of Euler, which states that 2r R, with equality if and only if 4 is equilateral. Ed. This can be found, for example, in x 5.1 on p. 48 of the book mentioned in the previous paragraph. 1997: 47 Proposed by Joaqu
n G omez Rey, IES Luis Bu~ uel,
n Alcorc
Madrid, Spain. on, Given a0 = 1, the sequence fan g n = 1; 2; : : : is given recursively by
2210?.
n
n
n
an , n , 1 an,1 + n , 2 an,2 , : : : bnc ab n c = 0: n n 2
2
Which terms have value 0? No solutions have been received to date. The problem remains open.
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
129
THE ACADEMY CORNER
No. 18 Bruce Shawyer
All communications about this column should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7
Memorial University Undergraduate Mathematics Competition
September 25, 1997
We present in this issue, two solutions from Bob Prielipp, University of Wisconsin Oshkosh, Wisconsin, USA, to the above competition, which was printed in Academy Corner No. 15 1997: 449 . Determine the radius and height of the cylinder given that the radius and height are both integers. Solution. Let r be the radius and h be the height of the closed cylinder, where r and h are both positive integers. If the surface area of the closed cylinder is twice the volume, then It follows that h + r = rh. Thus
2. The surface area of a closed cylinder is twice the volume.
2rh + 2r = 2r h:
2 2
r , 1h , 1 = rh , r , h + 1 = 0 + 1 = 1: Hence r , 1 = 1 and h , 1 = 1 since r and h are both positive integers, so that r = 2 and h = 2. It is easily checked that if r = 2 and h = 2, then the surface area of
the closed cylinder is twice the volume.
130
3. Prove that
Solution. Slightly shortened by the editor. We use mathematical induction to prove, for each positive integer n, The result clearly holds for n = 1. Assume that the result holds for n = k. Then
2 2
1 1 1+ 4 + 1 +::: + n 9
2
2:
1 1 + 1 + 1 + ::: + n 4 9
2
1 2, n :
1 1 1 + 1 + 9 + : : : + k1 + k + 1 4 1
1 2 , k + k + 1 , +k+ = 2 , k + 1 1 k = 2 , k k + 1 1 kk + k kk + 1 = 2 , 1 : 2 , kk + 1 k+1
2 2 2 2 2 2
Hence, by induction, the result holds. Equality holds if and only if n = 1.
Advance Notice
At the summer 1999 meeting of the Canadian Mathematical Society, to be held in St. John's, Newfoundland, there will be a Mathematics Education Session on the topic What Mathematics Competitions do for Mathematics". Invited speakers include Edward Barbeau, Toronto; Tony Gardner, Birmingham, England; Ron Dunkley, Waterloo; and Rita Janes, St. John's. Anyone interested in givinga paper at this session should contact one of the organizers, Bruce Shawyer or Ed Williams, at the Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. email addresses:
bshawyer@math.mun.ca ewilliam@math.mun.ca
131
THE OLYMPIAD CORNER
No. 189 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. We begin this issue with an exam set sent in by one of our newer correspondents. My thanks go to Enrique Valeriano, National University of Engineering, Lima, Peru.
THE XII IBEROAMERICAN OLYMPIAD August 1997  3.5 hours
PERU'S SELECTION TEST FOR
0 0 1 2
1. Given an integer a follows:
+1 +1 2
Prove that there is a nonnegative integer p such that ap ap ap . 2. A positive integer is called almosttriangular" if the number is itself triangular or is the sum of di erent triangular numbers. How many almosttriangular numbers are there in the set f1; 2; 3; : : : ; 1997g? Note: The triangular numbers are a ; a ; a ; : : : ; ak ; : : : ; where a = 1, and ak = k + ak, , for all k 2. 3. An n n chessboard n 2 is numbered with n nonzero numbers. This chessboard is called an Incaican Board" if, for each square the number written on the square is the di erence between two of the numbers written on two of the neighbouring squares sharing a common edge. For which values of n can one obtain Incaican Boards? 4. Let ABC be a given acute triangle. Give a ruler and compass construction of an equilateral triangle DEF with D on BC , E on AC , and F on AB such that the perpendiculars to BC at D, to AC at E , and to AB at F , respectively, are concurrent.
+1 +2 1 2 3 1 1 2
2, the sequence a ; a ; a ; : : : is de ned as ak = ak1 + ak ; if ak is an odd number; ak = ak ; if ak is an even number:
Next we give the Third and Fourth Grade problems of the 38 Mathematics Competition for Secondary School Students of the Republic of Slovenia. My thanks go to Richard Nowakowski, Canadian Team Leader to the IMO in Hong Kong, for collecting this contest and forwarding it to me.
th
132
38
Mathematics Competition for Secondary School Students April, 1994
REPUBLIC OF SLOVENIA th
Third Grade
3 2
1. Let n be a natural number. Prove: if 2n + 1 and 3n + 1 are perfect squares, then n is divisible by 40. 2. Show that cossin x sincos x holds for every real number x. 3. The polynomial px = x + ax + bx + c has only real roots. Show
3 2 2
that the polynomial q x = x , bx + acx , c has at least one nonnegative root. 4. Let the point D on the hypotenuse AC of the right triangle ABC be such that jAB j = jCDj. Prove that the bisector of BAC , the median through B , and the altitude through D, of the triangle ABD have a common point.
1. Prove that there does not exist a function f : Z ! Z, for which f f x = x + 1 for every x 2 Z. 2. Put a natural number in every empty eld of the table so that you
get an arithmetic sequence in every row and every column.
74 186 103 0
Fourth Grade
3. Prove that every number of the sequence
is a perfect square in every number there are n fours, n , 1 eights and a nine. 4. Let Q be the midpoint of the side AB of an inscribed quadrilateral ABCD and S the intersection of its diagonals. Denote by P and R the orthogonal projections of S on AD and BC respectively. Prove that jPQj = jQRj.
49; 4489; 444889; 44448889; : : :
133 As a nal contest for your puzzling pleasure in this number, we give the VIII Nordic Mathematical Contest. Again my thanks go to Richard Nowakowski, Canadian Team leader to the IMO in Hong Kong, for collecting this contest and forwarding it to me.
VIII NORDIC MATHEMATICAL CONTEST th
March 17 , 1994 Time: 4 hours
1. Let O be a point in the interior of an equilateral triangle ABC with side length a. The lines AO, BO and CO intersect the sides of the triangle at the points A , B and C respectively. Prove that
1 1
2. A nite set S of points in the plane with integer coordinates is called a twoneighbour set, if for each p; q in S exactly two of the points p +1;q , p; q + 1, p , 1; q, p; q , 1 are in S . For which n does there exist a twoneighbour set which contains exactly n points? 3. A square piece of paper ABCD is folded by placing the corner D at some point D0 on BC see gure. Suppose AD is carried into A0 D0 , crossing AB at E . Prove that the perimeter of triangle EBD0 is half as long as the perimeter of the square. A E A B
0
jOA j + jOB j + jOC j a:
1 1 1
1
D
a perfect square.
0
4. Determine all positive integers n
D
200 such that n + n + 1 is
2 2
C
Turning now to comments and solutions related to the February 1997 number of the corner, we welcome two alternate solutions to problems of the Sixth Irish Mathematical Olympiad sent in by another new contributor whom we also welcome. 3. 1995: 151 152; 1997: 9 13 Sixth Irish MathematicalOlympiad. , For nonnegative integers n, r the binomial coe cient n denotes the r number of combinations of n objects chosen r at a time, with the convention ,n ,n that = 1 and r = 0 if n r. Prove the identity
0
1 X
d=1
n , r + 1
r , 1
= n
d d,1 r
134 for all integers n, r with 1 r n. Alternate Solution by Mohammed Aassila, UFR de Math
matique et e d'Informatique, L'Universit
Louis Pasteur, Strasbourg, France. e We have
1 + xn = 1 + xn,r 1 + xr, 2 3 "m,r r, X n , r + 1
X r , 1
= xi 4 xj 5
+1 1 +1 1
=
The coe cient of xr is
r X
i i=0 j =0 m,r+1 r,1 n , r + 1
r , 1
X X
i=0 j =0
j
i
j
xi
+
j
d=0
r n , r + 1
r , 1
= X n , r + 1
r , 1
d r,d d d d,1
=1
1995:151 152; 1997: 9 13 Sixth Irish Mathematical Olympiad. Let x be a real number with 0 x . Prove that, for all natural numbers n, the sum is positive. Alternate solution by Mohammed Aassila, UFR de Math
matique et e d'Informatique, L'Universit
Louis Pasteur, Strasbourg, France. e We know that
4.
, sin x + sin33x + sin55x + + sin2n, 11x 2n
2 sin x sin2k , 1x = cos2k , 2x , cos 2kx:
Hence
, 2 sin x sin x + sin3x + + sin2n, 11x 3 2n 2x = 1 , cos 2x + cos 2x , cos 4x + + cos2n , n , , cos 2nx 3 2 1
1
= 1 , cos 2x 1 , 1 , cos 4x 3 , 1 , , cos 2nx 3 5 2n , 1 1
+ 1 , 1
+ + 1 = 0: 1, 1, 3 3 5 2n , 1
135 Next we turn to solutions to problems of the Latvian 44 Mathematical Olympiad given in the February number of the Corner last year 1997: 78 .
LATVIAN 44 MATHEMATICAL OLYMPIAD rd
Final Grade, 3 Round
Riga, 1994
Solutions by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; by Mansur Boase, student, St. Paul's School, London, England; by Panos E. Tsaoussoglou, Athens, Greece; and by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We give the solutionof Covas. More generally, we prove that sin mx + sin my = 0 where m is an integer.
Since sin mx + sin my = 2 sin m x y cos m x,y , it is su cient
to show that sin x y = 0 since sin m x y = 0 follows easily by induction from
+ 2 2 + 2 + 2
1. It is given that cos x = cos y and sin x = , sin y. Prove that sin 1994x + sin 1994y = 0.
Now,
sin m + 1 x + y = sin m x + y cos x + y 2
2 2
+ cos m x + y sin x + y : 2 2
and
cos x = cos y cos x , cos y = 0 ,2 sin x + y sin x , y = 0 2 2 sin x = , sin y sin x + sin y = 0 2 sin x + y cos x , y = 0 : 2 2
2 2 2
1
2
Squaring each of 1 and 2 and adding, we nd
4 sin x + y sin x , y + cos x , y = 0: 2  2 2 z
=1 + 2
Hence sin x y = 0. 3. It is given that a 0, b 0, c 0, a + b + c = abc. Prove that at least one of the numbers a, b, c exceeds 17=10.
136 Solutionsby Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; by HeinzJurgen Sei ert, Berlin, Germany; by Panos E. Tsaoussoglou, Athens, Greece; and by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We give two solutions of Shan and Wang. First Solution. We show, in general, that if xi 0 for i = 1; 2; : : : ; n, such that Pn Qn i xi = i xi , then maxfxi : i = 1; 2; : : : ; ng n = n, . In particular, when n = 3 we get
=1 =1 1 1
By the arithmeticgeometric mean inequality we have
p maxfx ; x ; x g 3 1:7 = 17 : 10
1 2 3
i=1 and thus i=1 xi nn=n,1.
Pn
=1
n ! n n 1 Xx n Y x = Xx ; i i i n i=1 i=1
1 x x : : : xi : : : xn = 1; ^ i where xi indicates that the factor xi is missing. Hence for some j we have ^
=1 1 2
Without loss of generality, we may assume that maxfxi : i = 1; 2; : : : ; ng = xn . Then nxn Pn xi nn= n, from which xn n = n, follows. i Second Solution without the AMGM inequality. P Q Suppose xi 0 Q= 1; 2; : : : ; n such that n xi = n xi . Then in i i dividing both sides by i xi we get
1 1 1 =1 =1
n X
=1
1 1 n or n x x : : : xj : : : xn: ^ x x : : : xj : : : xn ^
1 2 1 2 1 2 1 17 10
Without loss we may assume that xn = maxfxi : i = 1; 2; : : : ; ng. Then xn, x x : : : xj : : : xn n, from which xn n = n, follows. In ^ n p particular, for n = 3, we get x 3 1:7 = . 4. Solve the equation 1!+2!+3!+ + n! = m in natural numbers. Solution by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We solve the more general problem of nding all solutions of the Diophantine equation 1! + 2! + 3! + + n! = mk in natural numbers, n, m, and k. For convenience, let Sn = 1! + 2! + 3! + + n!. When k = 1 clearly m = Sn is the only solution for any n. When k = 2 we claim that the equation Sn = m has exactly two solutions: n = m = 1 and n = m = 3. Note rst that d! 0mod 10
1 1 3 3 2
137 for all d 5 and S = 1 + 2 + 6 + 24 = 33 3mod 10. Hence Sn 3mod 10 for all n 4. However, it is easy to see that the last digit of a perfect square can never be 3 and so there are no solutions if n 4. Checking the cases when n = 1; 2; 3 directly reveals that there are precisely two solutions, as given above. When k 3 we show that n = m = 1 is the only solution. If n 2 then clearly Sn k 0mod 3. But mk 0mod 3 implies m 0mod 3 and so m 0mod 27 as k 3. Since d! 0mod 27 for all d 9 and since
4
S = 1 + 2 + 6 + 24 + 120 + 720 + 5040 + 40320 = 46233 6= 0 mod 27 there are no solutions if n 8. On the other hand, direct checking shows that for n = 2; 3; 4; 5; 6; 7, Sn = 3; 9; 33; 153; 873, and 5913, none of which is a perfect k power for any k 3. Finally, it is trivial to see that n = m = 1
8 th
is a solution. Remark: The special case of this problem when k = 2 was proposed by E.T.H. Wang as Quicky Q657 in the Mathematics Magazine, 52 1979, p. 47. The general case was also proposed by him as problem 4203 in Mathmedia in Chinese 42, 1980; p. 64 with solution in 43, 1980, p. 49. This is a journal published by the Institute of Mathematics, Academia Sinica, Taipei, Taiwan. 5. There are 1994 employees in the o ce. Each of them knows 1600 others of them. Prove that we can nd 6 employees, each of them knowing all 5 others. Solution by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. Let E denote the set of these 1994 employees. For each x 2 E , let S x denote the set of all employees whom x does not know. Then by assumption, jSxj = 393 for all x 2 E. Let a and b be any two employees who know each other. Since
jSa Sbj 2 393 = 786 1992; 9c 2 E such that a, b, and c form a triple of mutual acquaintances. Since jSa Sb Scj 3 393 = 1179 1991; 9d 2 E such that a, b, c, and d form a quadruple of mutual acquaintances. Since jSa Sb Sc Sdj 4 393 = 1572 1990; 9e 2 E such that a, b, c, d, and e form a quintuple of mutual acquaintances.
Finally, since
jSa Sb Sc Sd Sej 5 393 = 1965 1989;
138
9f 2 E such that a, b, c, d, e, and f form a sextuple of mutual acquaintances. 1st SELECTION ROUND 1. It is given that x and y are positive integers and 3x + x = 4y + y. Prove that x , y , 3x + 3y + 1 and 4x + 4y + 1 are squares of integers.
2 2
Solutions by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; by Panos E. Tsaoussoglou, Athens, Greece; and by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We give the solution by Shan and Wang and their comment. Note rst that the given equation implies the following two equations:
x , y3x + 3y + 1 = y 1 x , y4x + 4y + 1 = x 2 1 2 yields x , y 3x + 3y + 14x + 4y + 1 = xy which implies that 3x + 3y + 14x + 4y + 1 is a perfect square. But clearly gcd3x +3y +1; 4x +4y +1 = 1 since 43x +3y +1 , 34x +4y +1 = 1. Therefore, 3x + 3y + 1 and 4x + 4y + 1 are both squares, which, together
2 2 2 2
with 1 or 2, implies that x , y is also a square. Comment: This would be a much better problem had it asked to show only that x , y is a square. This is an example of a case when asking to prove too many things actually gives the solution away, in some sense. Shan and Wang also proposed the followingproblem inspired by this one. The Diophantine equation 3x + x = 4y + y is satis ed when x = 30 and y = 26. a Find another solution in positive integers. b Are there in nitely many solutions in positive integers? Is so, describe all of them.
2 2
the expression
1 2 3
2nd SELECTION ROUND 1. It is given that 0 xi 1, i = 1; 2; : : : ; n. Find the maximum of
2
x n + x x x x: : x + 1 + + x x : : :xx + 1 : x x : : : xn + 1 : n n,
1 3 4 1 2 1
Solutions by Murray S. Klamkin, University of Alberta, Edmonton, Alberta; and by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We rst give the solution of Shan and Wang. Clearly n 2 for the question to make sense. Let Sn denote the given sum. We prove that Sn n , 1. For n = 2, equality holds if and only if x = 0, x = 1 or x = 1, x = 0 or x = x = 1. For n 3, equality holds if and only if one of the xi 's is 0 and the others are all equal to 1. We rst establish a lemma:
1 2 1 2 1 2
139
Lemma. Suppose that the xi 's are reals such that 0 xi 1 for all i = 1; 2; : : : ; n, where n 2. Then x + x + + xn n , 1+ x x xn with equality holding if and only if xi 6= 1 for at most one i, i = 1; 2; : : : ; n. Proof. For n = 2, x + x x x + 1 , 1 , x 1 , x 0, which is clearly true. Equality holds if and only if x = 1 or x = 1. Suppose the assertion holds for some n 2. Then x + x + + xn + xn n , 1 + x x xn + xn n , 1 + x x xn xn + 1 = n + x x xn : The 2 inequality is by the n = 2 case.
1 2 1 2 1 2 1 2 1 2 1 2 1 2 +1 1 1 2 2 +1 +1 1 2 +1 nd
If equality holds, then it must hold in both inequalities above. By the induction hypotheses, we then have i at most one of x ; x ; : : : ; xn is di erent from 1 and ii either xn = 1 or x x xn = 1. Since x x xn = 1 clearly implies x = x = = xn = 1, our assertion about the equality follows. Now we proceed to prove the claim about Sn. For n = 2,
1 2 +1 1 2 1 2 1 2 2 1 2 1 2 2 1 2 1 1 2
S = x x+ 1 + x x+ 1 x x x + x x x = 1: + + It is readily seen that equality holds if and only if x = 0, x = 1 or x = 1, x = 0, or x = x = 1. For n 3 we apply the lemma above and obtain +x Sn x x+xx + x + 1 n n , 1 + xxx +1 xn n , 1: n xx n If equality holds, then the 2 inequality implies xj = 1 for at most one j and 6 the 3 inequality implies n , 1x x xn = x x xn , which implies that xi = 0 for at least one i. Hence xi = 0 for some i and xj = 1 for all j = i. It is obvious that this condition is also su cient. This completes the 6
1 2 1 2 1 2 1 2 1 2 1 2 1 2 nd rd 1 2 1 2
proof. And here is Klamkin's somewhat more abbreviated solution with generalization. Since the expression is convex in each of the variables, the maximum value is achieved when each variable takes on 0 or 1. Clearly this occurs when one variable is 0 and the rest are 1 giving the maximum value of n , 1. The same maximum occurs if any of the numerators xi are replaced by xi i where i 1. A similar result, using convexity, that
X
xu + Y1 , x v 1; i i 1 + s , xi
140 where 0 xi 1, u; v 1, s = xi and the sum and product are over i = 1; 2; : : : ; n, is given in 1 . Reference: 1 M.S. Klamkin, USA Mathematical Olympiads 1972 1986, M.A.A., Washington D.C., 1988, p. 82. 3. A triangle ABC is given. From the vertex B, n rays are constructed intersecting the side AC . For each of the n +1 triangles obtained, an incircle with radius ri and excircle which touches the side AC with radius Ri is constructed. Prove that the expression
P
depends on neither n nor on which rays are constructed. Solution by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain. If A, B , C are the angles of a triangle, then and ra = s tan ; 2 where r, ra are the inradius and the radius of the excircle opposite A, and s is the semiperimeter. It follows that
r r : : : rn R R : : : Rn
1 2 1 2
+1 +1
r = s tan A tan B tan C 2 2 2
A
r = tan B tan C : ra 2 2 Next we apply this result to each of n + 1 triangles obtained see gure at
the top of the next page. This yields
r1 R1 r2 R2
= tan A tan 1 = tan , 1 tan
Multiplying these equalities, we observe that the product of all the right hand members is
180 , n,1 rn tan 2n Rn+1 = tan 180 , 2 n r C n Rn+1 = tan 2 tan 2 :
2
2 180
2
2
2
A tan tan
2
and we get
r r rn, A C R R Rn = tan 2 tan 2 which depends on neither n nor on which rays are constructed.
1 2 1 1 2 +1
2 cot 2 tan 2 cot 2 = tan A tan C ; 2 2
1 1 2 2
tan n cot n 2
C 2 tan 2
141
B
A 1
2
n
C
Let ABCD be an inscribed quadrilateral. Its diagonals intersect at O. Let the midpoints of AB and CD be U and V . Prove that the lines through O, U and V , perpendicular to AD, BD and AC respectively, are concurrent. Solution by Toshio Seimiya, Kawasaki, Japan. Case 1. Neither is AC orthogonal to BD, nor is AD a diameter.
B S H N Y V D U M X O A
3.
3rd
SELECTION ROUND
C
Let M , N be the feet of the perpendiculars from U , V to BD, AC respectively, and let S be the intersection of UM and V N . Let X , Y be the feet of the perpendiculars from A, D to BD, AC respectively, and let H be the intersection of AX and DY . Since U is the midpoint of AB, and UM kAX , M is the midpoint of BX . Similarly N is the midpoint of CY . Since AXD = AY D = 90 , A, X , Y , D are concyclic. Therefore Y XD = Y AD = CAD = CBD. Thus we have XY kBC . Since M , N are the midpoints of BX , CY respectively, we have MN kXY . Since SM kHX , XN kHY , and MN kXY , MX , NY and SH are concurrent at O. Therefore S , H , O are collinear. Since AH ? OD and DH ? OA, H is the orthocentre of OAD, so that HO ? AD. Thus we have SO ? AD. Thus the lines through O, U and V , perpendicular to AD, BD and AC respectively, are concurrent at S .
142
Case 2. AC is orthogonal to BD.
B M C V D O U A
Let M be the midpoint of BC . Since U is the midpoint of AB , we get UM ? AC , so that UM ? BD. Similarly we have V M ? AC . Since AC ? BD and M is the midpoint of BC , by Brahmagupta's Theorem we have MO ? AD. See: H.S.M. Coxeter and S.L. Greitzer, Geometry Revisited, p. 59. Thus the lines through O, U and V , perpendicular to AD, BD and AC respectively, are concurrent at M . Case 3. AD is a diameter.
S C V
q
B O
U
q
A
Let S be the intersection of AB and CD. Since AB ? BD and CD ? AC , S is the orthocentre of OAD. Thus SO ? AD. Hence the lines through O, U and V , perpendicular to AD, BD and AC respectively, are concurrent at S. That completes this number of the Olympiad Corner. Send me your nice solutions and suggestions for future issues.
D
143
BOOK REVIEWS
Edited by ANDY LIU
Vita Mathematica edited by Ronald Calinger, published by the MAA, Notes and Reports Series, 1996, ISBN 0883850974, softcover, 350+ pages, $34.95. Reviewed by Maria de Losada, Bogot
, Colombia. a This collection of papers read at the August 1992 Quadrennial Meeting of the International Study Group on the Relations between History and Pedagogy of Mathematics at the University of Toronto and the Seventh International Congress on Mathematical Education at Universit
Laval in Quebec, e edited and refereed, begins with a somewhat ponderous re ection on the different tendencies in research in the history of mathematics, contrasting the approaches of cultural and mathematical historians. Looking at the role of problems in the history and teaching of mathematics, Evelyne Barbin reminds us that one of the perverse e ects of education : : : is that answers are given to questions that have not been asked. The history of mathematics shows us that questions must come rst, and that it is through questions that we make sense of mathematical concepts." In the historical studies from antiquity to the Scienti c Revolution can be found intriguing material that is not of easy access. Swetz's paper on the enigmas of Chinese mathematics is informative as it strives to provide a balanced presentation and Katz's treatment of medieval Hebrew and Islamic mathematics provides useful and little known information in a concise and well ordered manner. Noteworthy among the more recent historical studies is Judith Grabiner's paper which contrasts historical perspectives of the calculus, the geometric McLaurin and algebraic Lagrange, linking these naturally with education and culture and suggesting that progress in mathematics is made by those who sharpen their thinking by exercising the courage of their sometimes idiosyncratic convictions". On an entirely di erent note, the detailed history given by Ronald Calinger of the University of Berlin Mathematics Seminar examines every facet of that paradigmatically successful structuring of a research tradition, in which many of the recurrent themes of this book, such as starting from problems and learning from the masters, were put into practice. The third section of the collection, devoted to the integration of history with mathematics teaching recounts many valuable experiences, two titles of note being Mathematical Masterpieces: Teaching with Original Sources and A History of Mathematics Course for Teachers, Based on Great Quotations.
144 Although they are presented as summaries and have catchy titles, these articles refer to teaching experiences that are both substantive and well structured. In the rst of these, Reinhard Laubenbacher and David Pengelley reveal not only a polished list of original sources that can be used with undergraduates, but also aspects of their pedagogical approach. In the second, Israel Kleiner begins with quotations addressing the question What is mathematics?" that are arranged in antagonistic" pairs to bring across the underlying message of mathematics as an activity whose history is susceptible to chronological, thematical, topical and biographical study, as long as sight is not lost of Lakatos' remark: The history of mathematics, lacking the guidance of philosophy is blind, while the philosophy of mathematics, turning its back on the most intriguing phenomena in the history of mathematics, is empty." It is unfortunate that the book has so many typos! Not only are footnote numbers routinely formatted incorrectly, but there are also plenty of errors in the text. In general, this is good formative reading for those with an appetite for historical material, but especially useful in its treatment of interrelations between history and pedagogy.
The Canadian Mathematical Society has initiated a new series of booklets of enrichment material for interested and mathematically talented high school students. La Soci
t
math
matique du Canada vient de lancer une s
rie de livrets ee e e d'enrichissement pour les el ves du secondaire int
ress
s et forts en math
e e e e matiques.
A Taste Of Mathematics Aime T On les Math
ematiques
See the enclosed yer for more details of the ATOM series. Voir le d
pliant ciinclus pour plus de renseignements sur la collection ATOM . e
145
Sum of powers of a nite sequence: a geometric approach
William O.J. Moser
In this note we give a geometriccombinatoric derivation of a formula for the sum of r powers of a nite positive integer sequence:
th
ar + ar + + ar = n
1 2
r n XX
`=1 i=1
ai
r; ` ; `
1
where
n n = k n,k ; if 0 k n , 0; if k n 1 , k and the numbers r;` are determined by the recurrence r; 1 = 1; r = 1; 2; 3; ; 1; ` = 0; ` = 2; 3; ; r; ` = `r , 1; ` , 1 + r , 1; `; r 2; ` 2 :
! ! !
2
The derivation parallels, simpli es and generalizes the proof given in 1 , and seems to be more elementary than proofs given in 2 and 3 . r For xed but arbitrary positive integers m; r, let Lm denote the rdimensional integer lattice points fx ; x ; ; xr j xi integers, 0 x ; x ; ; xr mg ; r and for w 1 let Sm w denote the set of rdimensional cubes" with faces r parallel to the coordinate planes, vertices in Lm and width w. r w = ; if w r w is identi ed by its Of course Sm m. A cube in Sm vertex = ; ; ; r closest to 0; ; 0. Clearly
1 2 1 2 1 2
w + m; w + m; ; w + r m;
1 2
1
; ; ;
2
r 0;
or
m , w; ; 0 r m , w; 1 w m ; 3 r so the number of cubes in Sm w is r r jSm wj = m , w + 1 ; if 1 w m ,. 0; if w m 1
1 2
0
m , w; 0
Copyright c 1998 Canadian Mathematical Society
146 For given 1 w m we can also nd the number of 's satisfying 3 as follows. Any rtuple satisfying 3 determines the sequence consisting of the distinct values, in increasing order, of the entries in :
1 2
:0 ` m , w; ` r : 4 Given a sequence satisfying 4 there are many rtuples satisfying 3 which have entries with values . How many? Let r;` denote this number. It depends on ` and not on the particular numbers ; ; ` and satis es the recurrence 2. For, given , there are ` choices namely ; ; ; ` for the rst entry of ; can then be completed in r , 1; ` ways if the integer chosen for appears among the entries ; ; ; r, and in r , 1; ` , 1 ways otherwise; 2 then follows. The numbers r; ` are exhibited in the display below:
1 1 2 1 1 2 3
r n` 1 2
1 2 3 4 5
. . .
1 1 1 1 1
. . .
0 2! 6 14 30
. . .
3 0 0 3! 36 150
. . .
4 0 0 0 4! 240
. . .
5 0 0 0 0 5!
. . . . . .
6 0 0 0 0 0
, Since there are m,w ` 4 we nd that
+1
ways of choosing the sequence satisfying
and setting a = m , w + 1 1
r r = X m , w + 1 r; `; m , w + 1 ` `=1
1 w m;
1 ` r;
a r;`; a 1; r 1 : ` ` Summing over a set of a's we have 1. Taking ai = a + i , 1d, i = 1; 2; ; n in 1, we have r n X X a + i , 1d
ar + a + dr + + a + n , 1dr = r; ` ; ` ` i the sum of the r powers of an arithmetic progression. When a = d = 1
r r n = 1r + 2r + + nr = X n + 1 r; ` ; s `+1 ` ar =
=1 =1 =1 th =1
r X
147 since
n X
i=1
i = n + 1
. ` `+1
+1 2
The display below exhibits s r n for r = 1; 2; 3; 4; 5:
s s s s s
1
2 3 4 5
n 1,n , , n 1n + 2! n n 1,n + 6,n + 3!,n , , , , n 1n + 14 n + 36 n + 4! n n1,n + 30,n + 150,n + 240,n + 5!,n
+1 2 +1 3 +1 2 +1 3 +1 4 +1 2 +1 3 +1 4 +1 5 +1 2 +1 3 +1 4 +1 5
+1 6
References
1. 1 Moser, W., Sums of d powers. Math. Gazette 75 1991 332. 2. 2 Paul, J.L., On the sum of the k powers of the rst n integers. Amer. Math. Monthly 78 1971 271 273. MR 43 4092. 3. 3 Wagner, C., Combinatorial proofs of formulas for power sums. Arch. Math. Basel 68 1997, no. 6, 464 467.
th th
148
THE SKOLIAD CORNER
No. 29 R.E. Woodrow
In this number we give the rst round of the 1996 97 Alberta High School Mathematics Competition written in November of 1996. The top students from Round 1 are invited to take part in the second round competition written in February. There are book prizes for individuals and teams based on the Round 1 results, and scholarships and cash prizes for Round 2. See 1997:410411, 479481 where we gave Round 2 of the 1996 97 contest and the answers. More information about the contest can be found at the website: www.math.ualberta.ca ahsmc . My thanks go to the organizers of the committee chaired by T. Lewis, University of Alberta for permission to use these materials.
ALBERTA HIGH SCHOOL MATHEMATICS COMPETITION
Part I  November, 1996
1. An eightinch pizza is cut into 3 equal slices. A teninch pizza is cut into 4 equal slices. A twelveinch pizza is cut into 6 equal slices. A fourteen inch pizza is cut into 8 equal slices. From which pizza would you take a slice if you want as much pizza as possible? a 8inch b 10inch c 12inch d 14inch e does not matter 2. One store sold red plums at four for a dollar and yellow plums at three for a dollar. A second store sold red plums at four for a dollar and yellow plums at six for a dollar. You bought m red plums and n yellow plums from each store, spending a total of ten dollars. How many plums in all did you buy? a 10 b 20 c 30 d 40 e not enough information 3. Six identical cardboard pieces are A piled on top of one another, and B the result is shown in the diaD gram. F E C
The third piece to be placed is: a A b B c C
d D e E
149
4. A store o ered triple the GST in savings. A sales clerk calculated the selling price by rst reducing the original price by 21 and then adding the 7 GST based on the reduced price. A customer protested, saying that the store should rst add the 7 GST and then reduce that total by 21. They agreed on a compromise: the clerk just reduced the original price by the 14 di erence. How do the three ways compare with one another from the customer's point of view? a The clerk's way is the best. c The compromise is the best. b The customer's way is the best. d All three ways are the same. e The compromise is the worst while the other two are the same. 5. If m and n are integers such that 2m , n = 3, then what will m , 2n equal? a ,3 only b 0 only c only multiples of 3 d any integer e none of these 6. If x is x of y, and y is y of z, where x, y and z are positive real numbers, what is z ? a 100 b 200 c 10; 000 d does not exist e cannot be determined 7. About how many lines can one rotate a regular hexagon through some angle x, 0 x 360 , so that the hexagon again occupies its original position? a 1 b 3 c 4 d 6 e 7 8. AB is a diameter of a circle of radius 1 unit. CD is a chord perpendicular to AB that cuts AB at E . If the arc CAD is 2=3 of the circumference of the circle, what is the length of the segment AE ? p a b c d e none of these 9. One of Kerry and Kelly lies on Mondays, Tuesdays and Wednesdays, and tells the truth on the other days of the week. The other lies on Thursdays, Fridays and Saturdays, and tells the truth on the other days of the week. At noon, the two had the following conversation: Kerry : I lie on Saturdays. Kelly : I will lie tomorrow. Kerry : I lie on Sundays. On which day of the week did this conversation take place? a Monday b Wednesday c Thursday d Saturday e Sunday
2 3 3 2 2 3 2
150
10. How many integer pairs m; n satisfy the equation
mm + 1 = 2n ?
a 0 b 1 c 2 d 3 e more than 3 11. Of the following triangles given by the lengths of their sides, which one has the greatest area? a 5; 12; 12 b 5; 12; 13 c 5; 12; 14 d 5; 12; 15 e 5; 12; 16 12. If x y and x 0, which of the following numbers is never greater than any of the others? a x + y b x , y c x + jy j d x , jy j e ,jx + y j 13. An x by y ag, with x y, consists of two perpendicular white stripes of equal width and four congruent blue rectangles at the corners. If the total area of the blue rectangles is half that of the ag, what is the length of the shorter side of each blue rectangle? p2 2 p2 2 p2 2 b x,y x y c x y x y a x,y x y
+ +
y d x y e none of these 14. A game is played with a deck of ten cards numbered from 1 to 10. Shu e the deck thoroughly. i Take the top card. If it is numbered 1, you win. If it is numbered k, where k 1, go to ii. ii If this is the third time you have taken a card, you lose. Otherwise, put the card back into the deck at the k position from the top and go to i. What is the probability of winning? a b c d e none of these 15. Five of the angles of a convex polygon are each equal to 108 . In which of the following ve intervals does the maximum angle of all such polygons lie? a 105 ; 120 b 120 ; 135 c 135 ; 150 d 150 ; 165 e 165 ; 180 16. Which one of the following numbers cannot be expressed as the di erence of the squares of two integers? a 314159265 b 314159266 c 314159267 d 314150268 e 314159269
3 + + + 2 th 1 5 5 18 13 45 3 10
4
px2
+
+
3 + +
+
2
2
4
151 In the last number we gave the problems of the Final Round of the British Columbia Colleges Junior High School Mathematics Contest 1997. Here are the o cial solutions. My thanks for sending the materials to me go to John Grant McLoughlin, now of the Faculty of Education, Memorial University of Newfoundland, who participated in formulating the exams while he was at Okanagan College.
BRITISH COLUMBIA COLLEGES Junior High School Mathematics Contest
Final Round 1997  Part A
1. The buttons of a phone are arranged as shown at the right. If the buttons are one centimetre apart, centretocentre, when you dial the number 5927018 the distance, in centimetres, travelled by your nger is:
1 4 7 2 5 8 0 3 6 9
Solution. Each of the distances between successive digits in the phone number is the hypotenuse of a rightangled trianglep p p p p with integer sides. The 6 lengths p be easily computed as 2; 5; 5; 2; 10pand 5, for a can p p p ; pp total of 2 2 + 3 5 + 2 5. This can be rearranged into 53 + 2 + p 2 2. Answer is A 2. What is the total number of ones digits needed in order to write the integers from 1 to 100? Solution. Clearly we need only one 1 for all the single digit numbers, and since we are only interested in one 3digit number, namely 100 we need only one 1 for all the 3digit numbers under consideration. For the 2digit numbers there are nine which have a 1 in the units position 11; 21; : : : ; 91, and 10 which have a 1 in the tens position 10; 11; : : : ; 19, for a grand total of 21 ones needed. Answer is D 3. The number of solutions x; y; z in positive integers for the equation 3x + y + z = 23 is: Solution. Clearly, there are only 7 possible values for x, namely 1 through 7. When x = 7, we have y + z = 2, which has a single solution y = z = 1; when x = 6, we have y + z = 5, which has 4 solutions: y;z = 1; 4; 2; 3; 3; 2; 4; 1. In this way we see that by reducing the value of x by 1 we increase the number of solutions by 3. Thus the total number of solutions are 1 + 4 + 7 + 10 + 13 + 16 + 19 = 70.Answer is D
152
4. In the diagram below the upper scale AB has ten 1 centimetre divisions. The lower scale CD also has ten divisions but it is only 9 centimetres long. If the right hand end of the fourth division of scale CD coincides exactly with the right hand end of the seventh division of scale AB , what is the distance, in centimetres, from A to C ?
A B C D
Solution. Let E be the right hand end of the fourth division of scale CD which is also the right hand end of the seventh division of scale AB. The distance AE is 7 cm. The distance CE is 4 0:9 = 3:6 cm. Thus the distance AC = AE , CE = 3:4 cm. Answer is E 5. Triangle ABCpis equilateral with sides tangent to the circle with centre at O and radius 3. The area of the quadrilateral AOCB , in square units, is:
A O B
C Solution. First join the points B and O. Since an intersecting tangent and radius of a circle are perpendicular, this produces 2 congruent right angled triangles, namely BOA and BOC , each of which is a 30 , 60 , 90 triangle, which means that side BO is twice the length of side AO. Using the Theorem of Pythagoras we have p p BA = 2 3 , 3 = 12 , 3 = 9: p Thus BA = BC = 3. The area of triangle BOA is now 3 3. Since this p3. is half the area we seek, the answer is 3 Answer is A
2 2 2 1 2
6. Times such as 1:01, 1:11,: : : are called palindromic times because their digits read the same forwards and backwards. The number of palindromic times on a digital clock between 1:00 a.m. and 11:59 a.m. is: Solution. Let us rst consider single digit hours. The rst and last digits must be the same, but the middle digit can be anything. Thus there are 6 such palindromic times each hour since there are 6 possible middle digits, for a total of 54 such palindromic times between 1:00 am and 9:59 am. For two digit hours that is, 10 and 11, in order to be a palindromic time the minutes are completely determined by the hour; there are only two such times, namely 10:01 and 11:11. So the grand total number of palindromic times between 1:00 a.m. and 11:59 a.m. is 56. Answer is D
153
7. Ted's television has channels 2 through 42. If Ted starts on channel 15 and surfs, pushing the channel up button 518 times, when he stops he will be on channel: Solution. Ted's television set has a total of 41 channels. Thus every time he pushes the channel up button 41 times he returns to the channel he starts with. Since
518 = 12 41 + 26 we see that he has e ectively gone up 26 channels from his starting point at channel 15. He should thus be at channel 15+26=41. Answer is E
1. If its tens and ones digits are switched, its value would increase by 36. 2. Instead, if its hundreds and ones digits are switched its value would decrease by 198. Suppose that only the hundreds and tens digits are switched. Its value would: Solution. Let n be a threedigit number with the given properties. Let a, b, and c represent its hundreds digit, its tens digit and its ones digit, respectively. Then n = 100a + 10b + c. By property 1 we see that
8. Consider a threedigit number with the following properties:
100a + 10c + b = n + 36 = 100a + 10b + c + 36; or 9c , b = 36; that is, c , b = 4:
By property 2 we have
Note that the two relations established above when added yield a , b = 6. Now let us consider the switch of the hundreds and the tens digits of n; the new number is
100c + 10b + a = n , 198 = 100a + 10b + c , 198; or 99a , c = 198; that is, a , c = 2:
100b + 10a + c = 100a + 10b + c + 100b , a + 10a , b = n , 90a , b = n , 540:
Answer is B 9. Speedy Sammy Seamstress sews seventyseven stitches in sixtysix seconds. The time, in seconds, it takes Sammy to stitch fty ve stitches is: Solution. It takes Sammy seconds to sew a single stitch. For 55 stitches it will take 55 = 47 seconds. Answer is E
6 7 1 7 6 7
154
10. How many positive integers less than or equal to 60 are divisible by 3, 4, or 5? Solution. There are 20 integers between 1 and 60 which are divisible by 3, 15 divisible by 4, and 12 divisible by 5. If we simply add up these numbers to get 20 + 15 + 12 = 47, we will have counted those integers divisible by at least two of 3, 4, or 5 more than once. Thus we must subtract from this total the number of integers between 1 and 60 divisible by both 3 and 4 that is, divisible by 12, namely 5; the number divisible by both 3 and 5, namely 4; and the number divisible by both 4 and 5, namely 3. Thus our accumulated total is now 47 , 5 + 4 + 3 = 35. However, the integer 60, which is the only positive integer in our range divisible by all of 3, 4, and 5 was at rst counted 3 times, and now is not counted at all. Thus we must add 1 back into the total. So our total is 36. Alternate method: Simply enumerate all the numbers between 1 and 60 and test each one. Those that are divisible by 3, 4, or 5 are: 3, 4, 5, 6, 8, 9, 10, 12, 15, 16, 18, 20, 21, 24, 25, 27, 28, 30, 32, 33, 35, 36, 39, 40, 42, 44, 45, 48, 50, 51, 52, 54, 55, 56, 57 and 60. There are 36 numbers in this list. Answer is C
Final Round 1997  Part B
1. a The pages of a thick telephone directory are numbered from 1 to N . A total of 522 digits is required to print the pages. Find N . Solution. From 1 to 9 inclusive there are 9 single digit numbers, which uses up 9 digits. From 10 to 99 inclusive there are 90 2digit numbers, which uses up a further 180 digits. So far we have used up 189 digits before we get to 3digit numbers leaving us with 522 , 189 = 333 digits still to be accounted for. With this many digits we can make up 111 3digit numbers. Starting at 100 and proceeding for 111 numbers brings us to page 210. Thus N = 210. b There are 26 pages in the local newspaper. Suppose that you pull a sheet out and drop it on the oor. One of the pages facing you is numbered 19. What are the other page numbers on the sheet? Solution. Since the newspaper has 26 pages, the outer sheet has pages 1, 2, 25, and 26. The next outermost sheet has pages 3, 4, 23, and 24. The next sheet has 5, 6, 21, and 22. Finally the sheet we are interested in has pages 7, 8, 19, and 20. So the other page numbers are 7, 8, and 20. 2. Create expressions for the numbers 1; 2; 3; : : : ; 10 by using each of the digits 1, 9, 9, and 7. Note that the digits must appear separately; that is, numbers like 17 are not allowed. Only the basic operations +, ,, , and brackets if necessary may be used. Other mathematical symbols such as p are not allowed. Every expression must include one 1, two 9's, and one 7, in any order.
155 Solution. There are several acceptable answers for this question. Here are some:
b Show that x2x 2 for any real number x 0. Solution. Note that x , 1 0 for all real numbers x. This can be rewritten as x , 2x + 1 0, and thus we have x + 1 2x. Since we are given x 0 we can divide both sides of the inequality by x and preserve the direction of the inequality to get
+1 2 2 2
p p y = 5 + 9 + 2 45 = 14 + 2 45 p p45, we see that x y and thus it follows that x y since Since 48 p p both are positive real numbers. Therefore, the larger value is 6 + 8.
2 2 2
7 9 , 9 + 1 9 + 7 9 , 1 9 9 + 1 , 7 9 , 1 9 , 7 9 + 1 9 , 7 = 7 , 9 9 , 1 9,9+7,1 9 , 9 + 7 1 = 9 , 9 1 + 7 9,9+7+1 99+7+1 9+9,7,1 p 3. a Decide which is greater: p6 + p8 or p 5 + p9. p6 + p8 and y = p5 + 9. Then Solution. Let x = p p x = 6 + 8 + 2 48 = 14 + 2 48 1 2 3 4 5 6 7 8 9 10 = = = = = = = = = =
2
x + 1 2: x
2
4. In the plane gure shown below, ABCD is a square with AB = 12. If A0 , B0 , C 0 , and D0 are the midpoints of AO, BO, CO, and DO,
respectively, then:
A A
0
B B O D
0 0
C
0
D
C
156 a Find the area of the square A0 B 0 C 0 D0 . Solution. Triangles AOB and A0 OB 0 are similar since both are isosceles right angled triangles. Since A0 O is one half the length of AO we see that A0 B0 is also one half the 0length of AB; that is, A0B0 is 6 units in length, which makes the square A B 0 C 0 D0 have area 36 square units. b Find the area of the shaded region. Solution. Triangle BA0 B 0 has a base A0 B 0 of 6 units, and an altitude which is one half of the altitude of triangle AOB from its base AB ; that is, triangle BA0 B 0 has altitude 3 from its base A0 B 0 . Thus the area of triangle BA0 B0 is 6 3 = 9 square units. Since the shaded area comprises 4 such triangles, the total shaded area is 36 square units. Alternate method for b: Triangles A0 OB and A0 B 0 B have the same altitude from the base OB 0 B , and they have the same length for a base, since B0 is the midpoint of OB. Thus their areas must be the same.0 But the area of triangle A0 OB 0 is clearly of the area of the square A0 B 0 C D0 computed in part a; that is, 9 square units. Thus, triangle BA0 B 0 has area 9 and we nish as in the rst method. c Find the area of the trapezoid AA0 B 0 B . Solution. Since the parallel sides of the trapezoid have lengths 12 and 6, and since its altitude is 3 as seen in part b above, we see that the area of the trapezoid is:
1 2 1 4
1 36 + 12 = 27: 2
Alternate method for c: The area we seek is of the di erence between the areas of the squares ABCD and A0 B 0 C 0 D0 , which is:
1 4
1 12 , 6 = 1 144 , 36 = 27: 4 4
2 2
5. The gure below shows the rst three in a sequence of square arrays of dots. The numbers of dots in the three arrays are 1, 5, and 13.
r r r r r r r r r r r r r r r r r r r
a Find the number of dots in the next array in the sequence. Solution. There are several ways of looking at the arrays. One way is to notice that in going from one array to the next we build around the outside of the array a new set of dots. That is, we add 4 dots to the single dot to get
157 array 2. We then add 4 2 = 8 dots around array 2 to get array 3. We would then add 4 3 = 12 dots around array 3 to get array 4. This leaves us with 13 + 12 = 25 dots in the next array. b Find the number of dots in the sixth array in the sequence. Solution. To carry on to array 6 we need rst to get array 5 from array 4. Using the process described in part a we would get 25 + 4 4 = 41 dots in array 5 and 41 + 4 5 = 61 dots in array 6. c Find an expression for an in terms of n alone. Solution. By observing the dot pattern, the number of dots seems to be in the order of n . If we subtract n from an for the rst few values of n we see that what remains is n , 1 . Since an = n + n , 1 satis es the relation from part d above, and since it coincides with the rst few values of an , this must be the solution for an . d If an is the number of dots in the n array in the sequence, nd a relation between an , an , and n. Solution. From the above analysis, if we denote by an the number of dots in the n array, then
2 2 2 2 2 th +1 th
an = an + 4n:
+1
That completes the Skoliad Corner for this number. I need contest materials suited to this feature, so please send me your materials, as well as comments and suggestions for the future of the Skoliad Corner.
158
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only problems  the next issue will feature only solutions. We warmly welcome proposals for problems and solutions. We request that solutions from this issue be submitted by 1 February 1999, for publication in issue 4 of 1999. Also, starting with this issue, we would like to reopen the problems to all CRUX with MAYHEM readers, not just students, so now all solutions will be considered for publication.
Erratum
We regret to report that the High School porblems in volume 24, issue 1 1998: 42 , were mislabelled. Instead of H223, H224, H225 and H226, they should have been labelled H233, H234, H235 and H236 respectively. We kindly ask readers to respect the new labelling when submitting solutions.
159
High School Problems
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca
H237. The letters of the word MATHEMATICAL are arranged at random. What is the probability that the resulting arrangement contains no adjacent A's? H238. Johnny is dazed and confused. Starting at A0; 0 in the Cartesian grid, he moves 1 unit to the right, then r units up, r units left, r units down, r units right, r units up, and continues the same pattern inde nitely. If r is a positive number less than 1, he will be approaching a point Bx; y. Show that the length of the line segment AB is greater than . H239. Find all pairs of integers x; y which satisfy the equation y x + 1 + x y + 16 = 448. H240. Proposed by Alexandre Trichtchenko, Brook eld High School, Ottawa, Ontario. A Pythagorean triple a; b; c is a triple of integers satisfying the equation a + b = c . We say that such a triple is primitive if gcda; b; c = 1. Let p be an odd integer with exactly n prime divisors. Show that there exist exactly 2n, primitive Pythagorean triples where p is the rst element of the triple. For example, if p = 15, then 15; 8; 17 and 15; 112; 113 are the primitive Pythagorean triples with rst element 15.
2 3 4 5 7 10 2 2 2 2 2 2 2 1
Advanced Problems
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu A213. Show that the number of nonnegative integer solutions to the equation a + b + c + d = 98, where a b c d, is equal to the number of nonnegative integer solutions to the equation p + 2q + 3r + 4s = 98. A214. Show that any rational number can be written as the sum of a nite number of distinct unit fractions. A unit fraction is of the form n , where n is an integer. A215. For a xed integer n 2, determine the maximum value of k + + kn, where k , : : : , kn are positive integers with k + + kn 7n. Polish Mathematical Olympiad
1 1 1 3 1 3
160 tions:
A216. Given a continuous function f : R ! R satisfying the condif 1000 = 999; f x f f x = 1 for all x 2 R:
Determine f 500. Polish Mathematical Olympiad
Challenge Board Problems
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu Editorial Notes: Part b is the interesting part of C77. Part a is a commonly asked problem, but I think it is better to ask it again than simply to state it for use in b.
th 0 1 2 3 4
C77. Let Fn denote the n Fibonacci number, with F = 1 and F = 1. Then F = 2, F = 3, F = 5, etc. a Prove that each positive integer is uniquely expressible in the form Fa1 + + Fak , where the subscripts form a strictly increasing sequence of positive integers, no pairp which are consecutive. of b Let = 1 + 5, and for any positive integer n, let f n equal the integer nearest to n. If n = Fa1 + + Fak is the expression for n from part a, prove that f n = Fa1 + + Fak . C78. Let n be a positive integer. An n n matrix A is a magic matrix of order m if each entry is a nonnegative integer and each row and column P P sum is m. That is, for all i and j , k Aik = k Akj = m. Let A be a magic matrix of order m. Show that A can be expressed as the sum of m magic matrices of order 1.
1 2 +1 +1
161
Tips on Inequalities
graduate student, Yale University Inequalities can be di cult to solve because there are few systematic methods for tackling even the most simple formulations. Indeed, solving usually involves a trial and error of di erent approaches, before one hits the right combination of estimations and manipulations. In this article, we expose some useful standard approaches and techniques. We recall two basic and fundamental inequalities: AMGM Inequality. For all x , x , : : : , xn 0,
Naoki Sato
x + x + + xn px x x ; n n n with equality if and only if x = x = = xn . CauchySchwarz Inequality CSB. For all real xi , yi , i = 1, 2, : : : , n, x y + x y + + xn yn x + x + + xn y + y + + yn ; with equality if and only if the vectors x ; x ; : : : ; xn and y ; y ; : : : ; yn
1 2 1 2 1 2 1 1 2 2 2 2 1 2 2 2 2 1 2 2 2
1
2
are proportional. Manipulating the Expressions A typical approach in proving an inequality of the form A B is to nd intermediate expressions, so we have a chain
1
2
1
2
A P P Pn B:
1 2
These are usually found through using the classic inequalities, and by manipulating terms until we get what we want. As anyone who has worked with inequalities knows, this takes great care; one constantly has to make sure that the estimates are not too crude, and that the inequality signs are going the right way. In this kind of approach, there are several things one should keep in mind: 1. Is the inequality sharp or strict? An inequality is sharp if equality occurs at a point, and strict if equality never occurs. It is always a good idea to check which type it is, though it is usually given or obvious. A strict inequality may allow for generous estimates, but not always. A sharp inequality leaves no such allowance. 2. If equality does occur, when where does it occur?
Copyright c 1998 Canadian Mathematical Society
162 The points where equality occurs are points you must work around. In the chain above, each intermediate inequality must become an equality at these points. This is a good check of whether your intermediate expressions are the right ones. Pairing and Grouping In inequalities where several terms are involved, it might be possible to group terms together and prove smaller" inequalities. Problem 1. a Prove that x + y + z xy + yz + zx for all x, y , z 2 R. b Prove that x + y + z x y + y z + z x for all x, y , z 0. Solution. a No grouping is immediately obvious. We know x + y 2xy, but how can we incorporate this? By adding x + y 2xy , y + z 2yz, and z + x 2zx, and dividing by 2, we obtain the desired inequality. b Here, we know 2x + y = x + x + y 3x y by AMGM. We then add the other two corresponding inequalities, and then divide by 3. Problem 2. For all a, b, c, d 0, show that
2 2 2 3 3 3 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 2
a3 + b3 + c3 + a3 + b3 + d3 + a3 + c3 + d3 + b3 + c3 + d3 a2 + b2 + c2 + d2 : a+b+c a+b+d a+c+d b+c+d
Solution. We claim that
3
a +b +c a +b +c : a+b+c 3
3 3 2 2 2 3 3 2 2 2
Then, the problem follows by symmetry. Our inequality is equivalent to
3a + 3b + 3c a + b + c a + b + c ;
3
which, in turn, is equivalent to
2
a3 + 2b3 + 2c3 , a2 b , ab2 , a2 c , ac2 , b2 c , bc2 3 2 2 3 3 2 2 3 3 2 2 3 = a , a b , ab + b + a , a c , ac + c + b , b c , bc + c 2 2 2 = a , b a + b + a , c a + c + b , c b + c 0 ;
which is true. Note that this inequality also follows from Chebyshev's inequality. The CauchySchwarz Inequality The CauchySchwarz inequality is a good way to deal with squares and especially fractions. Problem 3. For x , x , : : : , xn 0, show that
1 2
+ x x x + + x xn x x + x + + xn : x +x + 2 n+
x
2 1
2 2
2
1
2
1
2
2
3
1
163
Solution. By CSB,
1
x + x + x + x + + xn + x x x x + x x x + + x xn x + + n+ x + x + + xn : The result then follows by dividing each side by 2 x + x + + xn . Problem 4. Prove that for a , a , : : : , an 0, a + a + + an a + a + + an : 2a + a + + an a + a a + a a +a
2 2 3 1 2 1 2 2 2 1 2 2 3 1 1 2 2 1 2 1 2 1 2 2 1 2 2 1 2 2 2 2 3 3 4 1 2
2 2 2 x2 y2 + + xn yn 2 ,x2 + x2 + + x2 ,y1 + y2 + + yn : 1 2 n p 4 Setting xi = pai , yi = 4 ai , we obtain a + , + + an ap p a + a + + pan ,a pa + a pa + + anpan : Setting xi = pai , yi = ai , we obtain 1 1+
xy
1990 1991 IMO Correspondence Solution. Recall that CSB states that for all real xi , yi, i = 1, 2, : : : , n,
3
1
2
2
1
2
1
1
2
2
,
Combining these two, we obtain
1 2 3 1 2
a pa + a pa + + a, pan n a + a + + an a + a + + an :
1 1 2 2 2 1 2 2 1 2 2 2 2
Finally, setting xi = ai + ai , yi = ai+1aiai+2 , we obtain
p
+1 +2 +
a + a + + an ,pa + pa + + pan ,a + a + + an :
2 1 2 2 2
q
,
pa + pa + + pan a + a + + an
: 2 a + a + + an a + a a + a a +a
1 2 2 1 2 1 2 2 3 3 4 1 2
The last two inequalities give the desired result. Elementary Symmetric Polynomials Given a set of variables X , such as X = fx; y;z g, a polynomial is symmetric in X if it is invariant under any permutation of the variables under X , and homogeneous of degree k if every term in the polynomial has degree k.
164 For example, x + y + z , xyz is symmetric but not homogeneous in X , and x + x y , xyz is homogeneous of degree 3 but not symmetric in X . The elementary symmetric polynomials in X are the polynomials obtained as the sum of the products of the variables in X , taken k at a time. For X = fx; y;z g, these would be x + y + z , xy + xz + yz , and xyz . Note that each of these is homogeneous. A theorem of Gauss states that any symmetric, homogeneous polynomial in X can be expressed as a polynomial in these elementary polynomials. After all these tedious de nitions, we nally get to the point that it can be useful to use these elementary polynomials. Problem 5. For nonnegative reals x, y , and z satisfying x + y + z = 1, show that
2 2 2 3 2
x
1 +1
y
1 +1
1 + 1 64 : z
Proof. Expanding, we rst must show
By AMGM, Therefore,
1 1 1 1 1 1 1 + x + y + 1 + xy + xz + yz + xyz 64 : z
3
1 xyz x + y + z = 27 3
so that
1 27 : xyz
1 1 1 1 1 1 1 + x + y + 1 + xy + xz + yz + xyz z 3 +p 3 + 1 1+
pxyz 3 x y z 3 1
= 1 + pxyz 3 4 = 64 :
2 2 3 3
2
xyz
Problem 6. Prove that
where x, y , and z are nonnegative real numbers for which x + y + z = 1. 1984 IMO, 1 Proof. Let S = xy + xz + yz , 2xyz , P = 1 , 2x1 , 2y 1 , 2z . Then
7 0 yz + zx + xy , 2xyz 27 ;
P = 1 , 2x + y + z + 4xy + xz + yz , 8xyz = 4S , 1 :
165 We must show that 0 S , or equivalently, ,1 P . Since 0 x; y;z 1, ,1 2x , 1; 2y , 1; 2z , 1 1, so ,1 P . Now, if one of the variables, say x, was greater than 1=2, then the other two would be less than 1=2, and we would have 1 , 2x 0, 1 , 2y , 1 , 2z 0, and P 0 . Otherwise, x, y , and z are at most 1=2, and all factors of P are nonnegative, so by AMGM,
7 27 1 27 1 27
1 , 2x + 1 , 2y + 1 , 2z
= 1 : P 3 27
3
Introducing and Removing Constraints Inequalities often come with constraints on the variables. Removing these constraints can simplify the problem. Alternately, introducing them may help as well. The most common way of removing a constraint is to homogenize" the given inequality. For example, suppose we are given the expression x + xy , 2, where xyz = 1. Then
3
3 x + xy , 2 = x + xy pxyz , 2xyz:
3 3
This new expression is homogeneous of degree 3. It's not pretty, but for the expressions given in problems, most of the time it will be nice and much easier to work with. At this point, we introduce an inequality that is not wellknown, but that seems to pop up from time to time. Schur's Inequality. For all x, y , z 0 and nonnegative integers n,
xnx , yx , z + yny , zy , x + znz , xz , y 0 : For n = 1, this becomes x + y + z + 3xyz x y + xy + x z + xz + y z + yz ; P P or in shorthand, x + 3xyz x y . This is a useful inequality to know,
3 3 3 2 2 2 2 2 2
with which we present another solution. Problem 6. Prove that
3
2
where x, y , z are nonnegative real numbers for which x + y + z = 1. 1984 IMO, 1 Solution. This is equivalent to the following problem: For x, y , z 0, prove that
7 0 yz + zx + xy , 2xyz 27 ;
7 0 yz + zx + xy x + y + z , 2xyz 27 x + y + z :
3
166 This is the homogeneous version of the original inequality. The expression P in the middle expands to x y + xyz , which is clearly nonnegative. We focus on the right inequality, which becomes
2
P
which implies
7 x y + xyz 27 P x + 7 P x y + 14 P xyz ; 9 9
2 3 2
This is where we must think backwards. What results do we know that P P we can use to prove this? By Schur's, 5 x y 5 x + 15xyz we try to eliminate the xyz term. Hence, to prove the above inequality, we must P P show that x y 2 x , which is left as an exercise for the reader it has been virtually done already in this article. We stated early in the solution that the modi ed problem was equivalent to the original problem. It is easy to see that the modi ed problem implies the original problem which is the only direction we actually needed, but what about the converse? What if x + y + z 6= 1? In such a case, we can normalize. A property of homogeneous polynomials, and an alternate de nition, is the following: px ; x ; : : : ; xn is homogeneous of degree k if
2 3 2 3
6 P x y 7 P x + 15xyz :
2 3
px; y; z = yz + zx + xyx + y + z , 2xyz : If x + y + z = 0, then all three variables must be 0, and the inequality follows. Otherwise, we can set = x y z , and so we must show
7 0 p x + x + z ; x + y + z ; x + z + z 27 : y y y This is the original problem. Thus, we can set x + y + z equal to 1, or
1 + +
for all 2 R. Going back to the original problem,
px ; x ; : : : ; xn = kpx ; x ; : : : ; xn
1 2 1 2
1
2
indeed anything we want to except 0. It can be vital to exploit this degree of freedom. We perform a similar setting in the next problem. Problem 7. Given 0 a b, and x , x , : : : , xn 0, show that
1 2
xa + xa + + xa 1=a xb + xb + + xb 1=b : 1 2 n 1 2 n Solution. If xb + xb + + xb = 0, then the problem is solved. 1 2 n Otherwise, by the arguments above, we can assume xb + xb + + xb = 1. 1 2 n
Then
xb 1 = xi 1 = xb,a 1 = xb xa i i i i = xa + xa + + xa xb + xb + + xb = 1 n n = xa + xa + + xa =a 1 = xb + xb + + xb =b: n n
1 2 1 2 1 2 1 1 2 1
167
Problems 1. Prove that if x, y , and z are nonnegative real numbers such that x + y + z = 1, then
2x + y + z + 9xyz 1 :
2 2 2 2 2 2 2
2. For any real numbers a, b, and c, show that
min a , b ; b , c ; c , a a + b + c : 2 3. Let a, b, and c be the sides of a triangle with perimeter 2. Prove that a + b + c + 2abc 2. 4. For nonnegative reals x, y , and z satisfying 2xyz + xy + xz + yz = 1, prove that x + y + z .
2 2 2 2 2
5. For all positive integers n, show that
3 2
1 3 5 2n , 1 p 1 : 2 4 6 2n 3n + 1
Hint: The inequality is almost certainly not sharp, so there is some room for approximation. The RHS suggests squaring. 6. Show that if x, y , and z are nonnegative reals such that x + y + z = 1, then
Note: The solution in Problem 5 does not work! 7. Given a, b, c, d, e 0, abcde = 1, show that
4 4 4 4 4
x
1 ,1
y
1 ,1
1 ,1 8: z
a + b + c + d + e a + b+ c +d + e:
A favourite of Ravi Vakil's. Find as many di erent solutions as you can. 8. Show that if nonnegative reals a, b, and c satisfy then abc 8.
1 + 1 + 1 = 1; 1+a 1+b 1+c
168
Riveting Properties of Pascal's Triangle
student, University of Waterloo Consider the following table of integers, known as Pascal's Triangle: 1 1 1 1 2 1 1 3 3 1 1 4 6 4 1 1 5 10 10 5 1 . . .
th
Richard Hoshino
This table is named after the French mathematician Blaise Pascal, who conceived of it at the age of thirteen although the Chinese discovered some properties of this triangle in the early 14 century, centuries before Pascal was even born!. To generate a row of Pascal's Triangle, look at the row immediately above it. Each element of the triangle is the sum of the two elements above it for example, the second element of the fourth row is 4 because 4 = 1+3. By convention, we denote the top row as the 0 row, and we denote the left most entry of each row as the 0 entry, even though it may seem a little awkward at rst. Let us show that this table is the same as the following table:
th th
,5
0
,4
0
,3
0
,2
0
,1
0
,0
0
,5
1
,4
1
,3
1
,2
1
,1
1
,5
2
,4
2
,3
2
,2
2
,5
3
,4
3
,3
3
,5
4
,4
4
,5
5
. . . ,n n In this table, k is determined by the formula k n,k . Thus, wewill ,n show that the k element of the n row of Pascal's Triangle equals, k ,for all integers n and k. Each of these elements in the triangle, namely , , , , , , , , : : : , is called a binomial coe cient, since the coe cients of the expansion of the binomial 1 + xn correspond to the entries of the n row
! ! ! th th 1 1 2 0 2 1 0 0 1 0 th
Copyright c 1998 Canadian Mathematical Society
169 of Pascal's Triangle for example, 1 + x = x + 4x + 6x + 4x + 1. Let us prove these statements, as they will be fundamental to our analysis of the properties of Pascal's Triangle. , Now n denotes the number of ways we may select k objects from a k , set of n objects. By convention, we let n = 1, since there is technically only one" way we may select nothing from a set of n objects. Let us show that
4 4 3 2 0
for all n and k. Note that the right side denotes the number of ways we may select a kmember committee from a class of n girls and one boy we choose k members fromna of n+1 people. Now, if the boy is on the committee, set , then we have k, ways of selecting the,remaining k , 1 members. And if he is not on the committee, then there are n ways of selecting the committee. k Hence,
1
n + n = n+1 k k,1 k
This is known as Pascal's Identity. , Since = 1, the table of binomial coe cients corresponds directly to Pascal's Triangle since the initial element is the same and, like Pascal's Triangle, each element is the sum of the two directly above it. Thus, we can determine any element in Pascal's Triangle with this formula. For example, , the thirty fth element in the seventyninth row of Pascal's Triangle is . We now prove that the entries in the n row of Pascal's Triangle are the coe cients in the expansion of 1 + xn . We proceed by induction. The case is trivial for n = 1. Suppose that
0 0 79 35 th
n + n = n+1 : k k,1 k
1 + xn =
+1
for some n. Then
n + n
x + n
x + + n
xk + + n
xn 0 1 2 k n
2
1 + x
= 1
xn1
x n + + n + n x + n x + + n
xn 1 + x = 0 1 2 n
2
=
Since n = n = n = n = 1, and using Pascal's Identity for n n all the other terms, we immediately arrive at the case for n + 1. Hence, the , coe cient of xk in 1 + xn is equal to n . k Now we illustrate some of the really neat properties of Pascal's Triangle.
,
0
n + n
+ n
x + + n
+ n
xn + n
xn : 0 1 0 n n,1 n
+1
,
+1 0
,
,
+1 +1
170
Theorem 1. i The sum of the coe cients in the n row of Pascal's Triangle is 2n . ii If we alternately add and subtract the digits in the n row of Pascal's Triangle, we always arrive at zero. For example, 1 , 4 + 6 , 4 + 1 = 0 for n = 4.
th th
Proof. i Since
1 + xn =
substituting x = 1 into this expression yields
n + n
x + + n
xk + + n
xn; 0 1 k n
ii The second part of the theorem is left as an exercise. It is the same technique as above; just substitute a di erent value for x. Can you guess which value of x we ought to substitute? Theorem 2. If the binary representation of n contains p ones, then there are 2p odd numbers in the n row of Pascal's Triangle. For example, since 9 = 1001 , there are 2 = 4 odd numbers in the ninth row. Proof. We shall analyze everything in modulo 2. For those of you not familiar with modular arithmetic, when we say that x is congruent to y modulo m, which is written x y mod m, we mean that x and y are numbers such that when both are divided by m, they give the same remainder. For example, 1998 is congruent to 2 modulo 4. Also, if r 0mod 2, then r is even. We rst show that
th 2 2
2n = n + n + + n : 0 1 n
for all nonnegative integers n. We proceed by induction. If n = 0, the claim is immediate. Suppose the claim is true for some n = k. Then
2 2 2 2 2 2 2 2
1 + x n 1 + x n mod 2
2 2
1+x k+1 1+x k 1+x k 1+2x k +x k+1 1+x k+1mod 2; and so it is also true for n = k + 1. Hence, by induction, the claim has been
veri ed. Here is an example for a speci c case. Let n = 50. Then 50 = 2 + 2 + 2 = 2 + 16 + 32. Hence,
1 4 5
1 + x
50
= 1 + x 1 + x 1 + x 1 + x 1 + x 1 + x 1 + x + x + x + x + x + x + x mod 2:
2 16 32 2 16 32 2 16 18 32 34 48 50
171 Hence, there are eight odd entries in the 50 row of Pascal's Triangle, , , , , , , , , namely , , , , , , , and . Since 50 = 11010 , this veri es that there are exactly 2 = 8 odd terms in this row. For a general n, suppose there are p ones in the binary representation of n. Then
th 50 0 50 2 50 16 50 18 50 32 50 34 50 48 50 50 2 3
where 0 a a ap and the ai digit of the binary representation of n is 1, starting from 0 at the right. The right side is congruent to
1 2 th
1 + xn = 1 + x a1 1 + x a2 1 + x ap ;
2 2 2
1 + x a1 1 + x a2 1 + x ap
2 2 2
modulo 2, and when we expand the right side, we will arrive at an expression with 2p terms. Note that there must be exactly 2p terms, since each exponent xk can be2formed inp only one way by multiplying coe cients from the set fx a1 ; x a ; : : : ; x a g. This is a direct result from the binary representation of k. Hence, if n has p ones in its binary representation, 1 + xn modulo 2 has 2p terms, and hence there are 2p odd entries in the n row of Pascal's Triangle. Theorem 3. The Fibonacci sequence is the sequence, 1, 1, 2, 3, 5, 8, 13, 21, 34, : : : , where each element in the sequence is the sum of the two before it. More formally, we say that the Fibonacci sequence fFn g satis es the conditions F = 1, F = 1, and Fn = Fn, + Fn, for all n 1. We can derive the Fibonacci sequence from Pascal's Triangle in the following manner.
2 2 2 th 0 1 1 2
F F F F F F
0 1 2 3 4 5
=1 =1 =2 =3 =5 =8

1 1 1 : : 1 2 1 : : 1 3 3 : 6 : 1 4 4 1 5 10 10
. . .
1 5
1
1
In other words, for all n, we have
2
2n
+ 2n , 1
+ 2n , 2
+ + n
Fn= 0 1 2 n
+1
and
2n + 1
+ 2n
+ 2n , 1
+ + n + 1
: Fn = 0 1 2 n
2
172 For example,
7
F = 7 + 6 + 5 + 4 = 1 + 6 + 10 + 4 = 21: 0 1 2 3
0 1 2
Proof. We prove this theorem using double induction: we know that the claim is true for F and F . Suppose the claim is true for some F k and F k . Then we want to show that
+
+
F k = 2k 0 2 + 2k 1 1 + 22k + + k + 1 ; k+1 that is, the right side is equal to the 2k + 2 Fibonacci number. But F k = F k
F
+ k
+ = 2k 0 1 + 20k + 21k + + k + 1 + k + 1 + k k,1
k k
+ + = 2k 0 1 + 2k 1 1 + 22k + + k + 2 + k ; k k
2 +2 th 2 +2 2 2 +1 2 +1
, by Pascal's Identity. Since k have
2 +2
2 +1 0
, =1= k
2 +2 0
, , and k = 1 = k k k
+1 +1
, we
as required. The second part of the induction follows similarly assume the claim is true for F k, and F k, and show that the claim is also true for F k . The proof is left as an exercise. Now we are done since the proposition is true for k = 0 and k = 1, it is true for k = 2. Since it is true for k = 1 and k = 2, it is true for k = 3, etc. Then by double induction, the claim is true for all nonnegative integers n. Exercises , , n 1. Show that n = n,k , and use this fact to show that Pascal's Triangle k is symmetric about the vertical line that separates the triangle into two equal halves. , n 2. For which n is n odd? 3. Let an represent the number of elements in the n row of Pascal's Triangle that are congruent to 1 modulo 3. Let bn represent the number of elements in the n row of Pascal's Triangle that are congruent to 2 modulo 3. Prove that for all n, an , bn is a power of two. This problem was on the IMO short list one year it is very tough!
2 1 2 2 +1 2 th th
+
+
F k = 2k 0 2 + 2k 1 1 + 22k + + k + 1 ; k+1
173
Swedish Mathematics Olympiad
1985 Qualifying Round 1. The real numbers a, b, and c satisfy the equations ab + b = ,1 bc + c = ,1 ca + a = ,1 Calculate the product abc.
2. 1985 runners reported for a marathon. They were assigned the numbers 1, 2, : : : , 1985. However, a number of runners dropped out before the race. In fact, among the starting runners, there were no two for whom one runner's number was 10 times the other. What is the greatest number of runners that could have participated? 3. In the system of equations
a x+b y+c z+d u = 0 a x+b y+c z+d u = 0 a x+b y+c z+d u = 0 a x+b y+c z+d u = 0 the coe cients a , b , c , and d are even integers and the other coe cients are odd integers. Prove that the only solution in integers is x = y = z = u = 0. 4. The nonnegative integers p, q , r, and s satisfy the equality p + q + p = r + s + r: Show that p = r and q = s. 5. Let f be de ned by sin f x = 4x x sinx + 9 : x Find the least value of f over the interval 0 x . 6. The point P lies on the perimeter or inside a given triangle T . The point P 0, in the plane of the triangle, lies at a distance d from P0 . Let r and r0 be the radii of the smallest circles, with centres P and P respectively, which contain T . Show that r + d 3r 0 :
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 2 2 2 2
Give an example where equality holds.
174
1985 Final Round
1. Let a
b 0. Prove that a , b a + b , pab a , b : 8a 2 8b
2 2
2. Find the least natural number such that if the rst digit is placed last, the new number is 7=2 times as large as the original number. The numbers are written in the decimal system. 3. A, B , and C are three points on a circle with radius r, and AB = BC . D is a point inside the circle such that the triangle BCD is equilateral. The line through A and D meets the circle at the point E . Show that DE = r. 4. The polynomial px of degree n has real coe cients, and px 0 for all x. Show that
px + p0x + p00x + + p n x 0:
5. In a rightangled coordinate system, a triangle has vertices Aa; 0, B0; b, and C c; d, where the numbers a, b, c, and d are positive. Show that if we denote the origin by O,
AB + BC + CA 2CO:
6. Xwich has a vibrant clublife. For every pair of inhabitants there is exactly one club to which they both belong. For every pair of clubs there is exactly one person who is a member of both. No club has fewer than 3 members. At least one club has 17 members. How many people live in Xwich?
175
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 "11" or A4 sheets of paper. These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 November 1998. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX.
1 2
PROBLEMS
2306.
Proposed by Vedula N. Murty, Visakhapatnam, India. CORRECTION to a Give an elementary proof of the inequality:
Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Prove or disprove that if A, B and C are the angles of a triangle, then
2326?.
x
sin 2
,
2
2x ; 1+x
2 2
0 x 1:
1
UK.
2
X
cyclic
1 , sin A ,1 + 2 sin A 9 : ,A
2 2
2327.
Proposed by Christopher J. Bradley, Clifton College, Bristol,
1 2 3 2
an = an , an, + aan ; n 3 : n, Prove that each an 2 N, and that no an is divisible by 4.
+1 1 2
The sequence fan g is de ned by a = 1, a = 2, a = 3, and
176 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. It is known from Wilson's Theorem that the sequence fyn : n 0g, n! + 1 , contains in nitely many integers; namely, y 2 N if and with yn = n n+1 only if n + 1 is prime. a Determine all integer members of the sequences fyna : n 0g, with yn = n! + a , in the cases a = 2, 3, 4.
2328?.
n+a b Determine all integer members of the sequences fyn a : n 0g, with +a yn = n!+ a , in the cases a 5. n
Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Suppose that p and t 0 are real numbers. De ne pt := tp + t,p + pp and pt := ,t + t, p + 2 :
1
2329?.
a Show that p t p t for p 2. b Determine the sets of p: A, B and C , such that 1. p t p t, 2. p t = p t, 3. p t p t.
2330.
where
e = 3 , 11!3 + 3 2!11 , 113!53 + 53 4! , 309 5! + : : : ; 309 2119
11 53 309 2119 = = = = 33 4 11 5 53 6 309 + + + + 2 1; 3 3; 4 11; 5 53;
Proposed by Florian Herzig, student, Perchtoldsdorf, Austria. Prove that
. . . Ed: There is enough information here to deduce the general term. 2331. Proposed by Paul Yiu, Florida Atlantic University, Boca Raton, Florida, USA. Let p be an odd prime. Show that there is at most one nondegenerate integer triangle with perimeter 4p and integer area. Characterize those primes for which such triangles exist.
177
2332.
2 2
Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Suppose x and y are integers. Solve the equation
x y , 7x y + 12x , 21xy , 4y + 63x + 70y , 174 = 0:
2 2 2
Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. You are given that D and E are points on the sides AC and AB respectively of 4ABC . Also, DE is not parallel to CB . Suppose F and G are points of BC and ED respectively such that
2333.
n +n+1 n + 2n + i + 1 F i , 1. F i = n +n+i Find F n.
2 2 2
Proposed by Toshio Seimiya, Kawasaki, Japan. Suppose that ABC is a triangle with incentre I , and that BI , CI meet AC , AB at D, E respectively. Suppose that P is the intersection of AI with DE. Suppose that PD = PI . Find angle ACB . 2335. Proposed by Toshio Seimiya, Kawasaki, Japan. Triangle ABC has circumcircle ,. A circle ,0 is internally tangent to , at P , and touches sides AB , AC at D, E respectively. Let X , Y be the feet of the perpendiculars from P to BC , DE respectively. Prove that PX = PY sin A . 2336. Proposed by Toshio Seimiya, Kawasaki, Japan. The bisector of angle A of a triangle ABC meets BC at D. Let , and ,0 be the circumcircles of triangles ABD and ACD respectively, and let P , Q be the intersections of AD with the common tangents to ,, ,0 respectively. Prove that PQ = AB AC . 2337. Proposed by Iliya Bluskov, Simon Fraser University, Burnaby, BC. n + 2n + 2 , and, for each i 1, let Let F 1 =
2 2 2
BF : FC = EG : GD = BE : CD: Show that GF is parallel to the angle bisector of BAC .
2334.
Some readers have pointed out that problem 2287 1997: 501 is the same as problem 2234 1997: 168 , and that problem 2288 1997: 501 is the same as problem 2251 1997: 299 . Also part a of problem 2306 1998: 46; 175 is the same as the rst part of 2296 1997; 503 . The editors missed these duplications. Proposers are asked not to submit the same problem more than once.
178
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems.
2219. 1997: 110 Proposed by Christopher J. Bradley, Clifton College, Bristol, UK. Show that there are an in nite number of solutions of the simultaneous equations:
x , 1 = u + 1v , 1 y , 1 = u , 1v + 1 with x; y;u; v positive integers and x 6= y .
2 2
I. Solution by Charles Ashbacher, Cedar Rapids, Iowa, USA; Edward J. Barbeau, University of Toronto, Toronto, Ontario; Charles R. Diminnie, Angelo State University, San Angelo, TX, USA; Florian Herzig, student, Perchtoldsdorf, Austria; Cyrus Hsia, student, University of Toronto, Toronto, Ontario; Walther Janous, Ursulinengymnasium, Innsbruck, Austria; and V
clav Kone n
, Ferris State University, Big Rapids, Michigan, USA. a c y For positive integers n, the quadruples
x; y;u; v = 1; 2n + 1; 2n + 2n + 1; 1 give an in nite set of solutions in which x 6= y , since x , 1 = u + 1v , 1 = 0
2 2
and
y , 1 = 4n + 4n = u , 1v + 1:
2 2
II. Solution by Edward J. Barbeau, University of Toronto, Toronto, Ontario; Digby Smith, Mount Royal College, Calgary, Alberta; and the proposer. For any positive integer n, the quadruple
x; y; u; v = 2n , n; 2n + n; 4n + n; n is a solution in which x 6= y , since x , 1 = u + 1v , 1 = 4n , 4n + n , 1
2 2 3 2 4 3 2
and
y , 1 = u , 1v + 1 = 4n + 4n + n , 1:
2 4 3 2
III. Solution by Charles R. Diminnie, Angelo State University, San Angelo, TX, USA; and Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario.
179 It is well known that the Pell equation s , 2t = 1 has in nitely many solutions in positive integers s, t. Clearly, s t 1. If we set u = s + t, v = s , t, x = t , 1 and y = t + 1, then u + 1v , 1 = s , t + 1 = t , 2t = t , 1 , 1 = x , 1, and u , 1v + 1 = s , t , 1 = t + 2t = t + 1 , 1 = y , 1. Also solved by MANSUR BOASE, student, St. Paul's School, London, England; FLORIAN HERZIG, student, Perchtoldsdorf, Austria a second solution; RICHARD I. HESS, Rancho Palos Verdes, California, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece two solutions; GERRY LEVERSHA, St Paul's School, London, England; and PANOS E. TSAOUSSOGLOU, Athens, Greece. The families given by I, II and III above, do not exhaust all the possible solutions. It is interesting to note that the smallest" solution produced by all these families is 1; 3; 5; 1. The next smallest ones are 1; 5; 13; 1, 6; 10; 34; 2 and 11; 13; 29; 5, respectively. Both Herzig and Leversha obtained another in nite set of solutions in which v = 2, by considering the Pell equation 3x , y = 8. Their smallest" solution is 6; 10; 34; 2 listed above. However, the next solution, 22; 38; 482; 2 is not obtainable from any of the families given in I, II and III.
2 2 2 2 2 2 2 2 2 2 2 2 2 2
2220. Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel, Alcorc
n on, Madrid, Spain. Let V be the set of an icosahedron's twelve vertices, which can be partitioned into four classes of three vertices, each one in such a way that the three selected vertices of each class belong to the same face. How many ways can this be done? Solution by Florian Herzig, student, Perchtoldsdorf, Austria. I will prove that there are 10 di erent ways of partitioning the vertices of an icosahedron in the described manner. Since only the topological properties of the icosahedron are important, consider the graph of the vertices: see gure on next page.
We want to nd four triangles in this graph such that each vertex is used exactly once. Consider the vertex A. There are ve triangles having A as vertex. Because of spatial symmetry we may assume that triangle ABM is chosen. Thus for the triangle containing point C only two choices remain: 4CFL and 4CFK . Without loss of generality we may assume that triangle CFL is chosen. Now notice that the only free" triangle containing D is 4DZY and nally 4EKX remains. We have covered all possibilities already. For this case there are two di erent ways of nding disjoint" triangles because we may choose from two equivalent triangles for vertex C . If all ve possible triangles at vertex
180
A
r
r
M E B
r
D
r
r
Z Y
r r
r
L F
r
r
X
r
K
r
C
A are considered, we get a total of 10 di erent con gurations as claimed.
Also solved by MANSUR BOASE, student, St. Paul's School, London, England; JORDI DOU, Barcelona, Spain; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; and the proposer. There were two incorrect solutions. Several solvers noted that the centres of the faces used in the partition are the vertices of a tetrahedron. 1997: 111 Proposed by Sefket Arslanagi
, University of Sarac jevo, Sarajevo, Bosnia and Herzegovina. Find all members of the sequence an = 3 n, + 2n, n 2 N which are the squares of any positive integer. Solution by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA. We have a = 4 and a = 29. For n 3, 3 n, 3mod 4 and n, 0mod 4. Thus, an 3mod 4. But a positive integer is a 2 square only if it is congruent to 0 or 1mod 4. Hence, a = 4 is the only square in the sequence. Also solved by SAM BAETHGE, Nordheim, Texas, USA; FRANCISCO
BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain and MARIA
ASCENSION LOPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MIGUEL ANGEL
CABEZON OCHOA, Logro~ o, Spain; ADRIAN CHAN, student, Upper Canada n College, Toronto, Ontario; GORAN CONAR, student, Gymnasium Vara din, z Vara din, Croatia; YEO KENG HEE, Hwa Chong Junior College, Singapore; z
2 1 1 1 1 2 2 1 1
2221.
181 FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; ROBERT B. ISRAEL, University of British Columbia, Vancouver, BC; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St Paul's School, London, England; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; BOB
PRIELIPP, University of Wisconsin Oshkosh, Wisconsin, USA; ISTVAN REIMAN, Budapest, Hungary; HEINZJURGEN SEIFFERT, Berlin, Germany; D.J. SMEENK, Zaltbommel, the Netherlands; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; PANOS E. TSAOUSSOGLOU, Athens, Greece; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; and the proposer.
2222. 1997: 111 Proposed by Shawn Godin, St. Joseph Scollard Hall, North Bay, Ontario. Find the value of the continued root:
v u u t
4 + 27 4 + 29 4 + 31 4 + 33p :
q
s
r
NOTE: This was inspired by the problems in chapter 26 Ramanujan, In nity and the Majesty of the Quattuordecillion", pp. 193 195, in Keys to In nity" by Cli ord A. Pickover, John Wiley and Sons, 1995. I. Solution by Robert B. Israel, University of British Columbia, Vancouver, BC. The answer is 29. More generally, for any positive integer n, we claim that s r
4 + n 4 + n + 2 4 + n + 4p = n + 2 ;
q v u u u t v u u t s r q
where the left side is de ned as the limit of
p F n;m = 4 + n 4 + n + 2 4 + n + 4 4 + m 4
as m ! 1 where m is an integer and m , n is even. If g n;m = F n;m , n + 2, we have
2 2
F n;m , n + 2 = 4 + nF n + 2; m , 4 + nn + 4 = nF n + 2; m , n + 4 ; so gn;m = F n;mn+ n + 2 gn + 2; m :
182 Clearly F n;m
2, so jgn;mj
nn jgn;mj mm+ 2 jgm;mj + 2 Therefore g n;m ! 0 as m ! 1.
s r
By iterating this, we obtain
n jgn + 2; mj : n+4 nn + 2 : m
II. Solution by Efstratios Rappos, Girton College, University of Cambridge, England Let
Sn = 4 + 2n , 1 4 + 2n + 1 4 + 2n + 3p
q
Sn satis es the recurrence relation q Sn = 4 + 2n , 1Sn
if and only if
1
+1
Sn , 2Sn + 2 = 2n , 1Sn : By inspection, this admits Sn = 2n + 1 as a solution. We only have to prove that S = 3 to make this induction complete. Let
+1
Tn = 4 + 4 + 3 2n , 3 4 + 2n , 1 2n + 3
and
s
v u u t
s
r
q
p
Un = 4 + 4 + 3 2n , 3 4 + 2n , 12n + 3 = 3 : Clearly Tn Un and the latter is identically equal to 3. Therefore, using the p p fact that B A 0 implies that 4 + A=4 + B A=B , r q 4 + + 2n , 1p2n + 3 T 1 T3n = Un = q p n 4 + 2n , 12n + 3 rq s + 2n , 1p2n + 3 n+1 qp 2 2n1 3 + + 2n , 12n + 3 1 = ,! 1 1 2n + 3 2 n+1
r
q
p
183 as n ! 1 for example, by rewriting as expf, ln2n + 3=2n g and using L'H^ pital's rule . This proves that S = limn!1 Tn = 3. The required o expression is precisely S and hence its value is 29. Also solved or answered by MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut, USA; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS,
Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St Paul's School, London, England; J.A. MCCALLUM, Medicine Hat, Alberta; HEINZJURGEN SEIFFERT, Berlin, Germany; and the proposer. Several readers made reference to the July 1996 issue of the Mathematical Gazette, which contained a similar problem, Problem 80E posed by Tony Ward: evaluate:
+1 1 14
s
1 + 1 1 + 2 1 + 3p :
q
r
Bradley notes that the solution appeared in the March 1997 issue where the value of the continued root was proved to be 2, and to be wellde ned". He also adds that the editor of the Problems section of the Gazette concludes Clearly, the problem raises some deep questions about the meaning of `wellde ned'." Along the same lines as this editorial comment, Murray S. Klamkin, University of Alberta, Edmonton, Alberta adds that the value can be anything since there is no de nition regarding the continuation of the root". Klamkin refers to A. Herschfeld, On In nite Radicals, Amer. Math. Monthly, 42 1935 419429 where Herschfeld notes that Ramanujan's solution for r q
1 + 2 1 + 3p1 + = 3
q p
is incomplete since one may write similarly that
4 = =
p r
1 + 2 15=2 =
q
1 + 2 1 + 3p1 + :
1 + 2 1 + 3 221=12
Despite these comments most solvers expressed no di culty in understanding the meaning of the continued root, and for that reason we have decided to print the above solutions". Another reference to similar problems given by several readers was to J.M. Borwein, G. de Barra, Nested Radicals, Amer. Math. Monthly, 98 1991 735739.
184
2224. 1997: 111 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. Point P lies inside triangle ABC . Triangle BCD is erected outwardly on side BC such that BCD = ACP and CBD = ABC . Prove that if the area of quadrilateral PBDC is equal to the area of triangle ABC , then triangles ACP and BCD are similar. Solution by Ian June L. Graces, Manila, The Philippines; and Giovanni Mazzarello, Firenze, Italy. Let A0 and P 0 be the respective images of A and P under re ection in the line BC . Note that B; D; A0 are collinear by the de nition of D. Denoting by XY Z the area of 4XY Z , we have
and
1 P 0CB = 2 P 0C BC sin P 0CB;
A0 CD = 1 A0C CD sin A0CD: 2 As a consequence of the given conditions PBDC = ABC and the e ect of the re ection, P 0 CB = A0 CD these are the complements of 4BDC in PBDC and in 4A0 CB and A0 CD = P 0 CB . Thus A0 C CD = P 0C BC: In other words, by SAS we have 4A0 CP 0 4BCD. From the re ection we have 4A0 CP 0 ACP , and the desired result =
follows. Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, and
MARIA ASCENSION LOPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; WALTHER JANOUS, Ursulinengymnasium,
Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece;
GERRY LEVERSHA, St Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
185
2225. 1997: 111 Proposed by Kenneth Kam Chiu Ko, Mississauga, Ontario. a For any positive integer n, prove that there exists a unique ndigit number N such that: i N is formed with only digits 1 and 2; and ii N is divisible by 2n . b Can digits 1" and 2" in a be replaced by any other digits? Solution by Robert B. Israel, University of British Columbia, Vancouver, BC. We can use any two nonzero digits whose di erence is odd. Let the digits be a and b, where a; b 2 f1; 2; : : : ; 9g and a , b is odd. There are 2n di erent n digit numbers formed with these two digits, and 2n residue classes modulo 2n . I claim that the 2n numbers are all in distinct residue classes. By the Pigeonhole Principle, exactly one of these numbers must be in the residue class 0. Consider two distinct n digit numbers, N and N , formed with the digits a and b. Suppose that the rst digit, counting from right to left, where they di er, is in the 10k position, 0 k n , 1, where N has a and N has b. Then N , N a , b10k = a , b5k2k mod 10k , and thus modulo 2k . Since a , b5k is odd, we have N 6 N mod 2n. Also solved by SAM BAETHGE, Nordheim, Texas, USA; CARL BOSLEY, student, Washburn Rural High School, Topeka, Kansas, USA; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CURTIS COOPER, Central Missouri State University, Warrensburg, Missouri, USA; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Aus
tria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St Paul's School, London, England; KATHLEEN E. LEWIS, SUNY Oswego, Oswego, NY, USA; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; JOEL SCHLOSBERG, student, Hunter College High School, New York, NY, USA; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; KENNETH M. WILKE, Topeka, Kansas, USA; YEO KENG HEE, student, Hwa Chong Junior College, Singapore; and the proposer. Besides Israel, only Lambrou, Leversha and Schlosberg proved the more general result presented above. Shan and Wang pointed out that if fa; bg = f2; 4g, f2; 8g, f4; 6g, f6; 8g or f4; 8g, then the existence" claim is still true and the uniqueness" claim would be true if one strengthens condition ii to 2n jN for the rst four pairs, and to 2n jN for the pair f4; 8g.
1 2 1 +1 1 2 +1 2 1 2 +1 +2
186 Diminnie asked whether any similar results are possible if 2n is replaced by kn for 3 k 9. 1997: 166 Proposed by K.R.S. Sastry, Dodballapur, India. An old man willed that, upon his death, his three sons would receive the u , v , w parts of his herd of camels respectively. He had uvw , 1 camels in the herd when he died. Obviously, their sophisticated calculator could not divide uvw , 1 exactly into u, v or w parts. They approached a distinguished CRUX problem solver for help, who rode over on his camel, which he added to the herd and then ful lled the old man's wishes, and took the one camel that remained, which was, of course, his own. Dear CRUX reader, how many camels were there in the herd? I. Solution by Robert B. Israel, University of British Columbia, Vancouver, BC. There were 41 camels. We may assume u v w. Of course u 2, and vw + uw + uv = uvw , 1. If u = 2, the equation becomes v , 2w , 2 = 5. Since the only factorization of 5 is 1 5, this means v = 3, w = 7, and uvw , 1 = 41. If u = 3, the equation becomes 2v , 32w , 3 = 11. Again there is only one factorization, and v = 2 which violates u v . Finally, if u 4 we have 1 , 1=uvw = 1=u + 1=v + 1=w 3=4 so uvw 4, which is impossible. II. Solution by Michael Lambrou, University of Crete, Crete, Greece. This solution is so much in keeping with the spirit of the problem that the editor felt a need to share it with all CRUX readers. In the holy name of the Almighty," said the distinguished problem solver who saw the CRUX of the matter, if you add my humble camel to your herd then there will be uvw camels to share. Your kind selves, whose renowned hospitality o ered me your well to quench my thirst, will receive uv, vw, and wu camels respectively. This amply ful ls the deceased's will, whose soul may rest in tranquility." So," interrupted the sophisticated calculator, a student of logistics the art of the practical arithmetician, perpetuator of the Pythagorean doctrine that whole numbers are the essence of nature, we must then have that uv + vw + wu +1 equals uvw, since numbers are the balance of ideas, the epitome of fairness, and we must take into account the camel of our distinguished guest. We cannot let him leave our oasis for his long journey to redeem his pilgrimage pledge, without a fair chance to cross the desert." The sophisticated calculator, well versed in the new art of Aljabr, could
th th th
2226.
187 easily rewrite the condition as
product of numbers, one of which is nothing!" He was perplexed although he had seen this `nothing' number in Ptolemy's Almagest, the monumental astronomical work, in the Table of Chords in Chapter One. I do not understand," he continued. How can nothing exist? It is contrary to nature. Nature abhors void because it would make motion impossible, as falling bodies would have to have in nite speed." On the contrary," said the distinguished problem solver, in my long travels I have heard that the wise men of the east have discovered a nothing number. They call it `assifr', and it comes from the Sanskrit `sunya'. It has the property that when multiplied by anything it gives assifr. So, mathematically speaking, we have to exclude equality of any of u, v , w to 1 because this would contradict the equation: u + v + w = u , 1v , 1w , 1." Perhaps mathematically we have to exclude this case but we have an inheritance problem," answered the calculator, trying to gain time, and this nothing is not, philosophically speaking, on solid ground." He then turned to the respectful cadi, the assessor of values and conservator of culture, whose judgment had a Rhadamanthean wisdom. What do you say, esteemed reverend?" The cadi replied that if any of u, v , w was 1, then one of the sons would take the entire herd, which is contrary to our sacred traditions. Surely it was not the intention of their late father to incite hatred in the thoughts of the two losers. Surely he did not want to upset the good values and bonds of his family." This answer satis ed the calculator. Fine! May your shadow never be less, but let us continue the analysis. We may assume that w v u 1 since birth rites allow shares to be larger or lesser. But could u be 4 or more?" I hope not," continued the calculator, because the u part would be too small, unworthy of the respect showed to his late father. Ah yes, if u 4 then
th
u + v + w = u , 1v , 1w , 1: What?" he exclaimed. If any of u, v , w is 1, then the right hand side is a
3w u + v + w = u , 1v , 1w , 1 3 3w , 1 giving 9 6w, which cannot be, since then w = 1, but then he would
take the entire herd, inciting hatred in the thoughts of the other two, as forewarned by the incontestable cadi." So we must have u 3. Let us then see what happens if u = 3. Here w v u = 3 gives
3 + 2w u + v + w = u , 1v , 1w , 1 2 2w , 1 ;
188 that is, 7 2w, giving w = 1, 2, or 3. The cases w = 1; 2 are excluded since w u = 3, leaving 3 = w u = 3; that is, all shares equal, u = v = w = 3. But this does not satisfy the original equation and must be dropped." The dropping of the equal shares possibility came as a relief to all. It must have been Almighty's wish since not all three sons deserved an equal share. One of the three was certainly more praiseworthy spiritually, as he attended prayers and consulted often the holy book. Last but not least we have to analyze the possibility u = 2." Everybody listened carefully, especially potential brides, because u = 2 meant that one son would take half the herd. That is, as much as the other two together. Wise is the Lord! So we have which, after some Aljabr, gives
2 + v + w = 1v , 1w , 1 v , 2w = 2v + 1."
At this point the problem solver interrupted again. Observe," he said, if v = 2 then the left hand side gives assifr, which is incompatible with the right, so this case must be dropped." I will do as you say," replied the calculator, although I think there is a deeper reason for that. Harmony with nature does not allow u = v = 2 because the original equation then becomes and, if anything is added to 4, be it something of substance or void, you get at least 4 more than the addend, and not one less than the addend." We, therefore, have," he continued: This is a di cult situation. How can an integer equal a fraction? Only when what seems a fraction is not really a fraction but an integer concealed. We are rescued from this di cult situation by appealing to the ideas of Diophantus. I am so glad I have a recent manuscript with a translation of his eternal book, because the original Greek is too di cult." This is how he approaches such problems: The denominator v , 2 must be a divisor of 5, a sacred number, the number of Platonic solids and the length of the hypotenuse of the eternal triangle. Divine wisdom arranged that 5 has the prime property that it possesses precisely two divisors, unity and itself. So v is either 3 or 7. It cannot be 7 because w would then be 3, a smaller number. This leaves v = 3 and w = 7."
4+w = w,1
+ 5 w = 2vv, 21 = 2 + v , 2
189 Thus the total number of camels in the original herd is 41. One camel is left over for our guest, which he can have back, as it is not counted in the 41. The shares are 21, 14, and 6 respectively," concluded the calculator boastfully. Everybody applauded the sagacity of this artful manipulator of numbers whose eurhythmic mind interpreted the inheritance laws with the infallible ways of the mathematician. The distinguished problem solver smiled to himself. He had succeeded again. In a true Socratic manner he led the dialogue by giving imperceptible hints, the CRUX of the matter, to his counterpart who then discovered the truth that had been known to the problem solver from the very beginning. He saddled his camel, thanked for the hospitality and the knowledge he acquired, savoured a sip of water and left for the next stage of his Promethean journey. Also solved or answered by SAM BAETHGE, Nordheim, Texas, USA;
FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, and MARIA
LOPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain;
ASCENSION MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; IAN JUNE L. GARCES, Ateneo de Manila University, The Philippines and GIOVANNI MAZZARELLO, Ferrovie dello Stato, Florence, Italy; DAVID HANKIN, Hunter College Campus Schools, New York, NY, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; JOHN G. HEUVER, Grande Prairie Composite High School, Grande Prairie, Alberta; D. KIPP JOHNSON, Beaverton, Oregon, USA; GERRY LEVERSHA, St Paul's School, London, England; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; REZA SHAHIDI, student, University of Waterloo, Waterloo, Ontario; D.J. SMEENK, Zaltbommel, the Netherlands; DAVID STONE and VREJ ZARIKIAN, Georgia Southern University, Statesboro, Georgia; PANOS E. TSAOUSSOGLOU, Athens, Greece; PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer. There were two incomplete solutions. Lambrou also considers the general problem where the number of camels is p , 1, and u, v , w all divide evenly into p. In addition to the solution given above he nds 13 other solutions u; v;w; p, where w v w: 2; 3; 8; 24, 2; 3; 9; 18, 2; 3; 10; 15, 2; 3; 12; 12, 2; 4; 5; 20, 2; 4; 6; 12, 2; 4; 8; 8, 2; 5; 5; 10, 2; 6; 6; 6, 3; 3; 4; 12, 3; 3; 6; 6, 3; 4; 4; 6, 4; 4; 4; 4:
190
2229. 1997: 167 Proposed by Kenneth Kam Chiu Ko, Mississauga, Ontario. a Let m be any positive integer greater than 2, such that x 1 mod m whenever x; m = 1. Let n be a positive integer. If mjn +1, prove that the sum of all divisors of n is divisible by m. ? Find all possible values of m. b
2
Solution by KeeWai Lau, Hong Kong modi ed by the editor. a We rst show that n cannot be a perfect square. Suppose that n = k . Then k ,1mod m. But kjn, mjn + 1 and n; n + 1 = 1 together imply that k; m = 1, and so, k 1mod m. Thus 1 ,1mod m, which is false since m 2. Therefore, all the divisors of n can be grouped into pairs s; t, where st = n and s 6= t. It then su ces to show that mjs + t. As above, s; m = 1 implies that s 1mod m. Adding st ,1mod m, we have that ss + t 0mod m, or s + t 0mod m. b We show that the possible values of m are precisely 3, 4, 6, 8, 12, and 24. For each m such that 3 m 24, direct checking of those x with 1 x m and x; m = 1 reveals that these values are indeed the only ones that satisfy the described condition. Assume then that m 24 and let p , p , p , : : : , denote the sequence of prime numbers. Then 2jm, for otherwise 2;m = 1 implies that 2 1mod m, which is false. Similarly, 3jm and 5jm. If 7; m = 1, then 7 1mod m, or mj48, which is impossible since 5jm. Thus 7jm. Suppose that pi jm for all i = 1, 2, : : : , k, for some k 4. If Qk m; pk = 1, then pk 1mod m, which implies that pk i pi. However, this contradicts the Bonse Inequality, which states that for all Qk k 4, pk i pi. See, for example, chapter 27 of The Enjoyment of Mathematics by H. Rademacher and O. Toeplitz; Dover, 1990. It follows that pk jm and so m is divisible by any prime, which is clearly impossible. This shows that if m 24, we cannot have x 1mod m whenever x; m = 1, and the proof is complete. Also solved by ADRIAN BIRKA, student, Lakeshore Catholic High School, Port Colbourne, Ontario; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; D. KIPP JOHNSON, Beaverton, Oregon, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; and HEINZJURGEN SEIFFERT, Berlin, Germany.
2 2 2 2 1 2 3 2 2 +1 2 +1 2 +1 =1 2 +1 =1 +1 2
191 Part a only was solved by JIMMY CHUI, student, Earl Haig Secondary School, North York, Ontario; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; SEAN MCILROY, student, University of British Columbia, Vancouver, BC; JOEL SCHLOSBERG, student, Hunter College High School, New York, NY, USA; and the proposer. Regarding the solution to b, the Bonse Inequality was also used, explicitly or implicitly, by Boase, Bradley, Herzig and Sei ert. This inequality is an easy consequence of Bertrand's Postulate as shown by Herzig and Seiffert. Johnson gave a solution using Bertrand's Postulate directly.
2230. 1997: 167 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. Triangles BCD and ACE are constructed outwardly on sides BC and CA of triangle ABC such that AE = BD and BDC + AEC = 180 . The point F is chosen to lie on the segment AB so that
Prove that
AF = DC : FB CE
DE = EF = FD : CD + CE BC AC
Solution by Toshio Seimiya, Kawasaki, Japan. Let G be a point on AE produced beyond E such that ECG = DCB . Since CEG = 180 , AEC = CDB , we have 4CEG 4CDB , directly similar from which we have 4CBG 4CDE . Thus
BGC = DEC: Since AF = CD = BD = AE ; we get FE k BG, so that FB CE EG EG AGB = AEF:
Hence we have from 1 and 2
1 2
FED =
= = = =
AEC , AEF + DEC AEC , AGB + BGC AEC , AGC ECG BCD: FDE = ACE:
3
Similarly we have
4
192 Let H be a point on CD produced beyond D such that DH = EC . Since BD = AE and BDH = 180 , BDC = AEC , we have
4BDH = 4AEC; so that BH = AC , and BHD = ACE: As FED = BCD = BCH , and FDE = ACE = BHD = BHC , we have 4FDE 4BHC .
DE = EF = FD : HC BC BH Since HC = CD + DH = CD + CE , and BH = AC , we have DE = EF = FD : CD + CE BC AC
Thus we get
Also solved by FLORIAN HERZIG, student, Perchtoldsdorf, Austria;
ISTVAN REIMAN, Budapest, Hungary; and the proposer.
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
193
THE ACADEMY CORNER
No. 19 Bruce Shawyer
All communications about this column should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7
Memorial University Undergraduate Mathematics Competition
September 25, 1997
Solutions by Solomon W. Golomb, USC, Los Angeles, CA, USA, who writes: It's reassuring to know that I can still do freshman high school mathematics, after all these years". 1. Determine whether or not the following system has any real solutions. If so, state how many real solutions exist. Solution. If 1 is to have real solutions, then xyz 6= 0. Hence
1 x + x = y;
1 y + y = z;
1 z + z = x:
1
so there are no real solutions. 2. The surface area of a closed cylinder is twice the volume. Determine the radius and height of the cylinder given that the radius and height are both integers. Solution. The surface area of a closed cylinder is twice the volume" is absurd. It is dimensionally incorrect. One has units of area, the other has units of volume. What was no doubt intended was 2rh + 2r2 = 2r2 h, h + r = rh, 1 + 1 = 1, with no integer solution except r = h = 2. But what are r h the units in which h + r = rh? No physical object corresponds to this solution, independent of the arbitrary choice of units.
x2 + 1 = xy; y2 + 1 = yz; z2 + 1 = zx: , , , Multiplying gives x2 + 1 y 2 + 1 z 2 + 1 = x2 y 2z 2 . But, for real x, y , z , we have x2 + 1 x2 , y 2 + 1 y 2, z 2 + 1 z 2,
194
3. Prove that
Solution a.
1 1 1 + 4 + 1 + + n2 9 1 1 1 + 1 + 9 + + n2 4
2:
1 X1 2 k=1 k 2 10 = 6 6 5 2: = 3
Solution b. For those who do not know the value of
2.
1 1 1 + 1 + 9 + + n2 4
Z n+1 dt 1+ 2 Z11 dtt 1 1+ = 1, 1 t1 1 t2 = 1 + 1 = 2:
4. Describe the set of points x; y in the plane for which
sinx + y = sin x + sin y: Solution. Since sinx + y = sin x cos y + sin y cos x, it follows that sinx + y = sin x + sin y has no solutions in the rst quadrant, in which cos x 1, cos y 1, so that sinx + y sin x + sin y .
There are no solutions in the third quadrant, in which
sinx + y = sin x cos y + sin y cos x 0 sin x + sin y: The line y = ,x, which bisects the second and fourth quadrants, is clearly a solution line: sinx + ,x = 0 and sin x + sin,x = 0.
To show that there are no other solutions, observe: any solution x; y in the second or fourth quadrant corresponds to a rstquadrant solution of sinx , y = sin x , sin y . If x 6= y , suppose without loss of generality that x y , so that sin x sin y while cos y cos x. But then
sinx , y = sin x cos y , sin y cos x sin x , sin y:
195
5. In a parallelogram ABCD, the bisector of angle ABC intersects AD
All angles are equal, because BCP is an isosceles triangle, BP bisects y 6 6 ABC , and alternate interior angles are equal. Thus BPA is also an isosceles triA P 5 D y angle, similar to triangle BCP , and z = 5 + y because BC = AD. 6 From similar triangles, we see that z = y , 36 = yz = y 5 + y , and 6 2 + 5y , 36 = 0. By the quadratic formula, the positive root is y = 4, y which is the length of side AB . Triangle CDP is a 4 5 6 triangle! 6. Show that, where k + n m,
n X n
m
m + n
= n+k : i=0 i k + i m + n
Solution. With k + n m, we represent n + k by taking j elements from n, and the remaining n + k , j elements from m, for every j , 0 j n. Thus m + n
n X n
m
= n+k j =0 j n + k , j 0 X n
m
= i=n n , i k + i n X n
m
= ; i=0 i k + i n
n
where we substituted i = n , j , and used a = n,a .
at P . If PD = 5, BP = 6 and CP = 6, nd AB . Solution. B z C We label the parallelogram as shown.
Solutions were also received from D. KIPP JOHNSON, Beaverton, Oregon, USA and D.J. SMEENK, Zaltbommel, the Netherlands.
196
THE OLYMPIAD CORNER
No. 190 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. Since the summer break" is coming up, we give three Olympiad contests from three di erent parts of the world. My thanks go to Bill Sands for collecting the contest materials for me when he was helping to coordinate marking of the IMO held in Toronto in 1995. We rst give the problems of the Grade XI and Grade XII versions of the Lithuanian Mathematical Olympiad.
44th LITHUANIAN MATHEMATICAL
OLYMPIAD 1995
GRADE XI
1. You are given a set of 10 positive integers. Summing nine of them in ten possible ways we only get nine di erent sums: 86, 87, 88, 89, 90, 91, 93, 94, 95. Find those numbers. 2. What is the least possible number of positive integers such that the sum of their squares equals 1995? Replace the asterisks in the 3. equilateral triangle" by the numbers 1, 2, 3, 4, 5, 6, 7, 8, 9 so that, starting from the second line, each number is equal to the absolute value of the di erence of the nearest two numbers in the line above.
Is it always possible to inscribe the numbers 1; 2; : : : ; n, in the way required, into the equilateral triangle with the sides having n asterisks? 4. A function f : N ! N is such that
f f m + f n = m + n for all m; n 2 N N = f1; 2; : : : g denotes the set of all positive integers.
Find all such functions.
197
5. In a trapezium ABCD, the bases are AB = a, CD = b, and the diagonals meet at the point O. Find the ratio of the areas of the triangle ABO and trapezium.
GRADE XII 1. Consider all pairs x; y of real numbers satisfying the inequalities ,1 x + y 1; ,1 xy + x + y 1: Let M denote the largest possible value of x. a Prove that M 3. b Prove that M 2. c Find M . 2. A positive integer n is called an ambitious number if it possesses the following property: writing it down in decimal representation on the right of any positive integer gives a number that is divisible by n. Find:
a the rst 10 ambitious numbers; b all the ambitious numbers. 3. The area of a trapezium equals 2; the sum of its diagonals equals 4. Prove that the diagonals are mutually orthogonal. 4. 100 numbers are written around a circle. Their sum equals 100. The sum of any 6 neighbouring numbers does not exceed 6. The rst number is 6. Find the remaining numbers. 5. Show that, at any time, moving both the hourhand and the minutehand of the clock symmetrically with respect to the vertical 6 , 12 axis results in a possible position of the clockhands. How many straight lines containing the centre of the clockface possess the same property? Next we give the problems of the Korean Mathematical Olympiad.
8th KOREAN MATHEMATICAL OLYMPIAD
First Round
Morning Session  2.5 hours
1. Consider nitely many points in a plane such that, if we choose any three points A, B , C among them, the area of 4ABC is always less than 1. Show that all of these nitely many points lie within the interior or on the boundary of a triangle with area less than 4. 2. For a given positive integer m, nd all pairs n; x; y of positive integers such that m, n are relatively prime and x2 + y 2m = xy n, where n, x, y can be represented by functions of m.
198
3. Let A, B, C be three points lying on a circle, and let P , Q, R be the midpoints of arcs BC , CA, AB , respectively. AP , BQ, CR intersect BC , CA, AB at L, M , N , respectively. Show that
For which triangle ABC does equality hold? 4. A partition of a positive integer n is a sequence 1; 2; : : : ; k of positive integers such that 1 + 2 + + k = n and 1 2 k 1. Each i is called a summand. For example, 4; 3; 1 is a partition of 8 whose summands are distinct. Show that, for a positive integer m with n 1 mm +1, the number of all partitions of n into distinct m summands 2 1 is equal to the number of all partitions of n , 2 mm + 1 into r summands r m.
AL + BM + CN 9: PL QM RN
5. If we select at random three points on a given circle, nd the probability that these three points lie on a semicircle. 6. Show that any positive integer n 1 can be expressed by a nite sum of numbers satisfying the following conditions: i they do not have factors except 2 or 3; ii any two of them are neither a factor nor a multiple of each other. That is,
n=
N X
Afternoon Session  2.5 hours
where i, i i = 1; 2; : : : ; N are nonnegative integers and, whenever i 6= j , the condition i , j i , j 0 is satis ed. 7. Find all real valued functions f de ned on real numbers except 0 such that
i=1
2 i3 i;
8. Two circles O1, O2 of radii r1; r2 r1 r2, respectively, intersect at two points A and B . P is any point on circle O1 . Lines PA, PB and circle O2 intersect at Q and R, respectively. i Express y = QR in terms of r1 , r2, and = APB . ii Show that y = 2r2 is a necessary and su cient condition that circle O1 be orthogonal to circle O2 .
1 f ,x + f 1 = x; x 6= 0: x x
199
Final Round
p2 1 + p2 jaj m; jbj m; 0 a + b m+2 : 2. Let A be the set of all nonnegative integers. Find all functions f : A ! A satisfying the following two conditions: i for any m; n 2 A, 2f m2 + n2 = ff mg2 + ff ng2; ii for any m; n 2 A with m n, f m2 f n2:
isfying
1. For any positive integer m, show that there exist integers a, b sat
First Day  4.5 hours
3. Let 4ABC be an equilateral triangle of side length 1, let D be a point on BC , and let r1, r2 be inradii of triangles ABD, ADC , respectively. Express r1r2 in terms of p = BD, and nd the maximum of r1 r2. 4. Let O and R be the circumcentre and the circumradius of 4ABC , respectively, and let P be any point on the plane of ABC . Let perpendiculars PA1, PB1, PC1, be dropped to the three sides BC , CA, AB. Express
A1B1C1 ABC in terms of R and d = OP , where ABC denotes the area of 4ABC . 5. Let p be a prime number such that i p is the greatest common divisor of a and b; ii p2 is a divisor of a. Prove that the polynomial xn+2 + axn+1 + n + a + b cannot be decomposed into the product of two polynomials with bx
integral coe cients, whose degrees are greater than one. 6. Let m, n be positive integers with 1 n m , 1. A box is locked with several padlocks, all of which must be opened to open the box, and all of which have di erent keys. A total of m people each have keys to some of the locks. No n people of them can open the box but any n + 1 people can open the box. Find the smallest number l of locks and in that case nd the number of keys that each person has.
Second Day  4.5 hours
200 Now, problems selected from the 1995 Israel Mathematical Olympiads.
ist di and dj among them, such that the denominator of the reduced fraction di dj is at least n. 2. Two players play a game on an in nite board that consists of 1 1 squares. Player I chooses a square and marks it with O. Then, Player II chooses another square and marks it with X . They play until one of the players marks a whole row or a whole column of 5 consecutive squares, and this player wins the game. If no player can achieve this, the result of the game is a tie. Show that Player II can prevent Player I from winning. 3. Two thieves stole an open chain with 2k white beads and 2m black beads. They want to share the loot equally, by cutting the chain to pieces in such a way that each one gets k white beads and m black beads. What is the minimal number of cuts that is always su cient? 4. and are two given circles that intersect each other at two points. Find the geometric locus of the centres of all circles that are orthogonal to both and . 5. Four points are given in space, in a general position that is, they are not contained in a single plane. A plane is called an equalizing plane" if all four points have the same distance from . Find the number of equalizing planes. 6. n is a given positive integer. An is the set of all points in the plane, whose x and y coordinate are positive integers between 0 and n. A point i; j is called internal" if 0 i; j n. A real function f , de ned on An , is called a good function" if it has the following property: for every internal point x, the value of f x is the mean of its values on the four neighbouring points the neighbouring points of x are the four points whose distance from x equals 1. If f and g are two given good functions and f a = g a for every point a in An which is not internal that is, a boundary point, prove that f g. 7. Solve the system
Selected Problems From ISRAEL MATHEMATICAL OLYMPIADS 1995 1. n positive integers d1; d2; : : : ; dn divide 1995. Prove that there ex
p x + logx + px2 + 1 = y y + logy + p y2 + 1 = z z + logz + z2 + 1 = x
201
8. Prove the inequality
1 + 1 + 1 + + 1 kn kn + 1 kn + 2 k + 1n , 1 n
r
n
k+1 ,1 k
!
for any positive integers k, n. 9. PQ is a diameter of a half circle H . The circle O is tangent to H from the inside and touches diameter PQ at the point C . A is a point on H and B is a point on PQ such that AB is orthogonal to PQ and is also tangent to the circle O. Prove that AC bisects the angle PAB . 10. is a given real number. Find all functions f : 0; 1 7! 0; 1 such that the equality holds for all real x
0.
1
x x2f x + f x = x + 1
Next we turn to solutions to problems posed in the February 1997 number of the Corner. Some new solutions arrived after we went to press last issue, and at least one other batch of solutions was incorrectly led and just turned up! Pavlos Maragoudakis, Pireas, Greece sent in solutions to problems 1, 3 and 4 of the Final Grade, Third Round, and also to Problem 1 of the 1st Selection Round. The misplaced solutions were from D.J. Smeenk, who gave solutions to Problem 3 of the Final Grade, 3rd Round and to Problem 3 of each of the second and third Selection Rounds. Because they are interesting and di erent, we give his two solutions to Problem 3 of the Final Grade. Look at all the 3's in the above! 1997: 78, 1998: 13 14 Latvian 44 Mathematical Olympiad. It is given that a 0, b 0, c 0, a + b + c = abc. Prove that at least one of the numbers a, b, c exceeds 17=10. Solution by D.J. Smeenk, Zaltbommel, the Netherlands. In any triangle ABC the following identity holds Hence, in this problem a, b, c can be considered to be the tangents of angles of an acutepangled triangle. At least one of the angles is at least , and 3 , 17 tan = 3 10 . 3
3.
tan + tan + tan = tan tan tan :
202 Second solution. We may suppose a b c,
abc = a + b + c 3c: p So ab 3, a b and a 3 17 . 10
Now we turn to readers' solutions of problems from the March 1997 Corner and the 3rd Mathematical Olympiad of the Republic of China Taiwan 1997: 66 .
3rd MATHEMATICAL OLYMPIAD OF THE
REPUBLIC OF CHINA Taiwan
First Day  April 14, 1994
1. Let ABCD be a quadrilateral with AD = BC and let A + B = 120 . Three equilateral triangles 4ACP , 4DCQ and 4DBR are drawn on AC , DC and DB away from AB . Prove that the three new vertices P , Q and R are collinear.
Solution by Toshio Seimiya, Kawasaki, Japan. P Q O R C D
`
A B Let O be the intersection of AD and BC . Since A + B = 120 , we get AOB = 60 . Let l be the exterior bisector of angle AOB . Since APC = 60 = AOC , we have that O, P , A, C are concyclic. Hence POA = PCA = 60 . The exterior angle of AOB is 120 , showing that PO bisects the exterior angle of AOB. Thus P lies on l. Similarly Q and R lie on l. Hence, P , Q and R are collinear. Comment. As is shown in the proof, the condition AD = BC is not necessary. 2. Let a, b, c be positive real numbers, be a real number. Suppose that
Determine the magnitude between f and g .
g = a
+2 b + c , a + b +2 a , b + c + c +2 a + b , c
f = abca + b + c
203 Solution by Panos E. Tsaoussoglou, Athens, Greece.
bca
+1 + acb +1 + abc +1 , a +2 b + c , a ,b +2a , b + c , c +2 a + b , c = a +1 bc , ab + c + a2 + b +1 ac , ba + c + b2 +c +1ab , ca + b + c2 +1 a , ba , c + b +1 b , ab , c + c +1 c , ac , b = a
0
which is an inequality of Schur. Next we move to solutions of Selected Problems from the Israel Mathematical Olympiads, 1994 1997: 131 .
SELECTED PROBLEMS FROM THE ISRAEL MATHEMATICAL OLYMPIADS, 1994 1. p and q are positive integers. f is a function de ned for positive numbers and attains only positive values, such that f xf y = xp y q . Prove that q = p2 . Solutions by Pavlos Maragoudakis, Pireas, Greece; and Michael Selby, University of Windsor, Windsor, Ontario. We give the solution of Maragoudakis. 1 , we get f y = y q=p . For x = 1=p
f y f 1 For y = 1, we get f 1 = 1, so f y = y q=p. Hence f x y q=p = xp y q . For y = z p=q we get f x z = xp z p or f x = xp . q Thus p = p, whence q = p2 . 2. The sides of a polygon with 1994 sides are ai = p4 + i2, i = 1; 2; : : : ; 1994. Prove that its vertices are not all on integer mesh points.
Solution by Michael Selby, University of Windsor, Windsor, Ontario. One assumes integer mesh points are lattice points. Assume that Pi , the ith vertex, has coordinates xi ; yi where xi , yi are ! integers. Let ,i = xi+1 , xi ; yi+1 , yi = i; i be a vector represend ! tation of the ith side. The indices are read cyclically. Then j,i j2 = a2 = d i 4 + i2. Also ,,! = x1 , x1994; y1 , y1994. d1994 We know that
1994 X
i=1
! ,i 2 = d
1994 X
i=1
4 + i2 ;
or
204
1994 X
2 + i2 = 4 1994 + 199419953989 i 6 i=1 = 4 1994 + 9976653989;
which is an odd integer. 1994 1994 1994 X X X! ! However, we know = = 0, since ,i = , . d 0 i i Therefore i + i i=1 Thus
1994 X
1994 X
i=1
!2
i=1
i=1
= 0.
i j+
i=1
0 X 2 + i2 + 2 @ i
ij i j+
X
i j
i j+
X
i j
i
1 jA = 0
and
0 X 2@
i;j
X
i j
i j+
X
i j
i
1 X ! , 2 + 2 ; jA = , i i
i
giving an odd integer on the right, and an even one on the left, a contradiction. Therefore, not all the vertices can be lattice points. 5. Find all real coe cients polynomials px satisfying for all x. Solutions by F.J. Flanigan, San Jose State University, San Jose, California, USA; and Michael Selby, University of Windsor, Windsor, Ontario. We give Flanigan's solution. We consider polynomials px with coe cients in a eld F of arbitrary characteristic and nd as follows: i If charF = 0, in particular, if F = R, then px = ax , 32, where a is any scalar possibly 0 in F; ii If charF = 2, then every px satis es the equation clear; iii If charF = an odd prime, l, then there are in nitely many solutions, including all px = ax , 32xl , x + c with a; c 2 F, and = 0; 1; 2; : : : . Note that px has the form ax , 32 if = 0. To prove this, observe that if charF 6= 2, then x , 1 and x , 3 are coprime, whence px = x , 32q x in F x . Thus our equation becomes
x , 12px = x , 32px + 2
x , 12 x , 32qx = x , 32x , 12q x + 2
205 whence q x = q x + 2, as polynomials; that is, elements of F x . Now if charF = 0, then has only constant solutions. The most elementary proof of this: without loss of generality, qx = xn + axn,1 + . Then qx + 2 , qx = 2nxn,1 + , and this is nonzero if n 1. Another proof: implies that q x is periodic, which forces equations q x = c to have in nitely many roots x, a contradiction. This establishes the assertion i. Re: assertion iii. Let charF = l and q x = xl , x + c. Then for x = 0; 1; : : : ; l , 1, that is for each element of the prime eld, we have q x = c and so q x = q x + 1 = q x + 2 = : : : , yielding polynomials of degree greater than or equal to l which satisfy . This establishes the assertion iii. We next turn to solutions to Problems From the BiNational Israel Hungary Competition, 1994 1997: 132 .
1. a1; : : : ; ak ; ak+1 ; : : : ; an are positive numbers k n. Suppose that the values of ak+1 ; : : : ; an are P How should one choose the values xed. a of a1 ; : : : ; an in order to minimize i;j;i6=j aji ? Solutions by F.J. Flanigan, San Jose State University, San Jose, California, USA; and Michael Selby, University of Windsor, Windsor, Ontario. We give Flanigan's solution. To minimize the given rational function, choose
! ak+1 + + an 1=2 = A H1=2; i = 1; 2; : : : ; k ai = 1 + + 1 ak+1 an where A is the arithmetic and H the harmonic mean of ak+1 ; : : : ; an . To prove this, we will be forgiven if we change notation: let xi = ai , i = 1; 2; : : : ; k and br = ak+r , r = 1; : : : ; m with k + m = n, and denote the given rational function F x1; : : : ; xk . Then we have F x1; : : : ; xk = X + Y + B, where X xi xj
X = xj + xi ; 1i j k X X xi br
Y = + ; 1ik 1rm br xi X br bs
B = + : 1r sm bs br
PROBLEMS FROM THE BINATIONAL ISRAELHUNGARY COMPETITION, 1994
206
00 1 0 1 1 X @@ X 1 A X A 1A @ Y = br xi + 1rm bi xi 1ik 1rm
X m = xi + mA xi i H where A is the arithmetic mean and H is the harmonic mean of the br . Now we recall that the simple function x + x with ; ; x all positive p = . Thus each of assumes its minimum when x = x ; that is x = the terms in Y and so Y itself assumes its minimum when we choose, for i = 1; 2; : : : ; k, s p m xi = m=A = AH; H
Note that B is xed and Y can be improved to
as asserted. But there is more. It is also known that each term in X , and so X itself assumes its minimum when xi = xj , with 1 i j k. Thus choosing p all xi = AH minimizes both X and Y and, since B is xed, minimizes F x1; : : : ; xk as claimed. Comments: 1 It is unusual to minimize the sum of two terms in the same variables by minimizing each term simultaneously. p 2 When m = 2, then AH = G, the geometric mean of b1, b2. p 3 Fmin = kk , 1 + 2kn , k2 A=H + B .
3. m, n are 2 di erent natural numbers. Show that there exists a real 2 number x, such that 1 fxng 3 and 1 fxmg 2 , where fag is the 3 3 3 fractional part of a. Solution by F.J. Flanigan, San Jose State University, San Jose, California, USA. We work in the rst quadrant of the standard uv plane, studying the ray
The key is to observe that the problem is equivalent to showing that the ray R contacts at least one of the small" 1 by 1 squares in the centres of the 3 3 large standard 1 by 1 lattice squares considered as a 3 3 checkerboard. For if xm; xn lies in one of these small squares then fxmg; fxng lies in 1 2 the small square fu; v : 3 u; v 3 g closest to the origin, as desired.
R = fu; v = xm; xn : x 0g:
207
n 0 n m, so that R is given by v = m u, u 0. 1 1 Consider the sequence of rays v = 2 u, v = 1 u, v = 1 u, v = 11 u; : : : , 5 8 with u 0. These rays are determined by the lower right corners of the 1st, 2nd, 3rd, 4th ; : : : , small central square.
To establish this contact, we assume, without loss of generality, that
v=u
1 1 2 3
1 v = 2 u typicalR v = 1u 5
1 1 v = 8 u v = 11 u
4
It is now apparent that our ray R lies between, or on the ray v = u 1 and v = 1 u, or v = 1 u and v = 5 u, or : : : , and hence R will contact the 2 2 rst or the second or : : : , small square, as required. Comment: We can now estimate the least x for given m, n in the sequence 1; 1 ; 1 ; 1 ; : : : of slopes and use this to determine which small square 2 5 8 is the rst to be contacted by the ray R. From this one can estimate the coordinates of xm; xn in various ways. 4. An nm society" is a group of n girls and m boys. Show that there exist numbers n0 and m0 such that every n0 m0 society contains a subgroup of 5 boys and 5 girls in which all of the boys know all of the girls or none of the boys knows none of the girls. Solution by Michael Lebedinsky, student, Henry Wise Wood School, Calgary, Alberta. We will show that we can take n0 = 9. For n0 9, observe that for each girl there must be at least 5 boys whom she knows, or 5 boys whom she does not know. We associate to each girl an ordered pair, the rst element of which is a subset of 5 of the boys all of whom she knows or all of whom she does not know, and the second element of which is 0 or 1 according as , she knows the boys or not. There are 9 2 = 252 such pairs. Invoking the 5 Pigeonhole Principle, if m0 4 252 + 1 = 1009, at least 5 girls must be assigned the same ordered pair, producing 5 girls and 5 boys for which each girl knows each boy, or no girl knows any of the boys. That completes this Corner for this issue. Enjoy solving problems over the next weeks  and send me your nice solutions as well as your Olympiad materials.
208
BOOK REVIEWS
Edited by ANDY LIU
Mathematical Challenge, by Tony Gardiner, published by Cambridge University Press, 1996, ISBN 0521558751, softcover, 138+ pages. More Mathematical Challenges, by Tony Gardiner, published by Cambridge University Press, 1997, ISBN 0521585686, softcover, 140+ pages. Reviewed by Ted Lewis, University of Alberta.
These two books are companion volumes which cover the recent phenomenal development in mathematics competitions in the United Kingdom. Although the country has a long and distinguished history in this endeavour, featuring the famed Cambridge Tripos, it was not until the late 1980's when a popularization movement, under the leadership of the author Tony Gardiner, made it a truly national event. This grassroots approach has nurtured the young talents into a force to be reckoned with consistently in the International Mathematical Olympiad. The rst book contains the problems for the UK Schools Mathematical Challenge Papers from 1988 to 1993. The target audience are children aged 12 to 14. Each paper consists of 25 multiplechoice questions, of which the rst 15 are relatively straightforward. The paper is to be attempted in one hour, and students are not expected to nish it. Random guessing is discouraged, and calculators are forbidden. Answers, statistics and brief comments are given, but the reader has to work out the solutions. The second book contains the problems for the UK Junior Mathematical Olympiad from 1989 to 1995. The target audience is children aged 11 to 15. Except in 1989, each paper consists of 10 problems in Section A and 6 problems in Section B. The 1989 paper consists of 13 problems, and may be regarded as a long Section B. The following paragraph is quoted from page 2 of the book. Section A problems are direct, closed" problems, each requiring a speci c calculation and having a single numerical answer. Section B problems are longer and more open". Thus, while the nal mathematical solution is often quite short, and should involve a clear claim, followed by a direct deductive calculation or proof, there will generally be a preliminary phase of exploration and conjecture, in which one tries to sort out how to tackle the problem. There is a section titled Comments and hints" and another titled Outline solutions". Thus a student who is frustrated by a particular problem has several recourses for assistance. The outline solutions are precisely that, with gaps for the reader to ll in, but revealing enough details to guide a diligent student to the full solution. We now present some sample problems.
1988 14.
Weighing the baby at the clinic was a problem. The baby would not keep still and
209
caused the scales to wobble. So I held the baby and stood on the scales while the nurse read o 79 kg. Then the nurse held the baby while I read o 69 kg. Finally I held the nurse while the baby read o 137 kg. What is the combined weight of all three in kg? a 142 b 147 c 206 d 215 e 284
1993 18.
Sam the super snail is climbing a vertical gravestone 1 metre high. She climbed at a steady speed of 30 cm per hour, but each time the church clock strikes, the shock causes her to slip down 1 cm. The clock only strikes the hours, so at 1 o'clock she would slip back 1 cm, at 2 o'clock she would slip back 2 cm and so on. If she starts to climb just after the clock strikes 3 pm, when will she reach the top? a 3:50 pm b 6:20 pm c 6:50 pm d 7:04 pm e 7:20 pm
1994 A6. 1995 B6.
A normal duck has two legs. A lame duck has one leg. A sitting duck has no legs. Ninety nine ducks have a total of 100 legs. Given that there are half as many sitting ducks as normal ducks and lame ducks put together, nd the number of lame ducks. I write out the numbers 1, 2, 3, 4 in a circle. Starting at 1, I cross out every second integer till just one number remains: 2 goes rst, then 4, leaving 1 and 3; 3 goes next, leaving 1  so 1" is the last number left. Suppose I start with 1, 2, 3, : : : , n in a circle. For which values of n will the number 1" be the last number left?
Comments and hints for 1995 B6.
The question For which values of n : : : ?" is more complicated than it looks. It is not enough to answer n = 4" even though n = 4 works, since we clearly want to know all possible values of n that work. Hence a solution must not only show that certain values of n work, but must also somehow prove that no other values of n could possibly work.
Outline solution of 1995 B6.
The easy step is to realize that, if n is odd and 3 then 1 will be crossed out at the start of the second circuit. Hence, for 1 to survive, either n = 1 or n must be e?e? say n = 2m. Suppose n = 2m. Then at the start of the second circuit, there remain exactly m numbers 1, 3, 5, : : : , 2m , 1. For 1 to survive next time it is essential that either m = 1 so n = 2 or m must be e?e? say m = 2p. Continuing in this way we see that 1 will be the last uncrossed number if and only if n is a ?o?e? of ??o. Ed. here ? indicates some number. As a bonus, the rst book provides 420 additional questions, and the second 40 Section A problems and 20 Section B problems. For one reason or another, these did not make it into the competition papers, but are nevertheless excellent questions. A lot of thought has gone into the planning of these two books. They are to be done actively rather than read passively. The problems are the important features, not the solutions. Particularly valuable are the author's comments on how things are done and why. This dynamic package is a must for anyone interested in mathematics competitions for youngsters.
210
A Note on Special Numerals in Arbitrary Bases
Glenn Appleby, Peter Hilton and Jean Pedersen
1 Introduction
In 1 , at the end of a section devoted to the Pigeonhole Principle, the author posed the following problem: 4.2.28 Show that, for any integer n, there exists a multiple of n that contains only the digits 7 and 0." Plainly the author is referring to the representation of the multiple as a numeral in base 10. Thus the reader of the problem is encouraged to believe that the validity of the statement depends in some way on the numbers 7 and 10 and their mutual relationship. Our analysis of the problem shows that, in fact, we may replace 10 by any base b 2, and 7 by any digit t in base b. Of course, we assume t 6= 0 to avoid absurd triviality. Moreover, we prefer to restrict ourselves to numerals consisting of a sequence of k t's followed by l 0's; and, if we assume, as we may without real loss of generality, that n 0, then we have k 1, l 0. We then give two arguments for the conclusion. The rst does employ the Pigeonhole Principle but does not give us a means of calculating all pairs k;l from the data. Our second approach yields the minimal pair k; l and shows, as expected, that an arbitrary pair K; L in the set of solutions is obtained from the minimal pair by taking K to be an arbitrary multiple of k, and L an arbitrary integer satisfying L l. We prefer to replace the modulus n of the problem as stated by m, thus freeing n to stand for the multiple we are seeking. Also we point out that, of course, the analysis we carry out does not depend on the condition t b imposed by the requirement that t be a digit in base b. Thus the number n we seek may be represented, for arbitrary t and base b, by the expression
k l n = t b b , 1b : ,1
1.1
2 The two main arguments
Let b be an arbitrary base and let t, 0 digit in base b.
t b , 1, be an arbitrary nonzero
Copyright c 1998 Canadian Mathematical Society
211
Theorem 1 For any integer m 0, there exists a positive integer n, written as a numeral in base b consisting of a sequence of t's followed by a possibly empty sequence of 0's, such that n 0 mod m. We give two proofs, the rst uses the Pigeonhole Principle, the second being an exercise in modular arithmetic. Proof A. Consider the sequence of numerals 0, t, tt, ttt, : : : . The remainders mod m of these integers lie in a set containing m elements. Thus after at most m + 1 such numerals, some remainder must have been repeated. If ttt : : : k + l occurrences, and ttt : : : , l occurrences, k 0 yield the same remainders, then the numeral ttt : : : 000 : : : k occurrences of t, l zeros ful ls the conclusion of the theorem. This elegant proof has the single defect that it gives no clue as to how one nds the minimal values of k and l as functions of b, t and m. Our second proof is not so neat, but gives more information. Proof B. Let z = b , 1. We show rst how to nd a smallest number n = nz , whose numeral in base b consists of a sequence of z 's followed by a sequence of 0's, such that n 0 mod m. Write m = vw, where v is prime to b and, for every prime p, pjw implies that pjb. Note that this factorization of m is unique. Let k be the order of b mod v and let l be the smallest nonnegative integer such that wjbl. Then the number n, represented in base b by a sequence of k z's followed by a sequence of l 0's, is obviously a positive integer nz of the form
nz = zzz : : : 000 : : : 2.1 divisible by m. We will show below that nz , given by 2.1, is minimal for
this property. It is now a trivial matter to complete the proof of the theorem. For if we replace m by mz in the argument above, we nd a number nz = zzz : : : 000 : : : such that nz 0 mod mz, so that n1, given by zn1 = nz gives us
n1 = 111 : : : 000 : : : 0 mod m:
2.2 2.3
But then
n = tn1 = ttt : : : 000 : : : 0 mod m:
However, if we want to make sure we have the smallest number n of the required form, we should proceed more cautiously.
212 We rst prove that if n = nz is chosen as in 2.1, then it is the smallest integer of the given form to satisfy n 0 mod m. Now if n is of the form zzz : : : 000 : : : , with K z 's and L 0's, then see 1.1 n = bK , 1bL. Moreover, since v , w are coprime, we have
n 0 mod m n 0 mod v and n 0 mod w: Since v is prime to b, n 0 mod v bK , 1 0 mod v kjK: Further, since, for all primes p, pjw implies that pjb, we have that w is prime to bK , 1, so that n 0 mod w bL 0 mod w L l: Now suppose that n is chosen to be the smallest integer of the form zzz : : : 000 : : : such that n 0 mod mz, and let n = zn1. Then since zn1 0 mod mz n1 0 mod m; 2.4 it follows that n1 , given by 2.2, is the smallest integer of the form 111 : : : 000 : : : to satisfy n1 0 mod m. We want now to nd the smallest integer nt represented in base b by a sequence of t's followed by a sequence of 0's compare 2.3 to satisfy nt 0 mod m. Our recipe for constructing nt is as follows. Let d = gcdm;t, m = m0d, t = t0 d. Then if n1 has the form 111 : : : 000 : : : , tn1 0 mod m t0 n1 0 mod m0 n1 0 mod m0: 2.5 Thus we construct n as in 2.1 to be minimal satisfying n 0 mod m0 z . Then n = zn1, and, by 2.4, we have that n1 is minimal of the required form to satisfy n1 0 mod m0 , so that nally, by 2.5, we have that nt = tn1 is minimal of the form ttt : : : 000 : : : to satisfy nt 0 mod m.
form
Example 1. Let us work in base b = 10 and look for the smallest number n6 of the
n6 = 666 : : : 000 : : : divisible by 99. We have b = 10, z = 9, m = 99, t = 6, so d = 3, m0 = 33, m0 z = 297. To nd n, minimal of the form 999 : : : 000 : : : , to satisfy n 0 mod 297, we factorize 297, as in the construction above, as 297 = 297 1, since 297 is prime to 10. We nd k = 6 that is, the order of 10 mod 297 is 6 and, of course, l = 0. Thus n = 999999, so n1 = 111111, n6 = 666666.
213
nt, we simply 0 nd the minimal n of the form zzz : : : 000 : : : to satisfy n 0 mod m z, and then replace z by t in the numeral for n. Indeed, all we have to do is to nd the minimal values of k and l which yield n see
Section 3. It is also plain that, having obtained our minimal nt , involving k t's followed by l 0's, we obtain all solutions of the congruence N 0 mod m of the required form by taking K t's followed by L 0's, where kjK and L l. Notice that this implies that every solution of the congruence N 0 mod m, of the required form, is a multiple of the minimum solution.
Example 1 brings out the important practical point that, to nd
3 The algorithm
We extract from the analysisin Section 2 the algorithm for nding the minimal pair k;l as functions of b, t and m. Given: b base, t digit and m modulus. Set gcdm; t = d, and m = m0 d. Write m0 b , 1 = vw where v is prime to b and, for all primes p,
pjw = pjb: Finally, let k be the order of b mod v that is, bk 1 mod v , with k positive minimal and let l be minimal such that wjbl. Then k  z times zl times  nt = ttt : : : 000 : : : is the minimal numeral n, in base b, consisting of a sequence of k t's followed by a sequence of l 0's, such that mjn.
Example 2. Given: b = 7, t = 5 and m = 2499. gcd2499; 5 = 1, so that m = m0 . Write 2499 6 = 306 49 thus, v = 306, w = 49. Finally, let k be the order of 7 mod 306 that is, 7k 1 mod 306, with k positive minimal, and let l be minimal such that 49j7l . Using modular arithmetic1 we see that k = 48, and it is clear that l = 2. Thus
1 306 = 2 9 17. The order of 7 mod 2 is 1; the order of 7 mod 9 is 3; the order of 72mod 17 is 16. To see8 the last, observe that, by Fermat's Theorem, 716 1 mod 17; but 7 ,2 mod 17, so 7 ,1 mod 17. Thus the order of 7 mod 306 is 48.
48  z times n5 = 555 : : : 00
214 is the minimal numeral, in base 7, consisting of a sequence of 5's followed by a sequence of 0's such that 2499jn5. Example 2 shows that it may sometimes be very tedious to apply the Pigeonhole Principle to obtain k and l. Notice that, in the special case t = b , 1, we may simplify the algorithm by cutting out the rst step and replacing m0 b , 1 by m in the second step.
Acknowledgment The authors would like to thank Professor Robert Bekes for bringing the original problem in 1 to their attention. Reference 1 Zeitz, Paul, The Art and Craft of Problem Solving, John Wiley & Sons, 1998. Addresses
Glenn Appleby Department of Mathematics Santa Clara University Santa Clara, California 950530290 email: glenn@mathrs.scu.edu Peter Hilton Department of Mathematics University of Central Florida Orlando, Florida 328161364 and Department of Mathematics and Computer Science University of New England Armidale, NSW 2351 Australia Jean Pedersen Department of Mathematics Santa Clara, California 950530290 email: jpedersen@scuacc.scu.edu and Department of Mathematics and Computer Science University of New England Armidale, NSW 2351 Australia
215
THE SKOLIAD CORNER
No. 30 R.E. Woodrow
In the last number, we gave the problems of Part I of the Alberta High School Mathematics Competition. Here are the o cial" solutions.
ALBERTA HIGH SCHOOL MATHEMATICS COMPETITION Part I  November, 1996 1. An eightinch pizza is cut into three equal slices. A teninch pizza is cut into four equal slices. A twelveinch pizza is cut into six equal slices. A fourteen inch pizza is cut into eight equal slices. From which pizza would you take a slice if you want as much pizza as possible? Solution. b The area of a circle is proportional to the square of its radius. It follows that we are comparing 16=3, 25=4, 36=6 and 49=8. 2. One store sold red plums at four for a dollar and yellow plums at three for a dollar. A second store sold red plums at four for a dollar and yellow plums at six for a dollar. You bought m red plums and n yellow plums from each store, spending a total of ten dollars. How many plums in all did you buy? + Solution. d The total expenditure is 10 = m + n + m + n = m2 n . 4 3 4 6 3. Six identical cardboard pieces are A piled on top of one another, and the result is shown in the diaD B gram. F E C
The third piece to be placed is: Solution. b Clearly, F , E , C , B and A are in that order from top to bottom. If D is pointing up, it is under A. If it is pointing down, it is under B . 4. A store o ered triple the GST in savings. A sales clerk calculated the selling price by rst reducing the original price by 21 and then adding the 7 GST based on the reduced price. A customer protested, saying that the store should rst add the 7 GST and then reduce that total by 21. They agreed on
216 a compromise: the clerk just reduced the original price by the 14 di erence. How do the three ways compare with one another from the customer's point of view? Solution. e Since 1:07 times 0:79 is 0:8103, both the customer's way and the clerk's way yield a discount approximately 19. 5. If m and n are integers such that 2m , n = 3, then what will m , 2n equal? Solution. c We have m , 2n = m , 2n +2m , n , 3 = 3m , n , 1. 6. If x is x of y, and y is y of z, where x, y and z are positive real numbers, what is z ? Solution. a Since y is y of z , z = 100. The situation is possible if and only if y = 100 also. 7. About how many lines can one rotate a regular hexagon through some angle x, 0 x 360 , so that the hexagon again occupies its original position? Solution. e The axes of rotational symmetry are the three lines joining opposite vertices, the three lines joining the midpoints of opposite sides, and the line through the centre and perpendicular to the hexagon. 8. AB is a diameter of a circle of radius 1 unit. CD is a chord perpendicular to AB that cuts AB at E . If the arc CAD is 2=3 of the circumference of the circle, what is the length of the segment AE ? Solution. b By symmetry, ACD is an equilateral triangle. Hence its centroid is the centre O of the circle. Since AO = 1, AE = AO + OE = 3=2. 9. One of Kerry and Kelly lies on Mondays, Tuesdays and Wednesdays, and tells the truth on the other days of the week. The other lies on Thursdays, Fridays and Saturdays, and tells the truth on the other days of the week. At noon, the two had the following conversation: Kerry : I lie on Saturdays. Kelly : I will lie tomorrow. Kerry : I lie on Sundays. On which day of the week did this conversation take place? Solution. b Kerry is clearly lying, and is the one who tells the truth on Saturday. Hence the conversation takes place Monday, Tuesday or Wednesday, and Kelly's statement is true only on the last day. 10. How many integer pairs m; n satisfy the equation: mm + 1 = 2n? Solution. c Either both m and m + 1 are powers of 2, or both are negatives of powers of 2. The two solutions are m;n = 1; 1 and ,2; 1.
217
11. Of the following triangles given by the lengths of their sides, which one has the greatest area? Solution. b Note that 52 + 122 = 132. With two sides of length 5 and 12, the area is maximum when there is a right angle between them. 12. If x y and x 0, which of the following numbers is never greater than any of the others? Solution. d If y 0 or y = 0, then x , y = x , jy j is the minimum. If y 0, then x + y = ,jx + y j = x , jy j is the minimum. 13. An x by y ag, with x y, consists of two perpendicular white stripes of equal width and four congruent blue rectangles at the corners. If the total area of the blue rectangles is half that of the ag, what is the length of the shorter side of each blue rectangle? Solution. a Let z be the length of the shorter side of a blue rectangle. Then the longer side has length z + y,x . From 8z z + y , x = xy , we 2 1 have z = 4 x , y + fx2 + y 2g1=2 or z = 1 x , y , fx2 + y 2g1=2 . Clearly, 4 the negative square root is to be rejected. 14. A game is played with a deck of ten cards numbered from 1 to 10. Shu e the deck thoroughly. i Take the top card. If it is numbered 1, you win. If it is numbered k, where k 1, go to ii. ii If this is the third time you have taken a card, you lose. Otherwise, put the card back into the deck at the kth position from the top and go to i. What is the probability of winning? Solution. c If card number 1 is initially in the rst or second position from the top, you will win. You can also win if it is in the third, and card 1 1 1 , number 2 is not in the rst. Hence the winning probability is 10 + 10 + 10 8 . 9 15. Five of the angles of a convex polygon are each equal to 108 . In which of the following ve intervals does the maximum angle of all such polygons lie? Solution. a The sum of the exterior angles of the ve given angles is 5180 , 108 = 360 . Hence these ve angles are the only angles of the convex polygon. 16. Which one of the following numbers cannot be expressed as the di erence of the squares of two integers? Solution. b Suppose k = m2 , n2 = m , nm + n. If k is odd, we can set m , n = 1 and m + n = k. If k = 4l, we can set m , n = 2 and m + n = 2l. However, if k = 4l + 2, then m , n and m + n cannot have the same parity.
218 As a contest this number we give the Fifteenth W.J. Blundon Contest, which was taken by students in Newfoundland and Labrador. The contest is supported in part by the Canadian Mathematical Society. My thanks go to Bruce Shawyer for forwarding me a copy.
15th W.J. BLUNDON CONTEST
February 18, 1998
1 1 log2 36 + log3 36 :
1. a Find the exact value of
b If log15 5 = a, nd log15 9 in terms of a. 2. a If the radius of a right circular cylinder is increased by 50 and the height is decreased by 20, what is the change in the volume? b How many digits are there in the number 21998 51988? 3. Solve: 32+x + 32,x = 82. 4. Find all ordered pairs of integers such that x6 = y2 + 53. 5. When one fth of the adults left a neighbourhood picnic, the ratio of adults to children was 2 : 3. Later, when 44 children left, the ratio of children to adults was 2 : 5. How many people remained at the picnic? 6. Find the area of a rhombus for which one side has length 10 and the diagonals di er by 4. 7. In how many ways can 10 dollars be changed into dimes and quarters, with at least one of each coin being used? 4 8. Solve: p2 + 10 + px + 10 = 12. 9. Find the remainder when the polynomial x135 + x125 , x115 + x5 +1 is divided by the polynomial x3 , x. 10. Quadrilateral ABCD below has the following properties: 1 The midpoint O of side AB is the centre of a semicircle; 2 sides AD, DC and CB are tangent to this semicircle. Prove that AB2 = 4AD BC . D F C E G
r r r r r r r r
A
O
B
219 Now a correction to a solution given in the October number of the Skoliad Corner. 1997:347 1995 P.E.I. Mathematics Competition. An autobiographical number is a natural number with ten digits or less in which the rst digit of the number reading from left to right tells you how many zeros are in the number, the second digit tells you how many 1's, the third digit tells you how many 2's, and so on. For example, 6; 210; 001; 000 is an autobiographical number. Find the smallest autobiographical number and prove that it is the smallest. Correction by Vedula N. Murty, Visakhapatnam, India. The answer given is 1201, but the smallest example is 1210. Editor's Note: Dyslexia strikes again! That is all for this issue of the Skoliad Corner. I need your suitable materials and your suggestions for directions for this feature.
4.
Advance Announcement
The 1999 Summer Meeting of the Canadian Mathematical Society will take place at Memorial University in St. John's, Newfoundland, from Saturday, 29 May 1999 to Tuesday, 1 June 1999. The Special Session on Mathematics Education will feature the topic The invited speakers are Ed Barbeau University of Toronto, Ron Dunkley University of Waterloo, Tony Gardiner University of Birmingham, UK, and Rita Janes Newfoundland and Labrador Senior Mathematics League. Requests for further information, or to speak in this session, as well as suggestions for further speakers, should be sent to the session organizers: Bruce Shawyer and Ed Williams CMS Summer 1999 Meeting, Education Session Department of Mathematics and Statistics, Memorial University St. John's, Newfoundland, Canada A1C 5S7
What Mathematics Competitions do for Mathematics.
220
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Shreds and Slices
An excerpt from Wilson's Theorem", by Oliver Johnson, Volume 5, Issue 2, Mathematical Mayhem: One of the more beautiful results in number theory is Wilson's Theorem. So named because Sir John Wilson didn't discover it, and Leibniz who did discover it didn't prove itLagrange did. This is of course an example of the lawyers doing the best out of everything; Wilson was a judge, and it is worth considering how many of Fermat's theorems weren't just won in lawsuits rather than proved by himself... So sue me Pierre. For the following we will need some de nitions, none of which should be too abstruse or unfamiliar. For a positive integer n, the polynomial xn , 1 has n distinct roots in the eld of complex numbers C. These roots are called the nth roots of unity, and denoted by n . An element
in n is called primitive if
k 6= 1 for 1 k n , 1; in other words, n is the smallest positive exponent k we must raise
to the power of to obtain 1. It can be shown that there are exactly n primitive nth roots of unity. If n is a prime p, then for
to be primitive, it su ces that
6= 1; in other words, any pth root of unity not equal to 1 is primitive. This turns out to be the case we are interested in.
From the Archives
Primitive Roots and Quadratic Residues, Part 1
221 Now, recall quadratic residues modulo n: these are the set of nonzero squares in modulo n. For example, the nonzero squares modulo 10 are 1, 4, 9, 16 6, 25 5, 36 6, and so on, so f1; 4; 5; 6; 9g is the set of quadratic residues modulo 10. Let An denote the quadratic residues modulo n, and Bn the remaining numbers, so B10 = f2; 3; 7; 8g. If n is a prime p, then both Ap and Bp contain exactly p , 1=2 elements, and this is where things nally get interesting. Let p be a prime, let
be a primitive pth root of unity, and let
x=
X
Then both x + y and xy are always integers. Actually x + y = ,1, and this is not hard to see. Since
p = 1, or
a2Ap
a; y =
X
b2Bp
b:
p , 1 =
, 1 ,
p,1 +
p,2 + : : : +
+ 1 = 0 and
6= 1, the latter factor must be 0. Also, Ap and Bp form a partition of f1; 2; : : : ; p , 1g. Hence, x + y =
+
2 + : : : +
p,1 = ,1: But why xy is an integer is a little deeper. Let us consider an example. For p = 7, x =
+
2 +
4 and y =
3 +
5 +
6. Then xy = ,
+
2 +
4 ,
3 +
5 +
6 =
4 +
6 +
7 +
5 +
7 +
8 +
7 +
9 +
10 = 3 +
+
2 +
3 +
4 +
5 +
6 = 2: In particular, this shows that to calculate xy , we do not need to resort
3 5 7 11 13 1 ,1 2 3 ,3 The absolute value of xy always seems to be p 1=4 the sign is the
to any messy cis notation, and that it is quite accessible by just pencil and paper. Playing around with other primes reveals the following values:
p xy
one which makes the resulting value an integer. Why is this? And what determines the sign of xy ? We encourage readers to play around with this, and to send in any interesting results. We will divulge the reasons behind this phenomenon, as well as some of the deeper theory it will lead to, in the next issue. Hint: If indeed x + y and xy are integers, then x and y are the roots of a quadratic equation with integer coe cients. What is this quadratic? What does this quadratic say about x and y ?
222
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only solutions  the next issue will feature only problems. We warmly welcome proposals for problems and solutions. We request that solutions from the previous issue be submitted by 1 February 1999, for publication in issue 4 of 1999. Also, starting with this issue, we would like to reopen the problems to all CRUX with MAYHEM readers, not just students, so now all solutions will be considered for publication.
High School Solutions
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca H221. Let P = 195 + 6605 + 13165. It is known that 25 is one of the fortyeight positive divisors of P . Determine the largest divisor of P that is less than 10; 000. Solution. If k divides a + b, then k divides a5 + b5 since the latter is a multiple of a+b. If k also divides c, then k divides a5 +b5 +c5, since k divides c5. Notice that 19 + 660 + 1316 = 1995 = 3 5 7 19. Since 3 divides 660 and 3 also divides 19 + 1316 = 1335, by our result, 3 divides 6605 + 195 + 13165 = P . Similarly, since 5 divides 660, 7 divides 1316 and 19 divides 19, we can show that P is also divisible by 5, 7 and 19. Therefore, P can be expressed in the form 3 52 7 19 R for some positive integer R, since we are given that 25 is a divisor of P . If R is prime, we know that P will have 2 3 2 2 2 = 48 divisors. Recall that if p1 , p2 , : : : , pn are distinct primes, then p1 a1 p2 a2 pn an has a1 + 1a2 + 1 an + 1 positive divisors. Thus, R must be prime; otherwise P will have more than 48 divisors, and so P = 3 52 7 19 R = 9975 R, for some large prime R. Since 13165 10005 = 1015, R will certainly be larger than 10; 000. Using Maple, one can show that R is 408; 255; 160; 853. Hence, it follows that the largest divisor of P less than 10; 000 is 9975.
223
H222. McGregor becomes very bored one day and decides to write down a three digit number ABC , and the six permutations of its digits. To his surprise, he nds that ABC is divisible by 2, ACB is divisible by 3, BAC is divisible by 4, BCA is divisible by 5, CAB is divisible by 6 and CBA is a divisor of 1995. Determine ABC . Solution by Evan Borenstein, student, Woodward Academy, College Park, Georgia. Since BCA is divisible by 5, A must be 0 or 5. But we are given that ABC is a threedigit number, so A = 5. Since ABC is divisible by 2 and CAB is divisible by 6, both B and C must be even. Since BAC is divisible by 4, 10A + C = 50 + C must be a multiple of 4. So C can be 2 or 6. If C = 2, B must be 2 or 8, since ACB is divisible by 3 and B is even. If C = 6, B must be 4. Hence, there are only three possibilities for ABC : 522, 582 and 546. And of these three, only the second one will make CBA a divisor of 1995, since 1995 = 285 7. Thus we conclude that ABC is 582. Also solved by Joel Schlosberg, student, Robert Louis Stevenson School, New York, NY, USA. H223. There are n black marbles and two red marbles in a jar. One by one, marbles are drawn at random out of the jar. Jeanette wins as soon as two black marbles are drawn, and Fraserette wins as soon as two red marbles are drawn. The game continues until one of the two wins. Let J n and F n be the two probabilities that Jeanette and Fraserette win, respectively.
1. Determine the value of F 1 + F 2 + + F 3992. 2. As n approaches in nity, what does J 2 J 3 J 4 J n approach? Solution. 1. If Fraserette wins, the balls must be drawn in one of the following three ways: red, red; red, black, red; or black, red, red. This must be the case, as otherwise two black balls will be drawn and Jeanette will win. Hence, the probability that Fraserette wins is the sum of the probabilities of each of three cases above. Thus,
2 1 2 n 1 n 2 1 F n = n + 2 n + 1 + n + 2 n + 1 n + n + 2 n + 1 n 2 2 2 = n + 1n + 2 + n + 1n + 2 + n + 1n + 2 6 = n + 1n + 2 :
224
6 6 Note that F n = n+1 , n+2 , so we can use a telescoping series to calculate the desired sum. We have
6 6
6 6
6 6
F 1 + F 2 + + F 3992 = 2 , 3 + 3 , 4 + 4 , 5 6 6
+ + 3993 , 3994
6 = 6 , 3994 2 3 = 3 , 1997 = 5988 : 1997
Thus,
n2 +3 6 2. Now J n = 1,F n = 1, n+1n+2 = n+1n,4 = n,1n+4 . n+2 n+1n+2
J 2 J 3 J 4 J n 6 3 = 1 4 2 7 5 8 4 9 n , 1 n + 4 3 4 5 6 6 7 n + 1 n + 2 , = 1 2 3 4 6 n , 2n n1 1 3 4 5 n + 6 7 8 9 n + 3 n + 4 4 5 6 7 n + 1 n + 2 12 = n n + 1 n + 3 n + 4 4 5 1 + = 10 n +3n1 4 n n+ 1 = 10 n + 3 n + 4 : n n+1 +4 3 3 Thus, as n approaches in nity, n+3 = 1 + n and n+1 = 1 + n+1 both n n 1 approach 1, and so J 2 J 3 J 4 J n approaches 10 . H224. Consider square ABCD with side length 1. Select a point M exterior to the square so that AMB is 90 . Let a = AM and b = BM . Now, determine the point N exterior to the square so that CN = a and DN = b. Find, as a function of a and b, the length of the line segment MN .
I. Solution by Adrian Chan, student, Upper Canada College, Toronto, Ontario. Let O be the center of the square. Consider a horizontal re ection through the line parallel to DA and passing through O. Let the image of M and N about this line be M 0 and N 0 , respectively. There exists a point where
225
D N
A a b
M
N0 N
D
r
A O B
M M0
C
B
C
MN intersects M 0N 0 . Let this point be K . Since K lies on both lines, this
point must lie on this horizontal line of re ection. Now, the diagram is also symmetrical about the line parallel to DC passing through O. Then K must lie on this line as well. This leaves us with K coinciding with O, since the two lines intersect at O. Now, because of the symmetry, NO = OM . That is, OM = MN . 2 Now consider quadrilateral OAMB . Since AOB is right and so is AMB , then the quadrilateral is cyclic. By Ptolemy's Theorem on this quadrilateral, we have OM AB = 1 1 AM OB + MB OA = a p p2 + b p2 , since OA = OB = p2 . Since 1 p2a + b, since OB = 1, we have OM = 22 a + b, and so MN = MN = 2OM . Thus the length of line segment MN is p2a + b. II. Solutionby Joel Schlosberg, student, Robert Louis Stevenson School, New York, NY, USA. Let BAM = x. Then a = cos x and b = sin x. Let O be the centre of ABCD and R be the midpoint of AB . Then BAM = RAM = RMA = x, and so ORM = ARO + ARM = + ,2x = 32 ,2x. 2 Now OR = RA = RM = 1 . By the cosine law on triangle ROM , we have: 2
OM 2 = OR2 + RM 2 , 2OR OM cos ORM
1 1 = 1 + 4 , 2 2 1 cos 32 , 2x 4 2
1 = 1 , 2 , sin2x 2 1 , 1 2 sin x cos x = 1 + ab: = 2 2 2
Thus using the fact that a2 + b2 = 1, we have 4OM 2 = 21 + 2ab = p = 2a + b2. Hence, 2OM = 2a + b, by taking the square root of both sides Note: a, b and OM are all positive. p Since 2OM = MN , we have MN = 2a + b.
2a2 + b2 + 2ab
226
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu A197. Calculate
Advanced Solutions
Z
sin2N + 1 d; sin ,2
2
where N is a nonnegative integer. Solution by D.J. Smeenk, Zaltbommel, the Netherlands, with minor modi cations. Let f N denote sin2N +1 . Then sin Now
f N + 1 , f N = sin2N + 3 ,sin2N + 1 : sin
sin2N + 3 , sin2N + 1 = sin2N + 1 cos 2 + cos2N + 1 sin 2 , sin2N + 1 = sin2N + 1cos 2 , 1 + cos2N + 1 sin2 = ,2 sin2N + 1 sin2 + 2 cos2N + 1 sin cos = 2 sin cos2N + 1 cos , sin2N + 1 sin = 2 sin cos2N + 2: These two equations imply f N + 1 , f N = 2 cos2N + 2. So,
Z
2
This means
,2
f N + 1 d ,
Z
2
,2
f N d = 2
Z
2
, 2
cos2N + 2 d = 0:
f N = f 0 =
Z
A198. Given positive real numbers a, b, and c such that a + b + c = 1, show that aa bb cc + ab bcca + ac bacb 1. Solution. Using the weighted AMGM inequality three times, we have the following: c a + a b + b c acbacb a+1b+c ; c+a+b b a + c b + a c abbcca a+1b+c ; b+c+a a a + b b + c c aabbcc a+1b+c : a+b+c
sin d = : , 2 sin
2
227 Adding these inequalities together gives
2 1 = a + b + c = a + bb+ cc a+ + 2 + b2 + c2 + 2ab + 2ac + 2bc = a a+b+c a bb cc + ab bc ca + ac bacb : a
A199. Let P be a point inside triangle ABC . Let A0 , B0, and C 0 be the re ections of P through the sides BC , AC , and AB respectively. For what points P are the six points A, B , C , A0 , B 0 , and C 0 concyclic?
Solution.
C0
r
r
B
r
A
r
P B0
r
A0
r
r
C
We show that AP must be perpendicular to BC . Similar arguments will show the same for BP and CP . Thus, P must be the orthocentre of triangle ABC . To see that AP is perpendicular to BC , consider angles C 0 A0 A and 0 A0 P as shown in the diagram. Then C 0 A0 A = C 0 BA = PBA = C PBF = FDP = C 0A0 P from the quadrilaterals AC 0BA0 and BFPD being concyclic and the last equality from similar triangles PFD and PC 0A0 . Thus, P lies on AA0 so AP is perpendicular to BC . A200. Given positive integers n and k, for 0 i k , 1, let
n
Sn;k;i = : j i mod k j X
Do there exist positive integers n, k 2, such that Sn;k;0, Sn;k;1, : : : , Sn;k;k,1 are all equal? Solution. The answer is NO. To see this, consider the kth roots of unity. In particular, since k 2, there is a k such that ! k = 1, ! 6= ,1. Now consider
228 the expansion of 1 + ! n:
n X n
j ! j =0 j X n
X n
= +! j 0 mod k j j 1 mod k j X n
k,1 + +! j k,1 mod k j = Sn;k;0 + Sn;k;1 ! + + Sn;k;k,1! k,1: Now if all the Sn;k;i are equal, say to A, then we have 1 + ! n = A1 + ! + + !k,1 = A 0 = 0. Thus ! = ,1, contradiction.
1 + ! n =
Challenge Board Solutions
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu
C74. Prove that the kdimensional volume of a parallelepiped in Rn spanned by vectors ~ 1, : : : , ~k is the square root of the determinant of the v v k k matrix f~i ~j gi;j . v v
Solution. First, note that by restricting to any kdimensional subspace of the nspace which contains ~1, : : : , ~k, we may assume without loss of generality v v that k = n. Let M be the n n matrix whose ith column is ~i, and let P be v the parallelepiped spanned by the ~i. Under the coordinate transformation v sending ~ to M~ , the ith elementary basis vector ~i = 0; : : : ; 1; : : : ; 0 x x e is sent to ~ i, and so transforms the unit cube 0; 1 n onto P . We nd then, v that Z Z
VolumeP =
P
1 dV =
0;1
n
j det 0j dV:
Since 0 = M , it follows that VolumeP = j det M j = det M T M 1=2, and since the i; j thentry of M T M is indeed ~i ~j , we are done. v v Remark. The above proof that VolumeP = j det M j is actually a bit bogus, since use is usually made of that result when deriving the changeofvariables formula for integration. So, for those of you who are dissatis ed: let V ~1; : : : ;~n be the oriented volume of the parallelepiped spanned by v v ~1, : : : , ~n in that order. One can then verify that: v v 1 V ~1; : : : ;~n = 1, e e
229 2 V ~1; : : : ;~i; : : : ;~j ; : : : ;~ n = ,V ~1; : : : ;~j ; : : : ;~i; : : : ;~n , and v v v v v v v v 3 V ~1; : : : ;~i + c~i0 ; : : : ;~n = V ~1; : : : ;~i ; : : : ;~ n v v v v v v v
+cV ~1; : : : ;~i0 ; : : : ;~n , v v v
where the last assertion follows because V ~1; : : : ; w; : : : ;~ n is directly v ~ v proportional to the component of w which lies perpendicular to the hyper~ plane spanned by ~ 1, : : : , ~i,1 , ~ i+1, : : : , ~n. However, we know from v v v v linear algebra that the determinant is the unique function satisfying conditions 13, so we conclude that V = det, the oriented volume of P is det M , and indeed VolumeP = j det M j. In clearing out the Mayhem archives, we also dug up this solution to an old problem. S21. Proposed by Colin Springer. There are n houses situated around a certain lake, in a circle. Each is painted one of k colours, chosen at random. Find the probability that no two neighbouring houses are of the same colour. Solution by Philip Oppenheimer, South Norwalk, CT. For convenience, we use the following notation: Let P n;k denote the desired probability. Let ha denote the colour of the ath house, starting from a xed house, going around the circle. Without loss of generality, we may x h1 = 1. Under this condition, let pn; a; b denote the probability that ha = b, if indeed no two adjacent houses are painted the same colour. With h1 = 1, there are k , 1 ways to colour the second house out of k colours, the third house, etc., until the n , 1th house. If hn , 1 = 1, then there are k , 1 ways to colour the nth house. However, if hn , 1 6= 1, then there are k , 2 ways to colour the nth house. Hence,
k , 1
n,2 k , 1 pn; n , 1; 1 + k , 2 1 , pn; n , 1; 1
: P n; k = k k k
But for all a,
1 pn;a; 1 = k , 1 1 , pn; a , 1; 1 1 1 = k , 1 , k , 1 pn;a , 1; 1
230 so that
1 1 pn; a; 1 , k = kk 1 1 , k , 1 pn;a , 1; 1 , ,1 pn;a , 1; 1 , 1
= k,1 k
,1
2 1 = k , 1 pn;a , 2; 1 , k =
,1
a,1 pn; 1; 1 , 1
k,1 k a,1 k , 1 = k,1 a,1 k , 1 ,1a,1 : = kk , 1a,2
. = . .
Therefore, and
n pn; n , 1; 1 = kk,1 n,3 ; , 1
P n;k =
= = =
k , 1
n,2 k , 1 1 ,1n
k k k + kk , 1n,3 1 n
+ k , 2 1 , k , kk,1 n,3 k , 1 k , 1
n,2 k , 1 ,1nk , 1 k k2 + k2k , 1n,3 n 2
+ k , 1k , 2 , ,1 , k ,,3 k2 k2 k 1n k , 1
n,2 k , 12 ,1n
k
k2 + k2k , 1n,3 k , 1 n ,1nk , 1 + : k kn
231
Swedish Mathematics Olympiad
1986 Qualifying Round
1. Show that 2. Show that for t
1986!1=1986 1987!1=1987: 0,
t2 +
1 , 3 t + 1
+ 4 0: t2 t
3. A circle C1 with radius 1 is internally tangent to a circle C2 with radius 2. Let ` be a line through the centres of the circles C1 and C2 . A circle C3 is tangent to C1, C2, and `. Find the radius of C3. 4. In how many ways can 11 apples and 9 pears be shared among 4 children, so that every child gets 5 fruit? The apples are identical, as are the pears. 5. P is a polynomial of degree greater than 2 with integer coe cients and such that P 2 = 13 and P 10 = 5. It is known that P has a root which is an integer. Find it. 6. The numbers 1, 2, : : : , n are placed in some order at di erent points on the circumference of a circle. Form the product of each pair of neighbouring numbers. How should the numbers be placed in order for the sum of these products to be as large as possible?
1986 Final Round
1. Show that the polynomial
x6 , x5 + x4 , x3 + x2 , x + 3 4
has no real roots. 2. ABCD is a quadrilateral, and O is the intersection of the diagonals AC and BD. The triangles AOB and COD have areas S1 and S2 respectively, and the area of ABCD is S . Show that
p
Show also that equality holds if and only if the lines AB and CD are parallel.
p p S1 + S2 S:
232 3. Let N be a positive integer, N 3. Form all pairs a; b of consecutive b integers such that 1 a b N and consider the quotient q = a for every such pair. Remove all pairs with q = 2. Show that of the remaining pairs, there are as many with q 2 as there are with q 2. 4. Show that the only positive solution of
is x = y = z = 1. 5. In the arrangement of pn real numbers below, the di erence between the greatest and least numbers in every row is at most d, where d 0:
x + y2 + z3 = 3 y + z2 + x3 = 3 z + x2 + y3 = 3
a11 a12 a1n a21 a22 a2n
. . . . . . ... . . .
ap1 ap2 apn
In each column, the numbers are now ordered by size, so that the greatest appears in the rst row, the next greatest in the second row, and so on. Show that the di erence between the greatest and least numbers in each of the rows is still at most d. 6. A nite number of intervals on the real line together cover the interval 0; 1 . Show that one can choose a number of these intervals such that no two have any points in common and whose total length is at least 1=2.
J.I.R. McKnight Problems Contest 1982
ma4 + nc4 + pe4
1=4 mb4 + nd4 + pf 4 is equal to each of the given rationals. b Given that a4 + b4 = 7c and that a and b are the roots of x2 , 5x +3, nd c. 2. Consider AB , the major axis of an ellipse centred at the origin with focus F as shown. Let P be any point on the ellipse. Draw the lines BP and AP and extend them so that they cross the directrix of F at R and S respectively. Prove that RFS is a right angle.
c e 1. a Given the equal positive rationals a , d , f , prove that b
233
R P B OF A
S
3. Solve the system of equations:
xy + yz + zx = ,4 y + z , yz = 4 x,y,z = 3
4. If cos,1 x +cos,1 y +cos,1 z = , prove that x2 + y 2 + z 2 +2xyz = 1. 5. A shopkeeper orders 19 large and 3 small packets of marbles, all alike. When they arrive at the shop it is discovered that all the packets have come open with the marbles loose in the container. If the total number of marbles is 224, can you help the shopkeeper put up the packets with the proper number of marbles in each? 6. A radar tracking station is located at ground level vertically below the path of an approaching aircraft ying at 900 km h at a constant height of 10000 m. Find the rate in degree s at which the radar beam to the aircraft is turning at the instant when the aircraft is at a horizontal distance of 3 km from the station.
900 km h 10 000 m
x
234
PROBLEMS
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 December 1998. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX. Proposed by Toshio Seimiya, Kawasaki, Japan. Suppose ABCD is a convex cyclic quadrilateral, and P is the intersection of the diagonals AC and BD. Let I1 , I2 , I3 and I4 be the incentres of triangles PAB , PBC , PCD and PDA respectively. Suppose that I1 , I2 , I3 and I4 are concyclic. Prove that ABCD has an incircle. 2339. Proposed by Toshio Seimiya, Kawasaki, Japan. A rhombus ABCD has incircle ,, and , touches AB at T . A tangent to , meets sides AB , AD at P , S respectively, and the line PS meets BC , CD at Q, R respectively. Prove that 1 + 1 = 1 , a and b
2338.
PQ RS
BT
1 , 1 = 1 : PS QR AT
235
2340. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Let 0 be a real number and a, b, c be the sides of a triangle. Prove that Y
As usual s denotes the semiperimeter. 2341. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Let a, b, c be the sides of a triangle. For real 0, put
s + a 2 + 33: cyclic s , a
X " a
b
s := b , a cyclic and let 4 be the supremum of s over all triangles. 1. Show that 4 is nite if 2 0; 1 and 4 is in nite for 1. 2.? What is the exact value of 4 for 2 0; 1?
Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Given A and B are xed points of circle ,. The point C moves on ,, on one side of AB . D and E are points outside 4ABC such that 4ACD and 4BCE are both equilateral. a Show that CD and CE each pass through a xed point of , when C moves on ,. b Determine the locus of the midpoint of DE .
2342.
2 2 n + 1x2 + n2 + 1yn + n2 + nzn = 2pn ,nxn yn + pnxn zn + ynzn ; n and for n 2, we have xn + pnyn , nzn = xn,1 + yn,1 , pn , 1zn,1: Find limn!1 xn , limn!1 yn and limn!1 zn .
2343. Proposed by Doru Popescu Anastasiu, Liceul Radu Greceanu", Slatina, Olt, Romania. For positive numbers sequences fxn gn1 , fyngn1 , fzngn1 with conditions: for n 1, we have
2344. Proposed by Murali Vajapeyam, student, Campina Grande, Brazil and Florian Herzig, student, Perchtoldsdorf, Austria. Find all positive integers N that are quadratic residues modulo all primes greater than N .
236
2345.
Proposed by Vedula N. Murty, Visakhapatnam, India. Suppose that x 1.
2346. Proposed by JuanBosco Romero M
arquez, Universidad de Valladolid, Valladolid, Spain. The angles of 4ABC satisfy A B C . Suppose that H is the foot of the perpendicular from A to BC , that D is the foot of the perpendicular from H to AB , that E is the foot of the perpendicular from H to AC , that P is the foot of the perpendicular form D to BC , and that Q is the foot of the perpendicular from E to AB . Prove that A is acute, right or obtuse according as AH , DP , EQ is positive, zero or negative. 2347. Proposed by Sefket Arslanagi
c, University of Sarajevo, Sarajevo, Bosnia and Herzegovina. Prove that the equation x2 + y 2 = z 1998 has in nitely many solutions in positive integers, x, y and z . 2348. Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Without the use of trigonometrical formulae, prove that 2349. Proposed by V
aclav Konecn
y, Ferris State University, Big Rapids, Michigan, USA. Suppose that 4ABC has acute angles such that A B C . Prove that
UK. Suppose that the centroid of 4ABC is G, and that M and N are the midpoints of AC and AB respectively. Suppose that circles ANC and AMB meet at A and P , and that circle AMN meets AP again at T . 1. Determine AT : AP . 2. Prove that BAG = CAT .
3x2 , 1 . x2 + 4x + 1 a,b 1 2pab + a + b
; b Show that lna , lnb 3 2 where a 0, b 0 and a 6= b.
a Show that lnx
sin 54 = 1 + sin 18 : 2
2350.
sin2 B sin , A sin ,A + B 2 2
sin2 A sin , B sin ,B + A : 2 2
Proposed by Christopher J. Bradley, Clifton College, Bristol,
237
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems.
SOLUTIONS
2223. 1997: 111 Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel,
n Alcorc
Madrid, Spain. on, We are given a bag with n identical bolts and n identical nuts, which are to be used to secure the n holes of a gadget. The 2n pieces are drawn from the bag at random one by one. Throughout the draw, bolts and nuts are screwed together in the holes, but if the number of bolts exceeds the number of available nuts, the bolt is put into a hole until one obtains a nut, whereas if the number of nuts exceeds the number of bolts, the nuts are piled up, one on top of the other, until one obtains a bolt. Let L denote the discrete random variable which measures the height of the pile of nuts. Find E L + E L2 . Solution by Gerry Leversha, St. Paul's School, London, England. Editor's comment: Leversha noticed the typographical error in the original statement of the problem and recognized correctly that what was required was E L + E L2 . He found the equivalent: E L2 + L . Denote the sequences of bolts and nuts by 0's and 1's respectively; thus, for n = 2 there are six possible sequences: 1100, 1010, 1001, 0110, , n 0101, and 0011, and, in general, there are 2n di erent sequences. Call the maximum height of the pile of nuts represented by such a sequence the value of the sequence. In the sequences above the heights are respectively 2; 1; 1; 1; 0; 0. We approach this problem by deriving a probability generating function Gnx for the case of n nuts and n bolts. De ne gnx = ,2nnGnx; then n X gnx = anxr , where an is the number of sequences of length 2n which r r r=0 have value r. The rst few such polynomials are as follows:
g0x = 1 g1x = x2+ 1 g2x = x3 + 3x2+ 2 g3x = x4 + 5x3 + 9x + 5 g4x = x + 7x + 20x2 + 28x + 14 2n
Notice in each case that the sum of the coe cients is gn 1 = n , as
required. In fact, the polynomials can be calculated from the following inductive de nition:
238
xgn+1x = x + 12gnx , x + 1gn0 starting with g0 x = 1. It is easy to check that gnx is indeed a polynomial of degree n. follows by virtue of the following relationships between the coe cients:
an+1 = an n+1 n n+1 = 2an + an,1 an n n an+1 = an+1 + 2an + an,1 r r r r n+1 = an + an a0 1 0
1 r n , 1
The rst of these is immediate, since it merely states that the leading coe cient is 1, and there is obviously only one way to obtain a value of n + 1 in an n 1 sequence, namely by having all the 1's at the front. The second statement follows because a value of n can be obtained either by placing 10 in front of the unique sequence of value n, or by placing 01 in front of the sequence of value n, or by placing 1 in a sequence of value n , 1 in front of that part of the sequence which creates the value. The third statement is the most di cult to see. It states that a sequence of value r, where r is at least 1 and at most n , 1, can be made either by placing 01 or 10 in front of an existing r sequence, or by devaluing an existing r + 1 sequence by placing an extra 0 in front of the part which matters, and an extra 1 at the end, or by adding 1 to an existing r , 1 sequence. The nal statement says that a 0 sequence can be created either by putting 01 in front of an existing 0 sequence or by devaluing an existing 1 sequence by placing an extra 0 in front of the critical part. These claims are best checked by looking at, say, the sequences for n = 3 and seeing how the sequences for n = 4 are constructed. We can now proceed to a calculation of E L + L2 ; this is done by nding 0 g001 + 2gn1 . In fact, we can now make the value of G00 1 + 2G0n 1 = n n gn1 the following claim: E L2 + L = n. This is proved using induction on n. It is trivially true for n = 0, so we assume that it is true for n = k; that is, we assume that
00 0 gk 1 + 2gk1 = kgk1: By di erentiating twice the de ning equation for gk+1 x, we have 0 0 xgk+1x + gk+1x = x + 12gkx + 2x + 1gkx , gk0 ; 00 0 00 0 xgk+1 x + 2gk+1x = x + 12gk x + 4x + 1gkx + 2gkx :
239 Hence, putting x = 1, we have
00 0 00 0 gk+11 + 2gk+11 = 4gk 1 + 8gk1 + 2gk1 = 4k + 2gk1 by the inductive hypothesis 4k + 2gk1 . Hence, for n = k + 1; E L2 + L = g 1 2kk+1
However, we know that gk1 = k , and so k!k + 1! E L2 + L = 4k + 22k!2k + 2! k + 1! ; k! and it is straightforward to check that this reduces to k + 1. This nishes the
induction step and establishes the claim.
Also solved by the proposer by a completely di erent method.
2227. 1997: 166 Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel,
n Alcorc
Madrid, Spain. on, Evaluate 1 Y " X ,2k
where the product is extended over all prime numbers. Composite solution by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA and KeeWai Lau, Hong Kong. It is well known that for jxj 1 , we have 4
k 2p2k : p k=0
,1=2 1 1 ,2k X 2k
k ,1=2 ; and hence, X k = 1 , 1 x = 1 , 4x : 2k p2 k=0 k k=0 2p Y 1
,1 It is also known that
s = 1 , ps for s 1, where
is the p Riemann Zeta function, and that
2 = 62 . See, for example, Hardy and
Wright, An Introduction to the Theory of Numbers, 5th edition, p. 246. Therefore,
1 Y X ,2kk ! Y 1
,1=2 , 1=2 1 , p2 =
2 = p : 2k = 6 p k=0 2p p
Also solved by MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE
240
PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; RICHARD I. HESS, Rancho Palos Verdes, California, USA; ROBERT B. ISRAEL, University of British Columbia, Vancouver, BC; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; HEINZJURGEN SEIFFERT, Berlin, Germany; and the proposer. Most of the submitted solutions are similar to, or virtually the same as, the one given above. By using the same argument, one caneasily show that the value , ! 2k Y , 1=2 k of the slightly more general product P = . This was k is
p 4p pointed out by Janous and the proposer.
2228. 1997: 167 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. Let A be the set of all real numbers from the interval 0; 1 whose decimal representation consists only of 1's and 7's; that is, let
A=
X 1
k=1
ak : a 2 f1; 7g : 10k k
Let B be the set of all reals that cannot be expressed as nite sums of members of A. Find sup B . Solution by the proposer. We prove that sup B = 1. Fix a 2 1; 7 and consider b = a , 1=6. Since b 2 0; 1 , b can be easily expressed as a sum of nine reals from 0; 1 each consisting only of 0's and 1's in decimal representation. Now multiplying each by 6 and next adding :1111 : : : to the result, we obtain since 1 = 9 :1111 : : : an expression for a as a sum of nine reals, each consisting only of 1's and 7's. Now take any a 7 and subtract from a as many :1111 : : : 's as needed to get into the interval 1; 7 . This procedure shows that any a 1 is a nite sum of members of A; that is, sup B 1. Now let x 2 8=9; 1 be a nite sum of members of A so that x 62 B. Fixing our attention, assume that x is a sum of k members of A; since x 2 8=9; 1 and each member of A is at least :1111 : : : , we know that 2 k 8. Consider Then y is a sum of k real numbers, each consisting only of 1's and 0's. This implies, since k 8, that y does not contain a digit 9 in its decimal representation. Thus we have shown that a number x 2 8=9; 1 must be in B if the following property holds:
y = x , k:1111 : : : : 6
241
: all seven numbers 1 x , k:1111 : : : ; k = 2; 3; : : : ; 8, 6 contain a 9 in their decimal representation.
8 , k 9 , k
54 ; 54 by Jk k = 2; 3; : : : ; 8. The set Ck of all numbers from Jk having a unique decimal representation and containing at least one 9 is easily seen to be open and dense in Jk . Editorial note. For the bene t of readers not familiar with these terms, here are a couple of explanations. Ck is open" means that for any c 2 Ck , there is some interval c , ; c + 0 which is completely contained in Ck ; that is, all numbers close enough to c also have a 9 in their decimal representation. Ck is dense in Jk " means that for any number y 2 Jk , and for any 0, the interval c , ; c + must contain some member of Ck ; that is, there are numbers with at least one digit 9 as close as you like to y , even if y itself does not have a 9 in it. For each k = 2; 3; : : : ; 8, the map ' x = x , k:1111 : : : = 9x , k
Denote the interval
k
is increasing and continuous and takes 8=9; 1 onto Jk . Therefore the sets ',1 Ck, k = 2; 3; : : : ; 8 are open and dense in 8=9; 1, and thus so is the k intersection ,1 ,1 ,1 Editorial note. In other words, the intersection of nitely many open and dense sets is open and dense. Proof left to the reader! Or look at problem 16, Chapter 2 of Rudin's Principles of Mathematical Analysis. But all elements x in this intersection have the property , and thus lie in B, which means that B is dense in 8=9; 1. This completes the proof that sup B = 1. Remark. In the last part of the proof we have used Baire's Theorem, in fact its easier version for the intersection of nitely many open and dense sets instead of countably many. Also, the argument used in the solution shows that there are plenty of elements of B in 8=9; 1! So I believe that a reader not familiar with topology might nd an explicit example of a sequence of elements of B which get arbitrarily close to 1.
Also solved by GERALD ALLEN, CHARLES DIMINNIE, TREY SMITH and ROGER ZARNOWSKI jointly, Angelo State University, San Angelo, Texas, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; and DAVID STONE and VREJ ZARIKIAN, Georgia Southern University, Statesboro, Georgia, USA. Two other readers sent in incomplete solutions. The other solutions to this problem actually contain the explicit examples that the proposer asks for in his remark above. Allen, Diminnie, Smith and Zarnowski give bn = :99 : : : 944999, where there are 3n + 1 9's before the two 4's. Stone and Zarikian use bn = :99 : : : 9546, where there are 3n , 2 9's before the 546.
6
54
'2 C2 '3 C3 '8 C8:
242
Lambrou uses somewhat more complicated bn 's. In all cases the solvers then must do some calculation to show that the bn 's all lie in B .
P1 M 13 := jjMP jj = cot 1 cot 1 ; cot 2 cot 2 3 where the +, sign holds if the line segment P1 P3 is located inside outside
2231. 1997: 167 Proposed by Herbert Gulicher, Westfalische WilhelmsUniversitat, Munster, Germany. In quadrilateral P1P2 P3 P4 , suppose that the diagonals intersect at the point M 6= Pi i = 1; 2; 3; 4. Let MP1 P4 = 1, MP3 P4 = 2, MP1P2 = 1 and MP3P2 = 2 . Prove that
the quadrilateral. Preliminary comment. No solver made use of directed angles and segments, which would have reduced the problem to a single case. The proposer's angles are assumed to be positive, and nobody explicitly went through all four resulting possibilities: M can lie between P1 and P3 or not, and between P2 and P4 or not. Since P1P2 P3 P4 need not have an interior, a more careful statement of the proposer's two cases would distinguish whether or not P2 and P4 lie on opposite sides of the line P1 P3 . We shall feature the oppositeside case, and leave the other to the reader. Solution by Michael Lambrou, University of Crete, Crete, Greece modi ed by the editor to explicitly distinguish the two subcases. Assume M to be any point of the line P1 P3 except P1 ; P3 , and let P2P4 be any other line through M with P2; P4 on opposite sides of M . Let ! = P1MP4. Then sin P1P4M = sin , 1 + ! = sin 1 + !, while sin P3P4 M = sin! 2 where `+' corresponds to the subcase where M is outside the segment P1 P3 , while `,' corresponds to where M is between P1 and P3. From the sine rule applied to triangles MP1P4 and MP3P4 we have Hence,
sin 1 sin P1M sin + ! = MP4 = MP3 sin! 2 :
1 2
where we divided both the numerator and the denominator by the product of three sines to get the last equality.
P1M sin 2 sin 1 + ! MP3 = sin! 2 sin 1 sin cos cos sin = sin ! 2cos 1sin ! + sin !2sin 1sin ! sin cos 2 1 2 1 cot ! + cot 1 = cot cot ! 2
1
243 Similarly from triangles MP1 P2 and MP3 P2 we have
The result follows by combining 1 and 2 using the following property of a c e c+ proportions: if = = then a = d+e . b f
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; JOHN G. HEUVER, Grande Prairie Composite High School, Grande Prairie, Alberta; WALTHER JANOUS, Ursulinengymnasium, Inns
bruck, Austria; GERRY LEVERSHA, St. Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer.
P1M = sin 2 sin! , 1 MP3 sin 1 sin! 2 = cot 1 + cot ! : cot 2 cot ! f
2
b
d
2232. 1997: 168 Proposed by Sefket Arslanagi
c, University of Sarajevo, Sarajevo, Bosnia and Herzegovina. Find all solutions of the inequality:
n2 + n , 5
Note: If x is a real number, then bxc is the largest integer not exceeding x. Solution by Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. The only solution is n = 2. Let
n n + 1 n + 2 n2 + 2n , 2; n 2 N: 3 + 3 + 3
n n + 1 n + 2 + + : M=
3 3 3
Then
M n + n + 1 + n + 2 = n + 1: 3 3 3 Hence n2 + n , 5 n + 1, or n2 6, and so n = 1 or 2. But when n = 1, M = 1 = n2 + 2n , 2, and so n = 1 is not a solution. By inspection, n = 2
is a solution.
Also solved by CHARLES ASHBACHER, Cedar Rapids, Iowa, USA; MICHEL BATAILLE, Rouen, France; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; JOE HOWARD, New Mexico Highlands University, Las Vegas, NM, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D. KIPP JOHNSON, Beaverton, Oregon, USA; MICHAEL
244
LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; SEAM MCILROY, student, Vancouver, BC; ANNE MARTIN, Westmont College, Santa Barbara, California, USA; CAN ANH MINH, Berkeley, California, USA; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; CHRISTOS SARAGOTIS, student, Aristotle University, Thessaloniki, Greece; ROBERT P. SEALY, Mount Allison University, Sackville, New Brunswick; HEINZJURGEN SEIFFERT, Berlin, Germany; REZA SHAHIDI, student, University of Waterloo, Waterloo, Ontario, Canada; D.J. SMEENK, Zaltbommel, the Netherlands; DIGBY SMITH, Mount Royal College, Calgary, Alberta; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; and the proposer. There were also three incomplete and one incorrect solution submitted. Some solvers initially prove the equality jnk + n + 1 + n + 2 = n; 3 3 3 or some more general form of it. The following general form is wellknown. n,1 X i x + n = bnxc for n 2 N; n 6= 0; and x 2 R
See for example, D.O. Shklarsky, N.N. Chentzov, and I.M. Yaglom, The USSR Olympiad Problem Book, W.H. Freeman and Company, 1962, p. 24, problem 101, 3.
i=0
2233. 1997: 168 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Let x, y , z be nonnegative real numbers such that x + y + z = 1, and let p be a positive real number. a If 0 p 1, prove that
where
b? Prove the same inequality for p 1. Show that the constant Cp is best possible in all cases. Solution by G.P. Henderson, Garden Hill, Campbellcroft, Ontario. a We are to prove that F x;y; z 0 where
xp + yp + zp Cp ,xyp + yzp + zxp ; p log 2 3 if 3, Cp = 2p+1 if p log loglog 2 ;: 2 p
log 3,log 2
X xp , C xp yp the sums here and below are cyclic over x; y;z . Here C = min3p; 2p+1 = 3p
F=
X
245 for 0
p 1. For any real u; v;w,
uv + vw + wu 1 u + v + w2: 3
1 u + v + w2 = 3uv + vw + wu + 2 u , v 2 + v , w2 + w , u2 ;
Editorial note. This follows from the identity
as other solvers point out. Therefore
F
X
X p2
X p
p,1 X p x = x 1,3 x 0; xp , C 3
max
the last inequality following since For example, since p 1,
X
xp = 31,p:
by the power mean inequality.  Ed. This is the best possible C because F 1=3; 1=3; 1=3 = 0. b With no loss of generality we assume x y z , which implies Give z a xed value in 0; 1=3 and set y = 1 , x , z so that dy=dx = ,1 . Then F is a function of x only and
xp + yp + zp
1=p x + y + z 1 3 =3 3
z1 3
and
x 1,z : 2
dF = pxp,1 , pyp,1 , Cpxp,1yp + Cxppyp,1 + Cpyp,1zp , Cpzpxp,1 dx = pxp,1 , y p,11 , Cz p + pCxp,1 y p,1x , y : Since C 3p and z p 3,p and p 1, we see that dF=dx 0 and we only need to prove F 0 when x has its minimum value, 1 , z =2. That is, we are to prove F x;x; z 0 where x = 1 , z =2 = y and 0 z 1=3;
that is, This is equivalent to showing that
2xp + zp , C x2p + 2xp z p 0:
p p min=3 x22px+ +xzpz p C: 0z1 2
246 Calling the function to be minimized Gz , we nd using dx=dz = ,1=2
sgn dG = sgn x2p + 2xp z p,pxp,1 + pz p,1 dz
,2xp + zp,px2p,1 , pxp,1 zp + 2pxpzp,1 , = sgn "pxp,1 z2p + px2p,1 z p , 3px2p z p,1 + px3p,1 z
2p z
p z
p,1 = sgn x + x , 3 x +1 :
z t = x = 1 2z z ; ,
Set that is,
dG 0 for 0 t t and dG 0 for t t 1: 1 1 dz dz Hence, putting z1 = t1 =t1 + 2, G increases in 0 z z1 , is a maximum at z1 and decreases in z1 z 1=3. Therefore min G = min G0;G1=3 = min2p+1; 3p = C: Since the minimum is attained in both cases, C is best possible.
Both parts also solved by RICHARD I. HESS, Rancho Palos Verdes, California, USA; and MICHAEL LAMBROU, University of Crete, Crete, Greece. Part a only solved by HEINZJURGEN SEIFFERT, Berlin, Germany; and the proposer. One other reader sent in a comment. Sei ert and the proposer had the same proof for part a as Henderson and incidentally, the proof seems to hold for negative p as well. For part b both Hess and Lambrou used multivariable calculus.
The expression in brackets is a continuous, increasing function. It is negative at t = 0 and positive at t = 1. Therefore it has a unique root, t0 , in 0; 1 . It follows that H decreases to a minimum at t0 then increases. Since H 0 = 1 and H 1 = 0, H t0 0 and we see that H has a unique root t1 in 0; t0 . Therefore
t z = t + 2 ; 0 t 1: As t increases from 0 to 1, z increases from 0 to 1=3. Then sgndG=dz = sgn H t where H t = t2p + tp , 3tp,1 + 1: We have dH = tp,2 2ptp+1 + pt , 3p + 3: dt
247
2234. 1997: 168 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Given triangle ABC , its centroid G and its incentre I , construct, using only an unmarked straightedge, its orthocentre H . Solution by the proposer edited to t the rewording of the problem from his original submission. We will rst establish a lemma: Lemma. Given a line segment and its midpoint, and any other point O in the plane, a line passing through O and parallel to the given line segment may be constructed using only an unmarked straightedge. Proof: Let AD be the given line segment and let K be its midpoint. Let P be any point on AO extended. Connect P with D and with K . Draw segment OD. Let M be the intersection of OD and PK , and let N be the intersection of AM with PD. We will show that ON is parallel to AD. Suppose instead that ON1 is parallel to AD with N1 on PD. Let M1 be the intersection of AN1 and OD, let K1 be the intersection of PM1 and AD, and let L1 be the intersection of ON1 with PK1 . Then 4AM1 K1 4N1 M1 L1 and 4OM1 L1 4DM1 K1. Therefore,
AK1 K1M1 K1D L1N1 = M1 L1 = OL1 . Also 4APK1 4OPL1 and 4DK1 P 4N1 L1 P , which implies AK1 = K1 P = K1 D . OL1 L1 P L1N1
From 1 and 2 it follows that 1 2
AK1 OL1 = L1 N1 K1 D;
Consequently,
AK1 L1N1 = OL1 K1D:
AK1 2 OL1 L1 N1 = K1D2 OL1 L1 N1 ; from which we have AK12 = K1D2 , and AK1 = K1 D. Thus point
K1 coincides with point K , PK1 coincides with PK , point M1 with M and point N1 with N . Thus ON is parallel to AD.
P
r
O
r
M K
r
r
r
L1
r
r
N
r
M1
r
N1
r
A
r
K1
D
248 Using the centroid G draw the medians AA1 , BB1 , CC1, where A1 , B1 , C1 are the midpoints of BC , CA, AB, respectively. Similarly, using the incentre I draw the angle bisectors AA2 , BB2 , CC2 , where A2, B2, C2 lie on BC , CA, AB, respectively. Draw triangle A1B1C1 and denote C1A1 BB2 by B3 , A1 B1 CC2 by C3 , B1C1 AA2 by A3 . Thus A3 , B3 , C3 are the midpoints of AA2 , BB2 , CC2, respectively. By applying the lemma twice we may draw a straight line BE through B parallel to angle bisector CC2 and draw a straight line BF parallel to angle bisector AA2 with points E and F located on the line CA. We see that triangles FAB and BCE are isosceles FBA = BFA, EBC = BEC . Draw a straight line A1K parallel to CA with K on BE. Thus K is the midpoint of BE . Similarly get the midpoint N of FB . Thus AN ? FB and CK ? BE .
Tr
r
R
Q
r r
B B3 rA1 r I A2 r r rC3 A3
r r
Nr
r
C1 r C2 r
r r
K
r
Now we may draw lines parallel to FB and BE through the points C1 and A1 respectively, and get the midpoints of AN and CK . Then applying the lemma we may draw FT parallel to CK and ET parallel to AN . Denote ET FB by R and BE FT by Q. Thus FR ? ET , EQ ? FT and point B is the point of intersection of the altitudes of triangle FTE . Draw line TB , denote TB CA by P and get BP ? CA; that is, BP is an altitude of triangle ABC . We can construct a second altitude similarly. Their intersection is H , the orthocentre.
F
A P B2 B1
rr
C
r
E
Also solved or answered by JORDI DOU, Barcelona, Spain and TOSHIO SEIMIYA, Kawasaki, Japan. There was one incorrect solution.
249
2235. 1997: 168 Proposed by D.J. Smeenk, Zaltbommel, the Netherlands, not Walther Janous, Ursulinengymnasium, Innsbruck, Austria as was printed. Triangle ABC has angle CAB = 90 . Let ,1O; R be the circumcircle and ,2 T; r be the incircle. The tangent to ,1 at A and the polar line of A with respect to ,2 intersect at S . The distances from S to AC and AB are denoted by d1 and d2 respectively. Show that a ST kBC , b jd1 , d2 j = r.
For the bene t of readers who are not familiar with the term polar line", we give the following de nition as in, for example, Modern Geometries, 4th Edition, by James R. Smart, Brooks Cole, 1994: The line through an inverse point and perpendicular to the line joining the original point to the centre of the circle of inversion is called the polar of the original point, whereas the point itself is called the pole of the line. Solution by Istv
n Reiman, Budapest, Hungary. a Assume without loss of generality that AB AC . Should AB = AC the polar line and tangent would be parallel so that S would be at in nity, while AB AC would involve only minor changes in notation. Let P be the point where ,2 touches AB , and Q be where it touches AC . Thus AQTP is a square whose sides have length r. Next, let U and V be the feet of the perpendiculars from S to AB and to AC respectively, so that AUSV is a rectangle. The polar of A is the line PQ. Since PQT is an isosceles right triangle, PQ makes a 45o angle with AC , which implies that QSV is also an isosceles right triangle. Consequently SV = QV = d1 , and AV , SV = d2 , d1 = AV , QV = AQ = r which proves b. Now, QAS = CBA, since both angles are subtended by the chord AC of the circle ,1. Moreover, QAS = QTS because these angles are ,! symmetric about the line PQ. Since the corresponding side vectors TQ ,! and ,! ,! BA have the same direction, it follows that so do the vectors TS and BC , and we conclude that the lines ST and BC are parallel as desired.
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; ADRIAN BIRKA, student, Lakeshore Catholic High School, Niagara Falls, Ontario; MANSUR BOASE, student, St. Paul's School, London, England;; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; JIMMY CHUI, student, Earl Haig Secondary School, North York, Ontario; FILIP CRNOGORAC, student, Western Canada High School, Calgary, Alberta; JORDI
250
DOU, Barcelona, Spain; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; JOHN G. HEUVER, Grande Prairie Composite High School, Grande Prairie, Alberta; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece 2 solutions; JAMES LEE, student, Eric Hamber Secondary School, Vancouver, BC; GERRY LEVERSHA, St. Paul's School, London, England; DAVID NICHOLSON, student, Fenelon Falls Secondary School, Fenelon Falls, Ontario; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; TOSHIO SEIMIYA, Kawasaki, Japan; CHRISTOPHER SO, student, Franch Liberman Catholic High School, Scarborough, Ontario; KAREN YEATS, student, St. Patrick's High School, Halifax, Nova Scotia; and the proposer.
2236. 1997: 169 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Let ABC be an arbitrary triangle and let P be an arbitrary point in the interior of the circumcircle of 4ABC . Let K , L, M , denote the feet of the perpendiculars from P to the lines AB , BC , CA, respectively.
Prove that KLM 4 . Note: XY Z denotes the area of 4XY Z . Almost identical solutions were submitted by Niels Bejlegaard, Stavanger, Norway; Mansur Boase, student, St. Paul's School, London, England; Istv
n Reiman, Budapest, Hungary; Toshio Seimiya, Kawasaki, Japan; a and Panos E. Tsaoussoglou, Athens, Greece. For 4KLM , with O as circumcentre, we have Various di erent references were given.
2, 2 KLM = R 4ROP ABC ABC : 2 4
ABC
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK;
WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; HEINZJURGEN SEIFFERT, Berlin, Germany; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
251 1997: 169 Proposed by Meletis D. Vasiliou, Elefsis, Greece. ABCD is a square with incircle ,. Let ` be a tangent to ,. Let A0 , B0, 0 , D0 be points on ` such that AA0 , BB 0 , CC 0 , DD0 are all perpendicular C to `. Prove that AA0 CC 0 = BB 0 DD0 . I. Solution by Francisco Bellot Rosado, I.B. Emilio Ferrari, Valladolid, Spain. We give a solution with coordinates. Clearly, without loss of generality, we can take A,1; 1, B 1; 1, C 1; ,1, D,1; ,1, and the equation of the incircle is x2 + y 2 = 1. Then, if P cos t; sin t is any point of the incircle, the equation of the line `, tangent to the incircle, is The cases in which ` is parallel to either of the axes are trivial. Then, calculating
2237.
cos tx + sin ty = 1:
AA0 BB0 CC 0 DD0
= = = =
dA; ` = j , cos t + sin t , 1j dB; ` = j cos t + sin t , 1j dC; ` = j cos t , sin t , 1j dD; ` = j , cos t , sin t , 1j
we easily obtain
AA0 CC 0 = j sin2tj = j , sin2tj = BB0 DD0 ;
and we are done. II. Solution by Gottfried Perz, Pestalozzigymnasium, Graz, Austria. D0
r
D
r
r
,
r
Y C A0 T C0
r r r r r r
M
r
B0
`
r
A X U B First we note that the equation holds obviously if the point of tangency of ` and , is the midpoint of a side of the square. Otherwise, by rotational symmetry about the centre M of ABCD, we may assume that, without loss of generality, ` touches the arc of , next to C ", say at T . Let X and Y be
r r
252 the points of intersection of l with AB produced and CD, respectively, and U and V the midpoints of AB and CD, respectively. Note that quadrangles MUXT and Y V MT are similar, since their respective angles correspond, and XU = XT and MV = MT . Hence it follows that
AA0 = AX = AU + UX = MU + UX BB0 BX UX , UB UX , MU Y V + V M = Y V + V D = Y D = DD0 : = VM ,YV CV , Y V CY CC 0
III. Solution by HeinzJurgen Sei ert, Berlin, Germany. There exist vectors E and F with E F = 0 and kFk = 1, such that
` = fE + tF j t 2 Rg: We have A = xE + y F and B = uE + v F for some reals x; y;u; v . If A0 = E + tA F, where tA 2 R, then from A , A0 F = 0, it follows that tA = y, so that A0 = E + yF. Similarly, we nd B0 =0 E + vF, and since C = ,A and D = ,B, we have C0 = E , y F and D = E , v F. From 2r2 = kAk2 = kBk2, where r denotes the radius of ,, and A B = 0, we
obtain
2r2 = x2r2 + y 2 = u2 r2 + v 2 and xur2 + yv = 0: The rst equation gives v 2 = r2 2 , u2, so that, using the second equation,
we have
x2u2r4 = y2v2 = y2r22 , u2 or u2 x2 r2 + y 2 = 2y 2: Using the rst equation again, we get y 2 = u2 r2 and then 2 = x2 + u2 , which by AA0 = kA , A0 k = jx , 1jr, BB 0 = kB , B0 k = ju , 1jr, CC 0 = kC , C0k = jx + 1jr, and DD0 = kD , D0k = ju + 1jr, implies the
desired equation.
Also solved by SEFKET ARSLANAGIC, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; SAM BAETHGE, Nordheim, Texas, USA; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; DENISE CHEUNG, student, Albert Campbell Collegiate Institute, Scarborough, Ontario; JIMMY CHUI, student, Earl Haig Secondary School, North York, Ontario; MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut, USA; JORDI DOU, Barcelona, Spain; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengym
nasium, Innsbruck, Austria; D. KIPP JOHNSON, Beaverton, Oregon, USA; VACLAV
KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece 4 methods; KEEWAI LAU, Hong
Kong; GERRY LEVERSHA, St. Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; K.R.S. SASTRY, Dodballapur, India; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
253
Most of the solutions were similar to either I or II above. Several solvers made generalizations in di erent directions. Janous considers a regular 2ngon P2n := A1 A2 A2n with incircle , and `, a tangent line to ,; then for the orthogonal projections A01 ; A02 ; : : : ; A02n of the vertices of P2n to ` we have
2n Y
k=1 k even
Ak A0k =
2n Y
k=1 k odd
Ak A0k :
Kone cny observes that the condition AA0 , BB 0, CC 0 , DD0 are all perpen
dicular to `" is not necessary; it is su cient to state that they make the same angle with `; that is, they are parallel to one another. Sastry shows that if we start with ABCD being a rhombus with incircle ,, and with A0 , B 0 , C 0, D0 de ned as before, we can prove that the equation holds for all lines ` tangent to , if and only if ABCD is a square. Finally, Seimiya comments that if we start with ABCD any quadrilateral having an incircle , with centre O, and with ` being a line tangent to ,, then when A0 , B0 , C 0, D0 are the feet of the perpendiculars from A, B, C , D, respectively, to `, we have AA0 CC 0 = AO CO ; a constant: BB 0 DD0 BO DO
2238. 1997: 242 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. A fourdigit number abcd is said to be faulty if it has the following property: The product of the two last digits c and d equals the twodigit number ab, while the product of the digits c , 1 and d , 1 equals the two digit number ba. Determine all faulty numbers! Solution by David R. Stone, Georgia Southern University, Statesboro, Georgia, USA. Implicit in the de ning properties is that a 6= 0, c 1, d 1, and b 6= 0. Translating the given conditions, if abcd is to be faulty we must have c d = ab = 10a + b; 1 and
These imply
c , 1 d , 1 = ba = 10b + a:
10b + a = cd , d , c + 1 = 10a + b , d , c + 1;
254
9a , b = c + d , 1: Thus 1 9a , b 17, forcing 9a , b = 9, or a = b + 1. Hence c + d = 10. Substituting into 1, we get c10 , c = 10b + 1 + b, or c2 , 10c + 11b + 10 = 0: p By the quadratic formula, c = 5 15 , 11b, which forces 15 , 11b = 4, or b = 1. Thus a = 2 and c = 7 or c = 3, which forces d = 3 or d = 7,
respectively. That is, the only two faulty numbers are 2137 and 2173: Checking, 3 7 = 21 and 3 , 1 7 , 1 = 12.
so
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain;
SEFKET ARSLANAGIC, University of Sarajevo, Sarajevo, Bosnia and Herzegovina; CHARLES ASHBACHER, Cedar Rapids, Iowa, USA; SAM BAETHGE, Nordheim, Texas, USA; FRANK P. BATTLES, Massachusetts Maritime Academy, Buzzards Bay, Massachusetts, USA; MANSUR BOASE, student, St. Paul's School, London, England;
CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MIGUEL ANGEL CABEZON OCHOA, Logro~ o, Spain; CON AMORE PROBLEM GROUP, The Royal Danish School n of Educational Studies, Copenhagen, Denmark; MAYUMI DUBREE, student, Angelo State University, San Angelo, Texas, USA; KEITH EKBLAW, Walla Walla, Washington, USA; JEFFREY K. FLOYD, Newnan, Georgia, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; JOHN G. HEUVER, Grande Prairie Composite High School, Grande Prairie, Alberta; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D. KIPP JOHNSON,
Beaverton, Oregon, USA; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; KATHLEEN E. LEWIS, SUNY Oswego, Oswego, NY, USA; J.A. MCCALLUM, Medicine Hat, Alberta; GRADY MYDLAK, student, University College of the Cariboo, Kamloops, BC; CHRISTOS SARAGIOTIS, student, Aristotle University, Thessaloniki, Greece; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; HEINZJURGEN SEIFFERT, Berlin, Germany; REZA SHAHIDI, student, University of Waterloo, Waterloo, Ontario; SKIDMORE COLLEGE PROBLEM GROUP, Saratoga Springs, New York, USA; D.J. SMEENK, Zaltbommel, the Netherlands; DIGBY SMITH, Mount Royal College, Calgary, Alberta; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer. Although nobody raised the issue, readers must have wondered why the proposer chose the term faulty" for numbers satisfying this condition. He reveals his reason at the end of his solution with the following remark: Crux problems 2137 and 2173 both have been corrected in Crux, 2137 even twice! They are not faulty anymore!" See 1996: 317 and 1997: 48 for 2137, and 1997: 169 for 2173.
255
2239. 1997: 242 Proposed by Kenneth Kam Chiu Ko, Mississauga, Ontario. Suppose that 1 r n and consider all subsets of r elements of the set f1, 2, 3, : : : , ng. The elements of these subsets are arranged in ascending order of magnitude. For i from 1 to r, let ti denote the ith smallest element in the subset. Let T n; r;i denote the arithmetic mean of the elements ti .
Prove that T n;r; i = i Solution by Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. , Let X n = f1; 2; :::; ng. Since there are n subsets of X n with r r elements, P
n + 1. r+1
T n;r; i = ,nti ; r where the summation is over all rsubsets. Thus we are to show that X + 1
+ 1
ti = i n + 1 n = i n + 1 : r r r
For each k 2 X n, let uk denote the number of rsubsets
1
of X n with k being the ith smallest element. Clearly, we must have i. , ,1 , k There are k,1 ways of choosing the i , 1 elements before k, and n,k ways i r,i of choosing the r , i elements after k. Thus from which we get
a1; a2; :::; ai,1 ; k; ai+1 ; :::; ar
k , 1
n , k
uk = i , 1 r , i
kuk =
X
ti =
X
ki
X k , 1
n , k
k : ki i , 1 r , i
2
From 1 and 2 we see that it remains to prove that
X k k , 1
n , k
n + 1
= r+1 ; ki i i , 1 r , i X k
n , k
n + 1
= r+1 : ki i r , i
3
or, equivalently,
The equation 3 is wellknown for example, formula 11 on page 207 of Applied Combinatorics, 2nd edition by Alan Tucker. For completeness we
256
+1 give a combinatorial proof of this identity. Note rst that n+1 is the numr ber of r + 1subsets of X n + 1. Again, we partition the family of all these subsets according to their i + 1th smallest element, where i is xed, 1 i r. Speci cally, for each k 2 X n + 1, we count the number, vk, of r + 1subsets with k + 1 being the i + 1th smallest element. Clearly, k i, and, by the same argument as before, with n; r; i; k replaced by n + 1; r + 1; i + 1, and k + 1, respectively, we have
,
Summing over k, we get
vk = k n , k : i r,i
n + 1
X X k
n , k
= vk = r + 1 ki ki i r , i
which establishes 3 and completes the proof.
Also solved by NIELS BEJLEGAARD, Stavanger, Norway; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; ADRIAN BIRKA, student, Lakeshore Catholic High School, Port Colbourne, Ontario; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; JIMMY CHUI, student, Earl Haig Secondary School, North York, Ontario; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; ALAN LING, student, University of Toronto, Toronto, Ontario; CHRISTOPHER SO, student, Francis Liberman Catholic High School, Scarborough, Ontario; and the proposer. There was also one incomplete solution submitted. Francisco Bellot Rosado noted that this problem is a generalization of the second problem from the IMO 1981 Washington, where the question was to nd the arithmetic mean of the smallest elements of the rsubsets. He also pointed out that a solution to the general question is given in a Romanian Olympiad book: Cuculescu, I., Olimpiadele Internationale de Matematica ale elevilor, Ed. Tehnica, Bucarest 1984, p. 315.
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
257
THE ACADEMY CORNER
No. 20 Bruce Shawyer
All communications about this column should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7
THE BERNOULLI TRIALS 1998
The Bernoulli Trials, an undergraduate mathematics competition, was held Saturday, March 7 at the University of Waterloo. This is the second year for this event, which is a double knockout competition with true" or false" as the answers on each round. The participants have 10 minutes for each question, and drop out after their second incorrect answer. There were 29 student participants in the competition, which lasted 4 hours and 16 rounds. The winner was third year student Frederic Latour. Second place went to rst year student Joel Kamnitzer, and third and fourth to Richard Hoshino and Derek Kisman. In keeping with the nature of the answers required, the prizes were awarded in coins: 100 toonies for rst place, 100 loonies for second, and quarters for third and fourth. Ian Goulden and Christopher Small 1. In the gure below, the two circles are tangent at A. The point C is the center of the larger circle, and FC is perpendicular to AB . The line segment DB is of length 9, and the line segment FE is of length 5. F E
A
C
D
B
TRUE OR FALSE? The diameter of the larger circle is less than or equal to 49.
258 2. One of the participants in the Bernoulli Trials answers by ipping a fair coin. He chooses True" when the coin lands heads and False" when the coin lands tails. TRUE OR FALSE? Using this method, on average he will last exactly 4 rounds until he drops out. 3. TRUE OR FALSE? There exist 11 distinct positive integers so that all integers from 8 to 1998 inclusive can be written as sums over subsets of the ai 's. 4. Let ABCD be a rectangle labelled clockwise as shown, and let P be any point in the plane. P
a1; a2; : : : ; a11
A D
B C
TRUE OR FALSE?
5. TRUE OR FALSE? The volume of a regular tetrahedron of side 1 is less than the volume of a sphere of radius 1= .
jPAj2 , jPBj2 + jPC j2 , jPDj2 = 0 : In this expression jEF j denotes the distance from point E to point F .
6. TRUE OR FALSE? The smallest perfect square ending in 9009 is 7. Let f : 0; 1 ! 0; 1 be a continuous strictly increasing function such that f 0 = 0 and f 1 = 1. TRUE OR FALSE? The inequality holds true for any such function.
i=1
9 9 X i
X ,1 i
99 f 10 + f 10 10
1265032 = 16003009009 :
i=1
259 8. Let k be the smallest positive integer such that kt + 1 is a triangular number whenever t is a triangular number. TRUE OR FALSE? There exists such a k and k 12. A number t is triangular if it can be written in the form where n 1. The rst 4 triangular numbers are 1, 3, 6, and 10. 9. TRUE OR FALSE? It is possible to construct a p planar quadrilateral with sides of length 1, 2, 3, and 4, and with area 2 6. 10. Let f : 0; 1 ! 0; 1 be a continuous strictly increasing function satisfying the equation for all x 0. TRUE OR FALSE? Then f x = x for all x 0. 11. TRUE OR FALSE? 23 32 12. TRUE OR FALSE?
t = 1+ 2+ 3 + +n
x
fx f 1 + x = 1 +f x
3 2
232
13. A triangle has squares constructed externally on each of its three sides. Suppose these squares have area 8, 13 and 41 square meters respectively. TRUE OR FALSE? The area of the triangle is greater than one square meter.
4 3
Z 1 x dx ex , 1 0
323 :
5: 3
41 13
14. TRUE OR FALSE? There exists a continuous function such that
8
f :R!R
f f x = ex , eex
260 15. Let A and B be two random points in the interior of a circle C . Let D be the circle with diameter AB , and let ` be the line tangent to D at B. C
A D
`
B
Let R1 be the area of the region that is interior to both C and D. Let R2 be the area of the region that is interior to C on the other side of ` from D. TRUE OR FALSE? The average value of R1 is strictly less than the average value of R2 . 16. The following game is played between two players using the petals of a daisy. Each player takes a turn plucking either one petal or two neighbouring petals from a daisy. The player who takes the last petal wins. TRUE OR FALSE? With best play by both sides for a daisy with thirteen petals, the rst player can force a win.
13 12 11 10 9 6 8 7 1 2 3 4 5
261
THE OLYMPIAD CORNER
No. 191 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. Welcome back from the summer break". I hope all of you have been taking some time to write up your nice solutions to problems given so far. We start this number with two traditional pieces. First we give the problems of the 1998 Canadian Mathematical Olympiad which we reproduce with the permission of the Canadian Mathematical Olympiad Committee of the Canadian Mathematical Society. Thanks go to Daryl Tingley, University of New Brunswick, and chair of the CMO Committee, for forwarding the questions to me.
1998 CANADIAN MATHEMATICAL OLYMPIAD
November 20
Final
1. Determine the number of real solutions a to the equation
Here, if x is a real number, then x denotes the greatest integer that is less than or equal to x. 2. Find all real numbers x such that
1 1 1 2 a + 3 a + 5 a = a:
1
1=2 + 1 , 1
1=2 : x = x, x x
3. Let n be a natural number such that n 2. Show that
1 1 + 1 + + 1
n+1 3 2n , 1
1 1 + 1 + + 1
: n 2 4 2n
4. Let ABC be a triangle with BAC = 40 and ABC = 60 . Let D and E be the points lying on the sides AC and AB, respectively, such that CBD = 40 and BCE = 70 . Let F be the point of intersection of the lines BD and CE . Show that the line AF is perpendicular to the line BC .
262
5. Let m be a positive integer. De ne the sequence a0; a1; a2; : : : by a0 = 0; a1 = m; and an+1 = m2an , an,1 for n = 1; 2; 3; : : : . Prove that an ordered pair a; b of nonnegative integers, with a b, gives a solution to the equation a2 + b2 = m2 ab + 1 if and only if a; b is of the form an ; an+1 for some n 0.
The next set of problems is from the twentyseventh annual United States of America Mathematical Olympiad written April 28, 1998. These problems are copyrighted by the committee on the American Mathematical Competitions of the Mathematical Association of America and may not be reproduced without permission. Solutions, and additional copies of the problems may be obtained from Professor Walter E. Mientka, AMC Executive Director, 917 Oldfather Hall, University of Nebraska, Lincoln, NE, USA 68588 0322. As always, we welcome your original, nice" solutions and generalizations which di er from the published solutions.
27th UNITED STATES OF AMERICA MATHEMATICAL OLYMPIAD
Part I
April 28, 1998
9 a.m. 12 noon
1. Suppose that the set f1; 2; ; 1998g has been partitioned into disjoint
pairs fai ; bi g 1 i 999 so that for all i, jai , bij equals 1 or 6. Prove that the sum ends in the digit 9. 2. Let C1 and C2 be concentric circles, with C2 in the interior of C1 . From a point A on C1 one draws the tangent AB to C2 B 2 C2 . Let C be the second point of intersection of AB and C1 , and let D be the midpoint of AB. A line passing through A intersects C2 at E and F in such a way that the perpendicular bisectors of DE and CF intersect at a point M on AB . Find, with proof, the ratio AM=MC . 3. Let a0; a1; ; an be numbers from the interval 0; =2 such that Prove that
ja1 , b1j + ja2 , b2j + + ja999 , b999j
tana0 , + tana1 , + + tanan , n , 1: 4 4 4 tan a0 tan a1 tan an nn+1:
263
usual way. One can select with a mouse any rectangle with sides on the lines of the chessboard and click the mouse button: as a result, the colours in the selected rectangle switch black becomes white, white becomes black. Find, with proof, the minimum number of mouse clicks needed to make the chessboard all one colour. 5. Prove that for each n 2, there is a set S of n integers such that a , b2 divides ab for every distinct a; b 2 S . 6. Let n 5 be an integer. Find the largest integer k as a function of n such that there exists a convex ngon A1 A2 : : : An for which exactly k of the quadrilaterals Ai Ai+1Ai+2 Ai+3 have an inscribed circle. Here An+j = Aj . As a third Olympiad for this issue we give the 11th Grade and 12th Grade problems of the 1994 Latvian Mathematics Olympiad. My thanks go to Bill Sands, of the University of Calgary, who collected these problems for me when he was assisting with the 1995 International Mathematical Olympiad held in Canada.
1 p.m.  4 p.m. 4. A computer screen shows a 98 98 chessboard, coloured in the
Part II
45th LATVIAN MATHEMATICAL OLYMPIAD, 1994
11th Grade
1. Prove for each choice of real nonzero numbers a1; a2 ; : : : ; c3, the stars" can be replaced by " and " so that the system
has no solution. 2. Solve in natural numbers:
8 a x+b y+c 0 1 1 1 a2x + b2y + c2 0 : a3x + b3y + c3 0
xx + 1 = y7
3. Given 4 noncoplanar points, how many parallelepipeds having these points as vertices can be constructed? 4. Let ABCD be a convex quadrilateral, M 2 AB, N 2 BC , P 2 CD, Q 2 DA; AM = BN = CP = DQ, and MNPQ is a square. Prove that ABCD is a square, too. 5. A square consists of n n cells, n 2. A letter is inserted into each cell. It is given that every two rows di er. Prove that there is a column which can be deleted from the square so that all rows are again di erent after this deletion.
264
2. All faces of a convex polytope are triangles. What can be the number of the faces? 3. Does there exist a polynomial P x; y in two variables such that a P x;y 0 for all x; y , b for each c 0 there exist such x and y that P x; y = c? 4. Let Sx be the digital sum of natural number x. Prove that S2n ! 1 when n ! 1, n natural. 5. The centres of four equal circles are the vertices of a square. How must A, B , C , D be chosen so that each circle contains at least one of them and the area of ABCD is as big as possible?
As a nal set of problems for you to puzzle over after the hiatus, we give the problems of the Dutch Mathematical Olympiad, Second Round, written 16 September 1994. My thanks again go to Bill Sands for collecting these problems for me while he was helping out at the IMO in Toronto.
12th Grade 1. Solve the equation cos x cos 2x cos 3x = 1.
DUTCH MATHEMATICAL OLYMPIAD
16 September, 1994
Second Round
1. A unit square is divided in two rectangles in such a way that the smaller rectangle can be put on the greater rectangle with every vertex of the smaller on exactly one of the edges of the greater.
Calculate the dimensions of the smaller rectangle. 2. Given is a sequence of numbers a1; a2; a3; : : : with the property: a = 2; a = 3 and an+1 = 2an,1 or for all n 2:
1 2
Prove that no number between 1600 and 2000 can be an element of the sequence.
an+1 = 3an , 2an,1
265
3. a Prove that every multiple of 6 can be written as the sum of four third powers of integers. b Prove that every integer can be written as the sum of ve third powers of integers. 4. Let P be any point on the diagonal BD of a rectangle ABCD. F is the projection of P on BC . H lies on BC such that BF = FH . PC intersects AH in Q. D C
Q P
H F
A B Prove: Area APQ = Area CHQ. 5. Three real numbers a, b and c satisfy the inequality: ax2 + bx + c 1 for all x 2 ,1; +1 : Prove: cx2 + bx + a 2 for all x 2 ,1; +1 .
We now give the solutions to the Canadian Mathematical Olympiad given earlier this number. I hope you have already solved them all! These o cial solutions" were selected from the most interesting student solutions by Daryl Tingley, University of New Brunswick and Chair of the Canadian Mathematical Olympiad Committee of the Canadian Mathematical Society. Of course the original submissions have been somewhat edited.
1998 CANADIAN MATHEMATICAL OLYMPIAD
Solutions of Students
Let a = 30k + r, where k is an integer and r is a real number between 0 and 29 inclusive. r 1 r Then 1 a = 1 2 1 a =26k + r . 30k + r = 15k + 2 . Similarly 3 a = 10k + 3 and 5 5
1. Solution by David Arthur, Upper Canada College, Toronto, Ontario.
266
1 a + 1 a + 1 a = a, so ,15k + r + ,10k + r + 5 3 ,6k + r =230k + r3 and hence k = r , r , r ,2 r : , 5 2 3 5 Clearly, r has to be an integer, or r , r , r , r will not be an 2 3 5 integer, and therefore, cannot equal k. On the other hand, if r is an integer, then r , r , r , r will also 2 3 5 be an integer, giving exactly one solution for k. For each r 0 r 29; a = 30k + r will have a di erent remainder mod30, so no two di erent values of r give the same result for a. Since there are 30 possible values for r 0; 1; 2; : : : ; 29, there are then 30 solutions for a.
Now,
2. Solution by Jimmy Chui, Earl Haig Secondary School, North York, Ontario. , , 1 , 1 1 Since x , x 1=2 0 and 1 , x 1=2 0; then 0 x , x 1=2 + ,1 , 1 1=2 = x. x 1 Note that x 6= 0. Else, x would not be de ned, so x 0. Squaring both sides gives,
1
1
s 1
1
x2 = x , x + 1 , x + 2 x , x 1 , x r 2 +2 x,1, 1 + 1 : 2 x = x+1, x x x2 Multiplying both sides by x and rearranging, we get p x3 , x2 , x + 2 = 2 x3 , x2 , x + 1 p x3 , x2 , x + 1 , 2 x3 , x2 , x + 1 + 1 = 0 p x3 , x2 , x + 1 , 12 = 0 p3 2 x ,x ,x+1 = 1 x3 , x2 , x + 1 = 1 xx2 , x , 1 = 0 x2 , x , 1 = 0 since x = 0 : 6 p5 Thus x = 12 . We must check to see if these are indeed solutions. p p Let = 1+2 5 ; = 1,2 5 . Note that + = 1; = ,1 and
0
.
267 Since 0; is not a solution. Now, if x = , then
3. Solution 1 by Chen He, Columbia International Collegiate, Hamilton, Ontario.
1 1 + 1 + : : : + 2n 1 1 = 1 + 1 + 1 + 5 + : : : + 2n1 1 3 , 2 2 3 ,
Since 1 gives 1
1=2 + 1, 1 = + 1=2 + 1 + 1=2 since = ,1 = 11=2 + 2 1=2 since + = 1 and = 1, since 0 = since + = 1: So x = is the unique solution to the equation.
,1
1=2
2
= + 1
1 3
1; 1 4 5
1; : : : ; 1 6 2n , 1
1 ; 2n
2
1 1 1 + 3 + : : : + 2n1 1 1 + 2 + 1 + 1 , 2 4 6 1 = 1 + 1 + 1 + 1 + : : : + 1
: + : : : + 2n 2 2 4 6 2n
Since then
1 2
1; 1 4 2
1 1 6; 2
1 ; ::: ; 1 8 2
1 2n
n = 1 + 1 + 1 + :::+ 1 2  2 2z 2 2
n
1 + 1 + 1 + ::: + 1 2 4 6 2n
so that
1 2
1 1 + 1 + 1 + : : : + 1
: n 2 4 6 2n
3
268
1 + 1 + : : : + 2n 1 1 3 , 1 1 + 1 + 1 + : : : + 1
+ 1 + 1 + 1 + : : : + 1
n 4 2n
2 4 6 2n 2 1
1 6 1 1 = 1 + n 2 + 4 + : : : + 2n 1
= n + 1 1 + 4 + : : : + 21 : n 2 n
1 1 ,1 1 1 1 Therefore n+1 1 + 3 + : : : + 2n1,1 n 2 + 4 + : : : + 2n for all n 2 N and n 2.
Then 1, 2 and 3 show that
Solution 2 by Yin Lei, Vincent Massey Secondary School, Windsor, Ontario. Since n 2; nn + 1 0. Therefore the given inequality is equivalent to
We shall use mathematical induction to prove this. , , For n = 2, obviously 1 1 + 1 = 4 1 1 + 1 = 3 . 3 3 9 2 2 4 8 Suppose that the inequality stands for n = k; that is,
1 + : : : + 1
n + 1 1 + 1 + : : : + 1
: n 1+ 3 2n , 1 2 4 2n
We know 1
k + 1 1 + 1 + : : : + 21k : 2 4 Now we have to prove it for n = k + 1.
1 + 3 + : : : + 2k 1 1 , 1 + 1 + : : : + 21k 1
, 1
2 1 4 1
= 1 , 2 + 1 , 4 + 5 , 6 + : : : + 2k 1 1 , 21k 3 , 1 + 1 + 1 + :::+ 1 = 12 34 56 2k , 12k :
34 56
k 1 + 1 + : : : + 2k 1 1 3 ,
1
Since
12
then
:::
2k , 12k
2k + 12k + 2
1 + 1 +::: + 1 12 34 2k , 12k
2k + 12k + 2
k
269 . Hence
1 + 1 + : : : + 2k 1 1 3 ,
Also
1 + 1 + ::: + 1 + k 2 4 2k 2k + 12k + 2 :
2
k+1 , k+2 2k + 1 2k + 2 2 2k + 2 , 2 k = 2k + 2k +2k + 122k+ , 4k , k , 2 = , 2k + 12k + 2 : k 2
Therefore
k+1 = k+2 , k 2k + 1 2k + 2 2k + 12k + 2 :
3
Adding 1, 2 and 3 gives
1 + : : : + 1
+ 1 + 1 + : : : + 1
+ k + 1 1+ 3 3 2
1 2k , 1 1 k
+ 1 12k , 1 k + 1 2 + 1 + : : : + 21k + 2 + 1 + : : : + 2k 4 4 k k + 2k + 12k + 2 + 2kk+ 22 , 2k + 12k + 2 +
.
Rearranging both sides, we get
1 +::: + 1
k + 1 1 + 3 2k + 1
1 1 1
: k + 2 2 + 4 + : : : + 2k + 2
This proves the induction. 4. Solution 1 by Keon Choi, A.Y. Jackson Secondary School, North York, Ontario. Suppose H is the foot of the perpendicular line from A to BC ; construct equilateral 4ABG, with C on BG. I will prove that if F is the point where AH meets BD, then FCB = 70 . Because that means AH , and the given lines BD and CE meet at one point, this answers the question. Suppose BD extended meets AG at I .
270
A E F D I
B G H C Now BF = GF and FBG = FGB = 40 , so that IGF = 20 . Also IFG = FBG + FGB = 80 , so that
FIG = 180 , IFG , IGF = 180 , 80 , 20 = 80 : Therefore 4GIF is an isosceles triangle, so GI = GF = BF : 1 But 4BGI and 4ABC are congruent, since BG = AB , GBI = BAC; BGI = ABC .
Therefore
GI = BC :
From 1 and 2 we get So in 4BCF ,
2
BC = BF :
, BCF = 180 ,2 FBC = 180 2 40 = 70 : Thus FCB = 70 and that proves that the given lines CE and BD and the perpendicular line AH meet at one point.
Solution 2 by Adrian Birka, Lakeshore Catholic High School, Port Colborne, Ontario. First we prove the following lemma: In 4ABC; AA0 ; BB 0 ; CC 0 intersect if and only if
271
sin 1 sin 1 = 1 ; sin 2 sin 2 2 where 1 ; 2 ; 1 ; 2 ; 1 ; 2 are as shown in the diagram just below.
1
sin sin
Editor: This is a known variant of Ceva's Theorem.
B c1 c2 A C0
2 1 1 2
a2 A0 a1
2
D b1 B0 b2
Proof: Let BB 0 C = x; then in 4BB 0 C yields
BB0A = 180
C , x. Using the Sine Law
1
1
b2 = a : sin 2 sin x Similarly using the Sine Law in 4BB 0 A yields b1 = c c = sin x : sin 1 sin180 , x
Hence,
2
c b1 : b2 = a sin sin
1 2 1
3
from 1,2. Editor: Do you recognize this when Similarly,
= 2?
4
By Ceva's Theorem, the necessary and su cient condition for AA0 , a2 b1 : b2 c1 : c2 = 1. Using 3, 4 on this yields:
sin b sin a1 : a2 = c sin 1 ; c1 : c2 = a sin 1 : b 2 2
BB0 , CC 0 to intersect is: a1 :
272
b sin c sin
so that
c a sin 1 a sin b sin 2 sin 2
1
1 2
= 1;
5
sin sin
1
sin 1 sin sin 2 sin 2
1 2
= 1:
This is just what we needed to show; therefore the lemma is proved. Now, in our original question, give BAC = 40 ; ABC = 60 . It follows that ACB = 80 . Since CBD = 40 ; ABD = ABC , DBC = 20 . Similarly, ECA = 20 . B
E
F
K
A C D Now let us show that FAD = 10 . Suppose otherwise. Let F 0 be such that F; F 0 are in the same side of AC and DAF 0 = 10 . Then BAF 0 = BAC , DAF 0 = 30 . Thus
sin ABD sin BCE sin CAF 0 = sin 20 sin 70 sin 10 sin DBC sin ECA sin F 0AB sin 40 sin 10 sin 30 = 2 sinsin20 20 cos 20 20 cos sin30 1 = 2 sin 30 = 1 : By the lemma above, AF 0 passes through CE BD = F . Therefore 0 = AF , and FAD = 10 , contrary to assumption. Thus FAD must AF be 10 . Now let AF BC = K . Since KAC = 10 ; KCA = 80 , it follows that AKC = 90 . Therefore AK ? BC AF ? BC as needed.
+ n Let us rst prove by induction that annan+1+1 = m2 for all n 0: +1
5. Solution by Adrian Chan, Upper Canada College, Toronto, Ontario. a2 a2
2+ 2 2 Proof: Base Case n = 0 : aa0a1a1 = 0+m = m2 : 0+1 0 +1
273 Now, let us assume that it is true for n = k, k 0. Then,
a2 + a2+1 = m2 k k ak ak+1 + 1 a2 + a2+1 = m2 ak ak+1 + m2 k k a2+1 + m4a2+1 , 2m2 ak ak+1 + a2 k k k = m2 + m4 a2+1 , m2 ak ak+1 k 2 2 2 2 2 ak+1 + m ak+1 , ak = m + m ak+1 m2ak+1 , ak a2+1 + a2+2 = m2 + m2 ak+1 ak+2 : k k
+ Therefore aak+1akak+2 = m2 , proving the induction. Hence an ; an+1 k+1 +2 +1 2 is a solution to aab+b2 = m2 for all n 0. +1
2
2
Now, consider the equation aab+b = m2 and suppose a; b = x; y +1 is a solution with 0 x y . Then
2
2
x2 + y2 = m2 : 1 xy + 1 If x = 0 then it is easily seen that y = m, so x; y = a0 ; a1 . Since we are given x 0, suppose now that x 0. Let us show that y m2 x. Proof by contradiction: Assume that y m2x. Then y = m2 x + k where k 1.
Substituting into 1 we get
x2 + m2x + k2 = m2 xm2x + k + 1 2 4 2 x + m x + 2m2xk + k2 = m4x2 + m2kx + m2 x2 + k2 + m2kx , 1 = 0:
Now, m2 kx , 1 0 since kx 1 and x2 + k2 x2 + 1 1 so x + k2 + m2kx , 1 6= 0. Thus we have a contradiction, so y m2 x if x 0. Now substitute y = m2x , x1 , where 0 x1 m2x, into 1.
2
274
x2 + m2x , x1 2 = m2 xm2x , x1 + 1 x2 + m4x2 , 2m2x x1 + x2 = m4x2 , m2x x1 + m2 1 x2 + x2 = m2x x1 + 1 1 x2 + x2 = m2 : 1 2 x x1 + 1 If x1 = 0, then x2 = m2 . Hence x = m and x1 ; x = 0; m = a0; a1. But y = m2x , x1 = a2 , so x; y = a1 ; a2. Thus suppose x1 0. Let us now show that x1 x. Proof by contradiction: Assume x1 x.
2 2 Then m2x , y x since y = m2x , x1 , and xxy+y x , y x +1 a2 +b2 = m2 . since x; y is a solution to ab+1 So x3 + xy 2 x2 y + xy 2 + x + y . Hence x3 x2 y + x + y , which is a contradiction since y x 0. With the same proof that y m2 x, we have x m2 x1 . So the substitution x = m2 x1 , x2 with x2 0 is valid. 2+ 2 Substituting x = m2 x1 , x2 into 2 gives xx1x2x2 = m2 . 1 +1 If x2 = 0, then we continue with the substitution xi = m2 i+1 , xi+2 * 6 x x2 +x2+1 = m2 and x = 0. The sequence x is decreasing, j j until we get xj xj+1 +1 j +1 i
nonnegative and integer. So, if xj +1 = 0, then x2 = m2 so xj = m and xj +1; xj = 0; m = j a0 ; a1. Then xj ; xj ,1 = a1 ; a2 since xj ,1 = m2 xj , xj +1 from *. Continuing, we have x1 ; x = an,1 ; an for some n. Then x; y = an; an+1 .2 Hence a2+b = m2 has solutions a; b if and only if a; b = an; an+1 for ab+1 some n.
We have
That is all we have room for this issue. Enjoy solving the problems  and send me your nice solutions as well as Olympiad contests for use in the corner.
275
BOOK REVIEWS
Models that Work, by Alan Tucker, published by Mathematical Association of America, 1995, ISBN 0883850966, softcover, 88+ pages, US$24.00. Reviewed by Jim Timourian, University of Alberta. At the University of Alberta, like many other institutions, we have large numbers of students enrolling in rst year calculus, but very few outside of engineering students taking more advanced courses. The number of students choosing a mathematical science as a major is small, given the number of students who attend the university. This report documents mathematics programs at a variety of institutions that are very successful in attracting mathematics majors and also students for advanced mathematics courses from other programs. In spite of the fact that the e ective undergraduate programs studied are at diverse institutions, from two year community colleges to Ph.D.granting institutions with outstanding reputations for graduate education, the investigators found a lot in common. Most of the report describes that commonality, while the rest discusses speci c site visits and answers to a set of questions. An e ective program is considered to be one that succeeds in attracting large numbers of students as majors, or prepares students to pursue advanced study in mathematics, or prepares future teachers or attracts and prepares underrepresented groups in mathematics. The departments are uni ed by some underlying philosophies, but otherwise they use diverse methods to achieve their goals. Even within departments there is a variety of approaches used by di erent faculty members. The report explicitly lists themes that are part of the department culture in each of the cases studied. These are respecting students, caring about their welfare, and enjoying the role of being college instructors. The respect for the students is characterized by the comment teaching for the students one has, not the students one wished one had." This is mirrored in the curriculum : : : geared toward the needs of the students, not the values of the faculty." The most important part of a successful department's culture is not explicitly mentioned in the report but is obvious throughout. As a whole a department has to adopt the philosophy that it is a good thing for more students to take more courses in the mathematical sciences. In many departments this idea is not shared by enough of the sta members. There are those who think there are no jobs" so we should not be attracting students. Others think that only the very best students, who have the potential to go on to graduate school in a mathematical science, are worthy and that the rest are weak" and can be ignored. A department that wants to improve its programs needs a sense of what it is trying to achieve. Read this report to learn what comparable institutions have done and how they have gone about doing it.
Edited by ANDY LIU
276
Pythagoras Strikes Again!
K.R.S. Sastry
Is it possible that a given triangle is similar to the triangle formed from its medians? In other words, is it possible that the side lengths of a triangle are proportional in some order to its medians' triangle? For a nontrivial example note the triangle with side lengths a; b; c p 23; 7; 17
p3 p3 =3 and the medians' lengths ma; mb ; mc = 7 2 ; 23 2 ; 17 2 in which a : mb = b : ma = c : mc is possible. You may use the formulas
4m2 = 2b2 + 2c2 , a2 a
1
etc. to calculate the lengths of the medians in the above example. The formulas 1 themselves are derived by means of the cosine rule. The type of triangle as described exists and shall be called a selfmedian triangle 1 . Is it possible, this time, that a given triangle is similar to the triangle formed from its altitudes? That is, the lengths of the sides of a triangle are proportional in some order to the lengths of its altitudes? Again, trivially it's so in an equilateral triangle but nontrivially we have the triangle with the lengths of the sides a; b; c = 6; 9; 4 and the lengths of the altitudes
4 4 4 ha; hb; hc = 26 ; 29 ; 24 , 4 being the area of the triangle. Observe that a : ha = b : hc = c : hb holds. This type of triangle too exists and matching the description of selfmedian triangles we will call the present type selfaltitude triangles 2 . Interestingly, as we shall see, with the exception of the equilateral triangle, both of these triangle types are simply reincarnations of appropriate right triangles. The theorems that follow illustrate this fact.
Theorems on selfmedian triangles
Theorem 1 tells us how an appropriate right triangle yields a selfmedian triangle. THEOREM 1: Let a0 ; b0; c0 be a right triangle in which a2 + b2 = c2 , 0 0 0 a0 b0 and c0 2b0 hold. Then a; b;c = a0 + b0; a0 , b0; c0 is a selfmedian triangle in which a c b holds. Proof: Let m1 ; m2 ; m3 denote the lengths of the medians drawn to the sides a0 + b0; a0 , b0; c0 respectively. Then from 1 we can easily deduce that
4m2 = 3a0 , b02; 4m2 = 3a0 + b0 2; 4m2 = 3c2 : 1 2 3 0
Copyright c 1998 Canadian Mathematical Society
277 Hence
In other words the hypotenuse must exceed twice the shorter leg to give us a selfmedian triangle. It is trivial to check that the other two triangle inequalities are satis ed. Moreover, in the right triangle a0; b0; c0 , we have a0 + b0 c0 a0 , b0 which is the same as a c b. Theorem 2 tells us how to recover the right triangle from a given selfmedian triangle. THEOREM 2: Let a; b; c,be a selfmedian triangle in which a c b 1 holds. Then a0 ; b0 ; c0 = 1 a + b; 2 a , b;c is a right triangle in which 2 a2 + b2 = c2; a0 b0 and c0 2b0 hold. 0 0 0 Proof: Since a c, we nd that rm2 ,m2 = 3c2 ,a2 0, or, ma mc . a c Thus a c b mb mc ma . From this and the de nition of selfb c a median triangle we have mb = ma = mc . We square two of these equations, use the formulas 1 and simplify. This yields a2 , b2 a2 + b2 , 2c2 = 0. The hypothesis a c b shows that this is equivalent to a2 + b2 = 2c2 . We consistently arrive at this same equation no matter which two equations are squared which is the same as 1 a + b 2 + 1 a , b 2 = c2 or a2 +b2 = c2 , 0 0 0 2 2 with a0 b0. Now in triangle a; b; c, we have c a , b so that c0 2b0 holds and the proof is complete. Remark 1: The proof of Theorem 2 says much more than its statement. Since the argument is reversible it has given two characterizations of selfmedian triangles: Suppose the side lengths of triangle a; b; c satisfy a c b. Then the triangle is selfmedian i if and only if a : mb = b : ma = c : mc or ii if and only if a2 + b2 = 2c2 . Also, it is easy to see that if triangle a; b; c has a : ma = b : mb = c : mc then it must be equilateral. This useful observation enables us to give an elegant proof of Theorem 3. Theorems 1 and 2 considered the determination of selfmedian triangles in each of which the sides have distinct length measures. Theorem 3 shows that if an isosceles triangle is selfmedian then it must be equilateral. THEOREM 3: The equilateral triangle is the only selfmedian triangle not covered by the Theorems 1 and 2. Proof: Any selfmedian triangle a; b; c not covered by the Theorems 1 and 2 must have at least two sides equal, say a = b. Then ma = mb . Let
a0 + b0 = a0 , b0 = c0 = p : 2 m2 m1 m3 3 It should be noted that the lengths a0 + b0 ; a0 , b0 ; c0 do not form a triangle for every given right triangle a0 ; b0 ; c0 . For example a0; b0 ; c0 = 4; 3; 5 yields a0 +b0 ; a0 ,b0 ; c0 = 7; 1; 5, and there is no triangle with these side lengths. However a0; b0 ; c0 = 12; 5; 13 yields the selfmedian triangle a0 + b0 ; a0 , b0; c0 = 17; 7; 13. In order to assure the formation of a selfmedian triangle, the lengths a0 + b0 ; a0 , b0 ; c0 have to satisfy the triangle inequality a0 , b0 + c0 a0 + b0. This simpli es to c0 2b0 .
278
m1; m2; m3 denote the medians ma; mb; mc in some order so that the defa = b = c: m1 m2 m3 If m1 = mc then m2 = m3 = ma = mb . But then a = b forces ma = mb = mc and hence the triangle is equilateral. If m1 = m2 = ma = mb then m3 = mc. This again forces the triangle to be equilateral as
inition of a selfmedian triangle enables us to write
mentioned in Remark 1. Exercise 1: Find the right triangle that generates the selfmedian triangle 23; 7; 17. Exercise 2: Suppose p b; c is a right triangle with a2 + b2 = c2 . Show that p a; the triangle a 2; b 2; c is selfmedian. Exercise 3: Show that there are exactly two selfmedian triangles in each of which i all the side lengths are integers and ii two side lengths are 7 and 17.
Theorems on selfaltitude triangles
Theorem 4 tells us how an appropriate right triangle yields a selfaltitude triangle. THEOREM 4: Let a0 ; b0; c0 be a right triangle in which a2 + b2 = c2 and 0 0 0 a0 2b0 hold. Then a; b; c = a0; c0 +b0; c0 ,b0 is a selfaltitude triangle in which b a c holds. Proof: Let ha ; hb ; hc denote the altitudes to the sides a; b; c and 4 denote the area of this triangle. Then aha = bhb = chc = 24 or
a0ha = c0 + b0hb = c0 , b0hc = 2 4 :
Now or
a0 = a2 = a2 = c2 , b2 = c2 , b2 0 0 0 0 0 0 ha a0ha 24 c0 , b0hc c0 + b0hb
a0 = c0 + b0 = c0 , b0 a = b = c ha hc hb ha hc hb implying that the triangle a; b; c is selfaltitude. Furthermore, in triangle a0 ; b0; c0, we have that c0 + b0 a0 c0 , b0 holds. This in triangle a; b; c is equivalent to b a c. Again, not any right triangle a0 ; b0 ; c0 yields a selfaltitude triangle. For example, a0 ; b0; c0 = 3; 4; 5 yields a; b; c = 3; 9; 1 and there is no triangle with these sidelengths. However, a0 ; b0 ; c0 = 12; 5; 13 yields a; b; c = 12; 18; 8. We may divide these side lengths by their gcd 2 and
279 take the primitive selfaltitude triangle 6; 9; 4. Here the triangle inequality to be satis ed is a0 + c0 , b0 c0 + b0 which simpli es to a0 2b0. It is trivial to check that the other two triangle inequalities are satis ed. Theorem 5 tells us how to recover the right triangle from a given selfaltitude triangle. THEOREM 5: Let a; b;c be a selfaltitude triangle in which b a c holds. Then
a0; b0; c0 = a; 1 b , c; 1 b + c 2 2 2 + b2 = c2 and a is a right triangle in which a0 0 0 2b0 hold. 0
Proof: From the hypothesis b a c we have hb de nition of selfaltitude triangle we have
ha
hc. From the
which is a2 + b2 = c2 . From a b , c follows a0 2b0 as required. 0 0 0 Remark 2: The proof of Theorem 5 gives two characterizations of selfaltitude triangles: Suppose the side lengths of triangle a; b; c satisfy b a c. Then the triangle is selfaltitude i if and only if a : ha = b : hc = c : hb , or ii if and only a2 = bc. It is easy to see that if in triangle a; b; c we have a : ha = b : hb = c : hc then it must be equilateral. This observation enables us to give an elegant proof of Theorem 6. Theorems 4 and 5 considered the determination of selfaltitude triangles in each of which the sides have distinct length measures. Theorem 6 shows that if an isosceles triangle is selfaltitude then it must be equilateral. THEOREM 6: The equilateral triangle is the only selfaltitude triangle not covered by the Theorems 4 and 5. Proof: We omit. It is similar to that of Theorem 3. Exercise 4: Find the right triangle that generates the selfaltitude triangle 35; 49; 25. Exercise 5: Let ABC be a right triangle with right angle at C . CD is drawn perpendicular to AB with the point D on AB . Prove that the triangle whose
a = b = c a2 = bc = bc ; ha hc hb 24 24 24 where 4 is the area of triangle a; b; c. Thus 1 2 2 a2 = bc = 2 b + c , 1 b , c 2
280 side lengths are AD, DB; CD is selfaltitude. In terms of a; b; c give an answer to the question: when do the lengths AD; DB; CD form a triangle? Remarks 1 and 2 suggest the following. OPEN PROBLEM: Suppose AD; BE; CF are three concurrent cevians of triBC angle ABC . Assume that AD = CA = AB holds. Prove or disprove that BE CF the triangle ABC is equilateral. Acknowledgement: I thank the referee for his painstaking e orts to improve the presentation. He also suggested Theorems 3 and 6.
References
1 K.R.S. Sastry, SelfMedian Triangles, Mathematical Spectrum, 22 1989 90, pp. 5860. 2 K.R.S. Sastry, SelfAltitude Triangles, Mathematical Spectrum, 22 1989 90, pp. 8890. K.R.S. Sastry Jeevan Sandhya Doddakalsandra Post Raghuvanahalli Bangalore 560 062, India
Advance Notice
At the summer 1999 meeting of the Canadian Mathematical Society, to be held in St. John's, Newfoundland, there will be a Mathematics Education Session on the topic What Mathematics Competitions do for Mathematics". Invited speakers include Edward Barbeau, Toronto; Tony Gardner, Birmingham, England; Ron Dunkley, Waterloo; and Rita Janes, St. John's. Anyone interested in givinga paper at this session should contact one of the organizers, Bruce Shawyer or Ed Williams, at the Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. Email addresses:
bshawyer@math.mun.ca ewilliam@math.mun.ca
281
THE SKOLIAD CORNER
No. 31 R.E. Woodrow
As a contest this month we give the U.K. Intermediate Mathematical Challenge which was written Thursday, 5th February 1998. This contest is organized from the United Kingdom Mathematics Trust from the School of Mathematics, the University of Leeds. The contest is open to students in School year 11 or below. My thanks go to John Grant McLaughlin of the Faculty of Education, Memorial University of Newfoundland, who sent me a copy of the questions.
U.K. INTERMEDIATE MATHEMATICAL CHALLENGE February 5, 1998
Time: 1 hour
Five marks are awarded for each correct answer to questions 1 15. Six marks are awarded for each correct answer to questions 16 25. Each incorrect answer to questions 16 20 loses 1 mark. Each incorrect answer to questions 21 25 loses 2 marks. 1. One quarter of a number is 24. What is one third of the original number? a 6 b 8 c32 d 72 e 96 2. 6 of 6 plus 8 of 8 equals a 0:14 b 1 c1:4 d 1:96 e 2 3. Starting at A, a point on a xed circle with centre, O, I rst move anticlockwise one quarter of the way round the circle to a point, W , hop across to X  the opposite end of the diameter through W , then travel one fth of the way round the circle clockwise to the point Y before hopping across to Z , the point at the opposite end of the diameter through Y . How big is AOZ ? a 18 b 22 c162 d 198 e 270 4. Which fraction is the odd one out? 20 02 8 a 1+4 b 140 c 1::4 d 111 e 56 7+4 711 5. J is the set of High Court judges; K is the set of living things beginning with K ; L is the set of all living creatures; M is the set of brilliant mathematicians. Kevin is a very ordinary kangaroo. In which of the ve regions AE of the diagram does Kevin belong?
282
J K
A E L
B D M C
a A
b B c C d D e E 6. ABCD is a square with sides of length 9 cm. How many points inside or outside the square are equidistant from B and from C , and are exactly 6 cm from A? a 0 b 1 c 2 d 3 e more than 3 7. Each person's birthday product is obtained by multiplying the day of the month in which they were born by the number of the month in which they were born, and then multiplying the answer by the year in which they were born. Here are ve English queens and their birthdays. Which of them has the same birthday product as someone born today? Ed. today" is February 5, 1998. a Mary I, 18 February 1516 b Elizabeth I, 7 September 1533 c Anne, 6 February 1665 d Victoria, 24 May 1819 e Elizabeth II, 21 April 1926 8. How large will an angle of 2 1 appear to be if you enlarge it by look2 ing through a stack of ve magnifying glasses, each one of which magni es by a factor of 2? a 2 1 b 12 1 c 25 d 40 e 80 2 2 9. What is the total number of letters in all the incorrect options for this question? a eleven b twenty two c thirty three d forty four e fty ve 10. On four tests, each marked out of 100, my average was 85. What is the lowest mark I could have scored on any one test? a 0 b 40 c 60 d 81 e 85 11. The World Wide Fund For Nature estimates that 54 acres of Brazilian rainforest are destroyed every minute of every day. Approximately how many acres are lost each week? a 50; 000 b 80; 000 c 200; 000 d 500; 000 e 2; 000; 000
283
12. If C Celsius is the same temperature as F Fahrenheit, then F = 9 C + 32. To avoid working with fractions and awkward numbers, some 5 people use the approximate formula F 0 = 2C +30. What is the temperature in degrees Celsius when the approximate formula gives an answer which is too large by 1? a 5 b 9 c 10 d 12 e 15 13. I fold a piece of paper in half, then in half again before cutting a shape from the folded paper as shown.
y X X
A
When I unfold the paper, what do I see?
B
C
D
E
a A
b B c C d D e E 14. In a sponsored Animal Streak" the cheetah ran at 90 km hr while the snail slimed along at 20 hr km. The cheetah kept going for 18 seconds. Roughly how long would the snail take to cover the same distance? a 9 months b 9 weeks c 9 days d 9 hours e 9 minutes 15. Wallace and Gromit are waiting in a queue. There are x people behind Wallace, who is y places in front of Gromit. If there are n people in front of Gromit, how long is the queue? a n , x + y + 2 b n + x , y c n , x + y , 1 d n + x , y + 1 e n , x + y 16. I made just enough sticky treacle mixture to exactly ll a square tin of side 12 inches. But all I could nd were two 8 1 inch square tins. How 2 well would the mixture t? a easily b just with a teeny bit of room to spare c an exact t d nearly with a small over ow e no way major over ow
284
17. Which of the four triangles W , X , Y , Z are rightangled?
5 42 6x 132 4x
x
W
X
Y
3x 4x
6
Z
4
5x
a only W b W and X c X and Z d Y and W e Y and Z 18. The integers from 1 to 20 are listed below in such a way that the sum of each adjacent pair is a prime number. Missing numbers are marked as s.
Which number goes in the place which is underlined? a 1 b 3 c 11 d 13 e 19 19. ABCDE is a regular pentagon. FAB is a straight line and FA = AB. What is the ratio x : y : z?
F x E D y A B z
20; ; 16; 15; 4; ; 12; ; 10; 7; 6; ; 2; 17; 14; 9; 8; 5; 18; :
C
a 1 : 2 : 3 b 2 : 2 : 3 c 2 : 3 : 4 d 3 : 4 : 5 e 3 : 4 : 6 20. The total length of all the edges of a cube is L cm. If the surface area of the cube has the same numerical value L cm2 , what is its volume in cm3 ? a 1 b L c 2 d L3 e 8 21. A piece of thin card in the shape of an equilateral triangle with side 3 cm and a circular piece of thin card of radius 1 cm are glued together so that their centres coincide. How long is the outer perimeter of the resulting 2 dimensional shape in cm? a 2 b 6 + c 9 d 3 e 9 + 2 22. Shape A is made from 6 unit squares; shape B is made from 8, C from 4, D from 8 and E from 3 unit squares. For four of these shapes, four exact copies can be tted together to make a rectangle. Which is the odd one out?
285
A
a A
B
C
D
E
b B c C d D e E 23. In this unusual game of noughts and crosses the rst player to form a line of three Os or three X s loses. It is X 's turn. Where should she place her cross to make sure that she does not lose?
A O B C X D E X O
a A b B c C d D e E 24. Each of the sides of this regular octagon has length 2 cm. What is the di erence between the area of the shaded region and the area of the unshaded region in cm2 ?
r
b 1 c 1:5 d 2 e 2 2 25. A square is inscribed in a 3 4 5 rightangled triangle as shown. What fraction of the triangle does it occupy? a 0
p
3
a b
5 4
12 25
24 49
1 c 2
d 25 49
e 13 25
That completes this number of the corner. Send me contest materials, nice solutions, and any suggestions about items you would like see covered in the Skoliad Corner.
286
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Shreds and Slices
Primitive Roots and Quadratic Residues, Part 2 Recall the work in Part 1 1998: 220 : Let p be an odd prime, and Ap , Bp the quadratic residues and nonresidues respectively. Let
be a primitive th pp,1root of unity; that is,
p = 1 and
6= 1. These imply 1 +
+
2 + +
= 0, an important fact we will use later. Let
x=
X
Then x + y = ,1, and we saw xy always seems to be an integer, but why, and which one? We introduce some basic quadratic residue theory. Let a be a nonzero integer modulo p. We de ne the Legendre symbol a as follows: p Although this notation has some deep consequences, we should think of it as just that: notation. We will now prove a useful relation, known as Euler's Criterion:
a2Ap
a;
y=
X
b2Bp
b :
a
p =
1 if a is a quadratic residue; ,1 if a is a quadratic nonresidue:
a ap,1=2 mod p : p If a is a q.r. quadratic residue, then x2 a for some x, so ap,1=2 xp,1 1 mod p by Fermat's Little Theorem. Otherwise, if a is a nonq.r.
287 quadratic nonresidue, then for all x, 1 x p , 1, there is a unique y , 1 y p , 1, y 6= x, such that xy a. All integers in f1; 2; : : : ; p , 1g can be so paired. Their product is ap,1=2 1 2 p , 1 p , 1! ,1 mod p, by Wilson's Theorem. As a corollary, note that ,1
1 if p 1 mod 4; p = ,1 if p 3 mod 4:
Consider the case p 3 mod 4, so p = 4k + 3 for some k, and p , 1=2 = 2k + 1. Let a1,1a2 , : : : , a2k+1 be the q.r.'s and b1 , b2, : : : , , b2k+1 the nonq.r.'s. Since p = ,1, a is a q.r. if and only if ,a p , a is a nonq.r.. Hence, for all i, there exists a unique j such that ai + bj = p. The expression we wish to calculate is
xy =
a1 +
a2 + +
a2k+1
b1 +
b2 + +
b2k+1 =
a1+b1 +
a1+b2 + +
a1 +b2k+1 +
a2+b1 +
a2+b2 + +
a2+b2k+1 + +
a2k+1 +b1 +
a2k+1+b2 + +
a2k+1+b2k+1 :
Note that there are 2k + 12 terms, and there is a
p = 1 term in each row. This leaves 2k + 12 , 2k + 1 = 2k2k + 1 terms. We claim that
,
2, : : : ,
p,1 =
4k+2 appear the same number of times among these 2k2k + 1 terms. Let n be a nonzero integer modulo p. We wish to nd the number of ways n can be expressed as the sum of a q.r. u and a nonq.r. v: n u + v. Assume 1 can be expressed in t distinct such ways: 1 u1 + v1 u2 + v2 = ut + vt. Then n nu1 + nv1 nu2 + nv2 nut + nvt. If n is a q.r., then so is nui, and nvi remains a nonq.r.. Otherwise, nui and nvi switch roles, and in either case, there are again t distinct ways of expressing n as the sum of a q.r. and a nonq.r.. Therefore, among the 2k2k + 1 terms, each power of
appears 2k2k + 1=p , 1 = k times. Hence,
xy = 2k + 1 + k
+
2 + +
4k+2 = 2k + 1 , k = k + 1 = 1 + p : 4
The case p 1 mod 4, say p = 4k + 1, is similar, and is only brie y sketched here. In this case, ,1 = 1, so a and ,a p , a both occur as p either ai or bj . Hence, when we expand xy to obtain 2k2 = 4k2 terms,
288 there are no terms of the form
p = 1, and using the same reasoning, the powers of
appear evenly among the 4k2 terms, so
k2 xy = p4, 1
+
2 + +
4k = ,k = 1 , p : 4
In general, we may write
Hence, x and y are the roots of the quadratic
1 , ,1 p p : xy = 4
which are namely
1 , ,1 p p t2 + t + 4
,1
x , y = x , 2xy + y = x + y , 4xy = p p:
Hence, it is possible to express ,1 p in terms of any primitive pth root of p
2 2 2 2
Which is x and which is y ? It depends on the choice of
. Note that the discriminant of this quadratic is
s ! 1 ,1 ,1
p : 2 p
unity.
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only problems  the next issue will feature only solutions. We warmly welcome proposals for problems and solutions.
289
High School Problems
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca
n
1 1998 + 2 1997 + 3 1996 + + 1997 2 + 1998 1 = 3 :
the inequality
H241. Find the integer n that satis es the equation
2
H242. Let a and b be real numbers that satisfy a + b
2 2
2
ja b + ab j 2 : Determine the values of a and b for which equality occurs. See how many
ways you can solve this! H243. For a positive integer n, let f n denote the remainder of n2 + 2 when divided by 4. For example, f 3 = 3 and f 4 = 2. Prove that the equation
p2
= 1. Prove
x2 + ,1yf z = 10y has no integer solutions in x, y , and z . H244. For a positive integer n, let P n denote the sum of the digits of n. For example, P 123 = 1 + 2 + 3 = 6. Find all positive integers n n satisfying the equation P n = 74 .
Advanced Problems
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu A217. Proposed by Alexandre Trichtchenko, OAC student, Brook eld High School, Ottawa. Show that for any odd prime p, there exists a positive integer n such that n, nn , nnn , : : : all leave the same remainder upon division by p where n does not leave a remainder of 0 or 1 upon division by p. A218. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. a Suppose f x = xn +qxn,1 +t, where q and t are integers, and suppose there is some prime p such that p divides t but p2 does not divide t. Show, by imitating the proof of Eisenstein's Theorem, that either f is irreducible or f can be reduced into two factors, one of which is linear and the other irreducible.
290 b Deduce that if both q and t are odd then f is irreducible. Generalization of Question 1, IMO 1993 A219. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. Solve the following system:
A220. Proposed by Waldemar Pompe, student, University of Warsaw, Poland. P is an interior point of triangle ABC . D, E, F are the feet of the perpendiculars from P to the lines BC , CA, AB , respectively. Let Q be the interior point of triangle ABC such that ACP = BCQ and BAQ = CAP : Prove that DEF = 90 if and only if Q is the orthocentre of triangle BDF .
1
1
3 x+ x = 4 y+ y = 5 z+ 1 ; z xy + yz + zx = 1 :
Challenge Board Problems
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu
C79. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. Let f : R+ ! R+ be a nonincreasing function, and assume that there are two constants p 0 and T 0 such that
Z1
t
f sp+1 ds Tf 0pf t
for all t 2 R . Prove that
+
for all t T . C80. Suppose a1 , a2, : : : , am are transpositions in Sn the symmetric group on n elements such that a1 a2 am = 1. Show that if the ai generate Sn, then m 2n , 2.
1 T + pt
, p f t f 0 T + pT
291
The Pentagram Theorem
teacher, Todaijigakuen Junior High School, Japan A pentagram ABCDE is a vepointed starshaped gure made by extending the sides of a convex pentagon FGHIJ until they meet, and is sometimes used as a mystic symbol.
A E I H D J F G
Hiroshi Kotera
B
AC DB EB FC = 1 : AB DC EC FB Proof. Let Hi i = 1; 2; 3; 4 be points on side BC such that DH1 , AH2, FH3, and EH4 are all perpendicular to BC .
A D F B E C
In my math class, Eiji Konishi, a Todaijigakuen Junior High School student in Japan, found a beautiful theorem concerning the pentagram. I would like to show the theorem he proved by himself. We need two lemmas. Lemma 1. In triangle ABC , let D and E be points on sides AB and AC respectively, and let F be the intersection of BE and CD. Then
C
H1 H2 H3 H4
Copyright c 1998 Canadian Mathematical Society
292 Since we have
FC FH3 = DH1 DC
and
DH1 = AH2 DB ; AB
1
Similarly, since we have
FC FH3 = AH2 DB DC : AB
and
FB FH3 = EH4 EB
EH4 = AH2 EC ; AC
2
Equations 1 and 2 imply It follows that
FB FH3 = AH2 EC EB : AC DB FC = EC FB : AB DC AC EB
AC DB EB FC = 1 : AB DC EC FB Lemma 2. Let D and E be two points on side AB of triangle ABC , and let F be a point on side BC . Let G be a point on side BC , extended. Let H be the intersection of DG and AC , let I be the intersection of DG and AF , and let J be the intersection of GE and AC . Then DE BF CJ HI = 1 : EB FC JH ID Proof. Let L be the point on AC such that BLkDG, and let N be the point on DG such that NB kAC . Let M be the intersection of AF and BL, and let P be the intersection of GE and BN . By Menelaus's Theorem in triangles BCL and BDN , we have BF CA LM = ,1 ; 3 FC AL MB BP NG DE = ,1 : 4 PN GD EB
By similar triangles,
LM = HI ; MB ID
293
N D P B E A I F M L
H J C G
and so by 3,
AH = AL ; AD AB BF LM = BF HI FC MB FC ID AL = , CA = , AH AB : AD AC BP = CJ ; PN JH GC = GB ; GH GN BP DE CJ DE PN EB = JH EB GD GD GC = , NG = , GH GB :
5
Similarly, and so by 4,
6
Equations 5 and 6, and Lemma 1 imply It follows that
BF HI CJ DE = AH AB GD GC = 1 : FC ID JH EB AD AC GH GB DE BF CJ HI = 1 : EB FC JH ID
294
Theorem. Let A1 A2 A3 A4 A5 be a pentagram with pentagon B1B2B3B4B5 as shown in the following gure. Let Ck be the intersection of the line Ak Bk with the side Bk+2 Bk+3 , k = 1, 2, 3, 4, 5. Then
B3C1 B4C2 B5C3 B1C4 B2C5 = 1 : C1B4 C2B5 C3B1 C4B2 C5B3
A1 A5 B2 B1 A4 A3
Proof. Lemma 2 yields
B3
C1
B4 B5
A2
B3C1 B4C2 = B1A4 B2B3 ; C1B4 C2B5 B5B1 A4B2
and
B5C3 B1C4 = B3A1 B4B5 : C3B1 C4B2 B2B3 A1B4
By Menelaus's Theorem in triangle A4 B5 C5 , we have
A4B1 B5A5 C5B2 = ,1 ; B1B5 A5C5 B2A4
and in triangle A1 B5 C5 , we have
A1B4 B5A5 C5B3 = ,1 : B4B5 A5C5 B3A1
295 It follows that
= = = =
B3C1 B4C2 B5C3 B1C4 B2C5 C1B4 C2B5 C3B1 C4B2 C5B3 B1A4 B2B3 B3A1 B4B5 B2C5 B5B1 A4B2
2 3 A1B4 C5
3 B1A4 B2C5 B BB3A1 B4B5 B B5B1 A4B2 A1B4 C5B3 A5C5 B5A5 B5A5 A5C5 1:
This result may be something already well known, but I presume the fact that its discovery has been made by a junior high school student is signi cant.
Have you heard about ATOM?
ATOM is A Taste Of Mathematics" Aime T On les Math
matiques. e The booklets in the series, A Taste of Mathematics, are published by the Canadian Mathematical Society CMS. They are designed as enrichment materials for high school students with an interest in and aptitude for mathematics. Some booklets in the series will also cover the materials useful for mathematical competitions at national and international levels. Publi
s par la Soci
t
math
matique du Canada SMC, les livrets de la e ee e collection Aimeton les math
matiques ATOM sont destin
s au perfectione e nement des etudiants du cycle secondaire qui manifestent un int
r^ t et des
ee aptitudes pour les math
matiques. Certains livrets de la collection ATOM e servent egalement de mat
riel de pr
paration aux concours de math
mati
e e e ques sur l'
chiquier national et international. e This volume contains the problems and solutions from the 1995 1996 Mathematical Olympiads' Correspondence Program. This program has several purposes. It provides students with practice at solving and writing up solutions to Olympiadlevel problems, it helps to prepare student for the Canadian Mathematical Olympiad and it is a partial criterion for the selection of the Canadian IMO team. For more information, contact the Canadian Mathematical Society.
The ATOM series
La collection ATOM
296
A Combinatorial Triad
student, University of Toronto One of the beautiful aspects of mathematics is that solving a problem can be done in so many ways. In combinatorics, the study of counting and ways of counting, this is more than evident in its many interesting problems. See any standard textbook on combinatorics 1 to convince yourself. Consider the following three standard combinatorial problems and see if you can nd the link between them. Problem 1. What is the number of ways to distribute 10 identical apples into 3 baskets labelled A, B , and C ? Problem 2. How many ways are there to select 10 icecream scoops from 3 di erent avours on an icecream cone? Problem 3. Find the number of nonnegative integer solutions to the diophantine equation x1 + x2 + x3 = 10. If you said the answers are all the same, then you are correct. The three problems above are di erent ways of counting something equivalent. Each way of thinking may be more helpful in certain situations than others. The following are solutions to approaching the apparently di erent, but similar, problems. Solution to Problem 1. The standard approach to solving this type of problem rests on the following key trick. Let the apples be represented by 1's since they are all identical. Now use two 0's to place between the 1's to split them into 3 piles. Put the rst pile into basket A, the second into basket B and the remaining pile into basket C . The problem then reduces to nding the number of ways of placing the 0's among the 1's. There are 10 + 1 = 11 positions to place the rst 0, namely 9 spots between the 10 1's and the front and back of the list of 1's. Now there are 11 numbers and 11 + 1 = 12 positions to place the second 0. However, we are counting everything twice! If we place 0 in front and then a 0 after the fth 1, this is the same as if we place the rst 0 after the fth 1 and the second 0 in the rst spot. Hence, there are ways in total. Solution to Problem 2. This is exactly the same as Problem 1 if we consider the icecream scoops to be 1's and the type of avour it is by its
Copyright c 1998 Canadian Mathematical Society
Cyrus Hsia
11 12 2
297 position in one of the three piles of numbers. We use two 0's again to be dividing points for the three piles. The number of ways of selecting scoops from three avours becomes a problem of arranging the 10 1's and two 0's. There are
ways of doing this. Solution to Problem 3. Here, the problem and solution have become completely algebraic. The standard way of solving this would be to use generating functions. Consider the following function: Now consider the coe cient of x10 . How do you get this term? Take xa in the rst factor, xb in the second, and xc in the third so that a + b + c = 10. Thus, we want all possible ways of doing this, so that a, b, and c are solutions to the equation in the problem: x1 + x2 + x3 = 10. The coe cient is the number of nonnegative integer solutions to the equation! , Of course, one must check that the coe cient of x10 is indeed 12 . 2 Here are some problems to drive this idea home.
1. Reformulate the following problems to the other two types and solve them. a How many ways are there to select 98 DNA codons if there are 4 di erent codons to choose from? b Solve the diophantine equation x1 + x2 + + x19 = 97, with xi 3 for all i = 1, 2, : : : , 19. c How many ways are there to put 98 pigeons into 97 pigeonholes? Aside: Must one pigeonhole have more than one pigeon? 2. How many ways are there to distribute 9 apples, 8 bananas and 7 oranges to 6 hungry school children? How would you reformulate this problem to the other two? 1. How many ways are there to select 3 numbers from the integers 1 to 20 so that no two are consecutive? 2. How many ways are there to distribute 10 distinct balls into three tubes labelled A, B , and C taking the order of the balls in each tube into account? Careful! This is harder than meets the eye.
Reference Other Exercises Standard Exercises
12 = 11 12 2 2
1 + x + x2 + x3 + 3:
1 Tucker, Alan. Applied Combinatorics. John Wiley & Sons, Inc., Toronto. 1995 pp. 205.
298
Swedish Mathematics Olympiad
1987 Qualifying Round
1. Kalle has a litre ask A full of orange juice and an empty litre ask B . He pours part of the juice from the full ask A to the empty ask B . He then adds water to B until it is full, and shakes it so that the mixture is thoroughly blended. Lastly, he adds the blended mixture from ask B to ask A until ask A is full. Show that ask A now contains at least 75 of the original juice. 2. Show that the number abc, which denotes a 3digit number written in the usual way, is divisible by 7 if and only if the number ,a + 2b , 3c is divisible by 7. 3. Solve the system of equations
x + yx2 , y 2 = 1176 x , yx2 + y 2 = 696:
4. In the triangle ABC , arbitrarily choose a point P on the side BC . Through the midpoint M of the same side, draw DM parallel to AP , where D is a point on one of the other two sides. Show that the line segment DP divides the triangle into two regions of equal area. 5. Sixteen light bulbs are connected in a square network as shown in the gure. Each row and column has a switch which works as follows: one halfturn of the switch extinguishes all the lighted bulbs in the row or column, and turns on all the bulbs that are o in that row or column.
In the initial state, six bulbs are on, as shown in the gure. Prove that it is impossible to turn on all 16 bulbs, using the switches. What is the maximum number of bulbs that can be turned on?
299 6. A small school with 50 pupils has put the pupils into 8 groups. There are 4 more pupils in the largest group than there are in the smallest group. These groups are combined two by two into larger groups as follows. The two smallest of the original groups are combined, numbers 3 and 4 form the next combined group, and so on. The largest of the four groups so formed now has 5 more pupils than the smallest. The two smallest of these groups form class A and the two largest form class B. Class B has 6 more pupils than class A. Find the numbers of pupils in the original groups.
1987 Final Round
1. Sixteen real numbers are arranged in a magic square" of side 4 in such a way that the sum of the numbers in each row, in each column and in each of the two main diagonals is always k. Prove that the sum of the four numbers in the corners of the square is also k. 2. A circle of radius R is divided into two equal parts by the arc of another circle. Show that the arc of this circle which lies inside the given circle is longer than 2R. 3. Assume that 10 closed intervals all of length 1 are placed in the interval 0; 4 . Show that there exists some point in the larger interval that belongs to at least 4 of the smaller intervals. 4. f is a di erentiable function de ned on the interval 0 x 1 and f 0 = f 1 = 1. Show that there exists at least one point y in the interval such that Z1
jf 0yj = 4
5. The numbers a, b, c, and d are positive and abcd = 1. Show that there exists a positive number t such that for all such a, b, c, and d, Find the largest t with this property. 6. A baker with access to a number of di erent spices bakes 10 cakes. He uses more than half of the di erent kinds of spices in each cake. No two of the combinations of spices are exactly alike. Show that there exist three spices a, b, c, such that every cake contains at least one of these.
0
jf xj dx:
1 + 1 + 1 + 1 1+a 1+b 1+c 1+d
t:
300
1. Prove that in any triangle ABC , where AD is a median and jADj = m, 4m2 = b2 + c2 + 2bc cos A. 2. Find the equations of the three normals to the parabola with the equation y 2 = 4x, from 21; 30. Note: A normal ` to a curve C at a point P is a line ` which intersects C at P and is perpendicular to the tangent of C at P . 3. Find the equations of the four common tangents of the circles having equations x2 + y 2 = 16 and x , 252 + y 2 = 121. 4. A tank is lled using two taps. It takes taps one and two t1 and t2 minutes respectively to ll the tank each on its own. From the following information, determine what the values t1 and t2 are: i If the rst tap is open for onethird of the time t2 and the second tap is open for onethird of the time t1 , then the fraction of the volume of the tank that is lled is 11=18. ii If both taps are used simultaneously, it would take 3 hours and 36 minutes to ll the tank. 5. Two bodies are moving with constant acceleration in a straight line going in the same direction. The rst body starts 20 km ahead of the other. It travels a distance of 25 and 50=3 km in 1 and 2 hours respectively. The second body travels a distance of 30 and 59=2 km in 1 and 2 hours respectively. How long does it take for the second body to catch up to the rst? Editor's comment. Problem 4 of the above contest, as stated, has no solution. The equation arising from condition i is not solvable for positive t1 and t2 . However, 11 the problem is solvable with nice values if the fraction 18 is replaced by 13 . 18
J.I.R. McKnight Problems Contest 1983
301
PROBLEMS
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 March 1999. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Prove or disprove that if A, B and C are the angles of a triangle, then
2326? Correction.
2
X ,1 , sin A ,1 + 2 sin A 2 2 29 : ,A cyclic
Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Suppose that p and t 0 are real numbers. De ne pt := tp + t,p + 2p and pt := ,t + t,1p + 2 : a Show that p t p t for p 2. b Determine the sets of p: A, B and C , such that
2329? Correction.
302 1. 2. 3.
pt pt, pt = pt, pt pt.
2351. Proposed by Paul Yiu, Florida Atlantic University, Boca Raton, Florida, USA. A triangle with integer sides is called Heronian if its area is an integer.
Does there exist a Heronian triangle whose sides are the arithmetic, geometric and harmonic means of two positive integers? UK.
2352.
Proposed by Christopher J. Bradley, Clifton College, Bristol,
Determine the shape of 4ABC if
UK.
2353.
cos A cos B cosA , B + cos B cos C cosB , C + cos C cos A cosC , A + 2 cos A cos B cos C = 1 :
Proposed by Christopher J. Bradley, Clifton College, Bristol,
2354. Proposed by Herbert Gulicher, WestfalischeWilhelmsUniversitat, Munster, Germany. In triangle P1P2 P3 , the line joining Pi,1 Pi+1 meets a line j at the point Si;j i; j = 1; 2; 3, all indices taken modulo 3, such that all the points Si;j , Pk are distinct, and di erent from the vertices of the triangle. 1. Prove that if all the points Si;j are noncollinear, then any two of the following conditions imply the third condition:
P1S3;1 P2S1;2 P3S2;3 = ,1; S3;2P2 S1;3P3 S2;1P1 S S S S S S b 1;2 1;1 2;3 2;2 3;1 3;3 = 1; S1;1S1;3 S2;2S2;1 S3;3S3;2
a
sin A sin B sinA , B + sin B sin C sinB , C + sin C sin A sinC , A = 0 :
Determine the shape of 4ABC if
c 1 , 2 , 3 are either concurrent or parallel. 2. Prove further that a and b are equivalent if the Si;i are collinear. Here, AB denotes the signed length of the directed line segment AB .
303 Proposed by G.P. Henderson, Campbellcroft, Ontario. For j = 1; 2; : : : ; m, let Aj be noncollinear points with Aj 6= Aj +1 . Translate every evennumbered point by an equal amount to get new points A02, A04, : : : , and consider the sequence Bj , where B2i = A02i and B2i,1 = A2i,1. The last member of the new sequence is either Am+1 or A0m+1 according as m is even or odd.
2355.
B1B2B3 : : : Bm to be greater than the length of the path A1A2A3 : : : Am
for all such nonzero translations.
Find a necessary and su cient condition for the length of the path
CRUX 1985 1994: 250; 1995: 280 provides an example of such a conguration. There, m = 2n, the Ai are the vertices of a regular 2n gon and A2n+1 = A1.
Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Five points, A, B , C , K , L, with whole number coordinates, are given. The points A, B , C do not lie on a line. Prove that it is possible to nd two points, M , N , with whole number coordinates, such that M lies on the line KL and 4KMN is similar to 4ABC . land.
2356.
2357.
Proposed by Gerry Leversha, St. Paul's School, London, Eng
An unsteady man leaves a place to commence a onedimensional random walk. At each step he is equally likely to stagger one step to the east or one step to the west. Let his expected absolute distance from the starting point after 2n steps be a. Now consider 2n unsteady men each engaging in independent random walks of this type. Let the expected number of men at the starting point after 2n steps be b. Show that a = b. land.
2358.
Proposed by Gerry Leversha, St. Paul's School, London, Eng
In triangle ABC , let the midpoints of BC , CA, AB be L, M , N , respectively, and let the feet of the altitudes from A, B , C be D, E , F , respectively. Let X be the intersection of LE and MD, let Y be the intersection of MF and NE , and let Z be the intersection of ND and LF . Show that X , Y , Z are collinear. Proposed by Vedula N. Murty, Visakhapatnam, India. Let PQRS be a parallelogram. Let Z divide PQ internally in the ratio k : l. The line through Z parallel to PS meets the diagonal SQ at X . The line ZR meets SQ at Y . Find the ratio XY : SQ.
2359.
304 Proposed by K.R.S. Sastry, Dodballapur, India. In triangle ABC , let BE and CF be internal angle bisectors, and let BQ and CR be altitudes, where F and R lie on AB, and Q and E lie on AC . Assume that E, Q, F and R lie on a circle that is tangent to BC . Prove that triangle ABC is equilateral. 2361. Proposed by K.R.S. Sastry, Dodballapur, India. The lengths of the sides of triangle ABC are given by relatively prime natural numbers. Let F be the point of tangency of the incircle with side AB. Suppose that ABC = 60 and AC = CF . Determine the lengths of the sides of triangle ABC . 2362. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. Suppose that a, b, c 0. Prove that
2360.
1 + 1 + 1 1 +3abc : a1 + b b1 + c c1 + a
2363. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. For natural numbers a; b; c 0, let
a+ qa;b; c := a + b+
+ b+::: a+ a+ c+::: b
a+::: +::: a+ a+::: b c+ b+ c+::: a+::: +::: c+ a+::: a+ a+ b +::: c a+::: c b+ a+ b+::: + c+::: a+::: a+::: a+:::
in the nth column" above, from the third one onwards, we have, from top to bottom, the sequence a; b; c; a repeated 2n,3 times, where it is assumed that the right side understood as an in nite process yields a wellde ned positive real number. The original, a Talent Search Problem, asked to determine q 1; 3; 5. p The value is 3 2 see Mathematics and Informatics Quarterly, 7 1997, No. 1, p. 53. Determine whether or not there exist in nitely many triples a; b; c such that q a; b;c is the cube root of a natural number.
305
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems. R.P. Sealy was accidentally omitted from the list of solvers for problem 2338.
2015. 1995: 53, 129 Proposed by ShiChangShi and Ji Chen, Ningbo University, China. Prove that
where A, B , C are the angles in radians of a triangle. Editor's comment. The rst solution printed was incorrect 1996: 47, solution I . Its correction" 1996: 125 was also incorrect. Finally, the second solution printed, 1996: 48, solution II , was also incorrect. It is time to correct these wrongs!! In solution I, solver Grant claims that both terms
1 1 1
sin A + sin B + sin C A + B + C 27 3 ; 2
p
sin A + sin B + sin C and 1=A + 1=B + 1=C are minimized when A = B = C = 60 . This is true for the second term, but NOT for the rst! It is obvious that for the degenerate triangle A = 180 , B = C = 0 , the sine sum is zero while the sum is positive for A = B =
C . And of course this means that for `real' triangles su ciently close to the
degenerate one, the sine sum will be less than the equilateral sine sum too. Thus the proof falls apart. In Solution II, the proposers rst show that the function yx = x,1=3 cos x is convex on 0;=2 . However, in the second inequality of their second displayed equation, they appear to apply Jensen's inequality to the function f x = log y x, which is NOT convex everywhere in this interval! That is, they want to prove that 6y A=2y B=2yC=2 6y=63, which by taking logarithms is equivalent to f A=2+ f B=2+ f C=2 3f =6; that is, For this we need that f is convex; it is not. In fact, the last inequality on 1996: 48 is incorrect, as can be seen again for the degenerate triangle A = 180, B = C = 0. However, this time we must use limits, so put
1 f A=2 + f B=2 + f C=2 f 1=3A=2 + B=2 + C=2 : 3
306
B=2 = C=2 = x and A=2 = =2 , 2x. Then you can nd that the limit of cos=2 , 2x cos2x
1=3 =2 , 2xx2 as x ! 0 is ZERO, so the inequality fails.
Thanks to Waldemar Pompe and Bill Sands for pointing these out. So, we must belatedly present a correct solution. III Solution by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. X s Since sin A = , the given inequality is equivalent to
R
Now, from item 6.60 on p. 188 in D.S. Mitrinovi
et al., Recent Advances c in Geometric Inequalities, Kluwer Academic Publishers, 1989, the following inequality due to V. Mascioni is known:
X 1 27p3 R A 2 s : r X1 9 R: A 2r
1
2
We now show that 2 is stronger than 1. Indeed, we have:
9 R 27 3 R ; 2r 2 s p2 3p3pRr; that is, since this is equivalent to s 2s2 27Rr :
r
p
3
4 Finally, 4 is simply item 5.12 on p. 52 in O. Bottema et al., Geometric Inequalities, Groningen, 1969.
A very similar solution was submitted by Bob Prielipp, University of Wisconsin Oshkosh, Wisconsin, USA. Remarks by Janous i We can also show the inequality
qY 3
sin
A
Because of the AM GM Inequality, 5 is stronger than 1. Y Since sin A = 2sr2 , we nd that 5 can be rewritten as R
X 1 9p3 A 2 :
5
X 1 9p3 r 2R2 3 A 2 sr :
6
307
We shall also show that
A simple algebraic manipulation shows that this is the same as 4. ii With the usual notation for the power mean
r R 9p3 3 2R2 : 2r 2 sr
9
r
7
Mt x; y; z
=
xt + yt + zt
1 t
3
; t 6= 0 and M0 x; y; z
3 3 2
=
3 xyz 1 ;
we nd that 5 is
This raises the questions: 1. Is t = 0 the minimum value of t such that 8 holds? 2. What is the maximum value of t such that 8 holds with " instead of "? 3. What about analogous questions if we replace the right side of 8 by p 3 3 MpA; B; C ? 2 iii Note that
p Mt sin A; sin B; sin C M,1 A; B; C with t = 0 :
8
Y
sin
A=2 = 4r : So, inequality 3 can be rewritten as R s X p6 Y sin A
: 1 sin A 3 2
This suggests a further question: Let 0 1 be a given real number. Determine the optimum constant C = C such that
A for all triangles. I conjecture that Ed. using A = B = C = 60 p
,3 C = 23 sin : 3
1 3
X
sin
AC
Y
sin
Similarly, inequality 7 is equivalent to
qY 3
s p6 Y sin A
; sin A 2
sin
leading to the analogous question of nding an optimal constant D = D such that where 0
Y A S sinA ; 1. I conjecture that D = C .
qY 3
308 Proposed by Ji Chen, Ningbo University, China. Let P be a point in the interior of the triangle ABC , and let 1 = PAB , 1 = PBC , 1 = PCA. p Prove or disprove that 3 1 1 1 =6. Solution by KeeWai Lau, Hong Kong slightly modi ed by the editor. We prove the inequality. We write = 1 , = 1 , = 1 . Applying the Sine Rule to the triangles PAB , PBC , PCA we obtain
220A?. 1996: 363
Therefore, we get
sin sin sin = sinA , sinB , sinC , : The function log sin x is concave for 0 x ; hence log sinA , + log sinB , + log sinC,
, 3 log sin A+B+C3 , , :
log sin + log sin + log sin 3 log sin , , , : 3 For ; ; 0 and + + , set
and de ne
1
f ; ; = log sin + log sin + log sin , 3 log sin , , , ; 3 n D = ; ; 2 R3 j ; ;
0; + +
;
3 3 o :
6
n 3o 3 : D" = ; ; 2 R3 j ; ; 0; + + ,"; 6 S D . Thus, it is enough It is easily seen that D"0 D" if " "0 and D = " " 0 to prove that there exists 0 such that, if " , then the inequality f ; ; 0 holds for all ; ; 2 D" .
In view of 1, it is enough to prove the following Proposition: If ; ; 2 D, then f ; ; 0. Proof: Fix " 0, and let
Editorial note: These " complications are caused by the fact that f is not de ned if + + = ; that is, on the closure of D in R3 . One may avoid these complications by putting f ; ; = +1 if + + = , and showing that such an f is continuous" on the closure of D. Nothing new, however, arises from this approach, nor does the proof become simpler! Since f is continuous on D" the closure of D" in R3 , and since D" is a compact subset of R3, the minimum value of f is attained on D" . But if
309
; ; 2 D" , then one of the numbers , , say is less than =2, so
that
@f ; ; = cot + cot , , ,
6= 0 : 2 @ 3 This means that f attains its minimum value on @D"  the boundary of D" in R3 . We have n o @D" = ; ; 2 R3 j ; ; 0; + + = ,"; 33 6 o n 3 ; ; 2 R j ; ; 0; + + ,"; = 33 6 0 00 =: @D" @D" : 0 Assume that ; ; 2 @D" . Then ; ; , sothat 2 33 , 6 which implies that 216 . Analogously: ; 216 . Also: 2 , , , 216 = 214 : 216 214 Analogously: ; . Since there exists a real number M such that 216 log sin x M for x 2 216 ; 214 ; 216
we obtain that
" 0 f ; ; 3M , 3 log sin 3 for ; ; 2 @D" : " " Since log sin 3 ! ,1 as " ! 0, there exists 0 such that M log sin 3 for all 0 " ; that is, if 0 " , then 0 f ; ; 0 for ; ; 2 @D" : 3
In order to complete the proof of the proposition, it is su cient to show 00 that, if 0 " , then f ; ; 0 on @D" because then 2 implies that f ; ; 0 for ; ; 2 D" . Set k = 3 =63 . De ne and
n E" = ; 2 R2 j ;
0; + + k
F ; = log sin + log sin + log sin k , 3 log sin
o ," , , , k !
3
:
00 If ; ; 2 @D" , then ; 2 E" and F ; = f ; ; . Therefore it is su cient to prove that F ; 0 for ; 2 E": Similarly, as before, if ; 2 @E" , where
n @E" = ; 2 R2 j ;
0; + + k = ," ;
o
310
0 then ; ; k 2 @D" , and, by 3,
F ; = f ; ; k 0 for ; 2 @E": Therefore it su ces to prove that F ; 0 for all stationary points ; 2 E". So, we assume that ; 2 E" . We have ! @F ; = cot + 1 , k
cot , , , k , k cot k 2 2 @ 3 and ! @F ; = cot + 1 , k
cot , , , k , k cot k : 2 2 @ 3
From the equations we obtain that
@F ; = @F ; = 0; @ @
4
= 0: 5 3 We show that the above equality implies that = . Suppose that 6=
and without loss of generality assume that
cot , cot + , cot
, , , k !
. Set
! cot , cot , cot , , , k : G ; = , 3 qk 2 + q q 2 k+ + k +
;
6
We have
k which gives 0 ,q . The function g1 = csc2 , cot is 2
k nondecreasing in 0; ,2 . This, together with 6, implies that
! 1 Z x csc2 x , cot x dx , cot , , , k G ; = , 3 0 qk 1 ,2 , A csc2 , cot , cot @ ,0:75 : 3
Ed: The last inequality can be veri ed by computer algebra using, for example, DERIVE. Hence from 5 we deduce that = .
311
! k
cot ,2 , k2 , k cot k = 0 : H := cot + 1 , 3 3 2 3 7 We know that , 2 , k2 0. The equation , 2 , k2 = 0 has two positive roots 0, 1 with 0:231 0:232 and 1:540 1 1:541, so 0 2 0; 1. We have H 0 = h1 , h2 + h3 + h4 , h5 ;
where
Now 4 becomes
2 3,k2 csc2 ,2 , k2 ; h = 2k2 csc2 k
: h4 = 3 5 5 5 2 3
We next verify using again DERIVE that
! ,2 , k2 ; h = csc2 ; h = 3 cot ; h1 = 3 cot 2 3 3 !
h1 , h2 2:97 and h3 + h4 , h5 0:995 for 2 0; 1. Therefore, the equation H = 0 has at most one solution. Thus, 7 implies that = . Hence ; = ; is the only 6 6 6 stationary point of F in E" and we have F ; = 0. 6 6
The proof is now complete.
2206. 1997: 46; 1998: 61, 62 Proposed by HeinzJurgen Sei ert, Berlin, Germany. Let a and b denote distinct positive real numbers.
a Show that if 0
p 1, p 6= 1 , then 2 ,apb1,p + a1,p bp 4p1 , ppab + 1 , 4p1 , p a + b : 1 2 2 a,b log a , log b
1 2pab + a + b
: 3 2
b Use a to deduce Polya's Inequality:
Note: log" is, of course, the natural logarithm. III. Solution to part a by the proposer, slightly adapted by the editor.
312 With r = ab and x = loga=r, we have x 6= 0, a = rex , and b = re,x . With q = 2p , 1 so that jqj 2 1, the desired inequality after dividing by r becomes coshqx 1 , q + q 2 coshx. Note that
p
1 , q2 + q2 coshx , coshqx =
since 1 , q 2k
,
0 for k 1. Thus the desired inequality is proved.
1 X x2k+1 2 , q 1 , q2k k=1 2k + 1!
0
DC FA EC FB ABC with equality if and only if two of the Prove that DEF 4 three points D, E , F , at least are midpoints of the corresponding sides. Note: XY Z denotes the area of triangle 4XY Z .
2240. 1997: 243 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Let ABC be an arbitrary triangle with the points D, E , F on the sides BC , CA, AB respectively, so that BD BF 1 and AE AF .
Editor's note: Most solvers noted that the condition for equality should actually be: F plus at least one of D or E be the midpoints of the corresponding sides. Solution by HeinzJurgen Sei ert, Berlin, Germany. There exist numbers u 2 0; 1; v 2 0; 1 ; w 2 0; 1 such that
D= 1 , uB + uC ; E = vA + 1 , v C ; F = 1 , wA + wB : Then BD = uBC , DC = 1 , uBC , AE = 1 , v CA; EC = vCA, AF = wAB and BF = 1 , wAB, so that the conditions BD BF and AE AF DC FA EC FB give u=1 , u 1 , w=w and 1 , v =v w1 , w or u + w 1 v + w: 1
As is wellknown, we have
0 1,u u DEF = v 0 1,v : ABC 1,w w 0
313 Editor's note: A triangle, X = x1; x2 ; Y = y1; y2 ; Z = z1; z2 ; has area see, for example, 13.45, H.S.M. Coxeter, Introduction to Geometry, John Wiley and Sons, Inc., London 1961. Then
x1 x2 1 XY Z = 1 y1 y2 1 ; 2 z1 z2 1
0 1,u u 1 d1 d2 1 0 1,v DEF = 2 e1 e2 1 = 1 v f1 f2 1 2 1 , w w 0
So
a1 a2 1 b1 b2 1 : c1 c2 1
DEF = w1 , w , 1 , u , wv + w , 1: 2 ABC Since w1 , w 1 ; 0 w 1, with equality if and only if w = 1 , from 4 2 1 and 2 we obtain the desired inequality. There is equality if and only if 1 w = 1 and u = 1 or v = 2 . 2 2
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; JIMMY CHUI, student, Earl Haig Secondary School, North York, Ontario; WALTHER JANOUS, Ursuli
nengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer. There was one incomplete solution.
1997: 243 Proposed by Toshio Seimiya, Kawasaki, Japan. Triangle ABC AB 6= AC has incentre I and circumcentre O. The incircle touches BC at D. Suppose that IO ? AD. Prove that AD is a symmedian of triangle ABC . The symmedian is the re ection of the median in the internal angle bisector. Solution by D.J. Smeenk, Zaltbommel, the Netherlands. Lemma. In a nonequilateral triangle ABC with side lengths a; b; c, incentre I and circumcentre O, let P be the point on the halfline starting at B in the direction of C for which BP = c, and Q be on the halfline from A to C with AQ = c; then PQ ? IO. Proof. Let O0 and O00 be the projections of O on BC and AC respectively, and let I 0 and I 00 be the projections of I on those sides. Let S be the point where OO0 intersects II 00 . Consider triangles CPQ and SOI . Since CO0 = 1 0 1 0 0 1 00 00 1 2 a and CI = 2 a + b , c; O I = 2 b , c; similarly, O I = 2 a , c.
2241.
314 Furthermore OSI = SOO00 = because the sides of the rst two angles are perpendicular to the sides of the third. It follows that
00 00 , O0 0 , OS = O I = 2asin c ; and IS = sinI = 2bsinc : sin This means that 4CPQ 4SOI by sideangleside. Since SO ? CP and SI ? CQ, it follows that OI ? PQ. For the solution to our problem we must show that CD : BD = b2 : c2 which, according to standard references, is a property that holds if and only if AD is a symmedian. Because we take AD ? IO, the lemma implies that AD k PQ, so that 4ACD 4QPC . Thus CP : CQ = CD : AC , or c , a = a + b , c , so that c,b 2b 2 2 a = bb + c : 1 +c From CD = s , c and BD = s , b we conclude CD a + b , c 2 BD = a , b + c : Plugging 1 into 2 gives the desired conclusion that CD : BD = b2 : c2 .
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MICHAEL LAMBROU, University of Crete, Crete, Greece; and the proposer. It is noteworthy that this result and that of 2246 which appears later in this issue follow immediately from the same lemma, even though they seem to have little else in common except for the same four solvers. Perhaps a peculiar kind of sunspot activity would account for mathematicians as far apart as the Netherlands and Japan conceiving such related problems at the same time. Smeenk believes that his lemma is known, but he found it easier to prove than to nd; in fact, he supplied two proofs.
1997: 243 Proposed by K.R.S. Sastry, Dodballapur, India. ABCD is a parallelogram. A point P lies in the plane such that 1. the line through P parallel to DA meets DC at K and AB at L, 2. the line through P parallel to AB meets AD at M and BC at N , and 3. the angle between KM and LN is equal to the nonobtuse angle of the parallelogram. Find the locus of P .
2242.
315 Editor's comment. Even though the number of solvers was relatively small, no two solutions were alike. This seemed to result, in part, from various interpretations of the proposal. For example, if one considers a parallelogram as the convex hull of its vertices M is on the line segment AD, for example and that the angle at vertex A is acute, then the locus of P is the empty set. Nonempty loci included various conic arcs according to the aforementioned interpretations. It would not seem instructive to try to report all of these; rather, we attempt to give a representative summary. The common approach was to use a standard analytic argument to derive the equation of a conic which passes through some, or all, of the vertices of the parallelogram, again, depending on interpretation. However, the actual derivation of the conic equations required the omission of the vertices so that they are not properly in the locus of P . Smeenk was the only solver to point this out speci cally. Most of the solvers assumed that BAD =2 which causes P to be outside the parallelogram. Smeenk, assuming further that LSM = BAD, where S = MK LN; derives the locus equation
xx , a + yy , b , 2 cos BADx , ay , b = 0; which is that of an ellipse through the vertices B a; 0, C a; b, D0; b. Con Amore pointed out that for P to be in the convex hull of the vertices BAD =2. Also, assuming that MK LN lies in the halfplane from BD containing A, they derive the equation of the composite quartic a , xx , b , yy a , xx + b , yy , 2a , xb , y cos v = 0 ; where v = ABC =2: The rst component gives the locus as the intersection of a
hyperbola with the interior of the parallelogram while the second similar to Smeenk's equation gives the locus as an elliptic arc also in the interior of the parallelogram. If MK LN lies in the halfplane from BD not containing A, then the resulting elliptic arc is symmetric to the derived one with respect to the centre of the parallelogram. The proposer points out that, since the locus is the circumcircle of a rectangle if the angle at vertex A is a right angle, this could be the springboard for a more elegant solution, presumably, via some a ne transformation. Kone cny remarks that
the parallelogram can, in fact, be obtained by rightangled projection of the rectangle onto the plane which has in common the diagonal, say AC of the parallelogram. Bradley derives the result that the lines KM , LN and AC are concurrent. Solutions were received from CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari,
Valladolid, Spain; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
316
2243. 1997: 243 Proposed by F.J. Flanigan, San Jose State University, San Jose, California, USA. Given f x = x , r1 x , r2 : : : x , rn and f 0 x = nx , s1 x , s2 : : : x , sn,1, n 2, consider the harmonic mean h of the nn , 1 di erences ri , sj . If f x has a multiple root, then h is unde ned, because at least one of the di erences is zero. Calculate h when f x has no multiple roots. All solutions submitted were essentially the same. n Y From f x = x , rk , we get, by logarithmic di erentiation, that
k=1
giving
n f 0x = X 1 ; f x k=1 x , rk
for k = 1, 2, : : : , n , 1. Now, the harmonic mean h is de ned by
n f 0sj = X 1 = 0 f sj k=1 sj , rk
1
X n 1 = 1 n,1 X 1 : h nn , 1 j=1 k=1 rk , sj
In view of 1, we may say, according to temperament, that h is in nite or unde ned.
Solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; DAVID STONE and VREJ ZARIKIAN, Georgia Southern University, Statesboro, Georgia, USA; and the proposer. The proposer said that the problem was inspired by A.G. Clark's solution to problem 3034 in the American Mathematical Monthly 1930, p. 317. See also the American Mathematical Monthly 1923, p. 276 and 1930, p. 94.
317 1997: 243 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a triangle and D is a point on AB produced beyond B such that BD = AC , and E is a point on AC produced beyond C such that CE = AB. The perpendicular bisector of BC meets DE at P . Prove that BPC = BAC . Solution by Stergiou Haralampos, Chalkis, Greece. Let jAC j = b and jAB j = c. We draw DF jjAC with jDF j = b and F on the opposite side of DE from A. Then ACFD is a parallelogram; so jCF j = b+c and CAD = CFD. Let H = CF ^DE. Since CH jjAD and 4ADE is isosceles, CHE = ADE = CEH . It follows that jCH j = c, so jHF j = b. If we draw BH , then ABHC is a parallelogram. Hence BHC = BAC = CFD above. It is easy to see that BHFD is a rhombus, so jPF j = jPB j = jPC j and PBH = PFH = PCH . But if PBH = PCH , it follows that BCHP is a cyclic quadrilateral, so BPC = BHC = BAC . Note that this proof works whether P is between D and E or not.
2244.
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; CHARLES DIMINNIE and TREY SMITH, Angelo State University, San Angelo, Texas; JORDI DOU, Barcelona, Spain; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; VICTOR OXMAN, University of Haifa, Haifa, Israel; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; K.R.S. SASTRY and JEEVAN SANDHYA, Bangalore, India; D.J. SMEENK, Zaltbommel, the Netherlands; KAREN YEATS, student, St. Patrick's High School, Halifax, Nova Scotia; PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer.
2245. 1997: 244 Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel,
n Alcorc
Madrid, Spain. on,
Prove that , 2n is divisible by 5 for n 2. 2 Solution by Florian Herzig, student, Perchtoldsdorf, Austria. I will prove that the assertion ,istrue for all nonnegative integers n. , Ed: with the usual convention that 0 = 1 = 0. 2 2 4 1 mod 5, since 24 1 mod 5 and since Since 3
2 3n + ,1n
n + 4n + 3 nn , 1 mod 2 = n
mod 2 ; = 2 2 2 2 2 3n + ,1n ,2n . we deduce that f n+4 f n mod 5, where f n = 2
n + 4
318 Since f 0 = f 1 = f 2 = 0 and f 3 = 5, it follows that f n is divisible by 5 for all nonnegative integers n.
Also solved by CHARLES ASHBACHER, Cedar Rapids, Iowa, USA; SAM BAETHGE, Nordheim, Texas, USA; MANSUR BOASE, student, St. Paul's School, London, England; PAUL BRACKEN, CRM, Universit
de Montr
al, Montr
al, Qu
bec; e e e e CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; JAMES T. BRUENING, Southeast Missouri State University, Cape Girardeau, MO, USA; MIGUEL ANGEL
CABEZON OCHOA, Logro~ o, Spain; THEODORE CHRONIS, student, Aristotle Unin versity of Thessaloniki, Greece; GORAN CONAR, student, Gymnasium Varadin, z Varadin, Croatia; CHARLES R. DIMINNIE, Angelo State University, San Angelo, z TX, USA; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut, USA; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ur
sulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; ALAN LING, student, University of Toronto, Toronto, Ontario; VEDULA N. MURTY, Visakhapatnam, India; VICTOR OXMAN, University of Haifa, Haifa, Israel; BOB PRIELIPP, University of Wisconsin Oshkosh, Wisconsin, USA; CHRISTOS SARAGIOTIS, student, Aristotle University, Thessaloniki, Macedonia, Greece; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; ROBERT P. SEALY, Mount Allison University, Sackville, New Brunswick; HEINZJURGEN SEIFFERT, Berlin, Germany; REZA SHAHIDI, student, University of Waterloo, Waterloo, Ontario; ZUN SHAN and EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; D.J. SMEENK, Zaltbommel, the Netherlands; DIGBY SMITH, Mount Royal College, Calgary, Alberta; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer.
2246. 1997: 244 Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Suppose that G, I and O are the centroid, the incentre and the circumcentre of a nonequilateral triangle ABC . The line through B , perpendicular to OI intersects the bisector of BAC at P . The line through P , parallel to AC intersects BC at M . Show that I , G and M are collinear. Solution by the proposer. Lemma. In a nonequilateral triangle ABC with side lengths a; b; c, incentre I and circumcentre O, let D be the point on the halfline starting at C in the direction of A for which CD = a, and E be on the halfline from B to A with BE = a; then DE ? IO. This lemma was proved as part of the foregoing solution to 2241 in the notation of that problem. We assume that all points are well de ned. See the remarks below, after the list of solvers. From the lemma,
319
AD = a , b; AE = a , c, and ED ? IO so that because we are given BP ? IO BP k ED. Let Q be the point where BP intersects AC . Then 4ADE AQB. It follows that AB : AQ = AE : AD = a , c : a , b. Since AP bisects BAQ; AB : AQ = BP : PQ, while PM k AC implies BP : PQ = BM : MC . Thus, BM : MC = a , c : a , b. We introduce trilinear coordinates with respect to 4ABC . The coordinates of I are 1; 1; 1, of G are bc; ca; ab, and of M are 0; MC sin ; MB sin = 0; ca , b; ba , c: Then I , G, and M are collinear if and only if
2 1 det 4 bc
0
ca ab ca , b ba , c
1
1
3 5 = 0;
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MICHAEL LAMBROU, University of Crete, Crete, Greece; and TOSHIO SEIMIYA, Kawasaki, Japan. All solvers used trilinear coordinates. Orthogonality is generally awkward with these coordinates, and each of the submitted solutions displayed clever insight to overcome the di culty that arises. The lemma in our featured solution simpli es matters considerably. Note that one must assume that 4ABC is not equilateral  otherwise O = I ; furthermore, as Bradley points out, IG must not be parallel to BC  otherwise M is not de ned. Smeenk's problem implies that the latter condition is equivalent to forbidding AI ? IO. Bradley determined that the required assumption should be that 2a 6= b + c. Putting these remarks together we deduce the unexpected consequence, For a triangle ABC with side lengths a; b; c, and with distinct centroid G, incentre I , and circumcentre O, IG k BC if and only if AI ? IO, if and only if 2a = b + c. This observation is closely related to problem 1506 in Mathematics Magazine 70 : 4 October, 1997 302303: Prove that AIO 90 if and only if 2a b + c, with equality holding only simultaneously.
which is easily con rmed.
Find the value of the sum dkk ; where dk is the number of positive 2 k=1 integer divisors of k. Solution by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA.
2248. 1997: 245 Proposed by Shawn Godin, St. Joseph Scollard Hall, North Bay, Ontario.
1 X
320
X !2 1 1 series may be rearranged in any order without changing the value 2 k=1 k of the sum 4=36. Thus X !2 X ! 0X 1 X X 1 1 1 1 1 1 1 X dk 1 @ j12 A = = 2 2 2j2 = 2 ; i=1 i j =1 k=1 ij =k i k=1 k k=1 k since there are exactly dk pairs i; j such that ij = k. Thus 1 X dk 4 2 = 36 : k=1 k
Also solved by NIELS BEJLEGAARD, Stavanger, Norway; MANSUR BOASE, student, St. Paul's School, London, England; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; ALAN LING, University of Toronto, Toronto, Ontario; BOB PRIELIPP, University of Wisconsin Oshkosh, Wisconsin, USA; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; HEINZJURGEN SEIFFERT, Berlin, Germany; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; and the proposer. Several solvers gave references to places where this problem or generalizations were treated. Indeed, DOSTER gives a reference to the best generalization with his comment: The series is a Dirichlet series. Hardy and Wright An Introduction to the Theory of Numbers, 5th ed., pp. 248250 work out the general theory of this type of 1 X kn sum and prove, more generally, that
s
s , k = k , where s 1, and n=1 n 1 X X 1 s k + 1, where k n = dk and
s = ns . n=1 djn Let s = 2 and k = 0 to get the required result.
The series
1 2 k2 converges absolutely to =6. Hence the terms of the k=1
1 X
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
321
Letter from the Editors
This edition is somewhat di erent from usual. A little explanation is necessary. At the end of July 1998, before all the material for issue 5, 1998, was complete, the EditorinChief, Bruce Shawyer, was admitted to hospital to await surgery. The Associate Editor, Clayton Halfyard, stepped into the breach, and, with the assistance of Bill Sands and Memorial University summer students Paul Marshall and Trevor Rodgers, completed issue 5. On 25 August 1998, the EditorinChief underwent a quadruple coronary artery bypass operation and is now at home for a two month recuperation period. The Associate Editor has prepared this issue, but since the Academy Corner is the sole responsibility of the EditorinChief, it is missing from this issue. The EditorinChief hopes to be able to respond to email messages in the near future. Bruce Shawyer Clayton Halfyard EditorinChief Associate Editor
Lettre de la r
daction e
Ce num
ro di ere quelque peu de l'ordinaire. Quelques explications e s'imposent. La n du mois de juillet 1998, avant que tous les textes du num
ro 5 e soient pr^ ts, le r
dacteur en chef, Bruce Shawyer, a et
admis a l'h^ pital e e
e o en vue d'une op
ration. Le r
dacteur en chef adjoint, Clayton Halfyard, a e e donc pris la rel ve. Avec l'aide de Bill Sands et des etudiants de l'Universit
e
e Memorial, Paul Marshall et Trevor Rodgers, il a r
ussi a terminer le num
ro 5. e e Le 25 ao^ t, 1998, le r
dacteur en chef a subi un quadruple pontage u e coronarien. Il est maintenant de retour a la maison, pour une convalescence de deux mois. Le r
dacteur en chef adjoint s'est occup
du pr
sent num
ro, e e e e mais, puisque la chronique Academy Corner rel ve enti rement du r
dacteur e e e en chef, elle ne para^tra pas dans ce num
ro. i e Le r
dacteur en chef esp re etre en mesure de r
pondre a son courriel e e ^ e sous peu. Bruce Shawyer Clayton Halfyard R
dacteur en chef e R
dacteur en chef adjoint e
322
THE OLYMPIAD CORNER
No. 192 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. As a rst Olympiad for this issue we give the problems of the 4th Mathematical Olympiad of the Republic of China Taiwan written April 13, 15, 1995. My thanks go to Bill Sands of the University of Calgary, who collected these problems when he was assisting with the 1995 International Mathematical Olympiad held in Canada.
4th MATHEMATICAL OLYMPIAD OF THE REPUBLIC
OF CHINA TAIWAN
First Day  Taipei
April 13, 1995
1 1 0 1
with complex coe cients. Suppose the roots of P x are 1; 2 ; : : : ; n with j 1j 1, j 2j 1; : : : ; j j j 1, and j j+1j 1; : : : ; j nj 1. Prove:
j Y i=1
1. Let P x = a + a x + + an, xn, + anxn be a polynomial
j ij ja0j2 + ja1j2 + + jan j2 :
p
2. Given a sequence of integers: x1; x2 ; x3; x4 ; x5; x6; x7; x8, one constructs a second sequence: jx2 , x1 j; jx3 , x2 j; jx4 , x3 j; jx5 , x4 j; jx6 , x5j; jx7 , x6j; jx8 , x7j; jx1 , x8j. Such a process is called a single operation. Find all the 8term integral sequences having the following property: after nitely many applications of the single operation the sequence becomes an integral sequence with all terms equal. 3. Suppose n persons meet in a meeting, and every one among them is familiar with exactly 8 other participants of that meeting. Furthermore suppose that each pair of two participants who are familiar with each other have 4 acquaintances in common at that meeting, and each pair of two participants who are not familiar with each other have only 2 acquaintances in common. What are the possible values of n?
323
Second Day  Taipei
April 15, 1995
4. Given n distinct integers m1; m2; : : : ; mn, prove that there exists a polynomial f x of degree n and with integral coe cients which satis es the following conditions: 1 f mi = ,1, for all i, 1 i n. 2 f x cannot be factorized into a product of two nonconstant polynomials with integral coe cients. 5. Let P be a point on the circumscribed circle of 4A1A2A3. Let H be the orthocentre of 4A1A2A3. Let B1 B2; B3 respectively be the point of intersection of the perpendicular from P to A2 A3 A3 A1 , A1 A2 respectively. It is known that the three points B1 , B2 , B3 are collinear. Prove that the line B1B2 B3 passes through the midpoint of the line segment PH . 6. Let a, b, c, d be integers such that ad , bc = k 0, a; b = 1, and c; d = 1. Prove that there are exactly k ordered pairs of real numbers x1; x2 satisfying 0 x1 , x2 1 and for which both ax1 +bx2 and cx1 +dx2 are integers.
As a second problem set this issue for your puzzling pleasure, we give the XI Italian Mathematical Olympiad written May 5, 1995 at Cesenatico. Thanks go to Bill Sands of the University of Calgary who collected them while at the 1995 IMO in Canada.
XI ITALIAN MATHEMATICAL OLYMPIAD
Cesenatico, May 5, 1995
Time: 4.5 hours
1. Determine for which values of the integer n it is possible to cover up, without overlapping, a square of side n with tiles of the type shown in the picture
where each small square of the tile has side 1.
324
2. In a class of 20 students no two of them have the same ordered pair written and oral examinations of scores in mathematics. We say that student A is better than B if his two scores are greater than or equal to the corresponding scores of B . The scores are integers between 1 and 10. a Show that there exist three students A, B and C such that A is better than B and B is better than C . b Would the same be true for a class of 19 students? 3. In a town there are 4 pubs, A, B, C and D, connected as shown in the picture.
C A B
A drunkard wanders about the pubs starting with A and, after having a drink, goes to any of the pubs directly connected, with equal probability. a What is the probability that the drunkard is at pub C at his fth drink? b Where is the drunkard more likely to be after n drinks? n 5 4. An acuteangled triangle ABC is inscribed in a circle with centre O. Let D be the intersection of the bisector of A with BC , and suppose that the perpendicular to AO through D meets the line AC in a point P interior to the segments AC . Show that AB = AP . 5. Two noncoplanar circles in Euclidean space are tangent at a point and have the same tangents at this point. Show that both circles lie in some spherical surface. 6. Find all pairs of positive integers x, y such that
D
x2 + 615 = 2y :
As a third set of problems for your attention we give the Third and Fourth Grade and IMO Team selection rounds of the Yugoslav Federal Competition for 1995. Thanks again go to Bill Sands, the University of Calgary, for collecting them for me.
325
YUGOSLAV FEDERAL COMPETITION 1995
Third and Fourth Grade
1. Let p be a prime number. Prove that the number
is divisible by p, where dots indicate that the corresponding digit appears p times consecutively. 2. A polynomial P x with integer coe cients is said to be divisible by a positive integer m if and only if the number P k is divisible by m for all k 2 Z. If the polynomial
11 122 2 99 9 , 123456789
P x = a0xn + a1xn,1 + + an,1x + an is divisible by m, prove that an n! is divisible by m. 3. A chord AB and a diameter CD ? AB of a circle k intersect at a point M . Let P lie on the arc ACB and let P 62 fA; B; C g. Line PM intersects the circle k at P and Q = P , and line PD intersects chord AB 6 at R. Prove that RD MQ. 4. A tetrahedron ABCD is given. Let P and Q be midpoints of edges AB and CD, and let O and S be the incentre and the circumcentre of the tetrahedron, respectively. If points P , Q and S belong to the same line, prove that the point O also belongs to that line. Selection of the IMO Team x y z and
1. Find all the triples x; y;z of positive rational numbers such that
1 1 1 x + y + z; x + y + z ; xyz 2 Z:
2. Let n be a positive integer having exactly 1995 1's in its binary representation. Prove that 2n,1995 divides n!. 3. Let SABCD be a pyramid such that all of its edges are of the same length. Let points M 2 BC and N 2 AS be such that the line MN is perpendicular to line AD as well as to the line BC . Find the ratios BM=MC and SN=NA. Editor's note: ABCD is the base of the pyramid.
We now turn to readers' solutions to problems of the Swedish Mathematics Contest, 1993 1997: 196 .
326
SWEDISH MATHEMATICS CONTEST 1993 1. The integer x is such that the sum of the digits of 3x is the same as the sum of the digits of x. Prove that 9 is a factor of x. Solutionsby Jamie Batuwantudawe, student, Sir WinstonChurchill High School, Calgary; by Michael Selby, University of Windsor, Windsor, Ontario; and by Enrico Valeriano Cuba, National University of Engineering, Lima, Peru. We give the solution of Valeriano. Let S n be the sum of the digits of n. Working modulo 9
Also Since S 3x = S x,
November 20
Final
3x S 3x mod 9 :
x Sx mod 9 :
2x 0 mod 9 ; and since 2; 9 = 1, we have x 0 mod 9. 2. A railway line is divided into 10 sections by the stations A, B, C , D, E, F , G, H , I , J and K . The distance between A and K is 56 km. A trip along two successive sections never exceeds 12 km. A trip along three successive sections is at least 17 km. What is the distance between B and G?
z 
A B C DE F G H I J K
Solutionsby Jamie Batuwantudawe, student, Sir WinstonChurchill High School, Calgary; by Michael Selby, University of Windsor, Windsor, Ontario; and by Enrico Valeriano Cuba, National University of Engineering, Lima, Peru. We give Batuwantudawe's solution.
B C D E F G H I J K Now AK = 56 and AK = AD + DG + GJ + JK . We know that AD; DG; GJ 17. Thus JK 5 to satisfy AK = 56. We know HK 17, and since JK 5, HJ 12. But, we also know HJ 12. Thus HJ = 12. Since HK 17 and HJ = 12, JK 5. The only possibility is that JK = 5. Symmetrically we nd that AB = 5 and BD = 12.
Now,
A
DH = AK , AB , BD , HJ , JK = 56 , 5 , 12 , 5 , 12 = 22 :
327 Now GJ 17 but HJ = 12. Hence GH 5. Since DG 17 and DH = DG + GH = 22, we obtain DG = 17 and GH = 5. Now
BG = BD + DG
x = y has an integer solution x, y if and only if the product ab is even.
2 2
3. Assume that a and b are integers. Prove that the equation a + b
2
= 12 + 17 = 29 :
2
+
Solutions by Bob Prielipp, University of Wisconsin Oshkosh, Wisconsin, USA; Michael Selby, University of Windsor, Windsor, Ontario; by Sree Sanyal, student, Western Canada High School, Calgary, Alberta; by Enrique Valeriano Cuba, National University of Engineering, Lima, Peru; and by Michael Lebedinsky, student, Henry Wise Wood High School, Calgary, Alberta. We give Selby's solution. First, we prove that this condition is necessary. Suppose ab is odd. Then a, b are odd and a2 b2 1 mod 4. Now x2 0 or 1 mod 4, and y2 0 or 1 mod 4. Therefore a22+ b2 + x2 = y2 is not possible, since if we consider this modulo 4, 2 + x y 2 mod 4, which is impossible since 2 + x2 2 or 3 mod 4. Therefore ab must be even. If ab is even, then, without loss of generality, a = 2k. Consider 4k2 + b2 + x2 = y 2. If 4k2 + b2 = 2t +1, t an integer, then set y , x = 1 and y + x = 2t +1, 2y = t + 12, y = t + 1 and x = t. Then 2t + 1 + t2 = t + 12. We are done. If 4k2 + b2 is even, then b = 2s and 4k2 + b2 = 4k2 + s2 = 4m. Again, y 2 , x2 = 4m. Set y , x = 2 and y + x = 2m. Then y = m +1 and x = y , 2 = m , 1. Now 4m + m , 12 = m + 12, and again we are done. Hence 2 + b2 + x2 = y 2 always has a solution when ab is even. a 4. To each pair of real numbers a and b, where a 6= 0 and b 6= 0, there is a real number a b such that
Solve the equation x 36 = 216. Solutions by Sree Sanyal and Aliya Walji, students, Western Canada High School, Calgary, Alberta; and by Enrique Valeriano Cuba, National University of Engineering, Lima, Peru.
a b c = a b c , a a = 1:
328 Now a a a = a a a, so
Finally
a1 = 1a = a: Also a b b = a b b and a = a 1 = a b b so ab= a: b
5. A triangle with perimeter 2p has sides a, b and c. If possible, a new triangle with the sides p , a, p , b and p , c is formed. The process is then repeated with the new triangle. For which original triangles can the process be repeated inde nitely? Solutions by Michael Selby, University of Windsor, Windsor, Ontario; by Enrique Valeriano, National University of Engineering, Lima, Peru; and by Sonny Chan, student, Western Canada High School, Calgary, Alberta. We give Valeriano's solution. Let a b c and be the di erence between the longest and the shortest side. Original Triangle New Triangle 1 = 3p , a + b + c = p Perimeter = 2p Perimeter
=c,a 1 = p , a , p , c = c , a
We can see that the perimeter of the new triangle is half the previous perimeter, but is the same. Then, if 0, repeating this process we can obtain 2p Perimeter k = k k = c , a :
x = 216 = x = 7776 : 36
If ck is the longest side we obtain the absurd relation ck Perimeterk k ck. Finally, only with an equilateral triangle as the original triangle = 0 can we repeat the process inde nitely. 6. Let a and b be real numbers and let f x = ax + b,1. For which a and b are there three distinct real numbers x1 , x2 , x3 such that f x1 = x2 , f x2 = x3 and f x3 = x1? Solutionsby Filip Crnogorac and Sonny Chan, students, Western Canada High School, Calgary, Alberta; and by Michael Selby, University of Windsor, Windsor, Ontario. We give Selby's writeup. Consider the functions of the form
2
gx = x + : x+
329 Lemma. g x has at least 3 distinct xed points if and only if = = 0, = 6= 0. Proof. If = = 0, = 6= 0, g x = x and it clearly has at least 3 distinct points x1; x2; x3 such that g xi = xi, i = 1; 2; 3. Conversely consider the equation for a xed point x, g x = x. This implies x2 + x = x + or x2 + , x , = 0. Suppose this has three distinct roots. Then the quadratic must be identically 0, or = = 0 and = . Now, if f x = ax1+b , then
The problem implies f f f has three distinct real xed points x1 ; x2 ; x3 . By the above lemma, this is true if and only if a + b2 = aa + b2 = 0 and ab + ba + b2 = ab 6= 0: This is true if and only if a = ,b2 and ab 6= 0. To complete this number of the Olympiad Corner we turn to readers' solutions to problems of the Dutch Mathematical Olympiad, second round, September 1993 1997: 197 .
abx + a 2 f f x = abxax +2b+ a and f f f x = aa + b2x + ab+ bbb2 + a : +b +
DUTCH MATHEMATICAL OLYMPIAD
Second Round
September, 1993
1. Suppose that V = f1; 2; 3; : : : ; 24; 25g. Prove that any subset of V with 17 or more elements contains at least two distinct numbers the product of which is the square of an integer. Solution by Sonny Chan and Filip Crnogorac, students, Western Canada High School, Calgary. The set of numbers A = f1; 2; : : : ; 24; 25g contains a total of ve perfect squares f1; 4; 9; 16; 25g. The product of any two of these will also be a perfect square. There is one triplet, the product of any two of its elements will result in a perfect square: f2; 8; 18g. The only other pairs of numbers from A whose product is a perfect square are f3; 12g, f5; 20g, f6; 24g. The other eleven elements of the set A are f7; 10; 11; 13; 14; 15; 17; 19; 21; 22; 23g and they cannot form a perfect square when multiplied with any other element of set A. Group the elements of set A as follows:
f1; 4; 9; 16; 25g; f2; 8; 18g; f3; 12g; f5; 20g; f6; 24g; f7g; f10g; f11g; f13g; f14g; f15g; f17g; f19g; f21g; f22g; f23g:
330 If more than one number is chosen from a given group, a perfect square will result. There is a total of 16 groups, so 16 numbers can be chosen without creating a perfect square product. However, if any 17 numbers are chosen, then two must be contained within the same group, and therefore will form a perfect square product. 2. Given is a triangle ABC , A = 90 . D is the midpoint of BC , F is the midpoint of AB , E the midpoint of AF and G the midpoint of FB . AD intersects CE, CF and CG respectively in P , Q and R. Determine the ratio P Q . QR C
q
D q Rq q qQ q P Eq Fq Gq qB A
Solution by Filip Crnogorac, student, Western Canada High School, Calgary. C
r
J
r K r r
P
r
A B E F G We know that two medians in a triangle divide each other in 2 : 1 ratio, or in other words the point of intersection is 2 the way from the vertex. 3 AQ 2 Since CF and AD are both medians in 4ABC , then QD = 1 , where Q is the point of intersection. Also, since D is the midpoint of the hypotenuse in the right triangle ABC , then it is the centre of the circumscribed circle with radius DA = DC = DB. Drop a perpendicular from D onto sides AB and CA. The feet of the perpendiculars will be F and J , respectively, where J is the midpoint of AC , since DF and DJ are altitudes in isosceles triangles 4ADB and 4ADC , respectively. Now consider 4CFB . The segments CG and FD are medians and therefore intersect at H say in the ratio 2 : 1 so, HD = 1 . From here FD 3 it can be seen that 4ARC and 4DRH are similar, since their angles are
r
r rD rR Q r
r
r
331 the same. Also, since we know that FD = JA, and 2JA = AC then HD = 1 CA and 4ARC is 6 times bigger than 4DRH . Now we can see 6 AR 6 that RD = 1 and since AR + RD = AD, then RD = 1 . AD 7 Similarly 4APE 4KPD, where medians DJ and CE meet at K . 1 We know that AE = 1 AB , so then JK = 4 JD , since JD is parallel to 4 AE = 2 , and from the similarity of the triangles AB. It now follows that KD 3 2 2 AP AP AD P D = 3 . Also, since AP + PD = 2 , then AD1 = 5 . Combining these 2 results we have AP = 5 AD , AQ = 3 AD , QD = 3 AD and RD = 1 AD . 7 Thus and
4 PQ = AQ , AP = 2 AD , 2 AD = 15 AD 3 5
From these P Q = 7 . QR 5 3. 1 A series of numbers is de ned as follows: u1 = a, u2 = b, un+1 = 2 un + un,1 for n 2. Prove that limn!1 un exists. Express the value of the limit in terms of a and b. Solution by the Editors. For the recurrence un+1 = 1 un + un,1 we obtain the associated 2 equation 22 , , 1 = 0, ,which has roots = , 1 ; 1. Thus we seek a 2 solution of the form un = X , 1 n + Y . From u1 = a and u2 = b we get 2
4 QR = QD , RD = 1 AD , 1 AD = 21 AD: 3 7
so that X + 4 b , a and Y = a+2b . It is now easy to check by induction 3 3 , 4 that un = 3 b , a , 1 n + 1+2b and as n ! 1, un ! 1 a + 2 b. 2 3 3 3 4. In a plane V a circle C is given with centre M . P is a point not on the circle C .
,1X + Y = a 2 1 4X + Y = b
qP
Bq Mq V
qA
332 a Prove that for a xed point P , AP 2 + BP 2 is a constant for every diameter AB of the circle C . b Let AB be any diameter of C and P a point on a xed sphere S not intersecting V . Determine the points P on S such that AP 2 + BP 2 is minimal. Solutionby Jamie Batuwantudawe, student, Sir Winston Churchill High School, Calgary. a With 4PAB , we can join P and M to create two new triangles, 4PMA and 4PMB. Let PMA = . Then PMB = 180 , . Because M is the centre of circle C and A and B both lie on circle C , we have MA = MB = r, the radius of the circle. By the Law of Cosines,
BP 2 = MP 2 + r2 , 2MPr cos180 , = MP 2 + r2 + 2MPr cos
and
The right hand side is a constant depending only on the radius of the circle and the distance of P from the centre. b From a, we know that AP 2 + BP 2 = 2MP 2 +2r2 . For any point P on sphere S, the radius of the circle will 2remain constant. Therefore the only variable a ecting the sum AP 2 + BP is MP , the distance from the point P to the centre of the circle. AP 2 + BP 2 will be a minimum when MP is minimum. Therefore we are looking for the point on the sphere closest to M . Let T be the centre of the sphere S , D be the point on the segment MT that lies on the sphere, and D0 be any other point on S. We know that MD + DT MD0 + D0 T because the shortest distance between M and T is a straight line. We know that DT = D0 T . Thus MD MD0 . Thus D is the point on the sphere which minimizes the sum. That completes the Corner for this issue. Send me your nice solutions and generalizations as well as Olympiad contests.
AP 2 = MP 2 + r2 , 2MPr cos so AP 2 + BP 2 = 2MP 2 + 2r2 .
333
BOOK REVIEWS
Edited by ANDY LIU
Juegos y acertijos para la ense~ anza de las Matem
ticas n a Games and riddles for the teaching of mathematics by Bernardo Recam
n Santos, a published by Grupo editorial Norma educativa, 1997, Bogot
, Colombia. a ISBN9580437319; softcover, 134 pages. Reviewed by Francisco Bellot Rosado, Valladolid, Spain. This book contains 75 games and riddles, uniformly distributed in ve chapters: I mathematical games; II arithmetical riddles; III geometrical riddles; IV logical riddles; V algebraic riddles. A last chapter with solutions not very detailed, comments and alternatives or variants, and a little Bibliography completes the work. The di culty level is indicated by one, two or three stars ?, ??, ? ? ?, meaning that the problem is for students of primary education, of the rst years of the secondary, either preuniversity level, respectively; although, as the author said in the preface, this classi cation is arbitrary. The origins of the problems are very variable; there are some classical examples the game of Nim, the snake, a game by Paul Erdos, the Kaprekar algorithm, the Egyptian fractions, the age of the 3 daughters; others can be found in popular books Martin Gardner, Le petit Archim de and some are e originals from the author. As he says, los acertijos matem
ticos, como los a chistes callejeros, no tienen due~ o" mathematical riddles, like street jokes, n are not copyrightable. The little book can be used with pro t in mathematical clubs, extracurricular activities and also in the classroom. I have used it myself in my classes of Taller de Matem
ticas", a class of 2 hours each week for students of 12 a years old in the new educational system of my country Compulsory Secondary Education. By the way, the book reveals the identity of one collaborator of CRUX: Ignotus.
334
THE SKOLIAD CORNER
No. 32 R.E. Woodrow
As a contest this issue we give the Junior High School Mathematics Contest, Preliminary Round 1998 of the British Columbia Colleges which was written March 11, 1998. My thanks go to the contest organizer, Jim Totten, the University College of the Cariboo, for forwarding the 1998 contest materials to me. Students are given 45 minutes to respond to the 15 questions.
BRITISH COLUMBIA COLLEGES Junior High School Mathematics Contest Preliminary Round 1998
Time: 45 minutes
1. A number is prime if it is greater than one and divisible only by one and itself. The sum of the prime divisors of 1998 is: a 5 b 14 c 42 d 66 e 122 2. Successive discounts of 10 and 20 are equivalent to a single discount of: a 15 b 25 c 28 d 30 e 32 3. Suppose that A = A2 and A 2 B = A , 2B. Then the value of 7 2 3 is: a 1 b 16 c 961 d 43 e 31 4. The expression that is not equal to the value of the four other expressions listed is: p p p bp + 9 , 9 + 8 1 c ,1 9 + 9 + 8 a 1 9 + 9 , 8 d 1 , 9 9 , 8 e 19 , 9 , 8 5. The sum of all of the digits of the number 1075 , 75 is: a 8 b 655 c 664 d 673 e 675
k k
k
335
6. A circle is divided into three equal parts and one part is shaded as in the accompanying diagram. The ratio of the perimeter of the shaded region, including the two radii, to the circumference of the circle is:
a 1
2 b 3 7. The value of
c 1 3
d 3+ 3
e 3
5 b 4 c 6 d 6 e 6 5 7 5 8. If each small square in the accompanying grid is one square centimetre, then the area in square centimetres of the polygon ABCDE is:
is: a 3 4
2 , 2, 1, 1 1 2
2
1
C B
D
A
a 38
E
b 39 c 42 d 44 e 46 9. A point P is inside a square ABCD whose side length is 16. P is equidistant from two adjacent vertices, A and B , and the side CD opposite these vertices. The distance PA equals: p a 8:5 b 6 3 c 12 d 8 e 10 10. A group of 20 students has an average mass of 86 kg per person. It is known that 9 people from this group have an average mass of 75 kg
336 per person. The average mass in kilograms per person of the remaining 11 people is: a 94 b 95 c 96 d 97 e none of these 11. In the following display each letter represents a digit:
3 B C D E 8 G H I
The sum of any three successive digits is 18. The value of H is: a 3 b 4 c 5 d 7 e 8 12. In the accompanying diagram ADE = 140 . The sides are congruent as indicated. The measure of EAD is:
D A
a 30
E
b 25 c 20 d 15 e 10 13. The area in square units of the triangle bounded by the xaxis and the lines with equations y = 2x + 4 and y = , 2 x + 4 is: 3 a 8 b 12 c 15 d 16 e 32 14. Two diagonals of a regular octagon are shown in the accompanying diagram. The total number of diagonals possible in a regular octagon is:
a 8
b 12
c 16
d 20
e 28
337
15. A local baseball league is running a contest to raise money to send a team to the provincial championship. To win the contest it is necessary to determine the number of baseballs stacked in the form of a rectangular pyramid. The fth and sixth levels from the base of the stack of baseballs are shown. If the stack contains a total of seven levels, the number of baseballs in the stack is:
a 100
b 112
c 166
d 168
e 240
In the May number of the Corner we gave the problems of the 15th W.J. Blundon Contest written by students in Newfoundland and Labrador. Next we give the o cial" solutions. My thanks go to Bruce Shawyer for forwarding the contest and solutions to me.
15th W.J. BLUNDON CONTEST February 18, 1998 1. a Find the exact value of
1 + 1 : log2 36 log3 36
Solution. log1 36 + log1 36 = log36 2 + log36 3 = log36 6 = 1 . 2 2 3 b If log15 5 = a, nd log15 9 in terms of a. Solution. 1 = log15 15 = log15 5 3 = log15 5+log15 3 = a +log15 3. log15 3 = 1 , a = log15 9 = log15 32 = 2 log15 3 = 21 , a. 2. a If the radius of a right circular cylinder is increased by 50 and the height is decreased by 20, what is the change in the volume? Solution. V2 = 1:5r2:8h = 1:8r2 h = 1:8V1 . So the volume is increased by 80. b How many digits are there in the number 21998 51988 ? Solution. 21998 51988 = 210 21988 51988 = 1024 101988, which has 1988 + 4 = 1992 digits. 3. Solve: 32+x + 32,x = 82.
338 Solution.
32+x + 32,x 9 3x + 39x 93x 2 , 823x + 9 9 3x , 13x , 9 3x = 1 ; 9 x = ,2
= = = =
82 82 0 0 3x = 9 x=2
6
4. Find all ordered pairs of integers such that x
Solution.
= y2 + 53.
x6 = y2 + 53 x6 , y2 = 53 x3 , y x3 + y = 53 x33 , y = 53 x33 , y = 1 x33 , y = ,53 x33 , y = ,1 x + y = 1 x + y = 53 x + y = ,1 x + y = ,53 x=3 x=3 x = ,3 x = ,3 y = ,26 y = 26 y = 26 y = ,26
The pairs are 3; ,26; 3; 26; ,3; 26; ,3; ,26. 5. When one fth of the adults left a neighbourhood picnic, the ratio of adults to children was 2 : 3. Later, when 44 children left, the ratio of children to adults was 2 : 5. How many people remained at the picnic? Solution. Let A be the number of adults and C be the number of children initially at the picnic. After one fth of the adults left, four fths remain. So 4
5
C , 44 = 2 = 8A = 25C , 1100: 4 5 5A Solving the two equations gives A = 50, C = 60. The number remaining is
then
After 44 children left
A = 2 = 6A = 5C: C 3
4 50 + 60 , 44 = 40 + 16 = 56: 5
339
6. Find the area of a rhombus for which one side has length 10 and the diagonals di er by 4. Solution. 10
b+2 b b+2
10
b
10
10
b + 22 + b2 2b2 + 4b , 96 b2 + 2b , 48 b , 6b + 8 b = 6;
= = = =
100 0 0 0 b 6= ,8
Since the area of a rhombus is one half the product of the diagonals we get
1 A = 1 2b2b + 4 = 2 1216 = 96: 2
7. In how many ways can 10 dollars be changed into dimes and quarters, with at least one of each coin being used? Solution. Let q be the number of quarters and d be the number of dimes. Then
Since d must be an integer, q must be even. Also d must be positive. So
25q + 10d = 1000 d = 100 , 5 q: 2 100 , 2 q 0 5
So q must be an even positive integer less than 40, of which there are 19.
q 40:
340 Solution. Let y = 4 x + 10. Then y 2 = x + 10, and the equation becomes 4 y2 + y = 12 Then: px + 10 = 3
8. Solve: px + 10 + px + 10 = 12. p
4
p
y2 + y , 12 = 0 y + 4y , 3 = 0 y 6= ,4; y = 3
3
x + 10 = 81 x = 71
135
is divided by the polynomial x , x. Solution.
9. Find the remainder when the polynomial x
+ x125 , x115 + x5 +1
x135 + x125 , x115 + x5 + 1 = x3 , xQx + ax2 + bx + c = xx , 1x + 1Qx + ax2 + bx + c This must be valid for all values of x. Substituting in x = 0, x = 1, and x = ,1 gives: x = 0: 1= 0+c = c = 1 x = 1 : 3 = 0 + a + b + c = a + b = 2 x = ,1 : ,1 = 0 + a , b + c = a , b = ,2
Solving the system
a+b = 2 a , b = ,2 gives a = 0, b = 2. So the remainder is 2x + 1. 10. Quadrilateral ABCD below has the following properties: 1 The midpoint O of side AB is the centre of a semicircle; 2 sides AD, DC and CB are tangent to this semicircle. Prove that AB2 = 4AD BC .
Solution. First join the obvious lines in the given gure: D F C E G
r
r
r r
r
A
r
O
r
rB
By the properties of tangents, DE = DF and CF = CG. Therefore EDO = FDO = and FCO = GCO = . Since OA = OB, we have EAO = GBO = .
341 Summing the angles of quadrilateral ABCD, we get +2 +2 + = 360 . Hence + + = 180 ; that is, they are the angles of a triangle. Considering triangles AOD, DOC and COB , we get AOD = , DOC = and COB = . Thus the three triangles are similar. OB Considering the triangles ADO and BOC , we have AD = BC , or AO AD BC = AO OB. Since AO = OB = 1 AB , we get the result. 2 Last issue we gave the problems of the U.K. Intermediate Mathematical Challenge. Here are the solutions.
1. 6. 11. 16. 21.
C C C D E
2. 7. 12. 17. 22.
B C E D D
3. 8. 13. 18. 23.
A A D C A
4. 9. 14. 19. 24.
A C D A A
5. 10. 15. 20. 25.
E B D E B
That completes the Skoliad Corner for this number. Send me your comments, suggestions, and most importantly, suitable contest materials for use in future issues.
Advance Announcement
The 1999 Summer Meeting of the Canadian Mathematical Society will take place at Memorial University in St. John's, Newfoundland, from Saturday, 29 May 1999 to Tuesday, 1 June 1999. The Special Session on Mathematics Education will feature the topic The invited speakers are Ed Barbeau University of Toronto, Ron Dunkley University of Waterloo, Tony Gardiner University of Birmingham, UK, and Rita Janes Newfoundland and Labrador Senior Mathematics League. Requests for further information, or to speak in this session, as well as suggestions for further speakers, should be sent to the session organizers: Bruce Shawyer and Ed Williams CMS Summer 1999 Meeting, Education Session Department of Mathematics and Statistics, Memorial University St. John's, Newfoundland, Canada A1C 5S7
What Mathematics Competitions do for Mathematics.
342
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only solutions  the next issue will feature only problems. We warmly welcome proposals for problems and solutions.
High School Solutions
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca First, we present a very elegant solution to a past problem: H224. Consider square ABCD with side length 1. Select a point M exterior to the square so that AMB is 90 . Let a = AM and b = BM . Now, determine the point N exterior to the square so that CN = a and DN = b. Find, as a function of a and b, the length of the line segment MN .
343 Solution by Hoe Teck Wee, Singapore. Locate point P exterior to the square so that BP = a and CP = b. Locate point Q exterior to the square so that DQ = a and AQ = b. Then MPNQ is a square of side length a + b, and it follows that MN is a diagonal p of this square. Thus, the desired length of MN is 2a + b. H225. In Cruxmayhemland, stamps can be bought only in two denominations, p and q cents, both of which are at least 31 cents. It is known that if p and q are relatively prime, the largest value that cannot be created by these two stamps is pq , p , q . For example, when p = 5 and q = 3, one can a x any postage that is higher than 15 , 5 , 3, or 7 cents, but not 7 cents itself. The governor of Cruxmayhemland tells you that 1997 is the largest value that cannot be created by these stamps. Find all possible pairs of positive integers p; q with p q . Solutionby Miguel Carrion Alvarez, Universidad Complutense de Mad
rid, Spain. If p and q are not relatively prime, then they have a common divisor d. Then any value that does not divide d cannot be created by these two stamps. Hence there are in nitely many values that cannot be created. Hence, we must have p and q being relatively prime. We are seeking all possible pairs p; q with p q and p and q relatively prime such that 1997 = pq , p , q . Adding 1 to both sides of this equation, we have 1998 = pq , p , q + 1 = p , 1q , 1. Since 1998 = 21 33 371 has 2 4 2 = 16 positive divisors, there are exactly eight ways to write down 1998 as the product of two positive integers, namely 1998 = 1 1998 = 2 999 = 3 666 = 6 333 = 9 222 = 18 111 = 27 74 = 37 54. Now p and q are at least 31, so we must have p , 1 and q , 1 both being at least 30. The only pair for which this is true is p , 1 = 54 and q , 1 = 37. And we see that 55 and 38 are relatively prime integers. Hence, the only solution is p;q = 55; 38. H226. In rightangled triangle ABC , with BC as hypotenuse, AB = x and AC = y, where x and y are positive integers. Squares APQB, BRSC , and CTUA are drawn externally on sides AB, BC , and CA, respectively. When QR, ST , and UP are joined, a hexagon is formed. Let K be the area of the hexagon PQRSTU . a Prove that K cannot equal 1997. Hint: Try to nd a general formula for K . b Prove that there is only one solution x; y with x y so that K = 1998. Solutionby Miguel Carrion Alvarez, Universidad Complutense de Mad
rid, Spain. a The hexagon PQRSTU is the union of the rightangled triangle ABC , the three squares, and triangles QRB, STC , and UPA. We will now
344 show that all four triangles have equal area. Consider triangle QRB . We have
ABC + QBA + RBQ + CBR = 360 = RBQ = 180 , ABC: Let P denote the area of triangle P . 1 Now, QRB = 1 QB RB sin RBQ = 2 AB CB sin ABC = 2 ABC . The argument does not rely on ABC being rightangled, so similarly, we have STC = UPA = ABC = xy=2. It follows that K = 4 xy + x2 + y 2 + BC 2, and BC 2 = x2 + y2 implies K = 2x2 + xy + y 2. 2 Therefore, K is necessarily an even integer, so it cannot be 1997.
b Using our formula from the previous part, we seek an integer solution to 999 = x2 + xy + y 2. Taking congruences modulo 3, we have x2 + xy + y2 0 mod 3. Suppose x 0 mod 3. Then y 2 0 mod 3, which implies that y 0 mod 3. Suppose x 1 mod 3. Then y2 + y + 1 0 mod 3, which implies that y 1 mod 3. Suppose x 2 mod 3. Then y2 + 2y + 4 y2 + 2y + 1 y + 12 0 mod 3, which implies that y 2 mod 3. Hence, it follows that we must have x y mod 3. Then we make the substitution x = 3u + a, y = 3v + a where a is 0, 1 or ,1, and this yields 999 = 9u2 + 9au + 3a2 + 9v 2 + 9av + 9uv. All terms are divisible by 9 except possibly 3a2 . If 3a2 does not divide 9, then the left side, 999, is divisible by 9, but the right side is not. Hence, we must have 3 dividing a2 , which implies that a = 0, since neither 12 nor ,12 is divisible by 3. Hence, a = 0, and so x = 3u and y = 3v , which implies that u2 + uv + v2 = 111. Multiplying by 42, we get 444 =24u2 + 4uv + 4v2 = 3u2 + 6uv + 3v 2 + u2 , 2uv + v = 3u + v + u , v 2. Since u , v2 u + v 2, it follows that
3u + v2 444 4u + v 2 = 111 u + v 2 148 = 11 u + v 12: Now if u + v = 12, then u , v 2 = 444 , 3u + v 2 = 12, and because u and v are integers, this has no solution. However, if u + v = 11, then u , v 2 = 81, and so u , v = 9, because u v which follows directly from x y . Solving for u and v , we get u = 10 and v = 1, which implies that x = 30 and y = 3. Checking, x; y = 30; 3 is a solution to 2x2 + 2xy + 2y 2 = 1998. Therefore, there is a unique solution to K = 1998, and it is x; y = 30; 3.
345
H227. The numbers 2, 4, 8, 16, : : : , 2n are written on a chalkboard. A student selects any two numbers a and b, erases them, and replaces them by their average, namely a+b=2. She performs this operation n,1 times until only one number is left. Let Sn and Tn denote the maximum and minimum possible value of this nal number, respectively. Determine a formula for Sn and Tn in terms of n.
Solution. We shall prove by induction that if we perform the operation on the numbers a1 , a2 , : : : , an,1 , an , where a1 a2 an , then
n,2 n,1 n Tn = a21 + a42 + a83 + + an,2 + an,1 + 2a,1 : n 2 2
For n = 2, we have two numbers, a1 and a2 , and thus T2 must be a1 + a2 =2. Thus the claim holds for n = 2. Assume that for n = 2, 3, : : : , k , 2, k , 1, the identity holds. Then for the case n = k, that is, when we start o with the numbers a1 , a2 , a3 , : : : , ak , we will perform the operation k , 2 times and be left with two numbers, To illustrate this, say we start o with 2, 4, 8, 16, and 32. Then we can replace 2 and 8 by 5, 16 and 32 by 24, then 4 and 24 by 14. Then we are left with two numbers, 5 and 14. We got 5 from performing the operation on the numbers 2 and 8, and we got 14 from performing the operation on the other three numbers, namely 4, 16, and 32. Let b1 , b2 , b3, : : : , bp be the set of numbers that we used to get the number x, where the bi's are in increasing order, and let c1 , c2 , c3 , : : : , ck,p be the other numbers from the set that we used to get y , where the ci 's are in increasing order. Then, by our induction hypothesis, we have
x and y. Say x comes from performing the operation on a set of p numbers from the set. Then y comes from performing the operation on the remaining set of k , p numbers.
b b bp x b21 + b42 + + 2p,2 + 2p,1 + 2p,1 ; p,2 p,1
and similarly,
c c ck p y c21 + c42 + + 2k,p,2 + 2k,p,1 + 2k,,,1 : k,p,2 k,p,1 p
Note that the bi's and the ci 's are just some permutation of the set a1 , a2 , a3, : : : , ak. Hence, to minimize the value of x + y=2, the average of the two numbers will be minimized when the denominators are as large as possible. Hence, we want either p or k , p to be 1, since we will then have k,1 a denominator of 2 in one of the terms.
346 Without loss of generality, assume that k , p = 1. Thus, we want to now show that
x + y c1 + b1 + b2 + + bk,3 + bk,2 + bk,1 k,2 k,1 k,1 2 2 4 a1 + a2 + 83 + +2ak,2 +2ak,1 +2 ak ; a 2 4 8 2k,2 2k,1 2k,1 because this will prove the case for n = k. Note that the bi 's are in increasing order, so if c1 = ar for some r, then b1 = a1 , b2 = a2 , : : : , br,1 = ar,1 , br = ar+1 , br+1 = ar+2 , : : : , bk,1 = ak . Noting that the ai 's are in increasing order, we have ar = ar + ar + ar + 2 8 16 a4r + ar + ar + + ar 4 8 16 2r a1 + a2 + a3 + + ar,1 : 4 8 16 2r a1 + a2 + a3 + + ar,1 + ar+1 + ar+2 + + ak,1 + ak
4 8 16 2r 2r+1 2r+2 2k,1
Adding to both sides, we get the desired inequality. Thus, the induction is proved and hence, for our question, we take ai = 2i for i = 1, 2, : : : , k, and we have
n,2 n,1 n
2k,1
4 2 Tn = 2 + 4 + 8 + + 2n,2 + 2n,1 + 2n,1 2 8 2 2 = n , 1 1 + 2 = n + 1: Thus, our formula for Tn is obtained by taking the ai 's in increasing order. In contrast, to get Sn, we want to take the ai 's in decreasing order, namely a1 = 2n, a2 = 2n,1 , : : : , an,1 = 4, an = 2. This is so that we can
get the maximum possible value at the end. Thus, we have
n n,1 n,2
3 2
2 2 2 Sn = 22 + 2 4 + 2 8 + + 2n,2 + 2n,1 + 2n,1 = 2n,1 + 2n,3 + 2n,5 + + 25,n + 23,n + 22,n = 23,n 1 + 22 + 24 + + 22n,4 + 22,n n,1 = 23,n 4 , 1 + 22,n
1
3 84n,1 , 1 + 4 = 3 2n 2n 22n+1 + 4 : = 3 2n
347
H228. Verify that the following three inequalities hold for positive reals x, y and z : i xx , y x , z + y y , z y , x+ z z , xz , y 0 this is known as Schur's Inequality, ii x4 + y 4 + z 4 + xyz x + y + z 2x2 y 2 + y 2z 2 + z 2x2 , iii 9xyz + 1 4xy + yz + xz , where x + y + z = 1.
x + y , zx , y2 + y + z , xy , z 2 + z + x , y z , x2 0:
Solution by Vedula N. Murty, Visakhapatnam, India. We shall show that all three inequalities are equivalent to showing that
And we shall then show that this inequality holds for all positive reals x, y , and z . Let
p = x + y , zx , y2 + y + z , xy , z2 + z + x , yz , x2:
Expanding and rearranging terms, we nd that
2xx , y x , z + 2yy , z y , x + 2z z , xz , y = 2x3 + 2y 3 + 2z3 + 6xyz , 2x2y + xy2 + x2 z + xz 2 + y2z + yz2 = x3 + y 3 + 2xyz , x2y , x2 z , y 2x , y 2z +y3 + z3 + 2xyz , y 2x , y 2z , z 2x , z 2 y +z3 + x3 + 2xyz , z 2x , z 2 y , x2 y , x2 z = x + y , zx , y2 + y + z , xy , z 2 + z + x , y z , x2 = p: Therefore, we have shown that xx , y x , z + y y , z y , x + zz , xz , y = p=2. Thus i holds if and only if p 0. Note that x4 + y 4 + z 4 , x2 y 2 , y 2z 2 , z 2 x2 = 1 x2 , y 22 + y 2 , 2 z22 + z2 , x22 . Furthermore, x2y2 + y2z2 + z2x2 , xyzx + y + z = 1 2 2 2 2 2 2 2 x y , z + y x , z + z x , y . Subtracting the second equation
from the rst, we have
= = = =
x4 + y4 + z 4 + xyz x + y + z , 2x2 y2 + y2 z 2 + z 2 x2 1 x2 , y2 2 + y2 , z 2 2 + z 2 , x2 2 , x2 y , z 2 , y2 x , z 2 2 ,z2x , y2 1 fx , y2 x + y2 , z 2 + y , z 2 y + z 2 , x2 2 +z , x2 z + x2 , y2 g x + y + z x , y2 x + y , z + y , z 2 y + z , x 2 +z , x2 z + x , y px + y + z : 2
348 Since x, y , and z are positive, x + y + z is also positive, and it follows that ii holds if and only if p 0. Suppose that x + y + z = a. Then if we replace x by x=a, y by y=a and z by z=a, then the inequality will remain the same. Thus, there is no loss in generality in assuming that x + y + z = 1, for we can always divide each term by some constant to ensure that the sum of the terms becomes 1. Thus, letting x + y + z = 1, we have:
p = 1 x + y , z x , y2 + y + z , xy , z 2 + z + x , yz , x2 2 2 = x3 + y3 + z 3 + 3xyz , x2 y + x2 z + y2 x + y2 z + z 2 x + z 2 y = x3 + y3 + z 3 + 3xyz , x2 + y2 + z 2 x + y + z + x3 + y3 + z 3 = 2x3 + y3 + z 3 + 3xyz , x2 + y2 + z 2 = 2x + y + z x2 + y2 + z 2 , xy , yz , zx + 6xyz + 3xyz ,x2 + y2 + z2 2 = x + y2 + z 2 , 2xy + yz + zx + 9xyz = x + y + z 2 , 4xy + yz + zx + 9xyz = 1 , 4xy + yz + zx + 9xyz : Therefore, iii holds if and only if p 0. Hence, we have shown that all three inequalities are equivalent to p 0. Thus, if we can show that this
inequality holds, then i, ii, and iii all hold. To show that p 0, we separate the problem into two cases. If x, y , and z are the sides of a triangle, then x + y , z , x + z , y , and y + z , x are all positive, and thus it immediately follows that p 0. Otherwise, the three sides do not form the sides of a triangle, and so if we assume without loss of generality that x y z , then x + y , z 0. And we have
p = x + y , zx , y2 + y + z , xy , z2 + z + x , yz , x2 = x + y , z x , z + z , y 2 + y + z , xy , z 2 +z + x , yz , x2 = x , z2x + y , z + z , y 2x + y , z +2x , zz , y x + y , z + y + z , xy , z 2 +z + x , yz , x2 = x , z22x + y , z22y + 2z , xz , y z , x , y : And this last expression is nonnegative, since x 0, y 0, z , x 0, z , y 0 and z , x , y 0. Therefore, p 0, and we
have proven that all three inequalities hold.
349
Advanced Solutions
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu The following problems rst appeared in Volume 23, Issue 5. A201. Consider an in nite sequence of integers a1, a2, : : : , ak, : : : with the property that every m consecutive numbers sum to x and every n consecutive numbers sum to y . If x and y are relatively prime, then show that all numbers are equal. Erratum. As noted by some of our solvers, the problem as stated is incorrect . The correct problem should read as follows: Consider an in nite sequence of integers a1 , a2 , : : : , ak , : : : with the property that every m consecutive numbers sum to x and every n consecutive numbers sum to y . If m and n are relatively prime, then show that all numbers are equal. Solution. Since every m consecutive numbers sum to the same value, we must have a periodic sequence with ak = am+k , for all natural numbers k. Likewise, ak = an+k , for all natural numbers k. What this means is that the sequence is periodic with periods m and n. It may be intuitively clear now why the sequence must be a constant sequence. Here is a rigorous proof. It is enough to show that the n numbers a0 , a1 , : : : , an,1 are equal since every other number is equal to one of these by periodicity. Now a0 , am , a2m , : : : are equal, so it is enough to show that ai, for i = 0, 1, : : : , n , 1, is equal to akm for some integer k. We show that each of the n numbers a0 , am , : : : , an,1m , leaves a di erent remainder upon division by n. In other words, the set of numbers f0; am ; : : : ; an,1mg is a complete residue system modulo n. To see this, im jm mod n holds if and only if i j mod n since m and n are relatively prime. Further, for 0 i; j n, we must then have i = j . In other words, the n values a0 , am , : : : , an,1m must leave a di erent remainder upon division by n. Now, putting it all together, suppose lm = nq + r, where 0 r n. Then alm = anq+r = ar . Now we know that all possible remainders r, from 0 to n , 1, are achieved for some integer l, so we can conclude that all ar are equal. A202. Let ABC be an equilateral triangle and , its incircle. If D and E are points on AB and AC , respectively, such that DE is tangent to ,, show that 8th Iberoamerican Mathematical Olympiad, Mexico '93
AD + AE = 1: DB EC
350 Solution I by Miguel Carrion Alvarez, Universidad Complutense de
Madrid, Spain.
A D F I B
,
H
E
G
C
Let I be the incentre and r the inradius. Let F , G, and H be the points where , is tangent to AB , AC , and DE . Then, DF = DH and EG = EH this is a wellknown property of the two tangents from a point to a circle. This implies that FID = DIH and GIE = EIH . Since FIG = 120 , FID + GIE = 60 . We can write FID = 30 + , and GIE = 30 , . Now, FD = r tan FID and AF = r tan 60 , so that
AD = AF , FD = tan 60 , tan FID DB AF + FD tan 60 + tan FID p3 , tan 30 + tan , tan = p 1tantan 30 tan 30 + 3 + 1 , tan 30 tan p3 , 1p p3 tan + p 3 p tan = 1 , 3 tan ; , = p 2 + 3 + 1p 3 tan 3 , tan AE = 1 + p3 tan : EC 2
and similarly We then have as required.
AD + AE = 1 , p3 tan + 1 + p3 tan = 1 DB EC 2 2
351 Solution II. Using the same diagram as above, assume, without loss of generality, that the sides of the equilateral triangle ABC have length 1. Let x = AD and y = AE, where 0 x; y 1=2. Now using the Cosine Law in triangle ADE , we have
1 , x , y2 = x2 + y2 , 2xy cos 60 :
This is equivalent to each of the following:
1 + x2 + y 2 , 2x , 2y + 2xy = x2 + y 2 , xy ; 2x + 2y , 1 = 3xy ; x , xy + y , xy = 1 , x , y + xy ;
and nally,
The last equation is valid since x and y cannot be equal to 1. This is our desired result. Also solved by Alexandre Trichtchenko, student, Brook eld High School, Ottawa, Ontario. A203. Let Sn = 1 + a + aa + + aa , where the last term is a tower of n , 1 a's. Find all positive integers a such that Sn = naS ,1=n . Solution. Claim: a = 1 is the only possibility. Since a is positive, we have by the AMGM inequality
:::a n
1 ,x + 1, y = 1:
x
y
1 + a + aa + + aa
:::a
with equality if and only if 1 = a = aa = = aa . Using the de nition of Sn, we have
:::a
n
a0+1+a++a
a:::a
1 n
;
Sn ,aS ,1 1=n : n Since equality occurs, we must have 1 = a = aa = = aa , Sn = n, for all natural numbers n and it follows that Sn = n1S ,1=n . A204. Given a quadrilateral ABCD as shown, with AD = p3, AB + CD = 2AD, A = 60 and D = 120 , nd the length of the line segment from D to the midpoint of BC .
n :::a n
Solution I by Miguel Carrion Alvarez, Universidad Complutense de
Madrid, Spain and Alexandre Trichtchenko, student, Brook eld High School, Ottawa, Ontario.
352
D
p3 E
60
: : :
B F
C
Observe that AB and CD are parallel. Then by Thales' Theorem, a line through the midpoints E and F of AD and BC respectively will also be parallel to AB , and the length of EF will be AB + CD=2 = AD. Now, applying the Cosine Law to triangle DEF , we have
A
60
DF 2 = DE2 + EF 2 , 2 DE EF cos DEF
2 = AD + AD2 , AD2 cos 60
2 = 3AD ; 4
2
so DF = 3=2. Solution II.
D x A
60
:
E?
C
U
D0
:
B 2x
Let the midpoint of BC be E . Note that lines AB and CD are parallel. Rotate triangle DEC about point E so that C coincides with B , and D coincides with D0 as shown in the gure. This is possible since E was chosen to be the midpoint of line segment BC . Now A, B , and D0 are collinear since lines CD and AB are parallel, so that D0 BE + ABE = DCE + ABE = 180 . Let x = AD, then AD0 = AB +BD0 = AB +CD = 2x. Thus triangle DAD0 is 0 similar to the ubiquitous 1: 2: p3 triangle. Hence, DE = DD =2 = p3=2AD = 3=2.
353
Challenge Board Solutions
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu C75. a Let n be an integer, and suppose a1 , a2 , a3 , and a4 are integers such that a1 a4 , a2 a3 1 mod n. Show that there exist integers Ai , 1 i 4, such that each Ai ai mod n and A1A4 , A2 A3 = 1. b Let SL2; Z denote the group of 2 2 matrices with integer entries and determinant 1, and let ,n denote the subgroup of SL2; Z of matrices which are congruent to the identity matrix modulo n. Two matrices are congruent modulo n if each pair of corresponding entries is congruent modulo n. What is the index of ,n in SL2; Z? Solution. a Without loss of generality, we suppose that the ai are nonzero. They may, of course, nevertheless be congruent to 0 mod n. Choose A1 = a1. Our rst order of business is to nd A2 a2 mod n with A1 and A2 relatively prime. Let d = gcda2; n. Then a2=d and n=d are relatively prime, and so by Dirichlet's Theorem on the in nitude of primes in arithmetic progressions, we can choose a prime p distinct from the prime divisors of A1 and satisfying
p a2 mod n : d d Set A2 = dp. It is immediate that A2 a2 mod n. Since a1 and a2 must not share any common divisors with n, we see that d and A1 are relatively prime, and therefore so are A1 and A2 . Select any x, y such that A1 x , A2 y = 1. We must nd k such that x + A2k a4 mod n and y + A1k a3 mod n. Rewriting these congruences as A2k a4 , x mod n
and
A1k a3 , y mod n; if we multiply the rst congruence by ,a3 , the second by a4 , and we add the
equations, then we nd
A1a4k , A2a3k a4a3 , y , a3a4 , x mod n; and since A1 a4 , A2 a3 a1 a4 , a2 a3 1 mod n, we obtain k a3x , a4y mod n:
354 One readily checks that with such a choice of k, setting A4 = x + A2 k and A3 = y + A1k gives the Ai the desired properties. b Let SL2; Z=nZ denote the group of 2 2 matrices with entries in Z=nZ the ring of integers modulo n and determinant 1. The reductionmodn homomorphism 'n : SL2; Z ! SL2; Z=nZ, via which the entries of 'n A are the congruence classes of the corresponding entries of A, evidently has kernel equal to ,n. Additionally, part a of this problem proves that 'n is surjective. So, to compute the index of ,n in SL2; Z, we need only compute the order of SL2; Z=nZ. Call this order C n. Let n = pa1 pa be the prime factorization of n. Given an ele1 k ment of SL2; Z=nZ, we obtain an element in each SL2; Z=pa Z by rei ducing modulo pa . Conversely, given one element in each SL2; Z=pa Z, i i we can use the Chinese Remainder Theorem to obtain a unique matrix whose modulo pa reductions a our given matrices. It follows, therefore, that are i C n = C pa1 C pk . 1 Let
k i i i i k
a b c d be a representative of an element of SL2; Z=pk Z with p prime and k 0, and suppose A B
C D k+1 Z whose reduction modulo pk is is an element of SL2; Z=p a b
: c d A B C D
= = = =
Choose representatives a0 , b0 , c0 , and d0 in Z=pk+1 Z of a, b, c, and d, and write
with a0 , b0, c0 , d0 in Z=pZ. Then 0 0 det a b in Z=pk+1 Z, and the condition that
a0 + a0pk; b0 + b0pk; c0 + c0pk; d0 + d0pk;
c0 d0
= 1 + mpk
in Z=pk+1 Z is transformed into the condition that
0 0 pk det a0 + a0pk c +c
b00 + b0pk d + d0pk
= 1
a0d0 + a0 d0 , b0c0 , b0c0 ,m mod p :
355 One of a0 , b0 , c0 , d0 must be invertible modulo p, say d0 , without loss of generality. Then for any arbitrary choice of b0 , c0 , and d0 , there exists exactly one a0 which satis es this condition. Consequently, we have shown that each element of SL2; Z=pk Z lifts to precisely p3 elements in SL2; Z=pk+1 Z, and so C pk+1 = p3 C pk. Inductively, C pk+1 = p3k C p. To calculate the number of elements
in Z=pZ, note that to have determinant 1, the rst row of
a b
c d a b
c d
must be nonzero. However, the reader may verify directly that to each of the remaining p2 , 1 possible choices for the pair a; b, there are exactly p choices for the pair c; d which make
a b
= 1: det c d
Thus,
and putting together all our work, we conclude that
C p = p2 , 1p = p3 1 , p12 ;
C n = n
3
Y
where the product is taken over the primes p dividing n. C76. Let X be any topological space. The nth symmetric power of X , denoted X n , is de ned to be the quotient of the ordinary nfold product X n by the action of the symmetric group on n lettersthat is, it is the space of unordered ntuples of points of X . Show that the symmetric power Cn is actually homeomorphic to the ordinary product Cn . Solution. Every unordered ntuple hr1; : : : ; rn i of points in C is the solution set to a unique monic polynomial, speci cally, the polynomial f z = z , r1 z , rn whose coe cients are given by the elementary symmetric polynomials in the ri . Furthermore, by the Fundamental Theorem of Algebra, every polynomial of degree n over C has exactly n solutions in C, counting multiplicity. Therefore, the map ' : Cn ! Cn is a bijection, where ' is de ned as
pjn
1
; 1 , p2
'hr1; : : : ; rni = s1; : : : ; sn
356 with the si given by
f z = z , r1 z , rn = zn , s1zn,1 + + ,1nsn: It is evident that ' is continuous, and so to prove that ' is a homeomorphism, it remains to show that ',1 is continuous as well. This amounts to
proving that the roots of a polynomial vary continuously with the coe cients of the polynomial. The reader may attempt to prove this using elementary methods; instead, we employ a useful theorem from complex analysis: Theorem. Rouch
Let D be a closed disk in C. Suppose that f z e and g z are complexvalued functions which are complexdi erentiable in the interior of D, and that at any point z on the boundary of D we have jf z , gzj jf zj. Then f z and gz have the same number of zeros in D, counting multiplicity. Given 0, let Di be a disk centred at ri and of radius at most , chosen so that no other roots of f lie on the boundary of Di . Since jf z j achieves its minimum on the boundary of Di, this minimum is positive. Calling this minimum mi, we can choose i 0 to be su ciently small such that on the boundary of Di . Let be the minimum of the i , and let g z be any monic polynomial whose coe cients each di er from the coe cients of f z by at most . By construction, the polynomials f z and g z satisfy the conditions of Rouch
's Theorem on each disk Di, and consequently f and g e have the same number of zeros in each Di. Thus, the roots of g z all are distant from corresponding roots of f z by no more than , and so indeed ',1 is continuous and ' is a homeomorphism.
Note for advanced readers: Without too much di culty, one may use the above result to prove, more generally, that if X is any 2dimensional manifold, then the symmetric product X n is always a manifold again. However, if X is a manifold of any positive dimension k other than 2 and if n 1, then X n is never a manifold. One argues this roughly as follows. Let U be a neighbourhood of an unordered ntuple of points of X with two of the points the same and the rest di erent. Then U looks like Rk n,2 Rk Rk =fx; y y; xg, where, in the last term of the product, the action of the symmetric group causes x; y and y; x to be identi ed. Changing variables in the last term of the product by putting u = x+y and v = x,y, we nd that U is in fact homeomorphic to Rk n,2 Rk V , where V is the quotient of Rk under the identi cation v = ,v. Then V with the origin removed is homotopic to the projective space RP k,1 , whereas Rk with any point removed is homotopic to the sphere S k,1 . Since RP k,1 and S k,1 are homotopic only for k = 2, we conclude that U is not Euclidean space if k 6= 2, and so X n can only be a manifold if X is 2dimensional. iz n,1 + + z + 1
mi jf zj
357
Pushing the Envelope
student, Yale University If C is a di erentiable curve in the plane, then we can form a family of lines F by taking the set of tangents at all points of C . This family F is called the envelope of C see Figure 1. Obtaining F from C is easy enough, but consider the converse problem; that is, given a family of lines F , does there exist a curve C which is tangent to every line in F , so making F the envelope of C . C
Naoki Sato
F
Figure 1 To solve the problem, we assume that F is a oneparameter family of lines; that is, the lines are parameterized by one parameter, say t, and furthermore, that the lines vary smoothly in t. It should be noted that with respect to the pointline duality in projective geometry, the concept of the envelope is dual to the concept of the locus. Before tackling the problem, we rst give a few examples. Example 1. The parabola y = x2 . The derivative is given by
dy dx = 2x : Hence, the tangent at a particular point x0 ; y0 = t0 ; t2 is given by 0 y , y0 = 2t0x , x0 ;
Copyright c 1998 Canadian Mathematical Society
358 or
y = 2t0x , t2 : 0 dy = dx
dy dt dx = dt
Example 2. The general ellipse is given by the following parameterization: x = a cos t, y = b sin t. The derivative is given by
b cos t = , b cot t : ,a sin t a Hence, the tangent at a particular point x0 ; y0 = a cos t0 ; b sin t0 is given by b y , y0 = , a cot t0x , x0 ; or b y = , a cot t0x + b csc t0 : Now, we tackle the problem. Assume F is of the form fmtx + btg, so in Example 1, mt = 2t and bt = ,t2 . A diagram makes the next step visually clear. If we x one particular line ` in F , and consider other lines in F that approach `, then the intersections converge to a point on the desired curve C see Figure 2.
`
C
Figure 2 Mathematically, let ` be the line corresponding to a xed t, namely y = mtx + bt. We consider the lines y = mt + hx + bt + h as h approaches 0. Solving for their intersection, we obtain
mtx + bt = mt + hx + bt + h ;
359 which implies that
bt+h,bt b x = , mt + h , bt = , mt+hh,mt : t + h , mt h
0
b By de nition, this approaches , m0tt as h approaches 0; hence, this is the xcoordinate of the point on C that is in fact tangent to `. The ycoordinate is 0 0 0
Therefore,
b , m0tt mt + bt = m tbtm,tmtb t : 0 C =
b , m0tt ; m tbtm,tmtb t 0
0 0 0
:
Checking, for Example 1, we have
C =
and for Example 2,
2t ; 2,t2 , 2t,2t
= ft; t2 g ; 2 2
!
a csc t cot t ; b21 + cot2 t csc t , b2 cot2 t csc t C = 1 + cot2 t b1 + cot2 t = fa cos t; b sin tg :
There is an alternative method for nding curves of envelopes. If the envelope is given in the form F x; y;t = 0, then the curve is found by solving the equations F = 0 and @F = 0. There are also some regularity @t conditions which we will ignore here. In our case, we can take F x; y;t = y , mtx , bt, so the curve is found by solving the system of equations
y , mtx , bt = 0 ; m0tx + b0t = 0 :
Solving for x and y , we nd the same solution as above. Why does this more general method work? Orthogonal Curves A related topic, orthogonal curves, can be interesting in their own right. If F and G are each a family of curves in the plane, then they are orthogonal if every curve in F is orthogonal to every curve in G . Two curves are orthogonal if their tangents are perpendicular at any point they intersect. For example, the family of concentric circles centred at the origin and the family of lines passing through the origin are orthogonal see Figure 3.
360
Figure 3 The natural question that arises is: For a family of curves F , does there exist a family G which is orthogonal to F . Most smooth oneparameter families will have a solution. Example 3. We take the family of parabolas fy = tx2 g; in this case, t is again a parameter. For each parabola, di erentiating we nd We wish to eliminate t from the expression, because we want to nd curves which are orthogonal to every such parabola, and so the expression must be independent of t. As every high school student knows, two lines are perpendicular if and only if the product of their slopes is ,1, so we want to solve the di erential equation
dy = 2tx = 2y : dx x
Z
x2 + y2 = C or 2
dy = , x dx 2y 2y dy = ,x dx Z 2y dy = , x dx 2 y2 = , x + C
2
The orthogonal family is a family of ellipses. In other words, each parabola is orthogonal to each ellipse. Example 4. We take the family of circles fx , t2 + y 2 = t2 , a2 g, where a is a xed nonnegative integer, and jtj a. This is a general family
x2 + y2 = 1 : 2C C
361 of coaxial circles. For each circle, di erentiating we nd
dy = , x , t ; dx y
and by rearranging the equation of the circle, we obtain so
2 y2 2 t = x + 2x + a ;
The orthogonal family must then satisfy
dy = y2 , x2 + a2 : dx 2xy
2xy dy = 2 , y 2 , a2 dx x 2 2 2 x , y , a dy = 2xy dx 2xy dx + ,x2 + y 2 + a2 dy = 0
By some elementary theory of di erential equations, we nd y ,2 is an integrating factor of the above equation, so the equation becomes
We put in a constant of 2C , because on simpli cation, the above equation becomes x2 + y , C 2 = a2 + C 2 . This is the family of circles, with centre on the y axis, passing through the points a; 0. In general, families of the same type of curve can have di erent orthogonal families; it depends on how the curves are stacked". Problems 1. For each of the following families of lines, nd the curve which it is the envelope of: a fy = , ta x + 2ta g. This is the family of lines for which the product 2 of the xintercept and y intercept is a constant, 4a b fy = cos tx , t cos t + sin tg c fy = ,tan tx + a sin tg. This is the family of lines for which the axes are cut o a chord of constant length a.
2x
dx + , x2 + 1 + a2
dy = d x2 + y , a2
= 0 y y2 y2 y y2 2 2 x + y , a = 2C y y
362 2. For each of the following families of lines, nd the orthogonal family of lines: a fy = tx3 g b fy = tex g. 3. Let c be a positive real. Consider the graphs of the equation as t varies. For jtj c, the graph is an ellipse; for jtj c, it is a hyperbola, all with foci c; 0. Show that the family of such ellipses is orthogonal to the family of such hyperbolae.
x2 + y2 = 1 ; t2 t2 , c2
363
PROBLEMS
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 April 1999. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX.
2364. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. A sequence fxn g is given by the recursion: x0 = p, xn+1 = qxn + q , 1 n 0, where p is a prime and q 2 is an integer. 1 Suppose that p and q are relatively prime. Prove that the sequence fxng does not consist of only primes. 2? Suppose that pjq . Prove that the sequence fxn g does not consist of only primes.
Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Triangle DAC is equilateral. B is on the line DC so that BAC = 70 . E is on the line AB so that ECA = 55 . K is the midpoint of ED. Without the use of a computer, calculator or protractor, show that 60 AKC 57:5 .
2365.
364 land.
2366.
Proposed by Catherine Shevlin, WallsenduponTyne, Eng
Triangle ABC has area p, where p 2 N. Let
where the minimum is taken over all possible triangles ABC with area p, and where 2 N. Find the least value of p such that = p2 2367. Proposed by K.R.S. Sastry, Dodballapur, India. In triangle ABC , the Cevians AD, BE intersect at P . Prove that
= min ,AB 2 + BC 2 + CA2 ;
ABC DPE = APB CDE : Here, ABC denote the area of 4ABC , etc.
BC.
2368.
Proposed by Iliya Bluskov, Simon Fraser University, Burnaby,
Let a1 ; a2 ; :::; an be a permutation of the integers from 1 to n with the property that ak + ak+1 + ::: + ak+s is not divisible by n +1 for any choice of k and s where k 1 and 0 s n , k , 1. Find such a permutation a for n = 12; b for n = 22. 2369?. Proposed by Federico Arboleda, student, Bogot
a, Colombia age 11. Prove or disprove that for every n 2 N, there exists a 2nomino such that every nomino can be placed entirely on top of it. An nomino is de ned as a collection of n squares of equal size arranged with coincident sides. 2370. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Determine the exact values of the roots of the polynomial equation
x5 , 55x4 + 330x3 , 462x2 + 165x , 11 = 0 :
2371. Proposed by Bill Sands, University of Calgary, Calgary, Alberta. For n an integer greater than 4, let f n be the number of veelement subsets, S , of f1; 2; : : : ; ng which have no isolated points, that is, such that if s 2 S , then either s , 1 or s + 1 not taken modulo n is in S . Find a nice" formula for f n.
365
2372. Proposed by Bill Sands, University of Calgary, Calgary, Alberta. For n and k positive integers, let f n;k be the number of k element subsets S of f1; 2; : : : ; ng satisfying: i 1 2 S and n 2 S ; and ii whenever s 2 S with s n , 1, then either s + 2 2 S or s + 3 2 S . Prove that f n;k = f 4k , 2 , n;k for all n and k; that is, the sequence
f k;k; f k + 1; k; f k + 2; k; : : : ; f 3k , 2; k of nonzero values of ff n;kg1 is a palindrome for every k. n=1
Proposed by Toshio Seimiya, Kawasaki, Japan. Given triangle ABC with AB AC . Let M be the midpoint of BC . Suppose that D is the re ection of M across the bisector of BAC , and that A, B, C and D are concyclic. AB , AC . Determine the value of Proposed by Toshio Seimiya, Kawasaki, Japan. Given triangle ABC with BAC 60 . Let M be the midpoint of BC . Let P be any point in the plane of 4ABC . Prove that AP + BP + CP 2AM . 2375. Proposed by Toshio Seimiya, Kawasaki, Japan. Let D be a point on side AC of triangle ABC . Let E and F be points on the segments BD and BC respectively, such that BAE = CAF . Let P and Q be points on BC and BD respectively, such that EP kDC and FQkCD. Prove that BAP = CAQ.
2373.
2374.
BC
366
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems.
2247. 1997: 244 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. a Suppose that n 3 is an odd natural number. Show that the only polynomial P 2 R x satisfying the functional equation:
is given by P x = x. b? Suppose that n 1 is a natural number. Show that the only polynomial P 2 R x satisfying the functional equation:
n,1
n = P xn + X n xk ; P x + 1 k=0 k
for all x 2 R;
P x + 1n = P xn +
n,1 n
X k=0
k k x;
for all x 2 R;
is given by P x = x. c? Suppose that n 1 is a natural number. Show that the only polynomial P 2 R x satisfying the functional equation:
P x + 1n = P xn +
is given by P x = x.
n,1 n
X k=0
k k x;
for all x 2 R;
I. Solution by David Stone and Vrej Zarikian, Georgia Southern University, Statesboro, Georgia, USA. a First observe that
k n n k x = x + 1 , x : k=0 Thus the equation satis ed by P can be written as Qx + 1 = Qx for all x 2 R; n,1 n
X
367 where Qx = P xn , xn . But Q 2 R x and is periodic implies that Qx = c for some constant c, so
P xn = xn + c: 1 By degree considerations, P x = ax + b for constants a and b; if P were of degree 2, for instance, the left side of 1 would be of degree 2n, which exceeds the degree of the right side. Thus ax + bn = xn + c. Since n 3 and the right side of this equation has no middle terms", it must be that b = 0. Hence axn = xn + c, so c = 0 and an = 1. Since n is odd, a = 1. In other words, P x = x.
b The proposition is not true as stated. Arguing as in part a, we reach the same conclusion  P xn = xn + c implies P x = ax + b. If n = 1, then P x = x + c so a = 1 and b = c. If n 1, it must be that b = 0, so c = 0 and an = 1. If n is odd, a = 1. If n is even, a = 1 or ,1. In summary:
n
c This proposition is not true as stated, either. For example, take n = 1 and P x = x + c c arbitrary. II. Correction by the editor. Unfortunately, part b of this problem was incorrectly transcribed into print: the functional equation should have been
k k x: k=0 That is, the rst term on the right side should be P xn instead of P xn. With this correction, P x = ,x is never a solution, although P x = x + c is still a counterexample when n = 1. We apologize to the proposer for
P x 1 x + c c arbitrary 1, odd x 1, even x or ,x
P x + 1n = P xn +
n,1 n
X
stating his problem incorrectly. It appears from the proposer's partial solution that he had intended n 2 to be assumed in part b and c as well as a, though his problem statement did not make this clear. In any case, readers are invited to prove or disprove the revised part b for n 2. For part c, see below! III. Solution to part c by Michael Lambrou, University of Crete, Crete, Greece. Lambrou rst solved a and b, giving the same answer for b as in Solution I.  Ed. In this case we show that for n 1 the only polynomial satisfying the
368 modi ed identity
P x + 1n = P xn +
that is, equivalently
n,1 n
X k=0
k k x ;
P x + 1n = P xn + x + 1n , xn; 2 is P x = x. But for n = 1 the polynomial could be of the form P x = x + c for any constant c.
For a start it is easy to verify identity 2 for the stated polynomials, so let us concentrate on the more interesting converse. The case n = 1 is easy: 2 becomes P x + 1 = P x + 1 which is the same equation as in part b for the case n = 1. So this case has already been dealt with, giving the stated conclusion for the form of P . Assume then that n 2. We rst observe that 2, valid for the real variable x, can be lifted to an identity valid also for complex numbers z :
P z + 1n = P zn + z + 1n , zn:
3
For a quick proof, set
Rz = P z + 1n , P zn , z + 1n + zn; which is a polynomial in z . By 2 this polynomial vanishes for an in nity of values of z all real z !. Hence it is identically zero, proving 3. We now show that any polynomial P satisfying 3 must be of degree 1. For this purpose, suppose on the contrary that it is of degree m 2. Write P z = am zm + am,1 zm,1 + + a0 am = 0: 6 Given any real x, set z = e2i=n1 + x , 1
so that
z + 1n = e2i=n1 + x = x + 1n and so P z + 1n = P x + 1n. Using 3 we nd P zn + z + 1n , z n = P xn + x + 1n , xn ;
that is,
n
P zn , P xn = z n , xn :
4
369 Hence
n,1
Y k=0
P z , e2ki=nP x = zn , xn:
5
We have used the fact that the equation Qn,1an = 0 has roots w = e2ki=na w , 0 k n , 1 so wn , an factors as n=0 w , e2ki=na. k The right hand side of 5 as a polynomial in x is
zn , xn = e2i=nx + e2i=n , 1 , xn which is of degree n or less. In fact it is of degree n , 1 or less, as the coe cient of xn is e2i , 1 = 0, leaving powers of x up to n , 1 at most. We shall derive a contradiction to 5 by showing its left hand side is of degree larger than n , 1. For 0 k n , 1 we have P z , e2ki=nP x = am
zm + + a0 , e2ki=namxm + + a0 m = am e2i=nx + e2i=n , 1 + + a0 ,e2ki=namxm + + a0; and the coe cient of xm is clearly
am e2mi=n , e2ki=n = am e2ki=n e2m,ki=n , 1 : 6 Observe that n divides exactly one of the n consecutive numbers m , 0, m , 1, : : : , m , n , 1. Say n j m , k0 for some k0 2 f0; 1; : : : ; n , 1g, but n does not divide any of the remaining n , 1 terms. Thus the coe cient in 6 is nonzero for k = k0 , 0 k n , 1. In other words for k = k0 , 6 6 0 k n , 1, the polynomial P z , e2ki=nP x in x is of degree m. Moreover the polynomial P z , e2k0 i=nP x is not identically zero, since if P z = e2k0 i=nP x for all real x then P z n = P xn for all real x, and so by 4 we0i=n get zn = xn for all real x, which is clearly false; would otherwise z = e2k x for some 0 k0 n , 1, which is incompatible 2i=n 1 + x , 1, as the constant term of this last is nonzero. with z = e
n
To summarize, we have shown: the degree of P z , e2ki=n P x for k 6= k0, 0 k n , 1, is m; P z , e2k0i=nP x is not identically zero. Hence the left hand side of 5 is of degree at least mn , 1 n , 1. This is a contradiction. Hence we conclude that P is a rst degree polynomial, P x = ax + b for some a; b. Recall that we are seeking a polynomial P 2 R x , so a; b are restricted to be real.
370 By 2 we have
ax + 1n + b = ax + bn + x + 1n , xn for all real x. Comparing coe cients of xn and xn,1 and the constant term recall n 2 we nd in turn a = an , na = nan,1 b + n that is, a = an,1 b + 1, and a + b = bn + 1. The second equation gives a = 0 6
so our equations become
an,1 = 1; a = 1 b + 1;
7 8
a + b = bn + 1: 9 By 7, we have a = 1 in fact for even n it only gives a = 1. But if a = ,1 then byn 8, we would get b = ,2 and in 9, we would have ,1 , 2 = ,2 + 1. This last is clearly impossible as ,2n is never equal to ,4. Hence a = 1 and 8 gives b = 0. To summarize, we have shown P x = 1 x +0, showing that P x = x is the only real polynomial
satisfying our condition, and the proof is complete. Remarks. 1 For even n = 2t we do not have to go through the consideration of complex numbers. By setting z = ,1 + x , 1 = ,x , 2 and making the obvious adaptations and simpli cations of the above we can arrive at the proof. For example the analysis into products in 5 can be replaced by and then estimating the degree of the right hand side as at least mt. The di culty of the above approach for odd n arises from the fact that 1 + xn is, for the real variable x, a 11 function, so cannot be put in the form 1 + z n for a di erent real z . Thus we had to go through complex numbers. 2 If one could somehow show P 0 = 0 as expected, then a quick way to complete the proof would be as follows. Set x0 = 0 and xk+1 = 1 + xk n for k = 0; 1; : : : . It is easy to see that each xk is a xed point of P ; that is, P xk = xk , inductively based on
P z2t , P x2t = P z t , P xt P z t + P xt
P xk+1 = P 1+ xkn = P xkn +1+ xk n , xn = 1+ xk n = xk+1: k But the values of xk are all distinct the xk 's are strictly increasing, so P x = x identically as required. Unfortunately, however, I do not have an easy proof of P 0 = 0, other than a proof resembling the proof given above. Can the readers nd such an easy proof of P 0 = 0 or, just as well, P 1 = 1 ?
371
Parts a and the incorrectly stated b were also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; and DIGBY SMITH, Mount Royal College, Calgary, Alberta. Hess and Smith found all functions satisfying the given equation, as did our featured solvers Stone and Zarikian Solution I and Lambrou; Bradley and Herzig just showed that P x = x is not the only such function. Part a only was solved by the proposer. There was also one incorrect solution sent in. Stone and Zarikian also consider the similar functional equation n,1 ! X n k P x + 1n = P xn + k x;
k=0
which, in view of the correction to part b, is the obvious fourth variation on the proposer's original three equations. They prove easily that the only polynomials satisfying this equation for all real x are of the form P x = x + c for c constant. Readers may like to check this out for themselves, and then try nishing o part b!
2249.
1997: 245 Proposed by K.R.S. Sastry, Dodballapur, India. How many distinct acute angles are there such that
1 cos cos 2 cos 4 = 8 ?
Solution by Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. There are exactly three such values: = , 27 , and . 9 3 Let P = cos cos 2 cos 4 . Then, for all such that sin 6= 0, we have P = 1 if and only if 8 integer k. That is, either i = 2k , or 7 ii = 2k+1 . 9 Note that = 0 is not a solution. Since 59 47 , we see that the 2 only possible values for k, to ensure that 0 , are k = 1 in i and 2 2 , respectively. Since sin 6= 0 k = 0; 1 in ii. These yield = 7 , 9 , and 3 for these values, the result P = 1 follows. 8
sin = 8P sin = sin8 : This is equivalent to having 8 = 2k + or 8 = 2k + 1 , for some
Also solved by NOTE: a dagger y before a name indicates that the solver's solution is very similar to the one above, HAYO AHLBURG, Benidorm, Spain; GERALD ALLEN, CHARLES DIMINNIE, TREY SMITH and ROGER ZARNOWSKI, Angelo State University, San Angelo, TX, USA; MIGUEL AMENGUAL COVAS, Cala
Figuera, Mallorca, Spain; SEFKET ARSLANAGIC, University of Sarajevo, Sarajevo,
372
Bosnia and Herzegovina; SAM BAETHGE, Nordheim, Texas, USA; y MICHAEL BATAILLE, Rouen, France; y FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; ADRIAN BIRKA, student, Lakeshore Catholic High School, Port Colbourne, Ontario; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; ADRIAN CHAN, student, Upper Canada College, Toronto, Ontario; JIMMY CHUI, student, Earl Haig Secondary School, North York, Ontario; y THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; y GORAN CONAR, student, Gymnasium Varadin, Varadin, Croatia; DAVID DOSTER, Choate Rosemary Hall, Wallingford, z z Connecticut, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria;
VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; y KEEWAI LAU, Hong Kong; y ALAN LING, student, University of Toronto, Toronto, Ontario; VEDULA N. MURTY, Visakhapatnam, India; VICTOR OXMAN, University of Haifa, Haifa, Israel; y BOB PRIELIPP, University of Wisconsin Oshkosh, Wisconsin, USA; DIGBY SMITH, Mount Royal College, Calgary, Alberta; y CHRISTOPHER SO, student, Francis Libermann Catholic High School, Scarborough, Ontario; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; PANOS E. TSAOUSSOGLOU, Athens, Greece; DAVID VELLA, Skidmore College, Saratoga Springs, NY, USA; and the proposer. There were also two incorrect and four incomplete submissions, two of which gave the correct answer without proof.
1997: 245 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a scalene triangle with incentre I . Let D, E, F be the points where BC , CA, AB are tangent to the incircle respectively, and let L, M , N be the midpoints of BC , CA, AB respectively. Let l, m, n be the lines through D, E , F parallel to IL, IM , IN respectively. Prove that l, m, n are concurrent. Solution by Michael Lambrou, University of Crete, Greece. ,! a ,! ~ ,! Introduce position vectors, ~ = BC , ~ = BA, d = BD, c ~ = ,!, f = ,!, all originating from B. If we denote, as usual, theclengths e BE ~ BF jBC j = a
jCAj b, and jABj = c,then as jBDj
jBF j = a+2,b , , = = we
a c ~ = a+2ca,b ~ and f = a+2cc,b ~ . Also ~ = a+b,c~2+b+c,a~ . ~ have d c a e b
2250.
,!
c c If the bisector, AI , cuts BC at P , then jBP j = bacc , so BP = b+c ~. + AI So, as jjIP jj = jjABjj = b+c , we have, BP a , = ,! + , =
a ~ +
c ~: ! ! BI BP PI a+b+c a a+b+c c ,! x
, , a
, , c , Consider X , where BX = ~ = 3aa+bb+cc ~ + 3ac+bb+ca ~. We show that X is on lines l; m and n, so these lines are concurrent, as is required. , ! All we need to do is to verify that: ~ , d, ~ , ~ and ~ , f are parallel to IL, x ~x e x ~ ,! and , , respectively. ! IM IN
373 Then and
~ = 3a , b , c ~ + 3c , b , a , a + c , b ~ ~ ,d x c a+b+c a a+b+c a
, = , , , = ~ , a ~ + c ~
! ! ! c IL BL BI 2 a + b + c a a + b + c c
c a = 2a + bb, cc ~ , a + a + c ~ a+ +
b ,a = 3a , b , c ~ , d: x ~ , ! That ~ , f is parallel to IN follows immediately from the above by x ~ interchanging the roles of c and a. Similar routine calculation shows ~ , ~ is x e ,! parallel to IM .
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer. There was one incomplete solution.
2251. 1997: 300 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. In the plane, you are given a circle but not its centre, and points A, K , B, D, C on it, so that arc AK = arc KB and arc BD = arc DC . Construct, using only an unmarked straightedge, the midpoint of arc AC . Preliminary comment. There are two arcs AC ; the midpoint of one of them will always be easily constructed; the other is more challenging. Most solvers including the proposer constructed only the easier one, which does not require the lemma. Note that it is essentially the same as the lemma in Oxman's solutions to his earlier problem: 2234 1998: 247 . Solution by Toshio Seimiya, Kawasaki, Japan. Lemma. If we are given points A, M , B , and C on a line with AM = MB, then using only an unmarked ruler we can construct D on the line such that BC = CD. Construction. Let P be a point not on the line AB , and let Q be a point on the segment AP . Let R = BQ MP; S = AR BP;
N = PM QS; D = QT AB: T = NC SB; Then BC = CD:
374
PQ AM BS = 1. Since QA MB SP PQ = PS , so that QS k AB. It follows that since AM = MB we get QA SB QD, NC , SB are concurrent at T , BC SN CD = NQ ; and since SA, NM , QB are concurrent at R, SN = AM : NQ MB Using these two equalities together with AM = MB , we conclude that BC = AM = 1; CD MB so that BC = CD.
Proof. By Ceva's Theorem we have
We now turn to the main construction. Construction. We denote the given circle by ,. Draw AD meeting CK at I . Let M be the second intersection of BI with ,; then M is the midpoint of the arc AC not containing B . Let N be the intersection of KD with BI ; then BN = NI . By the lemma we can construct the point IB such that IM = MIB . Let IC be the intersection of AIB with KC , and let L be the second intersection of IC B with ,. Then L is the midpoint of the arc ABC . Proof. Since arc AK = arc KB , we have that CK bisects ACB . Similarly AD bisects BAC , so that I is the incentre of 4ABC . Editor's comment. The statement of the problem might be construed as allowing C and K to lie on the same side of the line AB , in which case I would be an excentre. Seimiya's argument is based on the interpretation that the points lie on the circle in the prescribed order, namely AKBDC ; it easily can be modi ed to accommodate the alternative interpretation. Hence BM bisects ABC , so that M is the midpoint of the arc AC not containing B . Since I is the incentre and K is the midpoint of arc AB , we have KI = KB. Similarly we have DI = DB. Hence KD is the perpendicular bisector of BI , and N is the midpoint of BI . By the lemma we can therefore construct point IB such that IM = MIB using only an unmarked ruler. Since M is the midpoint of arc AC while I is the incentre and IM = MIB ; IB is an excentre. Thus IC is an excentre, IC B is the exterior bisector of ABC , and L is the midpoint of arc ABC .
Also solved by NIELS BEJLEGAARD, Stavanger, Norway; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; HANS ENGELHAUPT, Franz Ludwig Gymnasium, Bamberg, Germany; SHAWN GODIN, St. Joseph Scollard Hall, North Bay,
375
Ontario; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; D.J. SMEENK, Zaltbommel, the Netherlands; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer.
1997: 300 Proposed by K.R.S. Sastry, Dodballapur, India. Prove that the ninepoint circle of a triangle trisects a median if and only if the side lengths of the triangle are proportional to its median lengths in some order. The rst part of the solution is by Christopher J. Bradley, Clifton College, Bristol, UK. The rest is a compilation of ideas from other solutions liberally sprinkled with comments by editor, Cathy Baker. If l; m and n are the lengths of the medians from A; B and C respectively, then Apollonius' Theorem gives the formulae
2252.
Suppose a b c. Then clearly from these formulae, n m l. If the medians are proportional to the sides, these orderings must be preserved and so a constant k exists such that
4l2 = 2b2 + 2c2 , a2 4m2 = 2c2 + 2a2 , b2 4n2 = 2a2 + 2b2 , c2:
Note that if the triangle has two equal sides, then it must be equilateral. If, for example, a = b c, then 2b2 = a2 + c2 and 2a2 = b2 + c2 , so adding gives 2c2 = a2 + b2 , and subtraction, a = b = c. Conversely, if a2 + c2 = 2b2, then substitution in the formulae gives p l 4l2 = 3c2; 4m2 = 3b2 and 4n2 = 3a2, so c = m = n = 23 ; that is, b a the medians are proportional to the sides. Note that in this case, b must lie between a and c. If, for example, b c a c a b , then 2b2 = a2 + c2 2c2 2b2 = a2 + c2 2a2 ; a contradiction. We have shown that a triangle with sides of length a; b;c is self median if and only if 2b2 = a2 + c2 , where b lies between a and c. Let D and E be the midpoints of AC and BC , respectively; F the foot of the altitude from A; P the other point where the nine point circle meets
2a2 + 2b2 , c2 = ka2 2c2 + 2a2 , b2 = kb2 2b2 + 2c2 , a2 = kc2: Adding, one nds k = 3 and each of the three equations reduces to a2 + c2 = 2b2:
376
BD. Then BD = kBP , for some k 0. Then using the power of B with
respect to the ninepoint circle, we get:
2 2 2 But BF = c cos B and, by the Cosine Law, cos B = a +2cac,b . Since BD2 = m2 = 2c2+24a2,b2 and BE = a ; we have 2
BP BD = BF BE:
2c2 + 2a2 , b2 = aca2 + c2 , b2 4k 4ac k , 1b2 = k , 2a2 + c2:
If the ninepoint circle trisects BD, then k = 3, so 2b2 = a2 + c2 , b lies between a and c, and the medians are proportional to the sides. Conversely, if 2b2 = a2 + c2 , then k , 3b2 = 0 and k = 3. Hence if 2 = a2 + c2 , then the ninepoint circle trisects the median from B . 2b
Solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; GERRY LEVERSHA,
St. Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; TOSHIO SEIMIYA, Kawasaki, Japan; and the proposer.
1997: 300 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a triangle and Ib, Ic are the excentres of 4ABC relative to sides CA, AB respectively. Suppose that Ib A2 + IbC 2 = BA2 + BC 2 and that Ic A2 + IcB2 = CA2 + CB2. Prove that 4ABC is equilateral. Solution by Florian Herzig, student, Perchtoldsdorf, Austria. Let us turn the problem around and start from 4ABC with sides a; b; c and the triangle formed by the feet of the altitudes D, E, F . Then this 4ABC plays the role of the given 4Ia Ib Ic , while 4DEF corresponds to the original 4ABC since Ia A, etc. are the altitudes in 4Ia Ib Ic . In this notation our given condition is that BD2 + BF 2 = ED2 + EF 2 and CD2 + CE2 = FD2 + FE2, and we are to show that 4DEF is equilateral. Since AE : AF = c cos A : b cos A = AB : AC , we have 4AEF 4ABC and, hence, EF = a cos A. Similarly FD = b cos B and DE = c cos C . Hence we get
2253.
a2 + c2 cos2 B = a2 cos2 A + c2 cos2 C; a2 + b2 cos2 C = a2 cos2 A + b2 cos2 B:
1 2
377 Adding b2 cos2 B to the rst and c2 cos2 C to the second equation yields
a2 + b2 + c2 cos2 B + a2 + b2 + c2 cos2 C; and, hence, B = C since B + C 6= . Thus 2 becomes a2 + b2 cos2 B = a2 cos2 A + b2 cos2 B; whence A = B . Therefore 4ABC and also 4DEF are equilateral.
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain;
FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, and MAR IA ASCENSION LOPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; GORAN CONAR, student, Gymnasium Varadin, Varadin, z z Croatia; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece;
WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England;
VEDULA N. MURTY, Visakhapatnam, India; ISTVAN REIMAN, Budapest, Hungary; MARAGOUDAKIS PAVLOC, Pireas Greece; D.J. SMEENK, Zaltbommel, the Netherlands; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer. There was one incorrect solution. Chronis notes that the result can be generalized by replacing the exponent 2 by any positive real: 4ABC is equilateral if for some t 0, we have Ib At + Ib C t = BAt + BC t and Ic At + Ic B t = CAt + CB t . Indeed, our featured solution extends to the proof of Chronis's more general statement.
1997: 300 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is an isosceles triangle with AB = AC . Let D be the point on side AC such that CD = 2AD. Let P be the point on the segment BD such that APC = 90 . Prove that ABP = PCB . Solution by Florian Herzig, student, Perchtoldsdorf, Austria. Let Q be the midpoint of BC and complete the rectangle AQCS: It follows that B; D and S are collinear, as AS jjBC and AS : BC = AD : DC . Now observe that A; P; Q; C and S are concyclic by Thales' Theorem APC = AQC = ASC = 90 , so P; Q and S are all on the circle with diameter AC and AB jjSQ. Hence
2254.
ABP = PSQ = PCQ = PCB:
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; MANSUR BOASE, student, St. Paul's School, London, England; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; RUSSELL EULER and JAWAD SADEK, NW
378
Missouri State University, Maryville, Missouri, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D. KIPP JOHNSON, Beaverton, Oregon, USA; DAG JONSSON, Uppsala, Sweden; GEOFFREY A. KANDALL, Hamden, Connecticut, USA;
VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece 2 solutions; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; VICTOR OXMAN, University of Haifa, Haifa, Israel; PAVLOS MARAGOUDAKIS, Hatzikeriakio, Pireas, Greece; LUIS A. PONCE, Santos, Brazil; D.J. SMEENK, Zaltbommel, the Netherlands 2 solutions; PANOS E. TSAOUSSOGLOU, Athens, Greece; ENRIQUE VALERIANO, National University of Engineering, Lima, Peru; JOHN VLACHAKIS, Athens, Greece; PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer.
1997: 300 Proposed by Toshio Seimiya, Kawasaki, Japan. Let P be an arbitrary interior point of an equilateral triangle ABC . Prove that j PAB , PAC j j PBC , PCB j. Solution by Theodore Chronis, student, Aristotle University of Thessaloniki, Greece. Without loss of generality we suppose that PAB PAC .
2255.
A
P B C C0 D In the gure PC = PC 0 ; we deduce that BPD CPD since BPD C 0PD CPD . Therefore PAB + ABP PAC + ACP opposite interior angles ;
which implies
PAB + 60 , PBC PAC + 60 , PCB;
which in turn implies
PAB , PAC PBC , PCB:
379 Editor's comment. It is clear from Chronis's argument that the result continues to hold for any isosceles triangle ABC with B = C . Just replace the 60 angle by the base angle. Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; JORDI DOU, Barcelona, Spain; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D. KIPP JOHNSON, Beaverton, Oregon, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; VICTOR OXMAN, University of Haifa, Haifa, Israel; PANOS E. TSAOUSSOGLOU, Athens, Greece; ENRIQUE VALERIANO, Lima, Peru; and the proposer.
2256. 1997: 300 Proposed by Russell Euler and Jawad Sadek, Department of Mathematics and Statistics, Northwest Missouri State University, Maryville, Missouri, USA.
If 0
I. Solution by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA. 1 1 First we show that ln u 1 , u for u 1. Let g u = ln u , 1 + u . , Then g 0 u = uu21 0, which implies that g is strictly increasing on 1; 1. Hence g u g 1 = 0 for u 1. Now, hold y xed at y = a, where 0 a x
1, and let
, ln 1 y x 1, prove that lnxx , y y ln y .
x a ln x , ln a + ln a. Then f 0 x = 1 , x , ln a : f x = x , a x , a2 Since x 1, we have f 0x 0 by the inequality above. Hence f is a a ln strictly decreasing on a; 1 and so f x f 1 = ,1,a a 0.
II. Solution by HeinzJurgen Sei ert, Berlin, Germany. For a xed y 2 0; 1, consider the function f x = x ln y +ln x, where y x 1. Since 0 y 1, we have y ln y + ln y ln y ; that is, f y f 1. Since f 00 x = , x12 0, we have that f is strictly concave on y; 1 , and since y x 1, we get f x f y , which is readily seen to be equivalent to the proposed inequality.
Also solved by PAUL BRACKEN, CRM, Universit
de Montr
al, Montr
al, e e e Qu
bec; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE e CHRONIS, student, Aristotle University of Thessaloniki, Greece; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; GORAN CONAR, student, Gymnasium Varadin, Varadin, Croatia; LUZ M. z z DeALBA, Drake University, Des Moines, IA, USA; HANS ENGELHAUPT, Franz Lud
380
wig Gymnasium, Bamberg, Germany; NOEL EVANS, GERALD ALLEN, CHARLES R. DIMINNIE, TREY SMITH and ROGER ZARNOWSKI jointly, Angelo State University, San Angelo, TX, USA; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; JOE HOWARD, New Mexico Highlands University, Las Vegas, NM, USA; WALTHER JANOUS, Ursulinengym
nasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; NICK LORD, Tonbridge School, Tonbridge, Kent, England; PHIL McCARTNEY, Northern Kentucky University, Highland Heights, KY, USA; VICTOR OXMAN, University of Haifa, Haifa, Israel; M.PERISASTRY, Maharaja's College, Vizianagaram, Andhra Pradesh, India; RENATO ALBERTO RODRIGUES, Pra a Osc waldo Cruz, Paraiso, S~ o Paulo, Spain; DIGBY SMITH, Mount Royal College, Calgary, a Alberta; PANOS E. TSAOUSSOGLOU, Athens, Greece; UNIVERSITY OF ARIZONA PROBLEM SOLVING LAB, University of Arizona, Tucson, AZ, USA; VREJ ZARIKIAN and DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; and the proposers. Most of the submitted solutions are similar to either
or II given above. Herzig 1I proved the stronger result that ln x , ln y ln 1 + y for all distinct x, y with x,y 0 x; y 1 by using the following inequality known as Bernoulli's Inequality: 1 + x 1 + x, where x ,1, x 6= 0 and 1 or 0. Ed. See, for example, page 34 of D.S. Mitrinovi
et al., Recent Advances in Geometc ric Inequalities, Kluwer Academic Publishers, 1989.
2258. 1997: 301 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. In a rightangled triangle ABC with C = 90 , D lies on the segp ment BC so that BD = AC 3. E lies on the segment AC and satis es p3. Find the angle between AD and BE. AE = CD I. Solutionby D. Kipp Johnson, Valley Catholic High School,Beaverton, Oregon. Without loss of generality we put A at 0; 1, C at 0; 0, and D at x; 0. Then BD = AC p3 = p3, so B is at x + p3; 0. Since AE = CDp3 = xp3, E is at 0; 1 , xp3. The slopep the line AD of p is ,1=x and the slope of the line BE is x 3 , 1=x + 3. If is the angle between AD and BE , then:
xp3 p 1 + 1 , p x +
p x ! = 3x2 + 1 = p3; 3 tan = x2 + 1 1 , 1 + ,x x 3p 1 x+ 3
implying = 60 .
381
II. Solution by Niels Bejlegaard, Stavanger, Norway. ,!  ,!  ,! y Let CA = a~ and EA = y~. Then we have that CD = p ~ and
,! = ap3 + p
~. Thus y CB 3 ,! = ,! + ,! = ,! + ,! + ,! EB EC CB EA AC CB p y
= ,! , ,! + ,! = y , a~ + a 3 + p ~ ; EA CA CB  3 ,! = ,! + ,! = ,a~ + p ~ : y AD AC CD 
Then the angle between the vectors EB and AD satis es:
3
,!
3
,!
III. Solution by Toshio Seimiya, Kawasaki, Japan. Let P be a point such that AP k CB and DP k CA. Since CAPD is a rectangle, we p PD = AC , AP = p , and CDP = 90 = CAP . get CD As BD = AC 3 we get BD = PD 3, so that DPB = 60 . Since AE = CDp3 we have AE = AP p3, so that APE = 60 . Since 4PAE 4PDB, we have PA : PD = PE : PB, and APD = 60 + EPD = EPB. Therefore 4PAD 4PEB. Thus we have
p y
y ,! ,! y , a,a + a 3 + p p EB 3 3 q q cos = ,! AD = ,!j 4 2 1 2 jEBj jAD 4a2 + 3 y a2 + 3 y a2 1 2 = hq + 3 y i2 = 1 : 2 1 2 a2 + 3 y 2 Therefore, = 60 .
PAD = PEB: 1 Let O be the intersection of AD and BE . Then we know from 1 that EOA = EPA = 60 . Therefore the angle between AD and BE is 60 .
E C O D B
A
60
P
60
382
IV. Solution by Luiz A. Ponce, Santos, Brazil. This is a generalization to the case where BD = AC m and AE = CD m for some m 0. Let AC = b, CD = a, and AD = c. De ne F to be the point of intersection of AD and BE . Let = DFB , = CDA, and = CAD.
H
A 6 6
am b
? E
C
N
K
F
?
Since AE = CD m and BD = AC m, we have AE = am and BD = bm. Since ACD is a rightangled triangle, we have + = 90 : 1 Through the point B construct a line parallel to AD intersecting the line AC extended at H , and through the point A construct a line parallel to CB intersecting the line BH at K . It is now clear that EAK = 90 , EBK = , AK = BD = bm, and BK = AD = c : 2 Note that AE = am = m = bm = AK : CD a b AC Since we also have DCA = EAK = 90 , it follows that 4AEK 4CDA SAS. According to similarity we can write EK = AD m = cm ; 3 and also KEA = ADC = = NEA, where N is the point of intersection of AD with EK . Moreover, in triangle ANE we have N + NEA + EAN = 180 ; that is, N + + = 180 ; or N = 90
a
D 
bm
B
383 by considering 1. Thus AD ? EK , which implies that EK ? BK since AD k BK . Consequently we conclude that EKB is a rightangled triangle. This together with 2 and 3 yield Therefore, the acute angle between AD and BE is , such that tan = m, where m 0 is the given ratio. V. Solutionby Paul Yiu, Florida AtlanticUniversity, Boca Raton, Florida, USA. points D and if r BC is possible to p3Let r = AC .3.ItThen the anglelocate theAD and BE isE60 and only p 4 if between . This is 3 a special case of the following situation. Let ACB be a right triangle with r = BC 1. For any positive t r satisfying t2 , rt + 1 0, let D and E AC be points on the segments BC and AC respectively, so that BD = t AC , and AE = t CD. Then, the angle between AD and BE is arctan t.
EK tan = tan EBK = BK = cm = m: c
B
D
C
q q
K H
P0
E
A
Complete the right triangle into a rectangle ACBK . For t r, we choose points D on BC and H on KA such that DBKH is a rectangle with DB = t KB. Since t2 , rt + 1 0, AC tAK , t BK = t AH , there is a point E on AC such that AE = t AH . Note that HE and HB are perpendicular to each other, since the right triangles HAE and BKH are similar. Consider the circumcircles of the rectangle DBKH and the right triangle AEH . The centres of these circles lie on HB and HE respectively. Let P 0 be the intersection of these circles other than H . Note that HP 0 E and HP 0 B are both right angles. This means that B , P 0 , and E are collinear. Note also that
BP 0D = BKD = AHE = AP 0E:
384 From this we conclude that A, P 0 , and D are collinear. This means that P 0 is the intersection of the segments AD and BE , and the angle between them is the same as BKD, which is clearly arctan t.
Remark. If r 2, the condition t2 , rt + 1 0 is always satis ed forp r. t
r, r2 ,4 For r 2, this is satis ed by values of t in either of the ranges 0; 2 p2
r+ r ,4 ; r . For example, for r = 5 , t must satisfy either 0 1 and t 2 or 2 2 2 t 5. 2
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; SAM BAETHGE, Nordheim, Texas, USA; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; PAUL BRACKEN, Universit
de Montr
al, e e Qu
bec; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MIGUEL ANGEL e
CABEZON OCHOA, Logro~ o, Spain; THEODORE CHRONIS, student, Aristotle Unin versity of Thessaloniki, Greece; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; HANS ENGELHAUPT, Franz Ludwig Gymnasium, Bamberg, Germany; FLORIAN HERZIG, student, Perchtoldsdorf, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA;
WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; VICTOR OXMAN, University of Haifa, Haifa, Israel;
ISTVAN REIMAN, Budapest, Hungary; ANGEL JOVAL ROQUET, LaSeud'Urgell, Spain; HEINZJURGEN SEIFFERT, Berlin, Germany; D.J. SMEENK, Zaltbommel, the Netherlands; PANOS E. TSAOUSSOGLOU, Athens, Greece; ENRIQUE VALERIANO, National University of Engineering, Lima, Peru; JOHN VLACHAKIS, Athens, Greece;
PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer. Both Hess and Lambrou consider the case when the points D and E are external to the segments BC and AC . Neither of them consider all the possible combinations, however. Perhaps the interested reader would like to pursue this investigation.
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
385
THE OLYMPIAD CORNER
No. 193 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. As a rst Olympiad to give you puzzling pleasure, we give the 18th AustrianPolish Mathematics Competition written in Austria June 28 30, 1995. My thanks go to Bill Sands, University of Calgary, who collected this contest while assisting at the International Olympiad in Toronto in 1995, as well as to Walther Janous, Ursulinengymnasium, Innsbruck, Austria.
18th AUSTRIANPOLISH MATHEMATICS
Problems of the Individual Contest
June 2829, 1995 Time: 4.5 hours
COMPETITION
1. For a given integer n 3 nd all solutions a1; : : : ; an of the system of equations
in real numbers. 2. Let A1; A2; A3; A4 be four distinct points in the plane and let X = fA1; A2; A3; A4g. Show that there exists a subset Y of the set X with the following property: there is no disc K such that K X = Y . Note: All points of the circle limiting a disc are considered to belong to the disc. 3. Let P x = x4 + x3 + x2 + x + 1. Show that there exist polynomials Qy and Ry of positive degrees, with integer coe cients, such that Qy Ry = P 5y2 for all y. 4. Determine all polynomials P x with real coe cients, such that P x2 + P 1=x2 = P x2 P 1=x2 for all x 6= 0:
a3 = a2 + a1 ; a4 = a3 + a2; : : : ; an = an,1 + an,2 a1 = an + an,1 ; a2 = a1 + an
5. An equilateral triangle ABC is given. Denote the midpoints of sides BC , CA, AB respectively by A1 , B1 , C1 . Three distinct parallel lines p; q;r are drawn through A1, B1, C1, respectively. Line p cuts B1C1 at A2;
386 line q cuts C1A1 at B2 ; line r cuts A1 B1 at C2 . Prove that the lines AA2 , BB2, CC2 concur at a point D lying on the circumcircle of triangle ABC . 6. The Alpine Club consisting of n members organizes four highmountain expeditions for its members. Let E1 , E2, E3, E4 be the four teams participating in these expeditions. How many ways are there to compose those teams, given the condition that E1 E2 6= ;, E2 E3 6= ;, E3 E4 6= ;?
June 30, 1995 Time: 4 hours 7. For every integer c consider the equation 3y4 + 4cy3 + 2xy + 48 = 0, with integer unknowns x and y. Determine all integers c for which the number of solutions x; y in pairs of integers satisfying the additional conditions A and B is a maximum: A the number jxj is the square of an integer; B the number y is squarefree that is, there is no prime p with p2 dividing y . 8. Consider the cube with vertices f1; 1; 1; that is, the set fx; y; z : jxj 1; jyj 1; jzj 1g. Let V1; : : : ; V95 be points of that cube. Denote by vi the vector from 0; 0; 0 to Vi. Consider the 295 vectors of the form s1v1 + s2 v2 + + s95 v95, where si = 1 or si = ,1. a Let d = 48. Show that among all such vectors one can nd a vector w = a; b; c with a2 + b2 + c2 d. b Find a number d 48 with the same property. Note: The smaller d, the better mark will be attracted by the solution. 9. Prove that the following inequality holds for all integers n;m 1 and all positive real numbers x; y :
Problems of the Team Contest
n , 1m , 1xn+m + yn+m + n + m , 1xn ym + xm y n nmxn+m,1y + xyn+m,1:
The next contest we give was also collected by Bill Sands while he was assisting at the IMO in Toronto. These are the problems of the 9th Iberoamerican Mathematical Olympiad held September 20, 21 in Fortaleza, Brazil. Students were given 4 1 hours each day. 2
9th IBEROAMERICAN MATHEMATICAL OLYMPIAD
Fortaleza, Brazil, September 20
First Day  Time: 4.5 hours
21, 1994
1. Mexico: A natural number n is called brazilian if there exists an integer r, with 1 r n , 1, such that the representation of the number
387
n in base r has all the digits equal. For example, 62 and 15 are brazilian,
because 62 is written 222 in base 5 and 15 is 33 in base 4. Prove that 1993 is not brazilian, but 1994 is brazilian. 2. Brazil: Let ABCD be a cyclic quadrilateral. We suppose that there exists a circle with centre in AB , tangent to the other sides of the quadrilateral. i Show that AB = AD + BC . ii Calculate, in terms of x = AB and y = CD, the maximal area that such a quadrilateral can reach. 3. Brazil: In each cell of an n n chessboard is a lamp. When a lamp is touched, the state of this lamp, and also the state of all the lamps in its row and in its column, is changed switched from OFF to ON and vice versa. At the beginning, all the lamps are OFF. Show that it is always possible, with suitable sequence of touches, to turn ON all the lamps of the chessboard, and nd, in terms of n, the minimal number of touches in order that all the lamps of the chessboard are ON.
Second Day  Time: 4.5 hours
4. Brazil: The triangle ABC is acute, with circumcircle k. Let P be an internal point to k. The lines AP , BP , CP meet k again at X , Y , Z . Determine the point P for which triangle XY Z is equilateral. 5. Brazil: Let n and r be two positive integers. We wish to construct r subsets of f0; 1; : : : ; n , 1g, called A1 ; : : : ; Ar , with cardAi = k and such that, for each integer x, 0 x n , 1, there exist x1 2 A1 ; x2 2 A2; : : : ; xr 2 Ar an element in each subset, with
x = x1 + x2 + + xr : Find, in terms of n and r, the minimal value of k.
Brazil: Show that all natural numbers n 21000000 can be obtained beginning at 1 with less than 1100000 sums; that is, there exists a nite sequence of natural numbers x0 ; x1 ; : : : ; xk , with k 1100000, x0 = 1, xk = n, such that for each i = 1; 2; : : : ; k, there exists r, s, with 0 r i, 0 s i, and xi = xr + xs . As a nal problem set to challenge you, we present the problems of the IX, X and XI Grade of the Georgian Mathematical Olympiad, Final Round for 1995. It is interesting that 60 of the Grade XI problems come from the Grade IX paper. My thanks again go to Bill Sands, University of Calgary, for collecting these problems while he assisted with the IMO in Toronto in 1995.
6.
388
GEORGIAN MATHEMATICAL OLYMPIAD 1995
Final Round
GRADE IX
1. A threedigit number was decreased by the sum of its digits. Then the result was decreased by the sum of its digits and so on. Show that on the 100th step of this procedure the result will be zero, whatever the initial threedigit number is chosen. How many repetitions are enough to get zero? 2. Two circles of the same size are given. Seven arcs, each of them of 3 measure, are taken on the rst circle and 10 arcs, each of them of 2 measure, are taken on the second one. Prove that it is possible to place one circle on the other so that these arcs do not intersect. Is it or is it not possible to prove the same if the number of arcs with measure 2 is 11? 3. Prove that if the product of three positive numbers is 1 and their sum is more than the sum of their reciprocals, then only one of these numbers can be more than 1. 4. Prove that in any convex hexagon there exists a diagonal which cuts from the hexagon a triangle with area less than 1 of the area of the hexagon. 6 5. The set M of integers has the following property: if the numbers a and b are in M , then a + 2b also belongs to M . It is known that the set contains positive as well as negative numbers. Prove that if the numbers a, b and c are in M , then a + b , c is also in M . GRADE X 1. a Five di erent numbers are written in one line. Is it always possible to choose three of them placed in increasing or decreasing order? b Is it always possible to do the same, if we have to choose four numbers from nine? 2. Same as IX.2 3. Prove that for any natural number n, the average of all its factors lies between the numbers pn and n+1 . 2 4. The incircle of a triangle divides one of its medians into three equal parts. Find the ratio of the sides of the triangle. 5. The function f is given on the segment 0; 1 . It is known that f x 0 and f 1 = 1. Besides that, for any two numbers x1 and x2, if x1 0, x2 0 and x1 + x2 1, then f x1 + x2 f x1 + f x2. a Prove that f x 2x for any x. b Does the inequality f x 1:9x hold for every x?
389
GRADE XI
table. By one move, it is allowed to double all numbers of any row or subtract 1 from all numbers of any column. Prove that by repeating these moves several times, all numbers in the table become zeros. Next a bit of housekeeping. After the columns were set, and before they appeared in print form, I received solutions from Pavlos Maragoudakis to the six problems of the Swedish contest for which we published the solutions last issue 1997: 196; 1998: 328 329 . He also submitted solutions to problems 1, 2, 3 and 5 of the Dutch Mathematical Olympiad, Second Round, 1993 1997: 197; 1998: 329 332 . Last issue we gave solutions to all but the last problem. It was rather unfortunate timing in terms of acknowledging his contribution, but we are able to close out the le by having a complete set of solutions from the readers. 5. P1; P2; : : : ; P11 are eleven distinct points on a line. PiPj 1 for every pair Pi , Pj . Prove that the sum of all 55 distances Pi Pj , 1 i j 11 is smaller than 30. Solution by Pavlos Maragoudakis, Pireas, Greece. Without loss of generality we suppose that P1 ; P2 ; : : : ; P11 are adjacent.
1. Same as IX.3 2. Same as IX.1 3. How many solutions has the equation x = 1995 sin x + 199? 4. Same as IX.4 5. A natural number is written in each square of an m n rectangular
P1
Now if 1 i
X
1
P2
qqq
Pj P1 , PiP1
P11
j 11, then PiPj = Pj P1 , PiP1. So
X
1
i j 11
PiPj =
= 10P11P1 + 9P10 P1 , P10 P1 + 8P9 P1 , 2P9P1 +7P8 P1 , 3P8 P1 + 6P7 P1 , 4P7P1 + 5P6P1 ,5P6P1 + 4P5P1 , 6P5P1 + 3P4P1 , 7P4P1 +2P3 P1 , 8P3 P1 + P2 P1 , 9P2 P1
i j 11
390
= 10P11P1 + 8P10P1 + 6P9 P1 + 4P8P1 + 2P7P1 ,2P5P1 , 4P4P1 , 6P3P1 , 8P2P1 = 10P11P1 + 8P10P1 , P2P1 + 6P9P1 , P3 P1 +4P8P1 , P4 P1 + 2P7P1 , P5P1 = 10P11P1 + 8P10P2 + 6P9 P3 + 4P8P4 + 2P7P5 10 1 + 8 1 + 6 1 + 4 1 + 2 1 = 30
Also setting the record straight, I found amongst the solutions for another contest, the solution by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain to problem 2 of the Dutch Mathematical Olympiad, Second Round, for which we published a solution last issue 1997: 197, 1998: 330 331 . My apologies. While we do not normally give solutions to problems of the USAMO, I am giving two comments solutions from our readers to problems of the USAMO 1997 1997: 261, 262 . 2. Let ABC be a triangle, and draw isosceles triangles BCD, CAE, ABF externally to ABC , with BC; CA; AB as their respective bases. Prove ! ! ! that the lines through A; B; C perpendicular to the lines EF; FD; DE , respectively, are concurrent. Comment by Mansur Boase, student, St. Paul's School, London, England. The result is immediate from Steiner's Theorem: If the perpendiculars from the vertices A, B , C of a triangle ABC to the sides B1 C1 , C1 A1 , and A1 B1 , respectively, of a second triangle A1 ; B1 ; C1 are concurrent, then the perpendiculars from the vertices A1 ; B1 ; C1 of the triangle A1 B1C1 to the sides BC , CA, AB are also concurrent. 5. Prove that for positive real numbers a, b, c,
a3 + b3 + abc,1 + b3 + c3 + abc,1 + c3 + a3 + abc,1 abc,1:
Solutions by Mansur Boase, student, St. Paul's School, London, England and by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. We give Klamkin's presentation. Since the inequality is homogeneous, we can assume abc = 1. Then if we let x = a3 , y = b3, z = c3 , the inequality becomes where xyz = 1 and x, y , z are positive. On expanding, 1 is equivalent to
1 1 + 1+y+z + 1+z+x 1 1+x+y x + y + zxy + yz + zx , 2 3 :
1
1
391 This follows from the known elementary inequalities
There is equality if and only if x = y = z = 1. Comment: The inequality in the form 1 was also given in the Spring 1997, Senior ALevel Tournament Of The Towns competition. A generalization to
x + y + z yz + zx + xy
1=2 xyz1=3 : 3 3
1 + 1 + S , x + + 1 + S1, x 1 ; 1 + S , x1 2 n where S = x1 + x2 + + xn , x1 x2 : : : xn = 1, and xi 0 is due to Dragos
1
Hrimiuc, University of Alberta, and will probably appear as a problem in Math. Magazine. Next we give two solutions by our readers to two problems of the 3rd Ukrainian Mathematical Olympiad, March 26 27, 1994 given in 1997: 262 . 2. 9 10 A convex polygon and point O inside it are given. Prove that for any n ,,!there ! points,,!A2 ;,: : ; An on the sides of the polygon 1 exist A1; : such that OA1 + ,, 2 + : : : + OAn = !. OA 0 Solution by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. It follows by continuity that there always exists a chord A1 OA01 such ! that A1 O = A01 O and hence ,, 1 + ,,! = , . Similarly, there exists OA OA01 ! 0 a chord A2 A03 which is bisected by the midpoint O1 of OA01 . It follows by ! ! ! the parallelogram law that ,, 2 + ,,! = ,,! and hence ,, 1 + ,, 2 + OA OA03 OA0 OA OA ,,! = , . Again similarly there exists a chord1 A3A0 which is bisected by OA03 ! 0 ! ! ! 4 the midpoint of OA03 so that ,, 1 + ,, 2 + ,, 3 + ,,! = , , and so on OA OA OA OA04 ! 0 for any number of vectors n 1. 3. 10 A sequence of natural numbers ak , k 1, such that for each k, ak ak+1 ak + 1993 is given. Let all prime divisors of ak be written for every k. Prove that we receive an in nite number of di erent prime numbers. Solution by Pavlos Maragoudakis, Pireas, Greece. We suppose that there is a sequence of natural numbers such that ak ak+1 ak + 1993, k 1, and the set of all prime divisors of all ak is nite. Let p1a;1p2; : : a ; pr list all the prime divisors of all ak . Now every : ak has the form p1 : : : pr r , ai = 0; 1; 2; : : : , i = 1; : : : ; r. Let S = fpa1 pa2 : : : par j ai = 0; 1; 2; : : : ; i = 1; 2; : : : ; rg: 1 2 r De ne xn with x1 x2 : : : such that S = fxn = n 2 N g. We have that ak ak+1 and ak 2 S , k 1. Thus ak is a subsequence of
392
xk . But ak ak,1 + 1993 ak,2 + 2 1993 a1 + k , 11993 ka1 + 1993; k 1. Hence ak ka1 + 1993, k 1. Therefore 1 1 1 X 1 X X 1 1 1 = a + 1993 n = +1 1 n=1 an n=1 na1 + 1993 n=1
and
1 X
n=1 n
1 1 X 1 = a x
=
which is a contradiction.
1 2 4 = p p, 1 p p, 1 p p, 1 1 2 4
r n=1 n a1 ;:::;ar 0 p1 2 ! 1 1 1 1 ! X 1 ! 1 X X a2 : : : 1 ar a1 =0 p1 a2 =0 p2 r =0 pr
X
1 pa2 : : : pa4
1
+1 ;
We now turn our attention to the solutions by readers to problems of the Mock Test of the Hong Kong Committee for the IMO 1994 1997: 322 323 .
INTERNATIONAL MATHEMATICAL OLYMPIAD 1994
Hong Kong Committee  Mock Test, Part I
Time: 4.5 hours
1. In a triangle 4ABC , C = 2 B. P is a point in the interior of 4ABC satisfying that AP = AC and PB = PC . Show that AP trisects the angle A.
Solutions by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; and by D.J. Smeenk, Zaltbommel, the Netherlands. We give the solution by Amengual Covas. Let PAC and BAP be 2 and respectively. Then, since C = 2 B, we deduce from A + B + C = 180 that 2 + + 3B = 180 : 1 A
P B C
393 The angles at the base of the isosceles triangle PAC are each 90 , . Also 4BPC is isosceles, having base angles
C , 90 , = 2B + , 90 ;
and so
BPA = 180 , PBA + BAP = 180 , B , 2B + , 90 + 180 , 2 , 3B = 4B + 3 , 90 : As usual, let a, b and c denote the lengths of the sides BC , AC and AB . By the Law of Cosines, applied to 4BPA, where PA = b and PB = PC = 2b sin , c2 = b2 + 2b sin 2 , 2 b 2b sin cos4B + 3 , 90 ; c2 = b2 1 + 4 sin2 , 4 sin sin4B + 3 : 2 We now use the fact that C = 2 B is equivalent to the condition c2 = bb + a, which has appeared before in CRUX 1976: 74 , 1984: 278 and 1996: 265 267 . Since a = 2 PC cos2B + , 90 = 4b sin sin2B + , we have c2 = b2 1 + 4 sin sin2B + : 3 b2 1 + 4 sin2 , 4 sin sin4B + 3 = b2 1 + 4 sin sin2B + ;
which simpli es to Therefore, from 2 and 3, we get so that
sin , sin4B + 3 = sin2B + : Since sin , sin4B + 3 = ,2 cos2B + 2 sin2B + , this equation
may be rewritten as
sin2B + 1 + 2 cos2B + 2 = 0:
Since, from 1, 2B + 180 , we must have 1+2 cos2B +2 = 0, giving cos2B + 2 = ,1=2 ; that is, 2B + 2 = 120 4 since, again from 1, 2B + 2 180 . Finally, we may eliminate B between 1 and 4 to obtain = . The result follows.
394
Mock Test, Part II
Time: 4.5 hours
1. Suppose that yz + zx + xy = 1 and x, y, and z 0. Prove that
x1 , y 1 , z + y1 , z 1 , x + z1 , x 1 , y 4 9 3 :
2 2 2 2 2 2
p
Solutions by Murray S. Klamkin, University of Alberta, Edmonton, Alberta; and Pavlos Maragoudakis, Pireas, Greece. We give the solution by Klamkin. We rst convert the inequality to the following equivalent homogeneous one:
xT2 , y2T2 , z2 + yT2 , z2T2 , x2 + zT2 , x2T2 , y2 4p3=9T25=2 where T2 = yz + zx + xy , and for subsequent use T1 = x + y + z , T3 = xyz .
Expanding out, we get
T1 T22 , T2
or
X
p xy2 + z2 + T2T3 4 3=9T25=2 ;
p T1 T22 , T2T1T2 , 3T3 + T2T3 = 4T2T3 4 3=9T25=2 :
3T
3
Squaring, we get one of the known Maclaurin inequalities for symmetric functions: p p
2 T2=3 : There is equality if and only if x = y = z .
To nish this number of the Corner we give two solutions to problems of the 45th Mathematical Olympiad in Poland, Final Round 1997: 323 324 . 1. Determine all triples of positive rational numbers x; y; z such that x + y + z, x,1 + y,1 + z,1 and xyz are integers. Solution by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. 1 1 1 Let x + y + z = n1, x + y + z = n2, and xyz = n3 , where n1, n2, n3 are integers. Then yz + zx + xy = n2n3 and x, y , z are roots of the cubic
t3 , n1t2 + n2n3t , n3 = 0 :
395 As known, the only rational roots of the latter are factors of n3, and consequently x, y , z are integers. The only triples of integers x; y;z , aside from permutations, which 1 1 satisfy x + y + 1 = n2 are z
5. Let A1; A2; : : : ; A8 be the vertices of a parallelepiped and let O be its centre. Show that
4OA2 + OA2 + + OA2 OA1 + OA2 + + OA8 2 : 1 2 8
Solution by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. Let one of the vertices be the origin and let the vectors B + C , C + A, A + B denote the three coterminal edges emanating from this origin. Then the vectors to the remaining four vertices are S + A, S + B , S + C , and 2S where S = A + B + C and which is also the vector to the centre. The inequality now becomes
1; 1; 1; 1; 2; 2; 2; 3; 6; 2; 4; 4; and 3; 3; 3:
2S2 + A2 + B 2 + C 2 jS j + jAj + jB j + jC j2 ;
or
S2 + A2 + B2 + C 2 2jSjfjAj + jBj + jC jg +2fjBj jC j + jC j jAj + jAj jBjg:
Since the inequality now becomes Clearly, and
S2 = A2 + B2 + C 2 + 2B C + 2C A + 2A B;
S2 ,B C ,C A,AB jSjfjAj+jBj+jC jg+fjBj jC j+jC j jAj+jAj jBjg: S2 jSjfjAj + jBj + jC jg
There is equality if and only if the parellelepiped is degenerate, for example, B = C = O. That completes this number of the Olympiad Corner. Send me your nice solutions and Olympiad contests.
,B C , C A , A B jBj jC j jAj + jAj jBj :
396
BOOK REVIEWS
Dissections : Plane & Fancy by Greg N. Frederickson, published by Cambridge University Press, 1997, ISBN 0521571979, hardcover, 310+ pages, $34.95. Reviewed by Andy Liu, University of Alberta. This is a much awaited sequel to Harry Lindgren's 1964 classic work, Geometric Dissections, which the author G.N.F. revised and augmented in the 1972 Dover edition. Actually, the current volume is much more than just a sequel. It is the most comprehensive treatise on the subject of geometric dissections. It may be enjoyed on at least three levels. First and foremost, this book is a collection of interesting dissection puzzles, old and new. Only some background in high school geometry is needed to fully enjoy these problems. Can you cut an octagon into ve pieces and rearrange them into a square? How about turning a star into a pentagon? The solutions, which are both appealing but for somewhat opposing reasons, are shown below.
Edited by ANDY LIU
Octagon to square
Star to pentagon
This book is also an instructive manual on the art and science of geometric dissections. While one may admire the ingenuity which produced the spectacular solutions, the author probes into the underlying fabric which might have led to such incisive insight. Many techniques are discussed, too many to enumerate here. A favourite is that of tessellation. Below are two tilings which might have suggested the dissections above.
397
Tilings for octagon, square
Cut the star for tiling
Tilings for star, pentagon
Finally, this book is an important historical document, detailing the intercultural development of the subject. Travel from the palace school of tenthcentury Baghdad to the mathematical puzzle columns in turnofthecentury newspapers, from the 1900 Paris Congress of Mathematicians to the night sky of Canberra. Readers puzzled by this quote need look no further than the illustrious names of Ab 'l Waf , Henry Dudeney Sam Loyd, David u a Hilbert Max Dehn and Harry Lindgren. Biographical sketches of Waf , Dua deney, Loyd and Lindgren are provided, along with those of over forty other people who have made signi cant contributions to geometric dissections. The writing style is very engaging, and the book is good reading even if one skips over some of the more complicated technical details. In conclusion, the reviewer echoes Martin Gardner that this book will be a classic. It comes with the highest recommendation.
398
THE SKOLIAD CORNER
No. 33 R.E. Woodrow
As a contest this issue we give the Senior High School Mathematics Contest, Preliminary Round 1998 of the British Columbia Colleges. My thanks go to the organizer, Jim Totten, The University College of the Cariboo, for forwarding me the contest materials. Time allowed is 45 minutes!
BRITISH COLUMBIA COLLEGES Senior High School Mathematics Contest Preliminary Round 1998
Time: 45 minutes
integers. It follows that c equals: a 2 b 5 c 6 d 7 e 8 9 2. The value of the sum log 1 + log 2 + log 3 + + log 10 is: 2 3 4 a ,1 b 0 c 1 d 2 e 3 3. Four basketballs are placed on the gym oor in the form of a square with each basketball touching two others. A fth basketball is placed on top of the other four so that it touches all four of the other balls, as shown. If the diameter of a basketball is 25 cm, the height, in centimetres, of the centre of the fth basketball above the gym oor is:
1. The integer 1998 = n , 1nn10n + c where n and c are positive
a 25 2 b 25 2 c 20 d 25 1 + 2 e 251 + 2 2 2 4. Last summer, I planted two trees in my yard. The rst tree came in a fairly small pot, and the hole that I dug to plant it in lled one wheelbarrow load of dirt. The second tree came in a pot, the same shape as that of the rst tree, that was oneandathird times as deep as the rst pot and oneandahalf times as big around. Let us make the following assumptions: i The hole for the second tree was the same shape as for the rst tree.
p
p
p
p
399 ii The ratios of the dimensions of the second hole to those of the rst hole are the same as the ratios of the dimensions of the pots. Based on these assumptions, the number of wheelbarrows of dirt that I lled when I dug the hole for the second tree was: a 2 b 2:5 c 3 d 3:5 e none of these 5. You have an unlimited supply of 5gram and 8gram weights that may be used in a pan balance. If you use only these weights and place them only in one pan, the largest number of grams that you cannot weigh is: a 22 b 27 c 36 d 41 e there is no largest number of grams 6. If all the whole numbers from 1 to 1;000;000 are printed, the number of times that the digit 5 appears is: a 100;000 b 500;000 c 600;000 d 1;000;000 e 2;000;000 7. The perimeter of a rectangle is x centimetres. If the ratio of two adjacent sides is a : b, with a b, then the length of the shorter side, in centimetres, is: 2bx ax bx a abxb b x , b c a+b d 2a+b e 2a+b + 2 8. The sum of the positive solutions to the equation xxpx = xpxx is: 1 1 1 a 1 b 1 1 c 2 4 d 2 2 e 3 4 2 9. Two circles, each with a radius of one unit, touch as shown. AB and CD are tangent to each circle. The area, in square units, of the shaded region is:
A
B
C
a
D
b c 2 , d 4 , e none of these 4 2 10. A parabola with a vertical axis of symmetry has its vertex at 0; 8 and an x intercept of 2. If the parabola goes through 1; a, then a is: a 5 b 5:5 c 6 d 6:5 e 7 11. A ve litre container is lled with pure orange juice. Two litres of juice are removed and the container is lled up with pure water and mixed
400 thoroughly. Then two litres of the mixture are removed and again the container is lled up with pure water. The percentage of the nal mixture that is orange juice is: a 27 b 25 c 30 d 36 e 24 12. The lengths of the sides of a triangle are b + 1, 7 , b and 4b , 2. The number of values of b for which the triangle is isosceles is: a 0 b 1 c 2 d 3 e none of these 13. The number of times in one day when the hands of a clock form a right angle is: a 46 b 22 c 24 d 44 e 48 14. In my town some of the animals are really strange. Ten percent of the dogs think they are cats and ten percent of the cats think they are dogs. All the other animals are perfectly normal. One day, I tested all the cats and dogs in the town and found that 20 of them thought that they were cats. The percentage of the dogs and cats in the town that really are cats is: a 12:5 b 18 c 20 d 22 e 22:5 15. A short hallway in a junior high school contains a bank of lockers numbered one to ten. On the last day of school the lockers are emptied and the doors are left open. The next day, a malicious math student walks down the hallway and closes the door of every locker that has an even number. The following day, the same student again walks down the hallway and, for every locker whose number is a multiple of three, closes the door if it is open and opens it if it is closed. On the next day, the student does the same thing with every locker whose number is divisible by four. If the student continues this procedure for a total of nine days, the number of lockers that are closed after the ninth day is: a 4 b 5 c 6 d 7 e 8 Last issue we gave the Junior High School Mathematics Contest, Preliminary Round 1998 of the British Columbia Colleges. Here are the o cial solutions", which come our way from the organizer, Jim Totten, The University College of the Cariboo.
BRITISH COLUMBIA COLLEGES Junior High School Mathematics Contest Preliminary Round 1998
Time: 45 minutes
1. A number is prime if it is greater than one and divisible only by one and itself. The sum of the prime divisors of 1998 is: c
401 Solution. We can factor the number 1998 as follows: 1998 = 2 999 = 2 32 111 = 2 33 37. Hence, the sum of its prime divisors is 2 + 3 + 37 = 42. 2. Successive discounts of 10 and 20 are equivalent to a single
discount of: c Solution. If P denotes the initial price then the new price after deducting the two consecutive discounts is P 1 , 0:11 , 0:2 = 0:72P . This gives the total discount of 1 , 0:72 100 = 28. 3. Suppose that A = A2 and A 2 B = A , 2B. Then the value of 7 2 3 is: e Solution. According to our de nitions, 7 2 3 = 72 2 32 = 49 2 9 = 49 , 2 9 = 31. 4. The expression that is not equal to the value of the four other expressions listed is: d Solution. The values of the ve expressions are: p p a 1 9 + 9 , 8 = 2, p d 1 , 9 9 , 8 = ,2, b 1 + 9 p 9 + 8 = 2, , e 19 , 9 , 8 = 2. c ,1 9 + 9 + 8 = 2, Thus,the expression d has a di erent value than the other four expressions. 5. The sum of all of the digits of the number 1075 , 75 is: c Solution. Consider the procedure for subtracting 75 from 1075 by hand":
k k
k
k k
100 : : : 000 ,75 99 : : : 925 Hence, the decimal digits of 1075 , 75 are: 5, 2 and seventythree copies of 9. The sum of the digits is 5 + 2 + 73 9 = 664.
6. A circle is divided into three equal parts and one part is shaded as in the accompanying diagram. The ratio of the perimeter of the shaded region, including the two radii, to the circumference of the circle is: d
402 Solution. The ratio is given by
2
r + r + r
3
2r
+ 3 = 2r = 2rr 6r = 3+ 3 : 6
2
r+6r
7. The value of
is: b Solution. We can simplify this compound fraction by working successively from the bottom to the top of the expression:
2 , 2, 1, 1 1 2
2
1
2 , 2, 2, 1 1
1
1
2
1 = 2 , 1 1 = 2 ,1 1 = 2 ,1 1 = 2 , 3 = 1 = 4 : 5 5 4 4 2, 1 2, 2 4 3 3 3 2
8. If each small square in the accompanying grid is one square centimetre, then the area in square centimetres of the polygon ABCDE is: a Solution. We can nd the area by decomposing the polygon ABCDE into simpler gures, for example, into three triangles: ABE , BCE , and CDE.
C B D
A
E
If we choose AE , CE , and CD as bases of the triangles then the lengths of the corresponding perpendicular heights are 5, 3, and 8 cm. Hence, the area of the polygon is 1 4 5 + 1 8 3 + 1 4 8 = 38. 2 2 2 9. A point P is inside a square ABCD whose side length is 16. P is equidistant from two adjacent vertices, A and B , and the side CD opposite these vertices. The distance PA equals: e
403
x denotes the distance PA. D
Solution. The situation is illustrated by the following diagram, where
C x P x x E B
16
A
The Pythagorean Theorem applied to triangle PEB2gives 16,x2+82 = x2 , +8 so that 162 , 32x + x2 + 82 = x2 , and x = 16232 = 10. 10. A group of 20 students has an average mass of 86 kg per person. It is known that 9 people from this group have an average mass of 75 kg per person. The average mass in kilograms per person of the remaining 11 people is: b Solution. If m1; m2; : : : ; m denote the masses of the students then m1 +m2 ++m20 = 86. Hence, m120 m2 + + m20 = 20 86 = 1720. We + 20 can assume, without loss of generality, that the average mass of the rst nine + students is 75; that is, m1 +m29 +m9 = 75. Hence, m1 + m2 + + m9 = 9 75 = 675. The total mass of the remaining 11 people is 1720 , 675 = 1045. This gives the average of 1045 = 95. 11 11. In the following display each letter represents a digit:
3 B C D E 8 G H I
The sum of any three successive digits is 18. The value of H is: a Solution. We have 3 + B + C = 18. Consequently, B + C = 15. By subtracting this equation from B + C + D = 18 we get D = 3. Now, D +E +8 = 18 gives E = 10,D = 10,3 = 7. Finally, E +8+G = 18 gives G = 10 , E = 10 , 7 = 3, and 8 + G + H = 18 gives H = 18 , 8 , 3 = 7. 12. In the accompanying diagram ADE = 140 . The sides are congruent as indicated. The measure of EAD is: e
A
B C
D
E
404 Solution. If EAD = , then also ACB = , since triangle ABC is isosceles. Hence, CBD = 2 as an external angle of triangle ABC . Consequently, ADC = 2 , since triangle BCD is isosceles. Further, ECD = ADC + CAD as an external angle of triangle ADC . Hence, ECD = 2 + = 3 . Now, AED = ECD = 3 , because triangle CDE is isosceles. This implies that CDE = 180 , 6 . Finally, ADE = ADC + CDE = 2 + 180 , 6 = 180 , 4 . Thus, 180 , 4 = 140 . This yields = 10 . 13. The area in square units of the triangle bounded by the x axis and the lines with equations y = 2x + 4 and y = , 2 x + 4 is: e 3 Solution. Two vertices of the triangle lie on the x axis, so they are the x intercepts of the lines. The x intercept of the rst line, determined by the equation 2x + 4 = 0, is ,2. Similarly, the second x intercept, determined by , 2 x2 + 4 = 0, is 6. Consequently, the length of the base of 3 the triangle is 6 , ,2 = 8. Since both lines have the same y intercept 4, they intersect each other and the y axis at level 4 and, consequently, the length of the height of the triangle is 4. Therefore, the area of the triangle is A = 1 4 8 = 16. 2 14. Two diagonals of a regular octagon are shown in the accompanying diagram. The total number of diagonals possible in a regular octagon is: d
Solution. Let di , for i = 1; 2; : : : ; 8, denote the number of diagonals connected to the ith vertex. Then d1 = d2 = : : : = d8 = 5, since each vertex of the octagon is connected to ve diagonals. On the other hand, each diagonal joins two vertices. Therefore, in the sum d1 + d2 + + d8 = 8 5 = 40, each diagonal is counted twice. Hence, the number of diagonals in the octagon is 20. 15. A local baseball league is running a contest to raise money to send a team to the provincial championship. To win the contest it is necessary to determine the number of baseballs stacked in the form of a rectangular pyramid. The fth and sixth levels from the base of the stack of baseballs are shown. If the stack contains a total of seven levels, the number of baseballs in the stack is: d
405
Solution. The fth level has 3 4 = 12 balls, the sixth 2 3 = 6 balls, and the seventh 1 2 = 2 balls. We notice that the number of balls in both sides of the rectangle they form increases by one each time we move one level down. Thus, the total number of balls is 1 2 + 2 3 + 3 4 + 4 5 + 5 6 + 6 7 + 7 8 = 168. That completes the Skoliad Corner for this issue. We need suitable contests and solutions. I welcome any comments, criticisms, or suggestions for the future direction of this feature.
Advance Announcement
The 1999 Summer Meeting of the Canadian Mathematical Society will take place at Memorial University in St. John's, Newfoundland, from Saturday, 29 May 1999 to Tuesday, 1 June 1999. The Special Session on Mathematics Education will feature the topic
What Mathematics Competitions do for Mathematics.
The invited speakers are Ed Barbeau University of Toronto, Ron Dunkley University of Waterloo, Tony Gardiner University of Birmingham, UK, Rita Janes Newfoundland and Labrador Senior Mathematics League, and Shannon Sullivan student, Memorial University. Requests for further information, or to speak in this session, as well as suggestions for further speakers, should be sent to the session organizers: Bruce Shawyer and Ed Williams CMS Summer 1999 Meeting, Education Session Department of Mathematics and Statistics, Memorial University St. John's, Newfoundland, Canada A1C 5S7
406
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Shreds and Slices
An Algebraic Relation with a Geometric Twist
Cyrus Hsia Consider the following algebraic relationship between the positive real numbers a, b, and c: If we consider line segments with lengths of a, b, and c, then they are related to each other as shown in the following diagram.
1 = 1 +1: c a b
a
b
c
The diagram shows the three line segments parallel to each other and emanating from a common line. This gure and relationship between the line segments appear a lot. The reader is encouraged to prove this.
407 As a corollary of this fact, here is another geometric example. Let P , A, B, and C be points in the plane such that APC = CPB = 60 and A, C , and B are collinear. Show that 1=PC = 1=PA + 1=PB.
P
60 60
A
C
B
Another algebraic relation between three positive numbers with an interesting geometric interpretation is the following: The reader is encouraged to nd a geometric interpretation for the above relation before looking at the diagram below. Use a, b, and c as the length of three line segments, and determine a geometric gure that relates the three.
1 1 1 pc = pa + pb :
q
a
q qc
b
It turns out that if a, b, and c are considered to be the lengths of the radii of three circles, then the circles may all be tangent to a common line and to each other as shown. Again readers are encouraged to prove this themselves. Now what about a generalization? Consider the following algebraic relation between positive reals a, b, c, and a real number x:
cx = ax + bx: The rst case then corresponds to the value x = ,1 and the second case to x = ,1. 2
408 The case x = 1 is trivial, as we could interpret it geometrically as a line segment of length c is made up of the sum of its parts of lengths a and b.
b a c
If we wanted to get fancy, we could give the following geometric example instead. Consider an equilateral triangle ABC inscribed in a circle, as shown. P is a point on arc BC . Prove that PA = PB + PC .
A
B P
C
The reader is probably already familiar with the famous case x = 2 known as the Pythagorean Theorem: A triangle with sides a, b, and c is a rightangled triangle if and only if the lengths satisfy a2 + b2 = c2 .
b
c a
Of course, no discussion about algebraic relations in the above form is complete without mentioning the notorious Fermat's Last Theorem and the recent announcement that it has nally been laid to rest. If x is an integer with x 2, then the claim is that no solution in the natural numbers exists for a, b, and c. However, in our general case, the values are real, so we are not limited by the above result to nding wild and wacky geometric or other interpretations for it.
409 If the reader is curious, as are we, then try the following exercises to nd geometric interpretations for special cases of the above relation. The exercises are explorational and may not have nice solutions, if any. Readers are welcome to submit any interesting results they nd. Exercises 1. Let a, b, and c be the lengths of three line segments. Determine how these three line segments are related geometrically if they satisfy the relation a c3 = a3 + b3, p b pc = pa + b, c c12 = a12 + b12 . 2. It is clear that algebraically, all the relations are similar. However, the geometric interpretations do not appear to be related. Is there a general geometric description where each of the above geometric gures is a special case? 3. The algebraic relation clearly does not work for the case x = 0. Is there a way to de ne the relation so that it would be consistent with everything else mentioned so far?
We extend congratulations to Ravi Vakil and Alice Staveley, who were married at St. John's, NF on Monday, 12 October 1998.
410
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only problems  the next issue will feature only solutions. We warmly welcome proposals for problems and solutions. We request that solutions from this issue be submitted by 1 September 1999, for publication in issue 8 of 1999.
High School Problems
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca
H245. Determine how many distinct integers there are in the set
12 ; 22 ; 32 ; : : : ; 19982 : 1998 1998 1998 1998
H246. Let Sn denote the sum of the rst n positive integers. We say that an integer n is fantastic if both n and S n are perfect squares. For example, 49 is fantastic, because 49 = 72 and S 49 = 1+2+3+ +49 = 1225 = 352 are both perfect squares. Find another integer n 49 that is fantastic. H247. Say that the integers a, b, c, d, p, and r form a cyclic set a; b; c; d; p; r if there exists a cyclic quadrilateral with circumradius r, sides a, b, c, and d, and diagonals p and 2r. a Show that if r 25, no cyclic set exists. b Find a cyclic set a; b;c; d; p; r for r = 25. H248. Consider a tetrahedral die that has the four integers 1, 2, 3, and 4 written on its faces. Roll the die 2000 times. For each i, 1 i 4, let f i represent the number of times that i turned up. So, f 1 + f 2 + f 3 + f 4 = 2000. Also, let S denote the total sum of the 2000 rolls. If S 4 = 6144 f 1f 2f 3f 4, determine the values of f 1, f 2, f 3, and f 4.
411
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu A221. Construct, using straightedge and compass only, the common tangents of two nonintersecting circles. A222. Does there exist a set of n consecutive positive integers such that for every positive integer k n, it is possible to pick k of these numbers whose mean is still in the set? A223. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. Suppose p is a prime with p 3 mod 4. Show that for any set of p , 1 consecutive integers, the set cannot be divided into two subsets so that the product of the members of the one set is equal to the product of the members of the other set. Generalization of Question 4, IMO 1970 A224. Proposed by Waldemar Pompe, student, University of Warsaw, Poland. Let P be an interior point of triangle ABC such that PBA = PCA = ABC + ACB=3. Prove that
Advanced Problems
AC = AB : AB + PC AC + PB
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu
Challenge Board Problems
C81. Let fan g be the sequence de ned as follows: a0 = 0, a1 = 1, and an+1 = 4an , an,1 for n = 1, 2, 3, : : : . a Prove that a2 , an,1 an+1 = 1 for all n 1. n
b Evaluate
1 X
C82. Find the smallest multiple of 1998 which appears as a partial sum of the increasing sequence
1; 1; 2; 2; 2; 4; 4; 4; 4; 8; : : : ; in which the number 2k appears k + 2 times for k a nonnegative integer.
k=1
arctan 4a2 : k
1
412
This year's Canadian IMO team began with a week of training at the University of Calgary with lavish meal tickets. Then they were o to beautiful and rocky Kananaskis where we all had an adventurous" time. Once the team stepped o the airconditioned plane and into hot and muggy Taipei, Taiwan, it marked the team's o cial arrival to the 39th International Mathematical Olympiad. The team consisted of the following members: Adrian Oops I dropped my..." Birka, Adrian If You Will" Chan, Jimmy Nuclear Aerial Strike" Chui, Mihaela Baia" Enachescu, Jessie So Cute" Lei, and Adrian Nailing Radar" Tang. Team leader Dr. Christopher Focus" Small was driven to the edge, while deputy leader J.P. It's So Easy" Grossman calmly polished o old competitions one by one. Special thanks to leader observer Arthur Rubik's Cube" Baragar and deputy observer Dorette Dutch" Pronk for their coaching and experience. Also, thanks must go to Dr. Bill Sands of the University of Calgary for organizing such a fun training session. The contest itself seemed to continue the trend of di cult IMO's and low medal cuto s. With 76 countries competing, Canada fared extremely well, bringing back 1 gold, 1 silver, 2 bronze and an honourable mention. The scores were as follows: CAN 1 Adrian Birka 10 CAN 2 Adrian Chan 31 Gold Medal CAN 3 Jimmy Chui 14 Bronze Medal CAN 4 Mihaela Enachescu 30 Silver Medal CAN 5 Jessie Lei 13 Honourable Mention CAN 6 Adrian Tang 15 Bronze Medal In this year's contest, Canada placed 20th out of 76 countries, up from last year's 29th ranking. Best of luck to CAN 1, 4, and 6 as they continue university studies at MIT, Harvard, and Waterloo respectively. CAN 2, 3, and 5 are all eligible for next year's team. Hopefully there won't be as much moaning of Where did I go wrong?" next year around! Special thanks must also go to Dr. Graham Wright of the Canadian Mathematical Society for again taking care of the tab, and Professor Ed Barbeau for his hard work and dedication to the training of potential IMO candidates through his yearlong correspondence program. Although sometimes things didn't make sense, and the IMO ag somehow disappeared, the 39th International Mathematical Olympiad ran smoothly and was de nitely a success. The new experience of a place halfway around the world with a stimulating culture was new to most of us. Best of luck to all IMO hopefuls for the 1999 team, as yet another Canadian IMO journey begins next July in Bucharest, Romania.
Adrian Chan, student, UCC, Toronto
IMO Report
413
Bogus Arguments and Arcane Identities
Princeton University In Euler's time, mathematics was faster and looser than today, and niceties such as limits were blatantly ignored. Here is an argument of Euler's that seems to have no right to work, but does nonetheless. We conclude with some avenues for exploration and an open question. If r1, : : : , rn are the zeros of a polynomial a0 + a1 x + + an xn , and none of the ri are zero or a0 6= 0, then the negative of the linear term over the constant term is the sum of the reciprocals of the roots:
Ravi Vakil
, a1 = r11 + + r1n : a0 What about power series? For example, cos x = 1 , x2=2+ x4 =24 , , so that 2 cos px = 1 , x + x , :
2 24
1 1 = X 1 2 n=0 2n + 1=22
2 1 = : 2 8 n=0 2n + 1
The zeroes of this function are the squares of the odd multiples of =2: 2n + 1=22, n = 0, 1, 2, : : : . One might hope that the principle for polynomials given above still holds:
which can be rewritten as
1 X
1
This is actually true! Another possibility is to use sin x, which has power series expansion
x sin x = x , x + 120 , : 6 Can you use the power series for sin px=x to prove" the identity 1 2 X 1 = ? 2 6 n=1 n
3 5
2
Copyright c 1998 Canadian Mathematical Society
414 Can you relate 2 to 1 by arguing that 2 minus a quarter of 2 is 1? P 1 If you write a short computer program to compute 1000 n2 , and comn=1 pare it to 2 =6, you'll see that they are indeed very close. In fact, they di er by almost exactly 1=1000. Can you explain why this might be? Can you P 1 guesstimate the di erence between 1000 2n+12 and 2 =8? n=1 And nally, can you conjure up other examples of this sort of argument, to prove" other arcane identities? If so, please let us know! Acknowledgements. This note was inspired by the example of cos px given in A . References A S. Abhyankar, Historical ramblings in algebraic geometry and related algebra, Amer. Math. Monthly 83 1976, no. 6, 409 448. E L. Euler, Introductio in Analysin In nitorum, Berlin Academy, 1748.
415
The Fibonacci Sequence
student, University of Waterloo The sequence de ned by F0 = 0, F1 = 1, and Fn = Fn,1 + Fn,2 for n 2 is called the Fibonacci sequence. Named after Leonardo of Pisa, who is also known as Fibonacci unsurprisingly, it is one of the most widely studied sequences of all time. The Fibonacci sequence is an excellent topic with which to begin learning some basic number theory and various techniques for working with recurrence relations.
Theorem 1. For all n 1,
Wai Ling Yee
Basic Results
2 Fn , Fn,1Fn+1 = ,1n,1 :
Proof by Induction. When n = 1, F12 , F0 F2 = 12 , 0 1 = 1 = ,1 , , so the formula holds for n = 1. Assume that the formula holds for some n = k, k 1. For n = k + 1,
1 1
Fk2+1 , FkFk+2 = Fk2+1 , FkFk+1 + Fk = Fk+1 , Fk Fk+1 , Fk2 = Fk,1Fk+1 , Fk2 = ,,1k,1 by the induction hypothesis = ,1k; so the formula holds for n = k + 1. Therefore, by mathematical induction, the formula holds for all n 1. Theorem 2. For all n 1, F1 + F3 + + F2n,1 = F2n :
Proof. Using the recurrence relation n times, we have
F2n = F2n,1 + F2n,2 = F2n,1 + F2n,3 + F2n,4 = = F2n,1 + F2n,3 + + F3 + F1 + F0 = F2n,1 + F2n,3 + + F3 + F1 :
Copyright c 1998 Canadian Mathematical Society
416
Exercise 1. Prove
n X j =1
Fj2 = FnFn+1 :
Divisibility Theorem 3. For all m, n 1, Fn+m = FnFm+1 + Fn,1Fm : Proof. We will prove this by induction on m. When m = 1, Fn+1 = Fn 1 + Fn,1 1 = FnF2 + Fn,1F1; so the formula holds for all n when m = 1. Assume that the formula holds for all n when m = M . For m = M + 1, Fn+M +1 = Fn+M + Fn,1+M = Fn FM +1 + Fn,1FM + Fn,1FM +1 + Fn,2FM
= Fn FM +1 + Fn,1 + Fn,2 FM + Fn,1 FM +1 = Fn FM + FM +1 + Fn,1 FM +1 = Fn FM +2 + Fn,1FM +1; so the formula holds for all n for m = M + 1. By mathematical induction, the formula holds for all m, n 1. Corollary 4. For all m, n 1, FnjFnm . Proof. We will prove this by induction on m. When m = 1, Fn certainly divides itself for every positive integer n. Suppose the statement holds for all n when m = M . For m = M + 1, FnM +1 = FnM +n = FnM Fn+1 + FnM ,1Fn
by the induction hypothesis
by Theorem 3. Since FnM is divisible by Fn by the induction hypothesis, FnM Fn+1 + FnM ,1Fn is also divisible by Fn. This is equivalent to FnjFnM +1, so the result holds for m = M + 1. By mathematical induction, the formula holds for all m, n 1. Exercise 2. Prove that for every positive integer n, there exist n consecutive, composite Fibonacci numbers. We will now de ne a few terms in the interests of formality. For integers a and b, we say that a divides b if there exists an integer q such that b = aq, and a is called a divisor of b. Given two nonzero integers a and b, the largest number which divides both of them, denoted gcda; b, is called
Number Theory 101
417 their greatest common divisor. If gcda; b = 1, then a and b are said to be relatively prime. Theorem 5. The Division Algorithm Given a positive integer a and an integer b, there exist unique integers q and r such that b = aq + r and 0 r a. Then q is called the quotient, and r is called the remainder upon division of b by a. Proof. Consider the set
S = fs : s = b , aq 0; q 2 Zg: S cannot be empty. If b 0, then select q = 0 to give b 2 S. Otherwise, if b 0, select q = b so that b , ab = b1 , a 0, which means that b , ab 2 S. Since S is nonempty and contains only nonnegative integers, we can nd the smallest element in S . Call it r. Suppose r a. Then 0 r , a = b , aq , a = b , aq + 1 for some q, so r , a 2 S and it is smaller than r, contradiction. Thus 0 r a. Suppose that we can nd 0 r1 r2 a in S and corresponding q1 and q2. Then b = aq1 + r1 = aq2 + r2, which implies that aq1 , q2 = r2 , r1. Thus a divides r2 , r1 . However, we also know that 0 r2 , r1 a; that is, r2 , r1 lies between two consecutive multiples of a and thus cannot be divisible by a, contradiction. We have shown the existence and uniqueness of q and r. Exercise 3. Prove that gcda; b = gcda; b , aq for any nonzero integers a and b and any integer q . Exercise 4. Prove that if a and q are relatively prime, then gcda; qb = gcda; b, where a; b; q are nonzero integers. Exercise 5. Prove that Fn and Fn+1 are relatively prime. Number Theory 102
The Euclidean Algorithm. The Euclidean Algorithm is an algorithm used to determine the greatest common divisor of two numbers. Suppose we have two distinct positive integers a and b where, without loss of generality, a b. By the Division Algorithm, b = aq1 + r1 where 0 r1 a for unique integers q1 and r1. If r1 = 0, then a divides b so our greatest common divisor is a. Otherwise, by Exercise 3, gcda; b = gcda; b , aq1 = gcda; r1 = gcdr1; a. In this case, we then repeat the same argument using r1 and a where we used a and b before, respectively. We have a = r1q2 + r2 where 0 r2 r1 for unique integers q2 and r2 by the Division Algorithm, and gcdr1; a = gcdr1; a , q2r1 = gcdr2; r1. Continue applying this argument. Since the ri s are strictly decreasing and nonnegative, there must be a last remainder, say rn, that is bigger than 0. So we have
418
b = aq1 + r1 ; 0 r1 a; gcda; b a = r1 q2 + r2 ; 0 r2 r1 ; gcdr1 ; a r1 = r2 q3 + r3 ; 0 r3 r2 ; gcdr2 ; r1 rn,2 = rn,1 qn + rn ; 0 rn rn,1 ; gcdrn,1 ; rn,2 rn,1 = rn qn+1 ; gcdrn ; rn,1 We have found that gcda; b = rn . Theorem 6. For all a, b 1, gcdFa; Fb = Fgcda;b:
= gcdr1 ; a; = gcdr2 ; r1 ; = gcdr3 ; r2 ; = gcdrn ; rn,1 ; = rn :
Proof. If a and b are equal, the result is immediate, so assume that a b. Apply the Euclidean Algorithm to obtain b = aq1 + r1; 0 r1 a; a = r1q2 + r2; 0 r2 r1; r1 = r2q3 + r3; 0 r3 r2; rn,2 = rn,1qn + rn; 0 rn rn,1; rn,1 = rnqn+1:
by Theorem 3. Since Faq1 Fr1+1 is a multiple of Fa by Corollary 4,
gcdFa; Fb = gcdFa; Faq1+r1 = gcdFa ; Faq1,1Fr1 + Faq1 Fr1+1
We have
gcdFa ; Fb = gcdFa; Faq1,1 Fr1 + Faq1 Fr1+1 = gcdFa; Faq1,1 Fr1 by Exercise 3. By Exercise 5, gcdFaq1 ; Faq1,1 = 1. Since Fa divides Faq1 , gcdFa; Faq1 ,1 = 1 also. By Exercise 4, since Fa and Faq1,1 are relatively prime, gcdFa; Faq1 ,1 Fr1 = gcdFa; Fr1 : We conclude that gcdFa; Fb = gcdFr1 ; Fa. Repeating this argument, we
obtain
gcdFr1 ; Fa = gcdFr2 ; Fr1 = = gcdFrn ; Frn,1 : Since rn divides rn,1 , Frn divides Frn,1 , which implies that gcdFrn ; Frn,1 = Frn . Thus, gcdFa; Fb = Frn = Fgcda;b : Finding Fn Explicitly
The monic quadratic with roots and is
x , x , = x2 , + x +
:
419 Let and be the roots of x2 , x , 1 in particular. Then, comparing coefcients, + = 1 and = ,1. Using this, we can rewrite the recurrence relation Fn = Fn,1 + Fn,2 as Fn = + Fn,1 , Fn,2 . From this equation, we obtain
Fn , Fn,1 = Fn,1 , Fn,2 : Let sn,1 = Fn , Fn,1 for all n 2. Rewriting the above equation in terms of the si, we obtain sn,1 = sn,2 . In other words, the sequence fsn g is a geometric sequence with common ration . We conclude that sn = n,1 s1 . Similarly, Fn , Fn,1 = Fn,1 , Fn,2 , and if we let tn,1 = Fn , Fn,1 for n 2, then tn = n,1t1 for n 1. Hence, Fn = , Fn + , Fn,1 , , Fn,1 , Fn , Fn,1 , Fn , Fn,1 = , tn,1 , sn,1 = , n,1t1 , n,1 s1 = , n,1F2 , F1 , n,1 F2 , F1 = , n,11 , , n,1 1 , = : ,
Recall that + = 1, so the equation above is, in fact,
n, n
Solving x2 , x , 1 = 0 for the values of and , we have, without loss of generality, p p 1 + 5 and = 1 , 5 : = 2 2 Substituting these values, we conclude that
,
:
Fn =
to p :
1+ 5 2
p n
This is called Binet's Formula for the Fibonacci sequence. p Exercise 5. Let = 1 + 5=2. Prove that Fn is the integer closest n
, p5
1
,p5 n
2
:
5
420
Problems
1. Prove that the product of every four consecutive Fibonacci numbers is the area of a Pythagorean triangle. 2. Prove that every positive integer can be written as a sum of distinct Fibonacci numbers. 3. Prove n+1
k,1 : 4. Prove that if Fn is prime and n 5, then n is prime. 5. Prove that Fn + 1 is always composite for n 4. 6. Show that for any positive integer n, among the rst n2 Fibonacci numbers, there exists at least one that is divisible by n.
k=1
Fn =
b X c n , k
2
7. De ne a Fibonacci prime to be a Fibonacci number that is prime. Prove or disprove: There are in nitely many Fibonacci primes. Note that this is an open problem.
421
J.I.R. McKnight Problems Contest 1984
1. Find the real roots of the equation:
q
x + 3 , 4px , 1 + x + 8 , 6px , 1 = 1 :
q
Note: All square roots are to be taken as positive. 2. Consider a reservoir in the shape of an inverted cone as shown in the diagram below. Water runs into the reservoir at the constant rate of 2 m3 per minute. How fast is the water level rising when it is 6 metres deep?
r r
5m

6
6
10 m
6m
? ?
3. Three forces of magnitudes 10 N, 15 N, and 10 N act at angles of 30 , 70 , and 120 respectively, to the real axis Ox. Using the complex numbers and the imaginary axis Oy nd the magnitude and direction of the resultant force. , 4. Normals are drawn from the point 15 ; , 3 to the parabola whose 4 4 equation is y 2 = 4x. Find the coordinates of the points where the normals meet the parabola. 5. The horizontal base of a triangular pyramid is an equilateral triangle QRS, each of whose sides is 20 cm long. The sloping edges of the pyramid PQRS are respectively 20 cm, 20 cm, and 12 cm long. a Calculate the perpendicular height of the pyramid to the nearest millimetre. b Calculate the angle of inclination of each of the three edges with the base to the nearest tenth of a degree. 6. Prove that if tan A = tan3 B and tan 2B = 2 tan C , then A + B , C = n for some n 2 Z.
422
Swedish Mathematics Olympiad
1988 Qualifying Round
1. Show that the function
q
2. Find the rational root of the equation
f x = x , 4px , 1 + 3 + x , 6px , 1 + 8 is constant on the closed interval 5 x 10.
q
3. 4.
5. 6.
in the form p , where p and q are integers. q We will call two squares on a chessboard neighbours" if they have a side or corner in common. The numbers 1 to 64 are arbitrarily placed on the 64 squares of a chessboard. Show that there are always two neighbours" whose numbers have positive di erence at least 9. A car's tires wear proportionally with the distance driven. Furthermore, front tires last a km and back tires b km, where a b. If, after an appropriate distance is driven, the tires are rotated that is, back tires placed on front wheels and front tires on back wheels, the distance which can be driven without needing to replace any of the tires can be increased. What is the longest distance which can be driven with a set of tires, before any new tires must be bought? P , Q, and R are points on the circumference of a circle such that PQR is an equilateral triangle. S is an arbitrary point on the circumference of the circle. Consider the lengths of the line segments PS , QS , and RS. Show that one of them is the sum of the other two. Show that for every positive integer n, there exist positive integers x and y such that p is an integer.
2xlog 2 = 3xlog 3 ; x 0
x2 + nxy + y2
423
1988 Final Round
1. The sides of a triangle have lengths a b c, and the corresponding perpendiculars have lengths ha , hb , and hc . Show that
a + ha
b + hb
c + hc :
2. Six ducklings swim on the surface of a pond, which is in the shape of a circle with radius 5 m. Show that, at every instant, two of the ducklings swim at a distance of at most 5 m from each other. 3. Show that for aribtrary real numbers x1 , x2 , and x3 , if x1 + x2 + x3 = 0; then x1 x2 + x2 x3 + x3 x1 0 : Find all n 4 for which the statement if x1 + x2 + + xn = 0; then x1 x2 + x2 x3 + + xn,1 xn + xn x1 0 is true. Both sums have n terms. 4. Let P x be a polynomial of degree 3 with exactly three distinct real roots. Find the number of real roots of the equation
P 0 x2 , 2P xP 00 x = 0 :
5. Let m and n be positive integers. Show that there exists a constant 1, independent of m and n, such that
m p7 implies that 7 , m2 : n n2 n2 6. The sequence a1 , a2 , : : : , is de ned by the recursion formula
and a1 = 1. Show that one can choose such that
an+1 = a2 + a1 n 1 ; n
n
s
1 an 2 n 2
for all n 1:
424
PROBLEMS
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 April 1999. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX.
2. 4AC 2 MC 2 , AC 2 BD2 = 4 ACD 2 , 4 BCD 2 , where XY Z denotes the area of 4XY Z . This is a continuation of problem 1812, 1993: 48 .
2374. 1998: 365 Correction Proposed by Toshio Seimiya, Kawasaki, Japan. Given triangle ABC with BAC 60 . Let M be the midpoint of BC . Let P be any point in the plane of 4ABC . Prove that AP + BP + CP 2AM . 2376. Proposed by Albert White, St. Bonaventure University, St. Bonaventure, NY, USA. Suppose that ABC is a rightangled triangle with the right angle at C . Let D be a point on hypotenuse AB , and let M be the midpoint of CD. Suppose that AMD = BMD. Prove that 1. AC 2 MC 2 + 4 ABC BCD = AC 2 MB 2 ;
425 Proposed by Nikolaos Dergiades, Thessaloniki, Greece. Let ABC be a triangle and P a point inside it. Let BC = a, CA = b, AB = c, PA = x, PB = y, PC = z, BPC = , CPA = and APB = . Prove that ax = by = cz if and only if , A = , B = , C = . 3 2378. Proposed by David Doster, Choate Rosemary Hall, Wallingford, Connecticut, USA.
, 4 cos 3
: Find the exact value of: cot Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Suppose that M1 , M2 and M3 are the midpoints of the altitudes from A to BC , from B to CA and from C to AB in 4ABC . Suppose that T1, T2 and T3 are the points where the excircles to 4ABC opposite A, B and C , touch BC , CA and AB . Prove that M1 T1 , M2 T2 and M3 T3 are concurrent. Determine the point of concurrency. 2380. Proposed by Bill Sands, University of Calgary, Calgary, Alberta. When the price of a certain book in a store is reduced by 1=3 and rounded to the nearest cent, the cents and dollars are switched. For example, if the original price was $43.21, the new price would be $21.43 this does not satisfy the reduced by 1=3" condition, of course. What was the original price of the book? For the bene t of readers unfamiliar with North American currency, there are 100 cents in one dollar. 2381. Proposed by Angel Dorito, Geld, Ontario. Solve the equation log2 x = log4 x + 1. 2382. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. If 4ABC has inradius r and circumradius R, show that
cos2 B , C 2r : 2 R
2377.
2379.
22
22
2383. Proposed by Mohammed Aassila, Universit
e Louis Pasteur, Strasbourg, France. Suppose that three circles, each of radius 1, pass through the same point in the plane. Let A be the set of points which lie inside at least two of the circles. What is the least area that A can have? 2384. Proposed by Paul Bracken, CRM, Universit
e de Montr
eal, Qu
bec. e Prove that 23n , 1n 3n + 1n for all n 2 N.
426
2385. Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel, Alcorc
n on, Madrid, Spain. A die is thrown n 3 consecutive times. Find the probability that the sum of its n outcomes is greater than or equal to n +6 and less than or equal to 6n , 6. 2386?. Proposed by Clark Kimberling, University of Evansville, Evansville, IN, USA. Write
1 ! 1 ! 3 ! 4 1 ! 6 2 1 ! 8 1 3 2 1 ! 1 1 1 3 1 3 4 1 2 3 4 6
The last ten numbers shown indicate that up to this point, eight 1's, one 2, three 3's, two 4's and one 6 have been written. a If this is continued inde nitely, will 5 eventually appear? b Will every positive integer eventually be written? Note: 11 is a number and not two 1's. 2387. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. For xed p 2 N, consider the power sums
Spn :=
n X
so that Sp n is a polynomial in n of degree p + 1 with rational coe cients. Prove that a If all coe cients of Sp n are integers, then p = 2m , 1 for some m 2 N. b? The only values of p yielding such polynomials are p = 1 and p = 3 with S1n = n2 and S3n = 2n4 , n2 .
k=1
2k , 1p ;
where n 1 ;
427
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems.
1637. 1991: 114; 1992: 125; 1994: 165 Proposed by George Tsintsifas, Thessaloniki, Greece. Prove that X
sin B + sin C
where the sum is cyclic over the angles A; B; C measured in radians of a nonobtuse triangle. Comment by Waldemar Pompe, student, University of Warsaw, Poland. We can use Crux 2015 1998: 305 to derive the strengthening of Crux 1637 given on 1994: 165 , namely that
X
A
12
for all triangles ABC . Without loss of generality, we can assume that A B C . Then and
sin B + sin C 9 3 A 1 1 1 A B C
p
By Chebyshev's Inequality and Crux 2015, we have
sin B + sin C sin A + sin C sin A + sin B : sin B + sin C
2 X sin A X 1
27 3 ; A A
3
X
p
and the result follows.
2257. 1997: 300 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. The diagonals AC and BD of a convex quadrilateral ABCD intersect at the point O. Let OK , OL, OM , ON , be the altitudes of triangles 4ABO, 4BCO, 4CDO, 4DAO, respectively. Prove that if OK = OM and OL = ON , then ABCD is a parallelogram.
428
I. Solution by Con Amore Problem Group, Royal Danish School of Educational Studies, Copenhagen, Denmark. Let OK = OM = h, OA = a, OB = b, OC = c, OD = d, and AOB = v. Expressing in two ways the area of 4AOB, we get
1 2
AB h = 1 ab sin v; 2
and so
Similarly, so that
r 1 = pa2 + b2 , 2ab cos v = 1 1 + 1 , 2 1 1 cos v : h ab sin v sin v b2 a2 b a
1 = 1 1 + 1 , 2 1 1 cos v ; h sin v d2 c2 d c
r
r
r
and similarly,
r
1 + 1 , 2 1 1 cos v = b2 a2 b a 1 + 1 + 2 1 1 cos v = a2 d2 a d
1 + 1 , 2 1 1 cos v ; d2 c2 d c 1 + 1 + 2 1 1 cos v : b2 c2 b c
1 2
r
Now, consider another convex quadrilateral A1 B1 C1 D1 , with diagonals in1 tersecting in O1 , and such that A1 O1 B1 = v , O1A1 = a , O1 B1 = 1 , b 1 O1C1 = 1 , and O1D1 = d . The equalities 1 and 2 imply that the oppoc site sides of A1 B1 C1 D1 are equal in length, which means that A1 B1 C1 D1 1 is a parallelogram. So a = 1 , and 1 = 1 , implying a = c and b = d. This c b d proves that ABCD is a parallelogram. II. Solution by the proposer slightly edited. If ABCD is a trapezoid with AB k CD, then which means that ABCD is a parallelogram. Hence, assume that ABCD is not a trapezoid and set P = AB CD, Q = AD BC . Indeed, P 6= Q. The assumption on the given quadrilateral says exactly that PO bisects BPC and QO bisects AQB . Thus
AB OK CD = OM = 1 ;
AP = AO = AQ ; PC OC QC implying that P , O, and Q lie on the Apollonius circle with centre on the line AC . Similarly, since BP = BO = BQ ; PD OD QD
429
P; O, and Q lie on the Apollonius circle with centre on the line BD. This implies that O is the circumcentre of 4POQ; that is, points P and Q coincide,
a contradiction.
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; MICHAEL LAMBROU, University of Crete, Crete, Greece; TOSHIO SEIMIYA, Kawasaki, Japan; and D.J. SMEENK, Zaltbommel, the Netherlands. There were also ve incorrect solutions submitted. Most of the submitted solutions are similar to the proposer's solution.
2260. 1997: 301 Proposed by Vedula N. Murty, Visakhapatnam, India. Let n be a positive integer and x 0. Prove that
n+1 1 + xn+1 n +n1 x : n
Solution by Florian Herzig, student, Cambridge, UK; Gerry Leversha, St. Paul's School, London, England; Nick Lord, Tonbridge School, Tonbridge, Kent, England; and Panos E. Tsaoussoglou, Athens, Greece. By the AM GM Inequality applied to the n + 1 positive numbers 1 ; 1 ; : : : ; 1 , we have x + 1
n+1 x , with equality if and only x; n n n n+1 nn 1 if x = n . This is clearly equivalent to the given inequality.
Also solved by PAUL BRACKEN, CRM, Universit
de Montr
al, Montr
al, e e e Qu
bec; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE e CHRONIS, student, Aristotle University of Thessaloniki, Greece; KEITH EKBLAW, Walla Walla, Washington, USA; RUSSELL EULER and JAWAD SADEK, NW Missouri State University, Maryville, Missouri, USA; RICHARD I. HESS, Rancho Palos Verdes, California, USA; JOE HOWARD, New Mexico Highlands University, Las Vegas, NM, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; PAVLOS MARAGOUDAKIS, Hatzikiriakio, Pireas, Greece; PHIL McCARTNEY, Northern Kentucky University, Highland Heights, KY, USA; VICTOR OXMAN, University of Haifa, Haifa, Israel; MICHAEL PARMENTER, Memorial University of Newfoundland, St. John's, Newfoundland; HEINZJURGEN SEIFFERT, Berlin, Germany; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; JOHN VLACHAKIS, Athens, Greece; EDWARD T.H. WANG, Wilfrid Laurier University, Waterloo, Ontario; ROGER ZARNOWSKI, TREY SMITH, CHARLES DIMINNIE and GERALD ALLEN jointly, Angelo State University, San Angelo, TX, USA; and the proposer. There were also three incomplete solutions. The majority of solvers used a standard calculus approach to establish the given inequality. The only exceptions are the ve listed in the solutions above plus Lambrou and Maragoudakis, who used Bernoulli's Inequality. Both Janous and Lambrou noted that the given inequality holds for all positive real n.
430
Janous also generalized the problem by showing that if a 0, and 0 are given real numbers, then the largest constant C = C a; ;
such that a , . a + x Cx holds for all x 0 is given by C = , The given problem is the special case when a = = 1 and = n + 1. By applying the AM GM Inequality to the n + 1 positive numbers: x ; x ; : : : ; x , Lord and the proposer obtained 1 + xn+1 n + 1n+1 xn , 1; n n n nn which is stronger than the proposed inequality for x 1. Zarnowski et al. commented that when n is odd, the inequality is true for all real x, while if n is even, there is a number xn ,2 such that the inequality holds for all x xn .
2261.
1997: 301 Proposed by Angel Dorito, Geld, Ontario. Assuming that the limit exists, nd where every fraction in this expression has the form
::: 2 + N +::: 1+ 1+::: Nlim 1 + N + 2+::: ; !1
!
a + b+::: c+::: c b + a+::: +::: for some cyclic permutation a, b, c of 1, 2, N .
Proposer's comment: this problem was suggested by Problem 4 of Round 21 of the International Mathematical Talent Search, Mathematics and Informatics Quarterly, Vol. 6, No. 2, p. 113. Solution by Keith Ekblaw, Walla Walla, Washington, USA. It will be shown that
p 2 + N + 1+ 5 1+ 1 + N + 1+ ,! 2 2+
1+
as N ! 1:
2+ 1+
First, consider Note that Thus JN
JN = N + 2 + N + : N +
2+ 1 + N + 2 + N + 1+
0:
KN = 1 + N + 1+ = 1 + J KN = 1 + J2 + K1 : 1+ N N N 2+
N and hence JN ! 1 as N ! 1. Now let
2 + N + 2 + JN
431 Thus as N ! 1 and hence JN ! 1, KN ! 1 + 1=KN . Letting K = limN !1 KN , we have K = 1+1=K or K 2 , K , 1 = 0 and so the required limit is p
K = 1 +2 5 :
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; FLORIAN HERZIG, student, Cambridge, UK; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengym
nasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; and the proposer. Hess calculates that if N is replaced by 3:5, then the expression inside the limit which is itself a limit, actually is equal to 2. Readers may like to nd other nice" triples of numbers a; b; c so that the expression c+ a+ a + b + a+ c + b+ is rational, say. The proposer notes as can be seen from the above proof that the answer is still the golden ratio if the 2's in the given expression are replaced by any constant real number.
2262. 1997: 301 Proposed by JuanBosco Romero M
arquez, Universidad de Valladolid, Valladolid, Spain. Consider two triangles 4ABC and 4A0 B 0 C 0 such that A 90 and 0 90 and whose sides satisfy a b c and a0 b0 c0 . Denote the A altitudes to sides a and a0 by ha and h0a . 1 1 + 1 , b 1 1 + 1 . Prove that a hah0a bb0 cc0 hah0a bc0 b0c Solution by Christopher J. Bradley, Clifton College,Bristol, UK. 1 1
1 1
a By the CauchySchwarz inequality on the vectors ; and 0 ; 0 , b c b c we have
1
1
1 + 1 1 + 1 2 1 + 1 2: bb0 cc0 b2 c2 b02 c02 1 1 b2 + c2 a2 Now 2 + 2 = 2 2 2 2 , since A 90 . b2 c b 2c bc a a 1 Also 2 2 = h2 , since 1 aha and 1 bc sin A are both 2 2 2 c2 sin2 A bc b
a
formulae for the area of 4ABC . Similarly for 4A0 B 0 C 0 . 1 1 1 , as required. Hence 0 + 0 0
bb
cc
haha
432
1 ; 1
and 1 ; 1
.
The proof of part b is the same, except that it starts with the vectors
Also solved by CLAUDIO ARCONCHER, Jundia
, Brazil; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; FLORIAN HERZIG, student, Cambridge, UK; RICHARD I. HESS, Rancho Palos Verdes, California, USA; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; VICTOR OXMAN, University of Haifa, Haifa, Israel; HEINZJURGEN SEIFFERT, Berlin, Germany; TOSHIO SEIMIYA, Kawasaki, Japan; GEORGE TSAPAKIDIS, Agrinio, Greece; JOHN VLACHAKIS, Athens, Greece; and the proposer. Most of the submitted solutions are similar to the one given above. Several solvers pointed out that the restrictions on the sides are unnecessary, and that equality in a occurs if and only if A = A0 = 90 and b=c = b0 =c0 and in b if and only if A = A0 = 90 and b=c = c0=b0 . Janous proved more generally that for any real number p 1
1 ha h0 p a 1 1 bb0p + cc0p ; and 1 ha h0 p a 1 bc0p + b01 p . c
b c
c0 b0
EC = EF = ED = EC ; sin EFC sin ECF sin ECD sin EDC which gives sin EFC = sin EDC . It follows that we have either that EFC = EDC , or that EFC + EDC = 180 . Case 1. EFC = EDC . Then FEC = DEC = 1 FED = 30 : 2 Let I be the intersection of BD and CE . Then DIC = IED + IDE = 30 + 30 = 60 : 1 Since DIC = 90 , 1 A, we obtain 90 , 2 A = 60 , so A = 60 . 2 Case 2. EFC + EDC = 180 . Then FED + FCD = 180 , so that 60 + FCD = 180 . Thus FCD = 120 ; that is, ACB = 120 .
1997: 364 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a triangle, and the internal bisectors of B, C , meet AC , AB at D, E , respectively. Suppose that BDE = 30 . Characterize 4ABC . Solution by the proposer. Let F be the re ection of E across BD. Since EBD = CBD, it follows that F lies on BC . So BDF = BDE = 30 , and DF = DE . Then 4DEF is equilateral, so EF = ED and FED = 60 . By the Law of Sines for 4EFC and 4EDC we obtain that
2263.
433
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; FLORIAN HERZIG, student, Cambridge, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; VICTOR OXMAN, University of Haifa, Haifa, Israel; D.J. SMEENK, Zaltbommel, the Netherlands. There were also two incomplete solutions submitted. Herzig and Lambrou have also shown that the characteristic condition is su cient, that is, A = 60 or C = 120 implies BDE = 30 .
Therefore, ABC is a triangle with either A = 60 or C = 120 .
2265. 1997: 364 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. Given triangle ABC , let ABX and ACY be two variable triangles constructed outwardly on sides AB and AC of 4ABC , such that the angles XAB and Y AC are xed and XBA + Y CA = 180 . Prove that all the lines XY pass through a common point. Solution by Toshio Seimiya, Kawasaki, Japan.
X r Ar
rD
B
r
Z
r
r rE P rC
,
Y
r
We denote the circumcircle of 4ABC by ,. Let BX and CY meet at Z . Since XBA + Y CA = 180 , we get XBA = 180 , Y CA = ACZ , so that A; B; Z; C are concyclic, that is, Z lies on ,. Let D; E be the second intersections of AX; AY respectively with ,. Since AX and AY are xed lines, D and E are xed points. Let P be the intersection of BE and CD. Since hexagon ADCZBE is inscribed in ,, by Pascal's Theorem the intersections of AD and BZ , of DC and BE , and of CZ and EA are collinear. Therefore variable line XY always passes through the xed point P . Editorial note: if the diagram di ers from the one shown, for example if Z lies between X and B , the proof still works with minor changes.
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; FLORIAN HERZIG, student, Cambridge, UK; MICHAEL LAMBROU, University of
Crete, Crete, Greece; MAR IA ASCENSION LOPEZ CHAMORRO, I.B. Leopoldo Cano,
434
Valladolid, Spain; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer. One incorrect solution and one comment were sent in. Seimiya and the proposer had similar solutions. Herzig notes that, if the xed angles are chosen to be AXB and AY C instead, then the lines XY still pass through a xed point. Readers may like to show this themselves.
2266. 1997: 364 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. BCLK is the square constructed outwardly on side BC of an acute triangle ABC . Let CD be the altitude of 4ABC with D on AB , and let H be the orthocentre of 4ABC . If the lines AK and CD meet at P , show that
I. Solution by Florian Herzig, student, Cambridge, UK.
M N K B X Y D P H
HP = AB : PD CD
A
L
C
Construct a square ABMN outwardly on 4ABC . Let X and Y be the points of intersection of AK; BC and AB; CM respectively. The rotation through a right angle about B maps 4MBC onto 4ABK . Hence AK and CM are perpendicular and it follows that AP and CD are altitudes in 4AY C . Therefore P is the orthocentre in that triangle, and as a consequence Y P ? AC or Y P k BH . Thus
HP = BY = BM = AB : PD Y D CD CD
435
II. Solution by Toshio Seimiya, Kawasaki, Japan.
K S B P D H A
L
C
HP = AH : 1 PD DS Since DS k BK and BK = BC , we have AD = DS = DS : 2 AB BK BC Since AH ? BC and CD ? AB we get HAD = BCD. Moreover we have HDA = BDC = 90 , so that 4HAD 4BCD. Thus AH = AD : 3 BC CD AH = AH BC = AD AB = AB ; DS BC DS CD AD CD HP = AB : so that we obtain from 1 that PD CD
From 2 and 3 we have
Let S be a point on AK such that DS ? BC and so DS k AH k BK . Since AH k DS we get
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; WALTHER JANOUS,
Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; D.J. SMEENK, Zaltbommel, the Netherlands; JOHN VLACHAKIS, Athens, Greece; and the proposer. The proposer's solution was the same as Herzig's. Most other solvers used either similar triangles as in II, trigonometry, or coordinates.
436
2268. 1997: 364 Proposed by JuanBosco Romero M
arquez, Universidad de Valladolid, Valladolid, Spain. Let x, y be real. Find all solutions of the equation
2xy + x2 + y 2 = pxy + x + y : x+y 2 2
Solution s Nikolaos Dergiades, Thessaloniki, Greece. by Let A =
s
2 2A2 + B2 = x + y2
x2 + y2 and B = pxy. Then
and
2A2 , B 2 = x , y 2
xy A , B = x + y , x2+ y 2 2A2 , B 2 = x , y 2 or A+B x+y 2 x , y = x , y2 : A+B x+y Therefore, we have either x , y 2 = 0 which implies that x = y or A + B = x + y. Let us consider A + B = x + y: A + B = x + y A + B2 = 2A2 + B2 A , B2 = 0 A = B s x2 + y2 = pxy
2 x , y2 = 0 x = y In conclusion, all solutions have x = y 6= 0, because x + y 6= 0.
Also solved by HAYO AHLBURG, Benidorm, Spain; PAUL BRACKEN, CRM, Universit
de Montr
al; CHRISTOPHER J. BRADLEY, Clifton Cole e lege, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark 2 solutions; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut, USA; RUSSELL EULER and JAWAD SADEK, NW Missouri State University, Maryville, Missouri, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; VICTOR OXMAN, University of Haifa, Haifa, Israel; HEINZJURGEN SEIFFERT, Berlin, Germany; D.J. SMEENK, Zaltbommel, the Netherlands; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; ROGER
and the given equation yields
437
ZARNOWSKI, Angelo State University, San Angelo, Texas, USA; and the proposer. There were 14 incorrect solutions submitted, 11 of which simply did NOT exclude the origin from the solution set.
2270. 1997: 365 Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Given 4ABC with sides a, b, c, a circle, centre P and radius intersects sides BC , CA, AB in A1 and A2 , B1 and B2 , C1 and C2 respectively, so that
A1A2 = B1B2 = C1C2 = 0: a b c Determine the locus of P . P to BC is x = P to CA is y =
s
2
Solution by the proposer. Assume that no two sides of the triangle are equal. The distance from
, 4a ;
2 2
s
2
2 2 P to AB is z = 2 , 4c : x2 , y2 b2 , a2 It follows that 2 2 = 2 2 , or y ,z c ,b x2c2 , b2 + y2a2 , c2 + z2b2 , a2 = 0 : 1 Considering x; y;z to be the triangular coordinates of P with respect to 4ABC , we conclude that 1 represents a conic K . Note that K passes through the incentre I 1; 1; 1 and the excentres Ia ,1; 1; 1; Ib 1; ,1; 1, and Ic 1; 1; ,1, and also the circumcentre Ocos A; cos B; cos C . So K is the conic through O of the pencil determined by I; Ia ; Ib ; Ic . Since the degenerate conics of the pencil are degenerate orthogonal hyperbolas that is, pairs of perpendicular lines, K must be an orthogonal hyperbola.
s
, 4b ;
2 2
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK. Neither solver mentioned the obvious special cases: The locus is a pair of perpendicular lines when 4ABC is isosceles, and just the four points I; Ia ; Ib ; Ic when equilateral. Bradley points out that it is clear from the statement of the problem with no need for coordinates that the locus includes I; Ia ; Ib ; Ic when = 0 and O when = 1.
438
2271. 1997: 365 Proposed by F.R. Baudert, Waterkloof Ridge, South Africa. A municipality charges householders per month for electricity used according to the following scale: rst 400 units  4.5cj per unit; next 1100 units  6.1cj per unit; thereafter  5.9cj per unit. If E is the total amount owing in dollars for n units of electricity used, nd a closed form expression, E n. Solution by Michael Lambrou, University of Crete, Greece. We may view the charges as consisting of i 4.5cj per unit and, additionally, ii a surcharge of 6:1 , 4:5 = 1:6cj per unit, but with iii a refund of 6:1 , 5:9 = 0:2cj per unit for units consumed in excess of 400 + 1100 = 1500. So in dollars the amount owing for n units is:
16 2 45 1000 n + 1000 maxf0; n , 400g , 1000 maxf0; n , 1500g: 1 Now writing maxfa; bg = 2 ja , bj + a + b, this simpli es to: 1 f52n + 8jn , 400j , jn , 1500j , 1700g : 1000
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; HANS ENGELHAUPT, Franz Ludwig Gymnasium, Bamberg, Germany; FLORIAN HERZIG, student, Cambridge, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Aus
tria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; KEEWAI LAU, Hong Kong; and the proposer. There were two incorrect solutions submitted. Some solvers used unit step functions instead of absolute value to express the answer.
1997: 365 Proposer unknown please identify yourself! Write r  s if there is an integer k satisfying r k s. Find, as a function of n n 2, the least positive integer satisfying
2272?.
k  k  k   k  k: n n,1 n,2 2
439 Solution by Florian Herzig, student, Cambridge, UK modi ed by the editor. Let kn denote the least positive integer k satisfying
k  k  k  k  k 1 n n,1 n,2 2 $ n + xn
2 + 1 , where x = bn + 1 , 2pn , 1c. We claim that kn = n 2 p We rst show that kn satis es 1. To this end, let u = kn . Then for all p integers a = 1; 2; ; u , 1 we have, from a kn , 1 that aa + 1 pk pk , 1 k . Hence kn , kn 1, which implies kn  kn . n n n a a+1 a+1 a Therefore we have kn  kn   kn k : 2 n u u,1 2 Next we show that for all integers a = u + 1; u + 2; ; n n + xn , a kan n + xn , a + 1: 3 In fact, the left inequality
3 holds for all a = 1; 2; ; n. To see this,
in note n + xn 2 we get a2 , ax , an + k = a , n + xn 2 + that from kn n n 2 2
2 n + xn kn : kn , 2 0, and thus n + xn , a a
On the other hand, note that the right inequality in 3 is equivalent to
n + xn + 1
2 a,
2 2
jp
Since kn
n + xn
2, we have
n + xn + 1
2 , k : n 2
4
k kn + 1 n + xn + 1 n + xn + 1 : 2 2 Ed: The last inequality holds since n + xn is an integer. Hence it su ces to establish 4 for a = n. Substituting a = n into the right inequality of 3, we need to show that kn nxn + n. From n , 2pn , 1 xn n we get n
2 xn 2 4n , 1 , n + xn + 1 nx + n. or n + xn 2 + 4 4nxn + n. Hence kn n 2 Therefore 3 holds, and by setting a = u + 1; u + 2; ; n we get
a u+1=
kn n
xn + 1 nkn 1 xn + n , u , 1 ukn 1 xn + n , u kun : , +
440
kn  kn  kn  kn : 5 n n,1 u+1 u From 2 and 5 we conclude that kn satis es 1. Now we show that if k is any integer satisfying 1, then k kn . To k .We rst show that x x . Since there exists an this end, let x = n n k and x = k z, we have z , x 1 k z integer z such that n n,1 n k k k and hence n , 1 x + 1. Similarly, n , 2 x + 2, n , 3 x + 3, , k x + n , 1. That is, for all a = 1; 2; ; n we have k n , ax + a 1
or k n , ax + a + 1 = ,a2 + an , x + nx + 1: 6
2
2 k n , x + nx + 1 , a , n , x : 7 2 2 k k 1, contradicting Note that x 1. Ed: If k n , 1, then 0 n n,1 k  k . Hence k n . n n,1 On the other hand, it is clear that xn n , 1. Suppose, contrary to what we claim, that x xn . Then we have 1 x xn n , 1 and so 2 n , x n or 1 n , x n . Here we must assume that n 3. 2 2 The case when n = 2 can be treated separately, since it is easy to verify that n , x in 6 and 7 and obtain k2 = 3. Hence we may let a = 2
Hence
Hence
n , x
2 + nx + 1 , n , x , n , x
2 : k 2 2 2
8
Since the right side of 7 is an integer and since the last squared term in 8 1 is either 0 or we get
4
k
$
n , x
2 + nx + 1 = 2
$
$
n + x
2 + 1 : 2
9
k 1 Thus x = n n
$
n + x
2 + 1 = 1 n + x
2 + 1 . 2 n 2 n
$
441
bzc = z for all real Ed: It is known and easy to show that n n numbers z and positive integers n. $ $ 1 n + x
2 + 1 , x = 1 n , x
2 + 1 , which Hence 0 n 2 ! n n 2 n
2
2 1 n , x + 1 1 or n , x n , 1. implies that n 2 2 pn , 1 or x n , 2pn , 1, from which we get x Thus n , x 2 bn +1 , 2pn , 1c = xn , a contradiction. Hence x xn . Therefore we may replace x by xn in 6, 7, and 9 and conclude that $
2 k n + xn + 1 = kn. This completes the proof. 2
Also solved by PETER TINGLEY, student, University of Waterloo, Waterloo, Ontario. There was one incorrect solution. Tingley gave the answer pn , 1 + 1c + $ n , bn , 2pn , 1 + 1c
2 + 1 kn = nbn , 2 2
which is readily seen to be the same as the one obtained by Herzig. The proposer had conjectured that 2 2 kn = 1 + n , m2 + n , m if m n , 2 1 + n , m otherwise p 4 where m = 1 + 2 n , 7 and had veri ed it for 2 n 600 using a computer. In a private communication Tingley has actually proved that this conjectured formula is equivalent to the answer given by Herzig and himself. Interested readers may nd the proof of this fact quite challenging.
2273. 1997: 366 Proposed by Tim Cross, King Edward's School, Birmingham, England. Consider the sequence of positive integers: f1, 12, 123, 1 234, 12 345, : : : g, where the next term is constructed by lengthening the previous term at its righthand end by appending the next positive integer. Note that this next integer occupies only one place, with carrying" occurring as in addition: thus the ninth and tenth terms of the sequence are 123 456 789 and 1 234 567 900 respectively. Determine which terms of the sequence are divisible by 7. Solution by HeinzJurgen Sei ert, Berlin, Germany. The sequence fan g under consideration satis es the recurrence
a1 = 1
and
an = 10an,1 + n for n 2:
442 A simple induction argument shows that
Let n 2 N. Applying the Euclidean Algorithm twice, we see that there exist nonnegative integers j; k; r such that 0 k 6, 0 r 5, and n = 42j + 6k + r. Since 342j 36k 1 mod 7 by Fermat's Little Theorem, it follows that
81an = 10n+1 , 9n , 10; n 2 N:
4an 3r+1 + 2k , 2r , 3 mod 7:
The following table gives the remainder when the expression on the right hand side of the above congruence is divided by 7: 0 0 2 4 6 1 3 5 1 4 6 1 3 5 0 2 2 6 1 3 5 0 2 4 3 2 4 6 1 3 5 0 4 1 3 5 0 2 4 6 5 2 4 6 1 3 5 0 Inspecting this table and using the above congruence, we see that an is divisible by 7 if and only if n 0; 22; 26; 31; 39; or 41 mod 42.
r nk 0 1 2 3 4 5 6
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK;
MIGUEL ANGEL CABEZON OCHOA, Logro~ o, Spain; HANS ENGELHAUPT, Franz n Ludwig Gymnasium, Bamberg, Germany; FLORIAN HERZIG, student, Cambridge, UK; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; KEEWAI LAU, Hong Kong; GERRY LEVERSHA, St. Paul's School, London, England; J.A. MCCALLUM, Medicine Hat, Alberta; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; TREY SMITH, GERALD ALLEN, NOEL EVANS, CHARLES DIMINNIE, AND ROGER ZARNOWSKI jointly, Angelo State University, San Angelo, Texas; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer. There was one incorrect solution submitted.
2274. 1997: 366 Proposed by V
aclav Konecn
y, Ferris State University, Big Rapids, Michigan, USA.
A. Let m be a nonnegative integer. Find a closed form for B. Let m 2 f1; 2; 3; 4g. Find a closed form for
n m X Y k=1 j =0 n m X Y k=1 j =0
k + j .
k + j 2.
443 C?. Let m and j j = 0; 1; : : : ; m be nonnegative integers. Prove or m+1 n m X Y Y disprove that k + j j is divisible by n + j .
k=1 j =0 j =0
n m + k! = m + 1! X m + k
k + j = k k=1 j =0 k=1
, 1! k=1
k , 1
m + 1 + m + 2 + m + 3 + + m + n
: = m + 1! 0 1 2 n,1
n m X Y
I. Solution by Florian Herzig, student, Cambridge, UK. A. Note that
n X
This combinatorial sum inside the square brackets is wellknown and can , +3 be evaluated as follows. Note that the rst two terms add to m1 which in , +4 turn adds with the third term to m2 , and so on until we obtain the desired ,m+n+1 closed form n,1 in the end. Thus
+n+
k + j = m + 1! m n , 1 1 k=1 j =0 + 1! : = mm + nn , 1! = nn + 1 :m:+n2+ m + 1 : + 2 P B. The expressions reduce to sums of the form n=1 km . With the help k of a calculator I got
n m X Y
!
n + 2 23n2 + 6n + 1 for m = 1; 5 3
!
!
n + 4 835n4 + 280n3 + 685n2 + 500n + 12 for m = 3; 21 5
!
n + 3 122n + 35n2 + 15n + 1 for m = 2; 4 35
n + 5 40126n5 + 1575n4 + 6860n3 + 12075n2 + 7024n + 60 for m = 4: 6 77
C. I assume that is divisible by" means polynomial division. Clearly the j should be positive integers, and in this case I will prove that the claim is true. De ne
P n =
n m X Y
k=1 j =0
k + j j ;
444
P P a polynomial of degree n=1 j + 1 since each n=1 kl is a polynomial j k of degree l + 1. Then
P n =
n X
m Y
as in the rst of these sums all the terms for k 0 vanish. Hence, for all integers a such that ,m a 0 we get
k=,m j =0 m+a+1 X k=1
k + j
j
=
m+n+1 Y m X k=1 j =0
k , m , 1 + j j ;
P a =
k , m , 1 0 k , m 1 : : : k , 1
m
= 0;
since each term is zero. Thus n , a must be a factor of P n for each such a, so P n must be divisible by each of n; n + 1; : : : ; n + m.  Ed. For n = ,m , 1 the above sum is empty and so PP,mn, 1 = 0 as well to avoid empty sums one can instead write P n as m+ +1 pQ , p0 where k k=0 pk is the above product. Therefore P n is divisible by m+1n + j as j =0 claimed. II. Solution to part A by Michael Lambrou, University of Crete, Greece. A. From the identity
m+1 m+1 Y Y k + j = m 1 2 4 k + j , k , 1 + j 5 + j=0 j =0 j =0 m Y
2 3
Q easily veri ed by considering the common factor m k + j of the two j =0 products on the right, we obtain telescopically
m+1 m+1 m+1 Y Y Y j 5 = m 1 2 n + j : k + j = m 1 2 4 n + j , + j=0 + j=0 k=1 j =0 j =0 n m X Y All three parts also solved by G. P. HENDERSON, Garden Hill, Ontario. Parts A and B only solved by THEODORE N. CHRONIS, Athens, Greece; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; and the proposer. As Herzig mentions above, part A at least is a fairly familiar result. For example, see formula 2.50, page 50 of Concrete Mathematics by Graham, Knuth and Patashnik. In fact it was proposed for publication in CRUX in 1991 by an Edmonton high school student, Jason Colwell, but was not accepted by the then editor. Chronis notes that, in the solution for part B, when m = 4, the fth degree polynomial has 2n + 5 as a factor.
2
3
Editorial note: Lambrou also solved parts B and C.
445
2275. 1997: 366 Proposed by M. Perisastry, Vizianagaram, Andhra Pradesh, India. Let b 0 and ba ba for all a 0. Prove that b = e. I. Solution by Gerald Allen, Charles Diminnie, Trey Smith and Roger Zarnowski jointly, Angelo State University, San Angelo, TX, USA; Russell Euler and Jawad Sadek jointly, NW Missouri State University, Maryville, Missouri, USA; Michael Parmenter, Memorial University of Newfoundland, St. John's, Newfoundland; Reza Shahidi, student, University of Waterloo, Waterloo, Ontario; George Tsapakidis, Agrinio, Greece; and John Vlachakis, Athens, Greece. Let f x = bx , bx for x 0. Since f x 0 for all x 0 and f 1 = 0, it follows that f has a relative as well as an absolute minimum at x = 1. Since f 0 1 exists, we have f 0 1 = 0; that is, b ln b , b = 0. Since b 0, we get ln b = 1 or b = e. II. Solution by Theodore Chronis, student, Aristotle University of Thessaloniki, Greece; Florian Herzig, student, Cambridge, UK; Michael Lambrou, University of Crete, Crete, Greece; Gerry Leversha, St. Paul's School, London, England; Vedula N. Murty, Visakhapatnam, India; HeinzJurgen Sei ert, Berlin, Germany; and David R. Stone, Georgia Southern University, Statesboro, Georgia, USA. The given inequality is equivalent to ba,1 a for a
n Letting all 0. 1 where n 2 N, we get b n 1 + 1 , or b 1 + 1 . Hence 1 a = 1+
1
n b nlim 1 + n = e. !1
n
n
n
On the other hand, letting a = 1 +
1 we get from b1,a , that b n+1
Also solved by FRANK P. BATTLES, Massachusetts Maritime Academy, Buzzards Bay, MA, USA; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; LUZ M. DeALBA, Drake University, Des Moines, IA, USA; DAVID DOSTER, Choate Rosemary Hall, Wallingford, Connecticut, USA; KEITH EKBLAW, Walla Walla, Washington, USA; HANS ENGELHAUPT, Franz Ludwig Gymnasium, Bamberg, Germany; F.J. FLANIGAN, San Jose State University, San Jose, California, USA; RICHARD I. HESS, Rancho Palos Verdes, California, USA; JOE HOWARD, New Mexico Highlands University, Las Vegas, NM, USA; WALTHER JANOUS, Ur
sulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; KEEWAI LAU, Hong Kong; D.J. SMEENK, Zaltbommel, the Netherlands; DIGBY SMITH, Mount Royal College, Calgary, Alberta; and the
1
n b nlim 1 + n = e. Therefore, b = e. !1
a
1
1
,1 = n , where n 2 N, n n+1 1 , or b 1 + 1
n+1 . Hence 1+ n n
446
proposer. There was one incorrect solution. Although the problem did not ask to show that the condition b = e is both necessary and su cient, a few solvers did provide a proof of the simple fact that ex ex for all x 0.
2277. 1997: 431 Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel,
n Alcorc
Madrid, Spain. on, For n 1, de ne
n un = 11 ; 22 ; : : : ; nn ,1;1n ; n; n ; ;n ;n ,
n p, then since p n p jk we have p n p jjk; n and so pa, n p jj k;kn . k arises when a = n . This shows Thus the highest power of p in any p k;n Y that un = pnp , n p , where the product is over all primes. Note that Y p n , 1 = p n,1 p and hence un = n , 1un,1 is equivalent to
If a
p
where the square brackets and the parentheses denote the least common multiple and greatest common divisor respectively. For what values of n does the identity un = n , 1un,1 hold? Solution by Florian Herzig, student, Cambridge, UK. We rst introduce some notation: for any prime p, let np = maxf jp ng, n p = maxf jp divides ng. Thus p n p jjn. For each k = 1; 2; ; n we determine a = ak such that pajjk. If a n p, then since pajn, we have pajjk; n and so p 6 j k .
k; n
np , n p = n , 1p for all primes p
1
We distinguish two cases: Case i Suppose n is a prime power, say n = q b where q is a prime and b 0. For p 6= q, 1 is satis ed since np = n , 1p and n p = 0. For p = q we have n = pb and so np = n p = b and n , 1p = b , 1. Hence 1 holds if and only if b , 1 = 0; that is, b = 1. Case ii If n is not a prime power, then np = n , 1p for all primes p. Hence i holds if and only if n p = 0 for all primes p, and so n = 1. Ed: Clearly n = 1 is not a solution, since u0 is unde ned. Therefore un = n , 1un,1 if and only if n is a prime.
447
Also solved by ED BARBEAU, University of Toronto, Toronto; NIKOLAOS DERGIADES, Thessaloniki, Greece; WALTHER JANOUS, Ursulinengymnasium, Inns
bruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; DAVID R. STONE, Georgia Southern University, Statesboro, Georgia, USA; and the proposer. From the proof given above, it is not di cult to see that, in fact, we have un = 1; 2; ; n =n. This was explicitly pointed out by Kone cny, Lambrou, and
the proposer, but only Lambrou and the proposer actually gave a proof.
2278. 1997: 431 Proposed by Joaqu
n Gomez Rey, IES Luis Bu~ uel,
n Alcorc
Madrid, Spain. on, Determine the value of an , which is the number of ordered n tuples k2 ; k3; : : : ; kn ; kn+1 of nonnegative integers such that
2k2 + 3k3 + : : : + nkn + n + 1kn+1 = n + 1:
I. Solution by Michael Lambrou, University of Crete, Crete, Greece. We show that an = pn + 1 , pn for n 1 where pm denotes the number of partitions of m into positive integral parts. Our argument is based on the wellknown observation that to a partition of m where lk k's appear k = 1; 2; : : : ; m, so that 1l1 + 2l2 + + mlm = m ; 1 corresponds the ordered mtuple l1 ; l2 ; : : : ; lm . Conversely, to any given ordered mtuple l1 ; l2 ; : : : ; lm of positive integers satisfying 1, there corresponds a partition of m. For xed n 1 consider the partitions of n + 1 as above. They are of two types: a those for which the number 1 is absent in the decomposition; or b those for which the number 1 appears at least once. The number of partitions of type a is clearly an . Moreover, for each partition in case b, if we delete one 1, we get a partition of n. Conversely, every partition of n with an extra 1 added on gives a partition of n + 1 of type b. Clearly then pn + 1 = an + pn, as required. II. Solution by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. We deal more generally with the equation: k1 + 2k2 + + j , 1kj,1 + j + 1kj+1 + + n + 1kn+1 = n + 1 where j 2 f1; 2; : : : ; n + 1g is xed and determine the number an j of its nonnegative solutions in Zn .
448 For this we recall that since Euler's days such problems are dealt with best by generating functions, namely:
Ex = 1 + x1 + x21 + : : : 1 + x2 + x22 + : : : : : : 1 1 1 1 : : : = X pkxk = 1 , x 1 , x2 1 , x3
k=0
where pk denotes the number of partitions of k; that is, the number of unordered representations of k as k = s1 + s2 + + se with sj a positive integer for j = 1; 2; : : : ; e, or equivalently k = 1n1 +2n2 + + knk , where nj 0 is the number of appearances of summand j . Therefore, all partitions with summand j forbidden are obtained via
1,xj E x = 1,xj
1 X
Hence the desired amount an j equals:
k=0
pkxk =
j ,1 X
k=0
pkxk+
1 X,
k=j
pk,pk,j xk :
an j =
pn + 1; if n + 1 j , 1 pn + 1 , pn + 1 , j ; if n + 1 j
Also solved by HEINZJURGEN SEIFFERT, Berlin, Germany. There was one incorrect solution submitted. Janous remarks how his ideas above can be extended to include the case where the excluded summand can be a subset of the values from 1 to n + 1.
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
449
THE ACADEMY CORNER
No. 21 Bruce Shawyer
All communications about this column should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7
THE BERNOULLI TRIALS 1998
Hints and Answers
Ian Goulden and Christopher Small University of Waterloo The questions were printed earlier this year 1998: 257 . 1. False. The diameter of the larger circle is 50. Let CD = x. Then AC = 9 + x and CE = 4 + x. Consider similar triangles in AED.
F E A C D B
2. True. Let X be the number of rounds until his rst error, Y the number of subsequent rounds until his second error. Then But E X = 2, which can be found by summing the series. 3. True. Use 1, 2, 4, 8, : : : , 210 and a binary expansion of any number from 8 to 1998. 4. True. Write the points in coordinates and use the Pythagorean formula. Everything will cancel!
EX + Y = EX + EY = 2 EX
450 5. True. The volume of the tetrahedron is 2=12. To prove this, imbed the tetrahedron into a cube so that the vertices of the tetrahedron are four of the eight vertices of the cube. Show that the tetrahedron has a volume which is one third the volume of the cube. On the other hand, the volume of the sphere is 4r3 =3 where r = 1= . 6. False. Actually 15032 = 2259009 is the smallest. 7. True. Draw a graph of the function in the unit square.
p
11 00 11 00
8. True. The choice k = 9 works: +1 = . 2 2 9. True. The maximum area is achieved with a cyclic quadrilateral. The area of such a cyclic quadrilateral can be determined by Brahmagupta's formula
9nn + 1
3n + 13n + 2
= 4! = 2 6 . 10. False. Write x = x=1 + x. The equation can be written as f x = f x . This is clearly satis ed by f x = nx. In particular, the choice of f x = x=1 + x works. 11. True. 32 = 38 2 29 = 2 23 . So 23 222 = 42 32 . So 32 33 2 23 . Finally, we have 23 222 = 42 32 .
3 2 23 32 32 2 32 32 3 23 23 2 23 23 32 32 32 32 32 32
p s A = ps , ap, bs , cs , d
Obviously, this can be continued. 12. True. Write
Z 1 x dx Z 1 x e,x = ex , 1 1 , e,x dx 0 0 Z1 = xe,x + e,2x + e,3x + dx 0 2 = 1 + 1 + 1 + =
22 32 6
451 13. False. The answer is one square metre exactly. To prove this consider the gure to the right: The thin triangle has the required dimensions and its area is
3
2 2 2
1 4 5 , 2 2 + 1 2 3 + 1 2 2
= 1 . 2 2 2
14. False. Suppose there were such a function. The function ex , eex can be seen to be a onetoone decreasing function. So f must be onetoone: f x = f y = f f x = f f y = x = y . As f is continuous and onetoone, it must be strictly increasing or strictly decreasing. But either way, f f x must then be a strictly increasing function. Contradiction. 15. False. C Consider a third random point A X in the circle. The region R1 corresponds to X B the points for X where the angle at X is obtuse. The region R2 corresponds to the points for X where the angle at B is obtuse. As the probabilities are the same, the average areas of R1 and R2 must be the same. 16. False. The second player wins by forcing bilateral symmetry on remaining petals. For example, if the rst player starts by taking petal 1, the second player takes petals 7 and 8 together. If the rst player chooses 1 and 2 the second player takes petal 8, etc. This ensures that the rst player can never take the last petal.
2
2
452
THE OLYMPIAD CORNER
No. 194 R.E. Woodrow
All communications about this column should be sent to Professor R.E. Woodrow, Department of Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada. T2N 1N4. We begin this number with the problems from the BiNational Israel Hungary Competition from 1995. My thanks go to Bill Sands, University of Calgary, who collected these problems for me while assisting at the IMO in Canada in 1995.
there exists a whole square between Sn and Sn+1. 2. Let P , P1, P2, P3, P4 be ve points on a circle. Denote the distance of P from the line PiPk by dik . Prove that d12 d34 = d13 d24 . 3. Consider the polynomials f x = ax2 + bx + c which satisfy jf xj 1 for all x 2 0; 1 . Find the maximal value of jaj + jbj + jcj. 4. Consider a convex polyhedron, whose faces are triangles. Prove that it is possible to colour the edges by red and blue in a way that one can travel from any vertex to any other vertex, passing only through red edges, and also one can travel only through blue edges. Next we give two rounds of the 31st Spanish Mathematical Olympiad. Both these contests were collected for me by Bill Sands, University of Calgary, while he was assisting at the IMO in Canada in 1995. I also received them from Francisco Bellot Rosado, I.B. Emilio Ferrari, Valladolid, Spain, one of the contest organizers. Many thanks.
BINATIONAL ISRAELHUNGARY COMPETITION 1995 1. Denote the sum of the rst n prime numbers by Sn. Prove that
31st SPANISH MATHEMATICAL OLYMPIAD
Proposed by the Royal Spanish Mathematical Society First Day  Time: 4 hours
First Round: December 2 3, 1994
1. Let a, b, c be distinct real numbers and P x a polynomial with real coe cients. If: the remainder on division of P x by x , a equals a,
453 the remainder on division of P x by x , b equals b, and the remainder on division of P x by x , c equals c; determine the remainder on division of P x by x , ax , bx , c. 2. Show that, if x + px2 + 1y + py2 + 1 = 1, then x + y = 0. 3. The squares of the sides of a triangle ABC are proportional to the numbers 1, 2, 3. a Show that the angles formed by the medians of ABC are equal to the angles of ABC . b Show that the triangle whose sides are the medians of ABC is similar to ABC . 4. Find the smallest natural number m such that, for all natural numbers n m, we have n = 5a + 11b, with a, b integers 0. Second Day  Time: 4 hours
5. A subset A M = f1; 2; 3; : : : ; 11g is good if it has the following property: If 2k 2 A, then 2k , 1 2 A and 2k + 1 2 A" . The empty set and M are good. How many good subsets has M ? 6. Consider the parabolas y = cx2 + d, x = ay2 + b, with c 0, d 0, a 0, b 0. These parabolas have four common points. Show that these four points are concyclic. 7. Show that there exists a polynomial P x, with integer coe cients, such that sin 1 is a root of P x = 0. 8. An aircraft of the airline Air Disaster" must y between two cities with m + n stops. At each stop, the aircraft must load or unload 1 ton of goods. In m of the stops, the aircraft loads; in n of the stops, the aircraft unloads. Nobody from the sta has observed that the aircraft cannot handle a load of more than k tons n k m + n, and the stops where the plane loads and unloads are randomly distributed. If the aircraft takes o with n tons of goods, nd the probability of the aircraft arriving at its destination. 31st SPANISH MATHEMATICAL OLYMPIAD
Second Round: February 24 25, 1995
First Day  Time: 4 hours
1. Consider sets A of 100 distinct natural numbers, such that the following property holds: If a, b, c are elements of A distinct or not, there exists a nonobtuse triangle of sides a, b, c." Let S A be the sum of the
454 perimeters of the triangles considered in the de nition of A. Find the minimal value of S A. 2. We have several circles of paper on a plane, such that some of them are overlapped, but no one circle is contained in another. Show that it is impossible to form disjoint circles with the pieces which result from cutting o the nonoverlapped parts and reassembling them. 3. A line through the barycentre G of the triangle ABC intersects the side AB at P and the side AC at Q. Show that
PB QC 1 : PA QA 4
Second Day  Time: 4 hours
4. Find all the integer solutions of the equation
in which p is a prime number. 5. Show that, if the equations
px + y = xy
have a common root, then the rst equation would have two equal roots, and determine in this case the roots of both equations in terms of n. 6. AB is a xed segment and C a variable point, internal to AB. Equilateral triangles ACB 0 and CBA0 are constructed, in the same halfplane de ned by AB , and another equilateral triangle ABC 0 is constructed in the opposite halfplane. Show that: a The lines AA0 , BB 0 and CC 0 are concurrent. b If P is the common point of the lines of a, nd the locus of P when C varies on AB. c The centres A00 , B 00 , C 00 of the three equilateral triangles also form an equilateral triangle. d The points A00 , B 00 , C 00 and P are concyclic. As a nal problem set for your puzzling pleasure over the Christmas break we include the Final Round problems of the 46th Polish Mathematical Olympiad. Again these problems come to us from more than one source. Bill Sands collected them while assisting at the IMO in Canada. Marcin E. Kuczma, Warszawa, Poland, one of Poland's premiere contest people, also sends us copies of contest materials regularly. Again, many thanks.
x3 + mx , n = 0 ; nx3 , 2m2x2 , 5mnx , 2m3 , n2 = 0 ;
m 6= 0; n 6= 0
455
46th POLISH MATHEMATICAL OLYMPIAD 1994 5
Problems of the Final Round March 31April 1, 1995
First Day  Time: 5 hours
1. Find the number of those subsets of f1, 2, : : : , 2ng in which the equation x + y = 2n + 1 has no solutions. 2. A convex pentagon is partitioned by its diagonals into eleven regions: one pentagon and ten triangles. What is the maximum number of those triangles that can have equal areas? 3. Let p 3 be a given prime number. De ne a sequence an by an = n for n = 0; 1; 2; : : : ; p , 1 ; an = an,1 + an,p for n p : Determine the remainder left by ap on division by p. Second Day  Time: 5 hours
3
4. For a
xed integer n 1 compute the minimum value of the sum
given that x1 ; x2 ; : : : ; xn are positive numbers satisfying the condition
x1 + x2 + x3 + + xn ; 2 3 n
2 3
n
1 1 1 x1 + x2 + + xn = n :
5. Let n and k be positive integers. From an urn containing n tokens numbered 1 through n, the tokens are drawn one by one without replacement, until a number divisible by k appears. For a given n determine all those k n for which the expected number of draws equals exactly k. 6. Let k, l, m be three noncoplanar rays emanating from a common origin P and let A be a given point on k other than P . Show that there exists exactly one pair of points B , C , with B lying on l and C on m, such that PA + AB = PC + CB and PB + BC = PA + AC :
Next an apology. Somehow when preparing the list of solvers of problems for the October issue of CRUX with MAYHEM, I omitted listing Miguel Amengual Covas, Cala Figuera, Mallorca, Spain as a solver of problem 3 of the Swedish Mathematical Olympiad 1998: 327 and for problem 2 of the Dutch Mathematical Olympiad 1998: 330 . Sorry!
456 Now we turn to solutions by the readers to problems of the Irish Mathematical Olympiad 1994 1997: 388 389 . 1. Let x, y be positive integers with y 3 and Prove that x2 + y 4 = 1994. Solution by Pavlos Maragoudakis, Pireas, Greece. Rewriting we get
x2 + y4 = 2 x , 62 + y + 12 :
y2 , 12 = y4 , 2y 2 , 4y + 70 y = 69 62 Z ; 4 and y 4 , 2y 2 , 4y + 70 = y 4 y 2 + 2y , 35 = 0 y = 5 or y = ,7. Thus, y = 5. Now 1 gives x = 37 and x2 + y4 = 372 + 54 = 1369 + 625 = 1994 :
x2 , 24x , y4 + 2y2 + 4y + 74 = 0 : 1 Now 1 has integer solutions only if the discriminant 4y 4 , 2y 2 , 4y + 70 is a perfect square. It is easy to prove that for y 4, y2 , 22 y4 , 2y 2 , 4y + 70 y 2 + 12: Indeed y 2 , 2y + 33 0 and 4y y + 1 69. The rst inequality is true. Since y 4, 4y y + 1 4 4 5 = 80 69. The only perfect squares between y 2 , 22 and y 2 + 12 are y 2 , 12 and y 22. Now
Comment by Jim Totten, The University College of the Cariboo, Kamloops, BC. The result also works for y = 1 and y = 2 as well, but fails for y = 3 with x = 1. 2. Let A, B, C be three collinear points with B between A and C . Equilateral triangles ABD, BCE , CAF are constructed with D, E on one side of the line AC and F on the opposite side. Prove that the centroids of the triangles are the vertices of an equilateral triangle. Prove that the centroid of this triangle lies on the line AC . Solutions by Miguel Amengual Covas, Cala Figuera, Mallorca, Spain; and by Pavlos Maragoudakis, Pireas, Greece. We give the solution of Amengual Covas. We introduce a rectangular Cartesian system with origin at B and x axis along AC . Let AB = a and BC = b. The centroids O1, O2, O3 of equilateral triangles ABD, BCE and CAF are
,a a
G1 = 2 ; 2p3 ;
b b
G2 = 2 ; 2p3
and
457
+
G3 = ,a2+ b ; , a p3b : 2 Hence the centroid G of triangle 4G1 G2 G3 is ,a + b
G= ;0 :
3
It is straightforward to verify that
G1G2 = G2G3 = G3G1 =
s
a2 + ab + b2 :
3
Therefore 4G1 G2G3 is equilateral. Finally, since the y coordinate of G is 0, clearly G is on the x axis; that is to say, G lies on the line AC .
6
D G1
G2 B G G3
A ,a; 0
p
Cb; 0 xaxis
tion
3. Determine with proof all real polynomials f x satisfying the equaf x2 = f xf x , 1 :
Solution by Pavlos Maragoudakis, Pireas, Greece. We will prove that f x = 0 or f x = x2 + x + 1k , k = 0; 1; 2; : : : . Consider any possibly complex root p of f x. Then
f p2 = f p f p , 1 = 0 f p , 1 = 0
and
f p + 12 = f p + 1 f p = f p + 1 0 = 0 :
458 So p2 , p+12 are also roots of f x. Thus p2n and p+12n are roots of f x, n = 0; 1; 2; : : : . If jpj 6= 1 or jpj 6= jp + 1j then we get an in nite number of roots, so f x is a constant polynomial, and having a root p, f x 0. If jpj 6= 1 or jp + 1+ 6= 1; that is, if jpj = 1 = jp + 1j then p p = 1 and p p = p + 1p + 1, so p + p = ,1 and p = ,p + 1, and now p,p,1 = 1. Therefore p2+p+1 = 0. It follows that f x = x2 +x+1k, for some k 1. On the other hand if f x has no roots, then f x = c 6= 0, is a nonzero constant. Then f x2 = f x f x , 1 gives c = c c, and c 6= 0 gives c = 1. Thus f x = x2 + x + 10 . In any case f x = 0 or f x = x2 + x + 1k for some k = 0; 1; 2; : : : . 6. A sequence fxng is de ned by the rules and
x1 = 2
n = 2; 3; : : : : Prove that xn is an integer for every positive integer n. x2 = 2 2 x1 x3 = 2 5 x2 3 , xn,1 = 2 2nn,13 xn,2 xn = 2 2nn,1 xn,1
nxn = 22n , 1xn,1 ;
Solutions by Murray S. Klamkin, University of Alberta, Edmonton, Alberta; and by Pavlos Maragoudakis, Pireas, Greece. We give Maragoudakis' solution. Now 3
It follows that
, n since 2n is a binomial coe cient. 7. Let p, q, r be distinct real numbers which satisfy the equations
q = p4 , p ; r = q4 , q ; p = r4 , r : Find all possible values of p + q + r.
, 5 xn = 2n,1 2nn, 12n1 3 32 3 2 n , 2n! = 2n
2 Z = n!2 n
459 Solution by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. From the discriminant of each quadratic equation, it follows that p, q, r are all less than 4. If one of p, q, r is zero, all are zero, so we now assume none are zero. Also, it follows that p, q , r all have the same sign. From the product of the three equations we get 1 = 4 , p4 , q 4 , r so that p, q , r are all positive. We now let p = 4 sin2 , so then successively, q = 4 sin2 2, r = 4 sin2 4, p = 4 sin2 8. Hence, sin = sin 8 so that we have sin7=2cos 9=2 = 0 or sin9=2cos 7=2 = 0 : Solving for leads to only the following possible nonzero values of p + q + r :
4sin2 =7 + sin2 2=7 + sin2 3=7; 4sin2 =9 + sin2 2=9 + sin2 4=9 and 3 4 sin2 =3; that is, 9.
9. Let w, a, b, c be distinct real numbers with the property that there exist real numbers x, y , z for which the following equations hold:
x 2+y 2+z 2 xa3 + yb3 + zc3 xa4 + yb4 + zc4 xa + yb + zc Express w in terms of a, b, c.
1 1 1 = = = = 1
w2 w3 w4 :
Solution by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. As known, the following determinant must vanish:
a2 b2 c2 w2 a3 b3 c3 w3 a4 b4 c4 w4
Also, from a known expansion theorem 1 for alternant" determinants, we have
1
w , aw , bw , ca , ba , cb , cwbc + ca + ab + abc = 0 : Hence, w = a, or b, or c, or ,abc= bc + ca + ab .
Reference 1 T. Muir, A Treatise on the Theory of Determinants, Dover, N.Y., p. 333, 337. Now we turn to readers' solutions of problems from the December 1997 number of the Corner. We give some solutions to problems proposed to the Jury but not used at the 37th International Olympiad at Mumbai, India 1997; 450 453 .
460 that
1. Let a, b and c be positive real numbers such that abc = 1. Prove
ca ab + b5 + bc + bc + c5 + a5 + ca 1 : 5 + ab 5 a +b c
5
When does equality hold? Solutions by Pierre Bornsztein, Courdimanche, France; and by Panos E. Tsaoussoglou, Athens, Greece. We give the solution by Tsaoussoglou. We note rst that
a5 + b5 a3 + b3
a2 + b2
; 2 2 2 since a5 , a3 b2 , a2 b3 + b5 = a , b2a + ba2 + ab + b2 0 with equality if and only if a = b. Similarly a3 + b3 a + b
a2 + b2
2 2 2 3 , a2 b , ab2 + b3 = a , b2 a + b 0, with equality if and because a only if a = b. Thus a5 + b5 a3 + b3
a2 + b2
ab a3 + b3
2 2 a + b
a2 2 b2
a2 b2a2+ b + : ab 2 2 2 ab ca + bcb + bcbc + bc + cac + aca + ca 1 ; aba + bab + ab c
1 1 + bcb + c + abc + cac + a + abc 1 : aba + b + abc 1 1 1 + bca + b + c + caa + b + c 1 ; aba + b + c 1
It is enough, therefore, to prove
or
Equivalently, or
Again, because abc = 1 we get
a b + abca + b + c + abca + b + c 1 : abca + b + c a + b+ c 1; a+b+c
c
461 which is true. The equality requires a = b = c = 1. 2. Let a1 a2 an be real numbers such that for all integers k 0,
ak + ak + + ak 0 : 1 2 n Let p = maxfja1 j; : : : ; jan jg. Prove that p = a1 and that x , a1 x , a2 x , an xn , an 1 for all x a1 .
Solution by Pierre Bornsztein, Courdimanche, France. Pour k 2 N , on note Sk = ak + + ak . 1 n On a S1 0, donc na1 s1 0 et alors a1 0. Si an 0, a1 an 0, donc p = a1 . Par contre, si an 0, par l'absurde, supposons que p 6= a1 . Alors p = jan j. Pour i 2 f1; : : : ; n,1g, si jai j
ai k 0. Cependant, si jaij = jan j, limk!1 an = 1. Par suite pour
jan j, limk!1
ai k
an
=
n,1 ai
k n X X ai
k = 1+ ; i=1 an i=1 an on a limk!1 Tk = l avec l 2 1; +1. En particulier, il existe k 2 N tel que T2k+1 0 et comme an 0 on a a2k+1 0, d'ou s2k+1 = a2k+1 T2k+1 0, n n qui est une contradiction. Par cons
quent, p = a1 . e Soit x a1 . Alors x , ai 0 pour i = 1; 2; : : : ; n.
Tk
Pn x , ai
n,1 i=2 x , ai n,1 i=2 n ! 1 X a n,1 : = x, n,1 i i=2 P Or s1 0, donc a1 , n=2 ai , et alors i n Y X a1
n,1 = n,1 n , 1
a1
i xn,1,i : x , ai x + n , 1 i n,1 i=2 i=0
n Y
D'apr s AM GM e
462
, 1 Mais pour i 2 f1; 2; : : : ; n , 1g, n,1 n,1i n,1n,2:::n,i 1: i n,1i Donc n,1 n Y X
et alors
i=2 n Y
x , ai
i=0
ai1xn,1,i
n,1 X i=0
i=1
x , ai x , a1
ai1xn,1,i
= xn , an : 1
3. Let a
2 be given, and de ne recursively:
a0 = 1; a1 = a; an+1
Show that for all integers k
! a2 , 2 a : = a2 n n n,1
0, we have 1 + 1 + 1 + + 1 1
2 + a , pa2 , 4 : a0 a1 a2 ak 2
Solution by Pierre Bornsztein, Courdimanche, France. Comme a 2, il existe b 1 tel que a = b + 1 et on a b c. .d. a
et comme le produit des racines est 1, d'ou
b2 , ab + 1 = 0 ; a + pa2 , 4 b 1 ; b= 2
1 = a , pa2 , 4 b 2
1 2 + a , pa2 , 4 = 1 + 1 : 2 b Pn 1 pour n 1. On pose Sn = i=0 ai On veut donc prouver que Sn 1 + 1 . b 1 1 2 + 1 et par une r
currence rapide Or a1 = b + b , donc a2 = b + b b b e pour n 1 : 1 b2 + 1
: : : b2n, + 1
: a = b +
2
n
b
b
1
2
b2n,
1
463 On en d
duit que e
n 2i ,1 X Sn = 1 + b2 b 1 + b2 + 1b b2i + 1 : + i=2 1 b2i Or = b2 + 1 +12i + 1 b2 + 1 b2i, + 1 b 2i = b2 + 1 b b2i + 1 + b2 + 1 1 b2i + 1 ; d'ou pour i 2 b2i,1 b2 + 1 b2i + 1
= 1 b2 + 1 1b2i, + 1 , b2 + 1 1 b2i + 1 b
1 1
et donc pour n 2
! b2 + 1 b4 + 1 : : : b2n + 1 an = b b2 b2n, b2 + 1b4 + 1 b2n + 1 = b2n,1
1
et ainsi
1 b + 1 1 , Sn = 1 + b2 + 1 b b2 + 1 b2 + 1 b2n + 1 1 + b2 b 1 + bb21+ 1 + b +1, b = 1 + b2 + 1 b b2 + 1
Sn
1+ 1 b
pour
d'ou et comme
n 2;
1 S1 = 1 + a = 1 + b2 b 1 1 + 1 ; + b le r
sultat est vrai pour tout n 1. e Remarque : Puisque b 1, limn!+1 b2n = +1, et donc limn!1 Sn = 1 1 + b.
a Prove that xn , a1 xn,1 , , an,1 x , an = 0 has precisely one positive real root.
4. Let a ; a ; : : : ; an be nonnegative real numbers, not all zero.
1 2
464 b Let A = n=1 aj and B = n=1 jaj , and let R be the positive real j j root of the equation in a. Prove that AA RB . Solution by Pierre Bornsztein, Courdimanche, France. a Pour x 0, xn , a1 xn,1 , : : : , an = 0 est equivalent a
a , , an = 0. 1, x n x P a Si l'on pose f x = 1 , n=1 xii , alors : i P + 0 i f est d
rivable sur R et pour x 0, f 0x = n=1 xiai e i i puisque ai 0, non tous nuls.Donc, f est continue, et strictement croissante sur R+ . ii De plus lim f x = ,1; x!+1 f x = 1 . If existe un lim x!0 unique r
el strictement positif, not
R, tel que f R = 0. e e P b Pour j 2 f1; : : : ; ng, on pose xj = aj 0 et alors n=1 xj = 1. j A + , on a Comme ln est une fonction concave sur R
1 +1 +
P
P
0n 1 A
X AA xj ln Rj ln @ xj Rj j =1 0j=1 1 n Xa = ln @ Rjj A j =1
n X
= ln1 cf. a = 0:
Donc c. .d. a ou encore
n X j =1
xj lnA , jxj lnR 0 ;
lnA A lnR B lnAA lnRB
d'ou le r
sultat suit. e 5. Let P x be the real polynomial, P x = ax3 + bx2 + cx + d. Prove that if jP xj 1 for all x such that jxj 1, then
jaj + jbj + jcj + jdj 7 :
Solution by Pierre Bornsztein, Courdimanche, France. On pose sP = jaj + jbj + jcj + jdj 0. Quitte a changer P en ,P et ou x en ,x on peut toujours supposer que a 0 et b 0.
465 On a P 0 = d, P 1 = a + b + c + d, P ,1 = ,a + b , c + d, 1 1 8P 2 = a + 2b + 4c + 8d, 8P , 2 = ,a + 2b , 4c + 8d, d'ou 1 1 a = 2 P 1 , 2 P ,1 + 4 P 2 , 4 P , 2 3 3 3 3 1 1 b = 2 P 1 + 2 P ,1 , 2 P 2 , 2 P , 2 3 3 3 3 1 1 c = , 1 P 1 + 1 P ,1 + 4 P 2 , 4 P , 2 6 6 3 3 1 1 d = , 1 P 1 , 1 P ,1 + 2 P 2 + 2 P , 2 = P 0 : 6 6 3 3 Maintenant si c 0 et d 0, S P = a + b + c + d = P 1 1. Cependant si c 0 et d 0, on obtient sP = a + b + c , d = a + b + c + d , 2d = P 1 , 2P 0 d'ou S P jP 1j + 2jP 0j = 3. On proc de au cas c 0 et d 0, ou e SP = a + b , c + d = a + b + c + d , 2c et d'apr s les relations pour a, b, c, d cihaut, e
1
SP = 4 P 1 , 1 P ,1 , 8 P 1 + 8 P , 2 3 3 3 2 3 4+1+8+8; 3 3 3 3 c. .d. a SP 7 : En n, si c 0 et d 0, S P = a + b , c , d, et suivant la m^ me proc
dure, e e on obtient 5 P 1 , 4P 1
+ 4 P , 1
SP = 3 2 3 2
5 +4+ 4; 3 3 c. .d: a SP 7 : Finalement, dans tous les cas S P 7.
Remarque : On peut montrer voir Olympiades su
doises 1965 nale que e pour un tel polyn^ me, on a toujours jaj 4 ce qui est imm
diat avec o e l'expression de a cihaut. Ces deux in
qualit
s sont simultan
ment v
ri ees par e e e e
P x = 4x3 , 3x, qui respecte jP xj 1 pour jxj 1.
466
7. Let f be a function from the set of real numbers R into itself such that for all x 2 R, we have jf xj 1 and
Prove that f is a periodic function that is, there exists a nonzero real number c such that f x + c = f x for all x 2 R. Solutions by Mohammed Aassila, Strasbourg, France; and by Pierre Bornsztein, Courdimanche, France. We give Aassila's solution. We will prove that f is 1periodic. We have
f x + 13 + f x = f x + 1 + f x + 1 : 42 6 7
l
f x+ l+1 +f x+ m + 7 = f x+ m +f x+ m + l+1 : 7 6 6 6 7 Similarly when l runs 1; 2; : : : ; n , 1, n 2 N and adding these equations, we obtain n
m
m n
f x + 7 + f x + 6 = f x + f x + 6 + 7 :
We choose n = 7 and m = 6, and nd
If k runs through 1; 2; : : : ; m , 1, where m 2 N, and adding these equations we obtain
k l + 1
k + 1 l
f x+ 6 + 7 +f x+ 6 + 7 k l
k + 1 l + 1
= f x+ 6 + 7 +f x+ 6 + 7 :
2f x + 1 = f x + f x + 2 : This means that the sequence f x + n is an arithmetic sequence with common di erence f x + 1 , f x. But, since f is bounded, we must have f x + 1 , f x = 0. Hence 1 is a period of f . Finally, 1 is the best"
period because the function
satis es all the hypotheses of the problem. Here, fxg denotes the fractional part of x. 8. Let the sequence an, n = 1; 2; 3; : : : ; be generated as follows: a1 = 0, and for n 1, Here t means the greatest integer less than or equal to t.
f x = f6xg + f7xg 2
an = a n=2 + ,1nn+1=2 :
467 a Determine the maximum and minimum value of an over n 1996 and nd all n 1996 for which these extreme values are attained. b How many terms an, n 1996, are equal to 0? Solution by Pierre Bornsztein, Courdimanche, France. Parti a On v
ri e facilement que a1 = 0, a2 = ,1, a3 = 1 et, pour e
k1
a4k = a2k + 1 9 a4k+1 = a2k , 1 = 1 a4k+2 = a2k+1 , 1 ; a4k+3 = a2k+1 + 1 Pour p 2 N , on pose Ep = f2p ; 2p + 1; : : : ; 2p+1 , 1g. Pour p = 1, on pose '1 2 = ,1, '1 3 = 1, et pour p = 2, on pose '24 = ,1; 1, '25 = ,1; ,1, '26 = 1; ,1, '27 = 1; 1. On proc de a construire 'p : Ep ! Fp pour p 2 comme suite ou e Fp = f"1; "2; : : : ; "p j 8i 2 f1; : : : ; pg, "i 2 f,1; 1gg. Alors pour n 2 Ep+1 n 2 E ; et on pose ' n = " ; : : : ; " ; "
ou
2
p
p+1
1
p p+1
"1; : : : ; "p = 'p
n
2
et
"p+1 = ,1nn+1=2 :
On construit ainsi une application 'p+1 : Ep+1 ! Fp+1 . Propri
t
1. Pour p 1, 'p est une bijection de Ep sur Fp. ee Preuve. Par r
currence sur p, '1 est clairement une bijection de E1 sur F1 . e Supposons que 'p soit une bijection de Ep sur Fp pour p 1 x
. e Soient n, k dans Ep+1 tels que 'p+1 n = 'p+1 k. On peut toujours supposer que n k. Par l'absurde, si n k, 'p+1 n = 'p+1 k entraine 'p n = 'p k et comme 'p est bijective, n = k , d'ou n = k + 1, car 2 2 2 n k 2. n, k sont entiers et si n k + 2 alors 2 2 Or si k est impair, k = 2q + 1 alors n = 2q + 2 et n = q + 1 6= q = k , 2 2 qui est contradictoire. Si, cependant, k = 4l, alors n = 4l + 1 d'ou kk2 = 2q 4q + 1, qui est +1 pair et nn2+1 = 2q + 14q + 1 qui est impair. Donc "p+1 k 6= "p+1 n qui est aussi contradictoire. Quand k = rq + 2, alors n = 4q + 3 d'ou kk2 = 2q + 14q + 3 qui +1 est impair et nn2+1 = 4q + 32q + 2 qui est pair, et on obtient la m^ me e contradiction.
468 Finalement n = k, et 'p+1 est injective. Car Ep+1 et Fp+1 ont la m^ me e cardinalit
, 2p , 'p+1 est bijective. e Propri
t
2. 8p 2 N, p 1, 8n 2 Ep, 'p n = "1; : : : "p . Alors ee an = Pp=1 "i. i Preuve. R
currence imm
diate sur p en utilisant la construction de 'p et e e an = a n=2 + ,1nn+1=2. Propri
t
3. 8p 2 N , 8n 2 Ep , an p mod 2. ee Preuve. Par r
currence sur p. C'est imm
diat pour p = 1. Supposons le e e r
sultat vrai pour p 1 x
. Alors pour n 2 Ep+1 on a n 2 Ep et e e 2 an = a n=2 + ,1nn+1=2 a n=2 + 1 p + 1 mod 2 d'ou le r
sultat e au rang p + 1. Propri
t
4. Pour n 1, a2n,1 = n , 1. ee Preuve. Par r
currence sur n. C'est imm
diat pour n = 1 et n = 2. Suppoe e sons le r
sultat vrai pour n 2 x
. Alors, d'apr s 1 e e e
a2n ,1 = a42n, ,1+3 = a22n, ,1+1 + 1 = a2n ,1 + 1 = n; d'ou la propriet
au rang n + 1. e 10 = 1024 On a 2 1996 2047 = 211 , 1. Or si n 29 , 1, n = 1 ou n 2 Ep avec p 8, donc an = 0 ou an = 'pn 8 cf P2. Cependant si n 2 E9 , an = '9 n 9 d'apr s P2 avec egalit
ssi n = e
e 210 , 1 d'apr s P4 et car '9 est bijective. e Si n 2 E10, an = '10 n 10 et an est pair cf. P2 et P3 avec an = 10 ssi n = 211 , 1 P4 et '10 bijective. d'ou an 9 pour n 1996. Finalement la valeur maximale de an , pour n 1996 est 9, avec an = 9 ssi n = 1023. On passe des valeurs de an pour n 2 Ep a celles de an pour n 2 Ep+1 en ajoutant ou en retranchant 1. On peut ainsi pr
voir que l'on obtiendra e les valeurs minimales en utilisant une suite Un d'indices avec U1 = 2 et pour n 1 2Un + 1 si Un est pair, Un+1 = 2Un si Un est impair.
+1 1 1
Propri
t
5. Pour p 1, ee a Up et p sont des entiers de parit
s contraires. e
469 b Up = 2p 3 ,1 si p est pair ; Up = sp 3 ,2 si p est impair. Preuve. R
currence sur p. e Propri
t
6. Pour n 1, aUn = ,n. ee h i Preuve. Il est clair que pour tout n 1, Un2 = Un par d
nition de e Un+1. Pour n 1, quand n est pair
+2 +2 +1
UnUn + 1 = 2n+2 , 1 2n+1 + 1 est impair ; 2 3 3 et de m^ me quand n est impair e UnUn + 1 = 2n+1 , 1 2n+2 + 1 qui est aussi impair :
Donc pour tout n 1 Alors pour tout n 1,
1
2
3
3
,1 Un Un
2 +1
+1
= ,1 :
et comme aU = a2 = ,1, une r
currence imm
diate sur n permet de e e conclure que si n 210 , 1, alors n = 1 ou n 2 Ep avec p 9 donc an = 0 ou an = 'pn ,9 cf. P2 si n 2 E10, alors an = '10n ,10 et an est pair cf. P2 et P3 . De plus U10 = 1365 cf. P5 donc U10 2 E10, et a1365 = aU = ,10 cf. P6 an = ,10 ssi n = 1365 :
10
aUn = aUn , 1
Finalement la valeur minimale de an , pour n 1996, est ,10, avec an = ,10 ssi n = 1365. Parti b Remarquons d'abord que a1 = 0. Si n 2 Ep , 'p n = "1; : : : "p ou "i 2 f,1; 1g et an = "1 + + "p cf. P2 . Donc an = 0 ssi il y a autant de "i = 1 que de "j = ,1. Comme 'p est une bijection de Ep sur Fp, il y a pautant de n 2 Ep pour p x
e tels que an = 0 que de fa ons de choisir 2 places pour les +1" parmi les p c places possibles. Ainsi, d'un part si p est impair 8n 2 Ep , an 6= 0 on retrouve P3. D'autre part, 1 p = 2 donne C2 = 2 possibilit
s, e 2 = 6 possibilit
s, p = 4 donne C4 e 3 = 20 possibilit
s, p = 6 donne C6 e
470
4 p = 8 donne C8 = 70 possibilit
s, e 5 p = 10 donne C10 = 252 possibilit
s. e 11 Mais 2 , 1 = 2047 1996. Il reste a d
terminer combien de e n 2 f1997; : : : ; 2047g v
ri ent an = 0. e
Or
a62 = a30 + 1 = a15 = a7 + 1 = 3 on utilise 1, et a63 = a2 ,1 = 5 cf. P4:
6
En utilisant 1 on en d
duit e Puis Et encore
a124 = 4 = a126; a125 = 2; a127 = 6 : a249 = a251 = a253 = 3; a250 = 1; a252 = a254 = 5; a255 = 7:
a499 = a503 = a505 = a507 = a509 = 4; a504 = a508 = a510 = 6; a511 = 8; a500 = a502 = a506 = 2; a501 = 0: Mais si n 2 f1997;: : : ; 2047g alors n 2 f998; : : : ; 1023g ; 2 1 n 2 f499; : : : ; 511g : c. .d. a
Donc an = 0 ssi ou
2 2
an = a501 + "9 + "10
avec
"9 + "10 = 0 deux possibilit
s, e
an = ak + "9 + "10 avec k 2 f500; 502; 506g et "9 = "10 = ,1 trois possibiliti
s : e Il ya a donc 5 indices n 2 f1997;: : : ; 2047g tels que an = 0 et comme 1 + 2 + 6 + 20 + 70 + 252 , 5 = 346 ; le nombre de n 1996 tels que an = 0 est N = 346. 9. Let triangle ABC have orthocentre H , and let P be a point on its circumcircle, distinct from A, B , C . Let E be the foot of the altitude BH , let PAQB and PARC be parallelograms, and let AQ meet HR in X . Prove that EX is parallel to AP .
471 Solution by Toshio Seimiya, Kawasaki, Japan.
Q A R X
E B H P C
Since AX kBP and A, B , P , C are concyclic we have
XAP = APB = ACB : As AH ? BC and BH ? AC , we get ACB = AHE , so that XAP = AHE : 1 Since APCR is a parallelogram and A, B , P , C are concyclic we have ARC = APC = ABC . As AH ? BC and CH ? AB, we get AHC + ABC = 180 ;
so that Hence A, H , C , R are concyclic. It follows that From 1 and 2 we have
AHC + ARC = 180 :
2
AHR = ACR = CAP : XAE = XAP , CAP = AHE , AHR = XHE : Hence X , A, H , E are concyclic, and AEX = AHX = AHR : Hence we have, from 2, that AEX = CAP . Thus EX kAP .
472
10. Let ABC be an acuteangled triangle with jBC j jCAj, and let O be the circumcentre, H its orthocentre, and F the foot of its altitude CH . Let the perpendicular to OF at F meet the side CA at P . Prove that FHP = BAC . Solutions by Toshio Seimiya, Kawasaki, Japan; and by D.J. Smeenk, Zaltbommel, the Netherlands. We give Smeenk's solution.
C
, 2
!
, 2
Q H F O
, 2
, 2
,
P! A
!
M
B
We denote CFP = OFB = . M is the midpoint of AB . Now
OM = R cos ;
FM = R sin , ;
so that
From the Law of Sines in 4CPF ,
cos tan ' = sin , :
1
so CP : CF = sin ' : cos , '.
CF = 2R sin sin FCP = , 2 + ,' FPC = 2
2
473 With 2
sin sin CP = 2R sin sin 'sin ' = cos2Rcot ' + sin : cos ,
From 1 and 3,
3
1
CQ = a , QB = R 2 sin , sin = ,R cos 2 : sin Furthermore, CH = 2R cos , CO = R. It is easy to verify that CP : CH = CO : CQ; and ,2R sin cos : 2R cos = R : ,R cos 2 :
sin cos 2 sin
We also have
Now
R sin cos CP = sin 2sin2 , cos2 = ,2R sin cos sin cos 2 OQ ? OB; OBQ = , = OQB = : 2
4
5
PCH = OCQ = , : 6 2 From 5 and 6, we have that 4PCH and 4OCQ are similar. Thus PHC = OQC = , . Thus FHP = . 11. Let ABC be equilateral, and let P be a point in its interior. Let the lines AP , BP , CP meet the sides BC , CA, AB in the points A1 , B1 , C1 respectively. Prove that A1B1 B1C1 C1A1 A1B B1C C1A :
Solutions by Toshio Seimiya, Kawasaki, Japan; and by Pierre Bornsztein, Courdimanche, France. We give Seimiya's solution. A b0 c B1 C1 b P c0 B a A1 C a0
474 We put A1 B = a, A1 C = a0 , B1 C = b, B1 A = b0 , C1A = c and C1B = c0. Then we have by Ceva's Theorem abc = a0 b0c0 : 1 Since B1 AC1 = 60 we have
2 B1C1 = b02 + c2 , 2b0c cos 60 = b02 + c2 , b0 c b0 c:
Similarly we have
2 and A1 B1 a0 b : Multiplying these three inequalities, we get 2 2 B1C1 C1A2 A1B1 b0c c0a a0 b: 1
C1A2 c0 a 1
2
From 1 and 2 we have Thus we have That is
2 2 B1C1 C1A2 A1B1 a2b2c2 : 1
B1C1 C1A1 A1B1 abc : B1C1 C1A1 A1B1 A1B B1C C1A :
That completes the Corner for this issue. Send me your nice solutions as well as Olympiad contests and materials.
Professor Dan Pedoe
The Editors of CRUX with MAYHEM are saddened to learn that Dan Pedoe died a few weeks ago, on 27 October 1998. He made major contributions to the success of CRUX during its rst ten years, and always maintained his interest in the journal. Readers may be interested to read his autobiographical notes in the recent COLLEGE MATH. JOURNAL 29:3 May 1998 170188.
475
BOOK REVIEWS
Edited by ANDY LIU
This is my last column as editor of the Book Reviews column. Looking back over the past ve years, I am happy that many books have been brought to the attention of the readers, obscure titles as well as famed tomes. While some are of marginal quality, and it is not our duty to just sing praises here, the great majority are outstanding publications. They are tremendous resources in our hands. Let us make good use of them. Grateful acknowledgement is due to the publishers who provided the volumes for review in the rst place. The Mathematical Association of America should be singled out for supplying the column with every title it has published in popular mathematics, and it is an ample and steady source. Gratitude is owed to all reviewers, solicited from as wide a spectrum as possible, particularly geographically. My friend and mentor Murray Klamkin regularly graced these pages with his muchesteemed opinion, and deserves a special thank you. Book Reviews columns as a rule do not generate much fan mail. However, I have been pleasantly surprised how often I hear remarks from satised readers, o ering corrections, suggesting titles, and providing constructive criticism. Recently, a reader from France called and pointed out that the Mathematical Association of America has not acquired Martin Gardner's Sixth Book of Mathematical Diversions from Scienti c American, as claimed in the Minireviews in the February issue, but the right belongs to the University of Chicago Press.
The Last Recreations by Martin Gardner, published by SpringerVerlag, New York, 1997, ISBN 0387949291, hardcover, 392+ pages, $25.00. Reviewed by Andy Liu. This is the fteenth and last of Martin's Scienti c American anthologies, but the rst with this publishing giant. This may have something to do with the move from W. H. Freeman to SpringerVerlag of Jerry Lyons, the outstanding editor of mathematics books. The subtitle is Hydras, Eggs, and Other Mathematical Mysti cations. The material is from the Scienti c American columns from December 1979 to his retirement in December 1981, though that particular column, titled The La er Curve", was in an earlier volume, Knotted Doughnuts. Included also are three chapters Martin later wrote as guest columnist, on August and September 1983 as well as June, 1986. The last chapter, titled Trivalent Graphs, Snarks, and Boojums", is a drastically revised version of
476 his April 1976 column, which he had originally intended to leave out, in view of rapid advancements on the status of the Four Colour Problem. The reviewer is glad to see its inclusion. Thus only the October 1975 column, titled Extrasensory Perception by Machines", is missing from the fteen volumes. It is anthologized in a science ction puzzle collection. By now, everything that can be said about Martin's writing has been said many times over. Buy this book at once and treasure it! Although Martin continues to contribute to various journals, this is The Last Recreations under his own banner! The EditorinChief, Bruce Shawyer, and his predecessor, Bill Sands, would like to take this opportunity of thanking Andy Liu for his long and sterling work in this column.
Thank you very much, Andy. All the best!"
Peter Hurthig, Columbia College, Burnaby, BC.
The 1998 Crossword Puzzle
Four Mathematicians, a Detective, a King and a French Lady
1. 4. 5. 9. 10. 11. 12.
Across: A word on a towel. An abbreviated diophantine solution. A Russian who solved number problems. King Zog's country. A graduate student who impressed Gauss! Ruth Rendell's inspector to his friends. One third of a communist or one half of a y.
1 4 5 9 10 11 12 6 7 8 2
Down: 1. A German whose problems were numbered. 2. Brands with dishonour. 3. British weight. 5. A bony sh. 6. A high priest. 7. She wrote Delta of Venus. 8. Professor Pedoe.
3
477
THE SKOLIAD CORNER
No. 34 R.E. Woodrow
We begin this issue with the problems of the Old Mutual Mathematical Olympiad 1991 Final Paper 1 and Final Paper 2. The Preliminary Round of this contest is in a multiple choice format and I plan to give an example with the 1992 paper in the new year. My thanks go to John Grant McLoughlin, of the Faculty of Education, Memorial University of Newfoundland, for collecting this contest and forwarding it for use in the Corner.
OLD MUTUAL MATHEMATICAL OLYMPIAD 1991 Final Paper 1
Time: 2 hours
1. In the gure shown ABC and AEB are semicircles and F is the midpoint of AC and AF = 1 cm. Find the area of the shaded region.
E B
A
2. form? 3.
What is the value of 17 , 12 2 + 17 + 12 2 in its simplest
p
F
p
p
C
p
In a certain mathematics examination, the average grade of the students passing was x, while the average of those failing was y . The average of all students taking the examination was z . Find the percentage who failed in terms of x, y and z . 4. In the gure shown AB = AD = p130 cm and BEDC is a square.
D A E B C
478 Also the area of 4AEB = area of square BEDC . Find the area of BEDC .
Final Paper 2
Time: 2 hours
1. If the pattern below of dot gures is continued, how many dots will there be in the 100th gure?
w w w w w w w w w w w w w w w w w w
w w
w
w
w
w
2. It is required to place a small circle in the space left by a large circle as shown. If the radius of the large one is a and that of the small one is b, nd the ratio a=b.
3. Find all solutions to the simultaneous equations
x + y = 2; xy , z2 = 1 ;
and prove that there are no other solutions. 4. If a, b, c and d are numbers such that
a+b c + d; b+c d + e; c+d e + a; and d + e a + b; prove that the largest number is a and the smallest is b.
479
5. The diagram below rotated through 90 shows a container whose lower part is a hemisphere and whose upper part is a cylinder.
20cm 6 ? 220cm

The cylindrical part has internal diameter of 20 cm and is 220 cm long. Water is poured into it and rises to a height of 20 cm in the cylindrical part. The top is then sealed with a at cover and the container is turned upside down. The water is now 200 cm high in the cylindrical part. i Calculate the volume of the hemisphere in terms of . ii Find the total height of the container. Note: The volume of a sphere of radius R is 4 R3 . 3 Last number we gave the problems of the British Columbia Colleges, Senior High School Mathematics Contest Preliminary Round, 1998. Next we give the o cial" solutions, courtesy of Jim Totten, The University College of the Cariboo, and an organizer of the contest.
BRITISH COLUMBIA COLLEGES Senior High School Mathematics Contest Preliminary Round 1998
Time: 45 minutes
integers. It follows that c equals: d. Solution. By comparing the expression 1998 = n , 1nn10n + c with 1998 = 2 33 37, we get n = 3, since 33 is the only power dividing 1998. Hence, 10n + c = 37, which gives c = 7. 9 2. The value of the sum log 1 + log 2 + log 3 + + log 10 is: a. 2 3 4 Solution. We have: 2 9 1 9 1 log 1 + log 3 + log 3 + + log 10 = log 2 2 3 10 = log 10 = ,1. 2 4 34
1. The integer 1998 = n , 1nn10n + c where n and c are positive
3. Four basketballs are placed on the gym oor in the form of a square with each basketball touching two others. A fth basketball is placed on top of the other four so that it touches all four of the other balls, as shown.
480
If the diameter of a basketball is 25 cm, the height, in centimetres, of the centre of the fth basketball above the gym oor is: d. Solution. Let us rst take a look at the four basketballs lying on the oor:
C
A
B
We nd AC = AB 2 + BC 2 = 252 + 252 = 25 2. Now, in order to nd the height of the centre of the fth ball, we analyze the crosssection of the pyramid by a vertical plane passing through A and C :
D A E F C
p
p
p
The height DF = DE + EF , where EF = 25 and DE = AD2 , 1 AC 2 2 2 q p p p2. 2 , 1 25 22 = 25 2. Hence, DF = 25 1 + = 25 2 2 2 4. Last summer I planted two trees in my yard. The rst tree came in a fairly small pot and the hole that I dug to plant it in lled one wheelbarrow load of dirt. The second tree came in a pot, the same shape as that of the rst tree, that was oneandathird times as deep as the rst pot and oneandahalf times as big around. Let us make the following assumptions: i The hole for the second tree was the same shape as for the rst tree. ii The ratios of the dimensions of the second hole to those of the rst hole are the same as the ratios of the dimensions of the pots.
q
481 Based on these assumptions, the number of wheelbarrows of dirt that I lled when I dug the hole for the second tree was: c. Solution. To solve the problem we need to nd the ratio of the volume of the second pot to the volume of the rst. We can assume that both pots are generalized cylinders with bases of the same shape. It is convenient to think that the second pot was obtained from the rst by stretching it horizontally and vertically. The change of the perimeter is proportional to the change of horizontal dimensions, so that both horizontal dimensions have been stretched oneandahalf times. Therefore, the area of the base has 1 1 increased 1 2 1 1 = 9 times. Since the height has been stretched 1 3 times, 2 4 9 1 the volume increased 4 1 3 = 3 times. 5. You have an unlimited supply of 5gram and 8gram weights that may be used in a pan balance. If you use only these weights and place them only in one pan, the largest number of grams that you cannot weigh is: b. Solution. If W is the weight in grams then W can be weighed if W = 5x + 8y, where x and y are nonnegative integers. Furthermore, y can be written in the form 5z + r, where z is a nonnegative integer and r is either 0; 1; 2; 3, or 4. Thus W = 5x + 8 + 80, W = 5x + 8 + 81, W = 5x +8+82, W = 5x +8+83, or W = 5x +8+85. These can be simpli ed to W = 5k, W = 5k +1+3, W = 5k +3+1, W = 5k +4+4 and W = 5k + 6 + 2. By putting k = x + 8z , we conclude that W can be weighed if W = 5k + 8r, with k 0 and r 4. Thus, according to the value of r, W = 5k + 80, W = 5k, W = 5k + 1 + 3, W = 5k + 3 + 1, W = 5k + 4 + 4, W = 5k + 6 + 2. On the other hand, W as well as every integer can take exactly one of the forms: 5K , 5K + 1, 5K + 2, 5K + 3, or 5K + 4. By comparing these forms with the expressions for W that can be weighed, we conclude: i if W is of the form 5K , then it can be weighed; ii if W is of the form 5K + 1, then it can be weighed if K = k + 3; that is, when K 3; iii if W is of the form 5K + 2, then it can be weighed if K = k + 6; that is, when K 6; iv if W is of the form 5K + 3; then it can be weighed if K = k + 1; that is, when K 1; v if W is of the form 5K + 4, then it can be weighed if K = k + 4; that is, when K 4. Consequently, the largest values of W that cannot be weighed in categories: ii, iii, iv, and v, are 52 + 1 = 11, 55 + 2 = 27, 50 + 3 = 3, 53 + 4 = 19, respectively. The largest of them is 27. 6. If all the whole numbers from 1 to 1,000,000 are printed, the number of times that the digit 5 appears is: c.
482 Solution. Method I. The number of times the numeral 5 appears will not change if instead of printing the numbers from 1 to 1,000,000 we will print the numbers from 0 to 999,999. Furthermore, the number of appearances of 5 will not change if we complete the decimal representation of each number to a sixdigit sequence, by writing some zeros in front of the number, if necessary. For example, 1998 will be represented by 001,998. The printout of all of the sixdigit sequences will have a total of 6 106 digits and each digit will appear the same number of times. This gives 610 = 600,000 10 appearances of 5. Method II. Standard, but more tedious. Again, as above, we represent the numbers by sixdigit sequences. At rst, we count the number of sequences that 6have precisely k copies of 5, for 1 k 6. The numeral 5 , can appear in k di erent positions in such a sequence, while in each of the remaining , , k positions we can have one of the nine ,digits other than 5. 66 6 This gives k 96,k di erent sequences. They provide k k 96,k copies of 5. Hence, the number of appearances of 5 in all of the sequences is given by the sum:
6
1 6 95 + 2 6 94 + 3 6 93 + 4 6 92 + 5 6 91 + 6 6 90 = 600,000 : 1 2 3 4 5 6
7. The perimeter of a rectangle is x centimetres. If the ratio of two adjacent sides is a : b, with a b, then the length of the shorter side, in centimetres, is: e. Solution. If a1 and b1 are the lengths of the longer and the shorter bx sides of the rectangle, then b1 : x = b : 2a + b . This gives b1 = 2a+b .
is: e p Solution. From x=x x = xpxx we get that that xxx = = xx1=2 x , which simpli es to xx = x3=2x . This has the trivial solution x = 1. If x 6= 1 then the exponents must be equal; that is x3=2 = 3 x. By squaring 2 9 both sides, x3 = 9 x2 . Therefore, x = 4 , since x is positive. The sum of the 4 solutions is 1 + 9 = 3 1 . 4 4
1 2 3 2
8. The sum of the positive solutions to the equation xxpx = xpxx
9. Two circles, each with a radius of one unit, touch as shown. See diagram on next page. AB and CD are tangent to each circle. The area, in square units, of the shaded region is: d. Solution. The square ABCD has area of 2 2 = 4 square units. To nd the shaded area we need to remove the area of two semicircles from the area of ACBD. Thus, the area of the shaded region is 4 , 2 1 = 4 , 2 square units.
483
A
B
C D 10. A parabola with a vertical axis of symmetry has its vertex at 0; 8 and an x intercept of 2. If the parabola goes through 1; a, then a is: c Solution. The parabola has another x intercept of ,2 symmetric to the x intercept of 2 with respect to the vertical axis of symmetry passing through 0; 8. This implies that the equation of the parabola is of the form y = Ax , 2x + 2. The parabola has the vertex at 0; 8: therefore 8 = A0 , 20 + 2. This gives A = ,2, so that the equation of the parabola is y = ,2x , 2x + 2. If a is the value of y corresponding to x = 1, then a = y = ,21 , 21 + 2 = 6. 11. A ve litre container is lled with pure orange juice. Two litres of juice are removed and the container is lled up with pure water and mixed thoroughly. Then two litres of the mixture are removed and again the container is lled up with pure water. The percentage of the nal mixture that is orange juice is: d. Solution. Initially, there were 5 litres of juice, and after the removal of the rst two litres, 3 litres of juice are left. After mixing this amount with 2 litres of pure water, each litre of the mixture will contain 3 litres of pure juice. 5 3 Thus, when we remove 2 litres of the mixture, 3 , 2 5 = 9 litres of pure 5 juice will remain. This gives the concentration of 9 5 100 = 36. 5 12. The lengths of the sides of a triangle are b + 1, 7 , b and 4b , 2. The number of values of b for which the triangle is isosceles is: b Solution. Potentially, we have three possibilities for the triangle to be isosceles: 1 b + 1 = 7 , b, 2 b + 1 = 4b , 2, and 3 7 , b = 4b , 2. The rst possibility gives b = 3 and, consequently, 4, 4, and 10 as possible lengths of the sides of the triangle. The second possibility gives b = 1, and 2, 2, and 6 as possible lengths. Finally, the third possibility gives b = 9 , and 5 14 26 , 5 , and 26 as the corresponding lengths. However, the sum of lengths of 5 5 any two sides in a triangle is greater than or equal to the length of the third side. This leaves b = 9 as the only possibility. 5
484
13. The number of times in one day when the hands of a clock form a right angle is: d. Solution. Suppose that we measure the time from midnight to the next midnight. If 1 is the angle made by the minute hand and 2 the angle made by the hour hand turned clockwise from 12 then 1 = 2t and 2 = 2 t = 1 t, where t represents time measured in hours. The angle between 12 6 1 both hands is 90 when 1 , 2 = 2 + 2k or 1 = 2 = 3 t + 2k . 2 6 12 12 By solving these equations for t, we get t = 11 + 11 k or t = 18 + 11 k. 11 Since 0 t 24, each of the equations yields 22 solutions in nonnegative integers k. This corresponds to the total of 44 solutions. 14. In my town some of the animals are really strange. Ten percent of the dogs think they are cats and ten percent of the cats think they are dogs. All the other animals are perfectly normal. One day I tested all the cats and dogs in the town and found that 20 of them thought that they were cats. The percentage of the dogs and cats in the town that really are cats is: a Solution. Let d and c denote the number of dogs and cats in the town, respectively. Then the number of animals who responded that they are cats is 0:2d + c. This corresponds to 0:1d + 0:9c, according to the truthfulness of our animals. By equating both expressions we get 0:1d = 0:7c, or d = 7c. c c Thus, the percentage of cats is c+d 100 = c+7c 100 = 100 = 12:5. 8 15. A short hallway in a junior high school contains a bank of lockers numbered one to ten. On the last day of school the lockers are emptied and the doors are left open. The next day a malicious math student walks down the hallway and closes the door of every locker that has an even number. The following day the same student again walks down the hallway and for every locker whose number is a multiple of three closes the door if it is open and opens it if it is closed. On the next day the student does the same thing with every locker whose number is divisible by four. If the student continues this procedure for a total of nine days, the number of lockers that are closed after the ninth day is: d. Solution. A locker number k changes its status from open to closed or viceversa nk times, where nk denotes the number of distinct divisors of k that are greater than 1 and less or equal to 10. Thus, n1 = 0, n2 = 1, n3 = 1, n4 = 2, n5 = 1, n6 = 3, n7 = 1, n8 = 3, n9 = 2, n10 = 3. Since all lockers are initially open, the kth locker is closed after the period of nine days if nk is odd. Therefore, seven lockers will be closed.
That completes the Skoliad Corner for this issue. Please send me your comments and suggests for the evolution of the Corner. Also, I need suitable contest material at the preOlympiad level for use in future Corners.
485
MATHEMATICAL MAYHEM
Mathematical Mayhem began in 1988 as a Mathematical Journal for and by High School and University Students. It continues, with the same emphasis, as an integral part of Crux Mathematicorum with Mathematical Mayhem. All material intended for inclusion in this section should be sent to the Mayhem Editor, Naoki Sato, Department of Mathematics, Yale University, PO Box 208283 Yale Station, New Haven, CT 06520 8283 USA. The electronic address is still mayhem@math.toronto.edu The Assistant Mayhem Editor is Cyrus Hsia University of Toronto. The rest of the sta consists of Adrian Chan Upper Canada College, Jimmy Chui Earl Haig Secondary School, Richard Hoshino University of Waterloo, David Savitt Harvard University and Wai Ling Yee University of Waterloo.
Mayhem Problems
The Mayhem Problems editors are: Richard Hoshino Mayhem High School Problems Editor, Cyrus Hsia Mayhem Advanced Problems Editor, David Savitt Mayhem Challenge Board Problems Editor. Note that all correspondence should be sent to the appropriate editor  see the relevant section. In this issue, you will nd only solutions  the next issue will feature only problems. We warmly welcome proposals for problems and solutions. We request that solutions from the previous issue be submitted in time for publication in issue 8 of 1999.
High School Solutions
Editor: Richard Hoshino, 17 Norman Ross Drive, Markham, Ontario, Canada. L3S 3E8 rhoshino@undergrad.math.uwaterloo.ca
H229. Here is a simple way to remember how many books there are in the Bible. Remember that there are x books in the Old Testament, where x is a twodigit integer. Then multiply the digits of x to get a new integer y, which is the number of books in the New Testament. Adding x and y , you end up with 66, the number of books in the Bible. What are x and y ?
486 Solution by Katya Permiakova, student, Lisgar Collegiate Institute, Ottawa, Ontario. Let x = 10a + b, where a and b are integers satisfying the conditions 1 a 9 and 0 b 9. Then y = ab. Now we are given that 10a + b + ab = 66. Rearranging terms76and solving for b, we get ba + 1 = 66 , 10a, so b = 66a,10a = ,10 + a+1 . Now in order for b to be an integer, a + 1 +1 must divide 76. The only positive divisors of 76 are 1, 2, 4, 19, 38, and 76. Since our choice for a is limited to the integers between 1 and 9, the only possibilities for a are 1 and 3 since that gives us a + 1 = 2 and a + 1 = 4, respectively. If a = 1, then we have b = ,10 + 76 = 28, but this does not satisfy 2 b 9. However, if a = 3, then b = ,10 + 76 = 9, and this is legitimate. 4 Hence x = 39 and y = 27.
Also solved by KEON CHOI, student, A.Y. Jackson Secondary School, North York, Ontario; LINO DEMASI, student, St. Ignatius High School, Thunder Bay, Ontario; and WENDY YU, student, Woburn Collegiate Institute, Scarborough, Ontario.
4 4 chessboard. Dick is telling too many bad jokes, so Cy decides to chase
after him. They take turns moving one square at a time, either vertically or horizontally on the board. To catch Dick, Cy must land on the square Dick is on. Prove that: i If Dick moves rst, Cy can eventually catch Dick. ii If Cy moves rst, Cy can never catch Dick.
H230. Dick and Cy stand on opposite corners on the squares of a
Can you generalize this to a 2m 2n chessboard? Solution. i Place coordinates on the board so that Cy is standing on 0; 0 and Dick is standing on 3; 3. We shall show that after a few moves, Cy can catch Dick on a turn. Regardless of what Dick does on his rst two turns, Cy can move to 1; 1 after two moves. Now it is Dick's turn. At that time, Dick must be on 1; 3, 2; 2, 3; 1, or 3; 3. So if on his next move, Dick goes to either 1; 2 or 2; 1, Cy is standing one square away and so Cy moves into Dick's square on his next move, and catches Dick. So Dick must move to one of 0; 3, 2; 3, 3; 2, or 3; 0. If Dick goes to 0; 3 or 2; 3, then Cy can go to 1; 2, and from here it is easy to see that Dick can last at most two moves before he gets caught since Cy can trap him into a corner. If Dick goes to 3; 2 or 3; 0, then Cy can go to 2; 1, by the same argument, Cy can catch Dick. Thus no matter what, if Dick moves rst, then Cy can eventually catch Dick.
487 i
3 1
;
2 2
;?
3 3
;
1 1
;
0 0
;
6
1 3
;
?
......... ......... ......... ......... Dick ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... Cy !......... ......... ......... ......... ......... ii
D
C
ii Colour the 4 4 board in black and white, as in a regular chessboard so adjacent squares are of di erent colours. Thus, if 0; 0 is a black square, then 3; 3 must be a black square as well. So for each move, if a person is on a square of a certain colour, then he will move to a square of the other colour. Hence, Dick and Cy both start o on a black square. If Cy moves rst, then Cy moves onto a white square while Dick remains on a black square. Then Dick moves to some white square. Now Dick and Cy are both on white squares, and so on Cy's next move he must move onto a black square while Dick remains on a white square. Thus whenever Dick is on a square of a certain colour, Cy is moving to a square of the other colour. And so, on any given move, Cy can never move to a square that Dick is currently on, and so Cy will not be able to catch Dick.
Also solved by KEON CHOI, student, A.Y. Jackson Secondary School, North York, Ontario; LINO DEMASI, student, St. Ignatius High School, Thunder Bay, Ontario; and WENDY YU, student, Woburn Collegiate Institute, Scarborough, Ontario.
H231. Let O be the centre of the unit square ABCD. Pick any point P inside the square other than O. The circumcircle of PAB meets the circumcircle of PCD at points P and Q. The circumcircle of PAD meets the circumcircle of PBC at points P and R. Show that QR = 2 OP . Solution by Wendy Yu, student, Woburn Collegiate Institute, Scarborough, Ontario and Keon Choi, student, A.Y. Jackson Secondary School,North York, Ontario. A D A B P Q P
O Q
q
L
O R
C D C B Construct line L through O parallel to AD and BC . All points on this line are the same distance from A as from B , and the same distance from C as
488 from D. Thus this line contains the centres of the circumcircles of PAB and PCD. Hence, the line L bisects segment PQ. So the point Q must be the re ection of P about the line L, and it follows that OP = OQ. Similarly, if we construct line M through O parallel to AB and CD, then R is the re ection of P about the line M . Hence, OP = OR. Because PQ and PR are perpendicular since the lines L and M are perpendicular, PQR is a rightangled triangle. Furthermore, OP = OQ = OR, which implies that O is the midpoint of the hypotenuse QR. Hence, we have QR = OQ + OR = OP + OP = 2OP , and so QR = 2OP , as desired.
Also solved by LINO DEMASI, student, St. Ignatius High School, Thunder Bay, Ontario.
H232. Lucy and Anna play a game where they try to form a tendigit number. Lucy begins by writing any digit other than zero in the rst place, then Anna selects a di erent digit and writes it down in the second place, and they take turns, adding one digit at a time to the number. In each turn, the digit selected must be di erent from all previous digits chosen, and the number formed by the rst n digits must be divisible by n. For example, 3, 2, 1 can be the rst three moves of a game, since 3 is divisible by 1, 32 is divisible by 2 and 321 is divisible by 3. If a player cannot make a legitimate move, she loses. If the game lasts ten moves, a draw is declared. i Show that the game can end up in a draw. ii Show that Lucy has a winning strategy and describe it. I. Solution by Lino Demasi, student, St. Ignatius High School, Thunder Bay, Ontario. The number 3,816,547,290 has the property that the number formed by the rst n digits is divisible by n, for n = 1, 2, 3, : : : , 10. Thus, if the moves are carried out in this order, then the game can end up in a draw. Here is Lucy's winning strategy. First note that Anna must play an even digit on each of her moves. So Lucy's goal is to play as many even numbers as possible. So Lucy plays a 6 to start. There are three cases to be considered for Anna's second move: 1 If Anna plays a 4 or a 2, then Lucy plays the other on the third move. Anna must now play an even number because her number now has to be divisible by 4, so if Anna plays an 8, then Lucy plays a 0, and Anna loses because on the sixth move, she would have to play an even number and there are none left. If Anna plays a 0, then Lucy plays a 5, and then Anna also loses because she must now play an even number on the sixth move, the only one of which is an 8, but neither 642,058 or 624,058 is divisible by 6. 2 If Anna plays a 0, then Lucy plays a 9. Then Anna must play a 2 to make the fourdigit number divisible by 4. Lucy then plays a 5. Anna must now play an 8 to make her number divisible by 6. Then Lucy can counter with a 3, since 6,092,583 is divisible by 7. The only even number Anna can now play is a 4, but 60,925,834 is not divisible by 8, so she loses.
489 3 If Anna plays an 8, then Lucy plays a 4. Anna's only choice now is a 0. Then Lucy plays a 5. Now Anna's only choice is a 2, but 684052 is not divisible by 6, so she loses. Note that this covers all the cases because Anna must play an even digit on the second move. Thus, Lucy can always force a win. II. Solution by Wendy Yu, student, Woburn Collegiate Institute, Scarborough, Ontario. As before, if the game is played in the following order: 3, 8, 1, 6, 5, 4, 7, 2, 9, 0, then the game will end up in a draw. For Lucy's winning strategy, she can start o with a 4. Then, Anna must counter with an even number. So if she responds with a 2 or an 8, then Lucy's next move is a 0. If Anna's response is a 0 or a 6, then Lucy's next move is a 2. Now, in the case where the number 480 has been written, Anna cannot nd a digit to make a fourdigit number divisible by 4, so she immediately loses. In the other three cases, there is at least one digit that Anna can pick to remain in the game. Thus, after four moves, if the game lasts that long, one of the following numbers will be on the board: 4028, 4208, 4620, or 4628. Then Lucy picks a 5, and in each of those four cases, it will be impossible for Anna to then make a move so that the new sixdigit number is divisible by 6, since the digits she needs are all taken. Thus, if Lucy follows this strategy, she can always force a win.
Also solved by KEON CHOI, student, A.Y. Jackson Secondary School, North York, Ontario.
Advanced Solutions
Editor: Cyrus Hsia, 21 Van Allan Road, Scarborough, Ontario, Canada. M1G 1C3 hsia@math.toronto.edu A205. Find all functions f : R ! R such that f x x and f x + y f x + f y for all reals x and y. Solution. We have f x x ; 1 f x + y f x + f y : 2 Let x = y = 0 in 2. We have f 0 2f 0 which implies 0 f 0. But for x = 0 in 1, we have f 0 0 so f 0 = 0. Now take y = ,x in 2 to get f 0 f x + f ,x for all real x. In other words, ,f x f ,x. But f ,x ,x by 1 so ,f x ,x
490 implies that f x x for all real x. Thus combining this with 1 we have f x = x for all real x. Now it can be easily checked that this function satis es the two conditions 1 and 2 in the problem. A206. Let n be a power of 2. Prove that from any set of 2n , 1 positive integers, one can choose a subset of n integers such that their sum is divisible by n. Solution. Let n = 2m , for some integer m. We will prove this result by mathematical induction on m. For m = 0, n = 1 and the result clearly holds. Now assume the result is true for some arbitrary k 0. In other words, for n = 2k , any set of 2k+1 , 1 positive integers has a subset of n integers whose sum is divisible by n. Now consider any set, A, of 2k+2 , 1 positive integers. Let A1 and A2 bekthe two subsets of A consisting of the rst 2k+1 , 1 integers and the last 2 +1 , 1 integers respectively. By the induction hypothesis, each of these sets has 2k integers whose sum is divisible by 2k . Call the subset of 2k numbers, from A1 , B1 and the subset of 2k numbers, from A2 , B2 . Now call the set of integers remaining from A when the 2k+1 integers from B1 and B2 are removed from A3. Now A3 also has 2k+2 , 1 , 22k = 2k+1 , 1 elements, so again by the induction hypothesis it has a subset, call it B3, of size 2k whose sum is divisible by 2k . It remains to show that two of the three sets B1 , B2 , and B3 can be combined to form a set of 2k+1 integers whose sum is divisible by 2k+1 . Let the sum of the three sets be s1 , s2 , and s3 respectively. Each is divisible by 2k , so let ti = si =2k , for i = 1, 2, and 3. Now by the Pigeonhole Principle, at least two of these numbers must have the same parity, even or odd. Without loss of generality, let the two sets be A1 and A2 . The sum of two numbers with the same parity is even, so t1 + t2 is even. Multiplying by 2k , we have s1 + s2 is divisible by 2k+1 . Thus there are 2k+1 integers from the original set A whose sum is divisible by 2k+1 . Our induction on m is complete. A207. Given triangle ABC , let A0 , B0, and C 0 be on the sides BC , AC , and AB respectively such that 40A00B00C 0 4ABC . Find the locus of the orthocentre of all such triangles A B C . Solution by Alexandre Trichtchenko, student, Brook eld High School, Ottawa, Ontario. Here we give the solution when the triangle ABC is acute. A similar argument can be given for an obtuse triangle. Let BAC = , ABC = , and BCA = . Let A00 , B 00 , and 00 be the feet of the altitudes from vertices A0 , B 0 , and C 0 respectively of C triangle A0 B 0 C 0 . Further, let H 0 be the point of the orthocentre of triangle A0 B0C 0, the point of intersection of its altitudes.
491
A C0 A00 B00 H0 C 00 B0
B C A0 Since triangles B 0 H 0 C 00 and B 0 A0 B 00 are similar B 0 H 0 C 00 = . Likewise, A0 H 0 C 00 = . Thus B 0 H 0 A0 = B 0 H 0 C 00 + A0 H 0 C 00 = + . Also, A0 H 0 B 0 + B 0 CA0 = + + = 180 . Hence, the quadrilateral CB 0 H 0 A0 is cyclic. Since B 0 A0 H 0 and B 0 CH 0 are inscribed in the circumcircle of CB 0 H 0 A0 and subtended by the same arc H 0 B 0 , we have H 00 CB00 = H00 A0B0 = 90 , 0 . Similarly, H 0AB0 = 90 , . So H CB = H AB0 , and so AH = CH00. By similar reasoning, we can show that AH 0 = BH 0 = CH 0 . Thus H is the circumcentre of triangle ABC and is independent of the choice 0of triangle A0B0 C 0. Thus the locus of the orthocentre of all triangles A0 B 0 C is just the single point H 0 = O, the
circumcentre of triangle ABC .
of the elements f1; 2; : : : ; p , 1g taken k at a time. For example, if p = 5, then S3 = 1 2 3 + 1 2 4 + 1 3 4 + 2 3 4 = 50. Show that pjSk for all 2 k p , 2. Solution. Consider the monic polynomial of degree p , 1, xp,1 , 1 0 mod p. There are precisely p , 1 incongruent solutions modulo p of this polynomial equation, namely, x = 1, 2, : : : , p , 1. Each of these follows from Fermat's Little Theorem, which states that ap,1 1 mod p, where p is a prime and a and p are relatively prime. Thus, modulo p, we have
Also solved by D.J. SMEENK, Zaltbommel, the Netherlands. A208. Let p be an odd prime, and let Sk be the sum of the products
xp,1 , 1 x , 1x , 2 x , p , 1 xp,1 , S1xp,2 + S2xp,3 , + ,1p,2Sp,2 + ,1p,1Sp,1 mod p : Equating coe cients, we have pjSk for all 2 k p , 2.
Note: We also get Wilson's Theorem for free. Why?
Also solved by ALEXANDRE TRICHTCHENKO, student, Brook eld High School, Ottawa, Ontario.
492
Challenge Board Solutions
Editor: David Savitt, Department of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA, USA 02138 dsavitt@math.harvard.edu Let Fi denote the i th Fibonacci number, with F0 = 1 and F1 = 1. Then F2 = 2, F3 = 3, F4 = 5, etc. a Prove that each positive integer is uniquely expressible in the form Fa + + Fak , where the subscripts form a strictly increasing sequence of positive integers no pair of which are consecutive. p b Let = 1 1 + 5, and for any positive integer n, let f n equal 2 the integer nearest to n . If n = Fa + + Fak is the expression for n from part a and if a2 6= 3, prove that f n = Fa +1 + + Fak +1 . c Keeping the notation from part b, if a2 = 3 so that a1 = 1, it is not always true that the formula f n = Fa +1 + + Fak +1 holds. For example, if n = 4 = F3 + F1 = 3 + 1, then the closest integer to n = 6:47 : : : is 6, not F4 + F2 = 5 + 2 = 7. Fortunately, in the cases where the formula fails, we can correct the problem by setting a1 = 0 instead of a1 = 1: for example, 4 = F3 + F0 = 3 + 1 as well, and indeed 6 = F4 + F1 = 5 + 1. Determine for which sequences of ai this correction is necessary.
1 1 1 1
C77.  Corrected problem not a solution!
C78. Let n be a positive integer. An n n matrix A is a magic matrix of order m if each entry is a nonnegative integer and each row and column P P sum is m. That is, for all i and j , k Aik = k Akj = m. Let A be a magic matrix of order m. Show that A can be expressed as the sum of m magic matrices of order 1. I. Solution by Christopher Long, graduate student, Rutgers University. Consider the magic matrix A as the adjacency matrix of a weighted bipartite graph G between two sets left" and right" of n vertices: If the i; j th entry of A is greater than 0, place an edge in G between the ith vertex on the left and the j th vertex on the right and give the edge a weight equal to the i; j th entry of A. If the i; j th entry of the A is 0, do not place an edge at all. The condition that the matrix A is a magic matrix implies that total weight of all the edges emanating from any single vertex of G, left or right, is equal to m. Given a subset S of the lefthand vertices of G, let us compute the size of its neighbourhood the collection of all vertices on the right which are joined by an edge to a vertex in S . Remove from G all of the edges whose lefthand vertices are not in S . Then the total weight of the remaining edges is exactly mjSj, and the neighbourhood of S is exactly the set of righthand vertices whose weight is still nonzero. By the weight of a vertex, we mean the weight of all the edges touching that vertex. But each righthand vertex has
493 weight at most m, so by the Pigeonhole Principle the number of righthand vertices with nonzero weight must be at least mjS j=m = jS j. That is, every subset on the left has a neighbourhood on the right which is at least as big. Thus, the conditions of the following famous theorem phrased traditionally, and thus thoroughly objectionable are satis ed, with the lefthand vertices as boys, the righthand vertices as girls, and edges as acceptable marriages: Theorem Hall's Marriage Theorem. Suppose there are n boys and n girls, and that each boy knows precisely which possibly more than one of the girls he is willing to marry. Suppose further that given any set S of boys, the total number of di erent girls that boys in S are willing to marry is at least S . Then there exists a way of pairing all the boys with the girls in such a way that each boy is willing to marry the girl to whom he is paired. The proof of the Marriage Theorem is an excellent exercise, and can also be found in almost any graph theory book, so we omit it here. In our case, if the pairing obtained from the Marriage Theorem pairs the vertex i on the left with the vertex i on the right, then we know that the i; i th entry of A is positive. Let A0 be the matrix whose i; i th entry is 1 for all i and whose other entries are all 0. Then A0 is a magic matrix of order 1 and A , A0 is a magic matrix of order m , 1, and the result follows by induction. II. Solution. As in the previous solution, we prove the result by induction by showing that there exists a permutation of f1; 2; : : : ; ng such that the i; i entry of A is positive that is, by constructing a magic submatrix" of order 1 in A. We do this for all magic matrices A by another induction, this time on NZ A, the number of nonzero entries of A. If the order of the magic matrix is m 0, then there is a nonzero entry in every row, so the total number of nonzero entries is at least n. However, if the number of nonzero entries is exactly n, then certainly A is m times a magic matrix of order 1, and this completes our base case. Now assume the result holds for NZ A k, where k n. Consider a magic matrix A of order m with exactly k nonzero entries. As k n, by the Pigeonhole Principle, there is a row with at least two nonzero entries, and each is less than m. Let i1; j1 be the position of one of them, and let i1 ; j2 be the position of the other. As the i1; j2th entry of A is less than m, there is a nonzero entry at some position i2; j2 in the same column and it is also less than m. By the same argument, there is a nonzero entry in the same row i2 ; j3 as i2; j2. We continue this process to get sequences ik , jk, such that ik 6= ik+1, jk 6= jk+1 , and such that the ik; jkth and ik ; jk+1th entries of A are all nonzero. Our goal is to nd a loop of an even number of distinct nonzero entries in the matrix, connected by alternating horizontal and vertical moves. Once we have such a loop, nd the point i; j in the loop whose entry is minimal. Suppose the i;j th entry is q . Decrease the i;j th entry by q to 0, increase
494 the next entry in the loop by q , decrease the next by q , and continue travelling once around the entire loop, alternately adding and subtracting q in this fashion. This will yield a magic matrix B with fewer nonzero entries than A, so by the induction hypothesis B contains a magic submatrix B 0 of order 1. However, by the construction of B , the nonzero entries of B all correspond to nonzero entries of A, so B 0 is also a magic submatrix of A. We will thus be done by induction. Let us proceed with nding this loop. As there are only a nite number of entries of A, at some point in the sequence i1; j1 , i1; j2 , i2; j2 , i2 ; j3, : : : there will be a repeated term. If the rst repeated term is of the form il; jl , and the rst appearance of this term is ik ; jk with k l, then the loop ik ; jk , ik ; jk+1 , : : : , il,1; jl , il; jl = ik; jk is exactly the kind of loop we are looking for, and we are done. We are similarly nished if the rst repeated term is of the form il; jl+1 and the rst occurrence of that term is of the form ik ; jk+1 . Suppose instead that the rst repeated term is of the form il; jl+1 and that the rst appearance of the term is ik; jk . Then, replacing jk by jl, we obtain the loop il; jl = ik; jl , ik ; jk+1 , : : : , il,1 ; jl, il; jl, and we are done. In the case when the rst repeated term is of the form il; jl and the rst appearance of the term is ik ; jk+1 , a similar trick works. Thus, we have exhausted all cases, and the proof is complete.
Shreds and Slices
There is a mistake in the 1987 Swedish Mathematical Olympiad, as printed in CRUX and MAYHEM 1998:298 . The expression ,a + 2b , 3c" in problem 2 of the Qualifying Round should read ,a + 2b + 3c". Thanks to Solomon Golomb for pointing this out. We regret to inform our readership that Richard Hoshino, long time Mayhem sta member and High School Editor, will be leaving us after this issue. We all thank Richard for his strong dedication and numerous contributions to Mayhem over the years, and we hope that he has gained as much in the experience as we have. We wish Richard the best of luck in his new responsibilities, and we will name his successor in the next issue.
Erratum
Goodbye, Richard!
495
1998 1999 Olympiad Correspondence Problems
Please mail your solutions to Professor E.J. Barbeau, Department of Mathematics, University of Toronto, Toronto, ON M5S 3G3. Set 1 1. ABC is an isosceles triangle with A = 100 and AB = AC . The bisector of angle B meets AC in D. Show that BD + AD = BC . 2. Let I be the incentre of triangle ABC . Let the lines AI , BI , and CI produced intersect the circumcircle of triangle ABC at D, E , and F respectively. Prove that EF is perpendicular to AD. 3. Let PQR be an arbitrary triangle. Points A, B , and C external to the triangle are determined for which
AQR = ARQ = 15 ; QPC = RPB = 30 ; PQC = PRB = 45 : Prove that: a AC = AB ; b BAC = 90 . 4. Let a and b be two positive real numbers. Suppose that ABC is a triangle and D a point on side AC for which BCA = 90 , jADj = a, and jDC j = b. Let jBC j = x and ABD = . Determine the values of x and for the con guration in which assumes its maximum value. 5. Let C be a circle with centre O and radius k. For each point P = O, we 6 de ne a mapping P ! P 0 where P 0 is that point on OP produced for which jOP j jOP 0 j = k2 :
In particular, each point on C remains xed, and the mapping at other points has period 2. This mapping is called inversion in the circle C with centre O, and takes the union of the sets of circles and lines in the plane to itself. You might want to see why this is so. Analytic geometry is one route. a Suppose that A and B are two points in the plane for which jABj = d, jOAj = r, and jOBj = s, and let their respective 2 im0 B0j = k d : ages under the inversion be A0 and B 0 . Prove that jA rs b Using a, or otherwise, show that there exists a sequence fXn g of distinct points in the plane with no three collinear for which all distances between pairs of them are rational.
496 6. Solve each of the following two systems of equations: p a x + xy + y = 2 + 3 2 ;
x2 + y2 = 6 : 2xy = 1 ; b x2 + y 2 + px + y x + y = x2 , y :
given two samples of three coins each and you know that each sample has exactly one counterfeit coin. What is the minimum number of weighings required to be certain of isolating the two counterfeit coins by means of an accurate scale not a balance? a Solve the problem assuming a and b are known. b Solve the problem assuming a and b are not known. 9. Similar isosceles triangles EBA, FCB , GDC , and HAD are erected externally on the four sides of the planar quadrilateral ABCD with the sides of the quadrilateral as their bases. Let M , N , P , and Q be the respective midpoints of the segments EG, HF , AC , and BD. What is the shape of PQMN ? 10. Given two points A and B in the Euclidean plane, let C be free to move on a circle with A as centre. Find the locus of P , the point of intersection of BC with the internal bisector of angle A of triangle ABC . 11. Let ABC be a triangle; let D be a point on AB and E a point on AC such that DE and BC are parallel and DE is a tangent to the incircle of the triangle ABC . Prove that
2 rn n for each integer n 7. 10 4 8. Counterfeit coins weigh a and genuine coins weigh b a 6= b. You are
7. For a positive integer n, let rn denote the sum of the remainders when n is divided by 1, 2, : : : , n respectively. a Prove that rn = rn , 1 for in nitely many positive integers n. b Prove that
Set 2
n2
8DE AB + BC + CA :
12. Suppose that n is a positive integer and that x + y = 1. Prove that
xn+1
n n X n + k
k n+1 X n + k
k y +y x = 1: k=0 k k=0 k
497
J.I.R. McKnight Problems Contest 1985
1. If Sn = 3+8x +15x2 +24x3 + +n2 +4n +3xn,1, determine Sn by rst evaluating 1 , xSn . Hence nd the limit of Sn as n approaches in nity, given x = 1 . 3 2. a P and Q are points ap2; 2ap and aq 2; 2aq on the parabola y2 = 4ax. Show that the equation of the chord PQ is 2x , p + qy + 2aq = 0. b If O is the origin and the chords OP and OQ are perpendicular, prove that the chord PQ cuts the xaxis in the same point for all possible positions of P and Q. 3. In the gure, angle A has a measure of 60 . At a distance of 10 cm from the vertex, a perpendicular is erected and a square is constructed on it with side s1 . In toward the vertex of the angle a second square of side s2 is formed. Then similarly a square of side s3 , and so on ad in nitum. Find the sum of the areas of these squares in simplest radical form and then give an approximation to the nearest hundredth of a square centimetre.
s1 s2 A s3
10 cm

4. A wire of length L is to be cut into two pieces, one of which is bent to form a circle and the other to form a square. How should the wire be cut if the sum of the areas enclosed by the two pieces is to be a maximum? 5. a Sketch the hyperbola represented by the equation and draw its asymptotes. b Draw a tangent to the hyperbola at any point on the hyperbola, and prove that the portion of the tangent between the points where it meets the asymptotes is bisected by the point of contact of the tangent.
x2 , y2 = 1 ; a b; a2 b2
498 c Prove that the segment of the tangent in b forms with the asymptotes a triangle of constant area. 6. Prove that if cos x + cos y = a and sin x , sin y = b, then
,b cosx , y = a2 + b2 : a
2 2
Swedish Mathematics Olympiad 1989
1989 Qualifying Round 1. Find the integer t and the hundreds digit a such that 3320 + t2 = 492a04 :
2. Form all possible six digit numbers, each using the digits 1, 2, 3, 4, 5, 6 exactly once. What is the sum of all these numbers? 3. Let ABC be an acuteangled triangle and let P be a point on the side BC . Let P 0 be the re ection of P in the side AB, and let0 P0000 be the re ection of P in the side CA. Show that the distance P P is least when P is the foot of the perpendicular from A to BC . 4. Show that if x, y , and z are positive real numbers and xy = y z = z x then x = y = z . 5. The equations x2 + px + q = 0 and qx2 + mqx + 1 = 0, where m, p, and q are real, and q 0, have roots x1 , x2 , and x1 , 1=x2 respectively. Show that mp 4. 6. Assume that a1 a2 a995 are 995 real numbers. Form all sums ai + aj , for 1 i j 995. Show that at least 1989 di erent numbers are obtained. Show also that exactly 1989 di erent numbers are obtained if and only if a1 , a2 , : : : , a995 is an arithmetic progression. The 1989 Final Round has already appeared in a previous issue of CRUX, in Olympiad Corner 125, 1991.
499
Dividing Points Equally
student, University of Toronto We start o with a simple problem and follow it through with some related questions that can be reduced to this problem. Problem 1. Given 2n points in the plane, is it possible to draw a straight line so that there are an equal number of points on either side of the line? Solution. The answer is yes. Since there are a nite number of points, namely 2n of them, there are a nite number of lines that pass through a , pair of points, at most 22n . Each line passing through a pair of points has a , certain direction. So, the 2n points have at most 22n distinct directions. It is not too hard then to pick a direction, call it , di erent from all of these. Intuitively, it is possible to take a line with direction and slide it along preserving its direction of course until half the points are on either side. This is possible since the line can pass through only one point at a time. If it passed through more than one point, then the direction between the two points must be the direction of this line. This contradicts the choice of to be di erent from any of those pairs from the 2n points. Hence, by sliding the line with direction along, we pass each point one at a time until we have passed exactly half, and then we are done. We follow this up with some generalizations and related problems for the reader to solve. Problem 1A. Prove that given mn points in the plane, we can nd m , 1 parallel lines that divide the plane into m regions with n points in each region. Problem 1B. Show that it is possible to divide 2n points in the plane by two intersecting lines so that for each line, half the points lie on either side of it. Show that it is possible to divide them by m concurrent lines so that for each line, half the points lie on either side of it. Problem 1C. Take 4n points in the plane and take any three distinct lines, each pair of which divides the set of points into equal quarters. Show that these three lines cannot be concurrent. Hint: Show that the three lines must form a triangle containing n of the points. Problem 1D. Is it possible to divide 2n points in the plane by a circle so that half of the points are inside and half are outside? Note: This does not follow immediately from Problem 1 by inversion.
Copyright c 1998 Canadian Mathematical Society
Cyrus C. Hsia
500
Problem 1E. Proposed by Loren C. Larson, St. Olaf College, North eld, Minnesota. Can every set of 4n points in the plane, no three of which are collinear, be evenly quartered by two mutually perpendicular lines? Mathematics Magazine, Problem 1513. Vol. 69, No. 5, December 1996, p. 385. In the next two problems, we consider three dimensions. Problem 2. There are 2n chocolate chips in a roll of frozen cookie dough. Show that it is possible to divide them into two sets of n chips by a plane cut. Solution. The 2n chocolate chips can be thought of as points in space. We want to roll the frozen cookie dough into a position where if the chocolate chips fell straight down, they would not hit each other. Essentially, we want to project the 2n points onto a plane so that each point is mapped to a distinct position. Then we are back to Problem 1 of dividing the points in the plane by a line. Why? This problem then becomes one of showing that there is such a projection. In 3 dimensions, we need something that is similar to slope". This is where the concept of a direction vector comes into play. The direction vector between two points from A = a1; a2 ; a3 to B = b1; b2; b3 is given by b1 , a1 ; b2 , a2 ; b3 , a3 or any multiple of this. To reduce two points having multiple direction vectors, we usually consider the unit direction vector which is found simply by taking the given direction vector dividing vector's norm pband a 2 + beachacoordinate,by theThe reader mayor length, namely 2 a3 2. wish to verify that 1, 1 2 , 2 + b3 for any line, any two points on the line will give the same direction vector up to a sign. Since there are a nite number of points, there are a nite number of unit direction vectors. We can then pick a unit direction vector di erent from all of them; call it = 1 ; 2 ; 3. By projecting each point in this direction onto a plane perpendicular to this vector, each point projects to a di erent point in the plane. Thus, take the dough so that the direction vector points vertically down on the kitchen table. Take a knife and cut along the line that would divide the points in half in the plane. Problem 3. Suppose the Earth has a population of 6 billion people in the near future. Is it possible to draw an imaginary equator around the world so that each hemisphere contains an equal number of humans? Here is a lemma that we will need. Lemma 3A. 2n points and a special point labelled S are given in the plane. Any line passing through S passes through at most one of the 2n points. It is then possible to divide the plane by a straight line that passes
501 through S so that n points are on either side of the line. Solution. Consider the lines passing through S and one other point. There are 2n of them. Pick any other line through S which is not one of these. Say there are k points on one side and 2n , k points on the other. If k = n then we are nished. Otherwise, rotate this line counterclockwise about S . This line will go through all of the points one at a time as it rotates. Let f be the di erence between the number of points on one side labelled A and the number of points on the other side labelled B as shown when the angle is from the original line. Points on the line are to be ignored.
q
1
q
qS
q q A q q
B
Thus, at the beginning, f 0 = 2n , k , k = 2n , k, and when it has completed a rotation of 180 , we have f 180 = k , 2n , k = ,2n , k. Make sure that sides A and B stay in the same orientation.
qS
q q
B
A line with sides A and B .
q
q
q
q
1
Thus these values have di erent signs and if this were a problem of a continuous function, then we could easily claim by the Intermediate Value Theorem that there is a value 0 180 such that f = 0. Nonetheless, we can still conclude this, since we know that when the value of f changes, it changes by a value of 1. To see this, note that each time the line passes through a point, the value on one side, say A decreases by 1 and the number of points on side B stays the same. The value of f then decreases by one. Once the line immediately passes by this point, the number of points on side B increases by 1 and the number of points on side A remains the same. Again, the value of f decreases by 1. This will happen for each point that the line crosses.
A line with sides A and B reversed from the above gure.
A
502 Note that when the line passes completely over a point the value of f changes by a value of 2. Thus the parity of f changes when the line goes through a point from the state when there are no points on the line and the parity stays the same when the line passes by a point. Since 2n , k and its negative are both integers, at some step in the value of f , 0 is reached. Now we must check that this value is not obtained when the line crosses through one of the 2n points. Since the values of 0 and 2n , k have the same parity, when the value of 0 is achieved, it is achieved in the state where the line does not pass through any of the 2n points. Hence, for this line, there is an angle counterclockwise away from the original line where the points are equally divided. Solution to Problem 3. We must assume, of course, that the people are points occupying a distinct and xed location on a spherical Earth. Now for each pair of people draw a great circle passing through them. That is, a circle given by the intersection of the sphere with a plane passing through the centre of the sphere; this would be our de nition of an equator. There are a nite number of such circles as there are a nite number of pairs. Thus we may always choose another point, call it N , not on any of the great circles. The point diametrically opposite N , call it S , must not be on any of the great circles either. Why? If a point was on a great circle then the point diametrically opposite it would be also, by de nition. Now place the spherical world on a plane with the point S tangent to it. From point N project each point onto the plane by drawing a line from N through each point to intersect the plane this projection is called the stereographic map. Now consider lines through S in the plane. Any such line can have at most one other point. If a line contains S and two other points then that would mean that S was on the great circle through the preimage of these points. By Lemma 3A there is a line through S which cuts the plane into two parts containing an equal number of points. This line projected back onto the sphere is a great circle dividing the world into two equal populations. Exercises 1. Give solutions to the problems listed above. 2. Is it possible to divide any 4n points in the plane by two intersecting lines so that each of the four sectors contains n points? 3. Is it possible to divide the Earth into 4 quarters with an equal number of people in each? Assume a population of 6 billion.
503
Problem proposals and solutions should be sent to Bruce Shawyer, Department of Mathematics and Statistics, Memorial University of Newfoundland, St. John's, Newfoundland, Canada. A1C 5S7. Proposals should be accompanied by a solution, together with references and other insights which are likely to be of help to the editor. When a submission is submitted without a solution, the proposer must include su cient information on why a solution is likely. An asterisk ? after a number indicates that a problem was submitted without a solution. In particular, original problems are solicited. However, other interesting problems may also be acceptable provided that 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 without the originator's permission. To facilitate their consideration, please send your proposals and solutions on signed and separate standard 8 1 "11" or A4 sheets of paper. 2 These may be typewritten or neatly handwritten, and should be mailed to the EditorinChief, to arrive no later than 1 April 1999. They may also be sent by email to cruxeditors@cms.math.ca. It would be appreciated if A email proposals and solutions were written in LTEX. Graphics les should be in epic format, or encapsulated postscript. Solutions received after the above date will also be considered if there is su cient time before the date of publication. Please note that we do not accept submissions sent by FAX. Proposed by Daniel Kupper, Bullingen, Belgium. Suppose that for each k 2 N, the numbers ak ; bk ; zk 2 C. Suppose that the polynomials
PROBLEMS
2388.
Anz =
n X
a1 + a2 a2 + a3
f a1; a2; a3 ; : : : ; an f 2 ; 2 ; a3; : : : ; an : a + a + : : : + an . Prove that Let m = 1 2
f a1; a2; a3; : : : ; an f m; m; : : : ; m :
2
2 are related by An zj = Bn zj = 0 for j 2 f1; 2; : : : ; ng. For each n 2 N, nd an expression for bn in terms of a0 ; a1 ; : : : ; an . 2389. Proposed by Nikolaos Dergiades, Thessaloniki, Greece. Suppose that f is continuous on Rn and satis es
k=0
ak zk
and
Bnz =
n X
k=0
bk zk
504
2390. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. n, X X n p q For 0 and p; q 1, let Sn; p; q := i j , where
n .
i=1 j =i+
Given the statement: Sn; p; q , understood as a polynomial in Q n , is always divisible by n , n , + 1n , + 2", a give examples for = 0; 1; 2; 3; 4; b? prove the statement in general. Proposed by G. Tsintsifas, Thessaloniki, Greece. Consider d + 1 points, B1 , B2, : : : , Bd+1 in the unit sphere in Rd , so that the simplex Sd B = B1 B2 : : : Bd+1 includes the origin O. Let P = fx j Bi x 1g for all i between 1 and d + 1. Prove that there is a point y 2 P such that jy j d. 2392. Proposed by G. Tsintsifas, Thessaloniki, Greece. Suppose that xi , yi, 1 i n are positive real numbers. Let
2391.
Prove that 1. An Cn , 2. Bn Dn , 3. 2An 2Bn Cn Dn .`
,Pn x ,Pn y n X xi yi i=1 An = x + y , Bn = Pn ix +i=1 i , i i i=1 i yi i=1 ,Pn x 2 + ,Pn y 2 n X x2 + yi2 i=1 ni i=1 i , Cn = P x + y Dn = xi + y . i=1 i i i=1 i i
2393.
Proposed by G. Tsintsifas, Thessaloniki, Greece. Suppose that a, b, c and d are positive real numbers. Prove that
4abcda + b + c + d2;
3 Y 2. a + bb + cc + dd + a 16abcd2 2a + b + c.
1. a + bb + cc + dd + a
a; b; c; d
cyclic
3=2
505 Proposed by Vedula N. Murty, Visakhapatnam, India.
a+b a bb a + b The inequality a , where a; b 0, is usually proved 2 using Calculus. Give a proof without the aid of Calculus. 2395. Proposed by Witold Janicki, Jagiellonian University, Krakow, Poland, Michael Sheard, St. Lawrence University, Canton, NY, USA, Dan Velleman, Amherst College, Amherst, MA, USA, and Stan Wagon, Macalester College, St. Paul, Minnesota, USA. Let P be such that A P 0 is true, and B P n = P n + 1. Find an integer n 106 such that P n can be proved without using induction, but rather using L the Law of Implication that is, X and X = Y yield Y ten times only. 2396. Proposed by Jose Luis Diaz, Universitat Politecnica de Catalunya, Colum, Terrassa, Spain n X k Suppose that Az = akz is a complex polynomial with an = 1,
k = n,k o. Prove that all the zeros of A lie in the and let r = 0max,1 jak j kn
=0 1
2394.
n
disk C = z 2 C : jz j
Proposed by Toshio Seimiya, Kawasaki, Japan. Given a rightangled triangle ABC with BAC = 90 . Let I be the incentre, and let D and E be the intersections of BI and CI with AC and AB respectively. Prove that Proposed by Toshio Seimiya, Kawasaki, Japan. Given a square ABCD with points E and F on sides BC and CD respectively, let P and Q be the feet of the perpendiculars from C to AE CP + CQ = 1. and AF respectively. Suppose that Prove that EAF = 45 .
2397.
2 =n , 1 .
1
r
2398.
BI 2 + ID2 = AB2 . CI 2 + IE2 AC 2
AE AF
506 Proposed by David Singmaster, South Bank University, London, England. In James Dodson's The Mathematical Repository, 2nd ed., J. Nourse, London, 1775, pp 19 and 31, are two variations on the classic Ass and Mule" problem: What fraction is that, to the numerator of which 1 be added, the value will be 1=3; but if 1 be added to the denominator, its value is 1=4?" This is easily done and it is easy to generalize to nding x=y such that x + 1=y = a=b and x=y + 1 = c=d, giving x = ca + b=ad , bc and y = bc + d=ad , bc. We would normally take a=b c=d, so that ad , bc 0, and we can also assume a=b and c=d are in lowest terms. A butcher being asked, what number of calves and sheep he had bought, replied, `If I had bought four more of each, I should have four sheep for every three calves; and if I had bought four less of each, I should have had three sheep for every two calves'. How many of each did he buy?" That is, nd x=y such that x + 4=y + 4 = 4=3 and x , 4=y , 4 = 3=2. Again, this is easily done and it is easy to solve the generalisation, x + A=y + A = a=b and x , A=y , A = c=d, getting x = A2ac , bc , ad=bc , ad and y = Aad + bc , 2bd=bc , ad. We would normally take a=b c=d so that bc , ad 0, and we can also assume a=b and c=d are in lowest terms. In either problem, given that a, b, c and d are integers, is there a condition simpler than computing x and y to ensure that x and y are integers? Alternatively, is there a way to generate all the integer quadruples a, b, c, d, which produce integer x and y? 2400. Proposed by V
aclav Konecn
y, Ferris State University, Big Rapids, Michigan, USA. cosx 1 + 2x for 0 x 1=2. a Show that 1 + , 2x Proposed by Bruce Shawyer, EditorinChief.
? Show that cosx , 2 , 2 x , 1 2 for 0 b
2399?.
1 , 2x
1 , 2x
2
2
x 1=2.
507
SOLUTIONS
No problem is ever permanently closed. The editor is always pleased to consider for publication new solutions or new insights on past problems. Hungary. For 0
2090.
1995: 307, 1997: 433 Proposed by Peter Iv
dy, Budapest, a
x =2 prove that sin x
2 2 , x2 x 2 + x2 :
In the remarks following the solution, Walther Janous gave the extension to: what is the value of that gives the best inequality of the type
sin x
2 x which is valid for all x 2 0; =2?
, x2 2 + x2
2
A From the Taylor series expansion, we have
sin x
2 2 , x2 2 + x2 x p is valid for all x 2 0; =2 if 6, and the optimal value is
Janous has now solved this question himself. We prove that
opt
= 6.
p
1
Thusp have as taking the limit as x ! 0 shows that we 6.
2 2 2 x2 x x2 2 , x2 , sin2 x = x2 2 , x2 + cos22 , 1 +x +X ! 1 2 x2k+2 = x2 + ,1k 2k k=1 1 X k+1 22k+1 2k+2! 2 + ,x + ,1 2k + 2! x k=1 ! 1 X k 1 22k = 2 ,1 2k , 2k + 2! x2k+2 k=1
2,6 4 2 = 2x 4 2 + x : : : 0: 2
, 6 0; that is
508 B Suppose that x
2
it is su cient to prove 1 for = opt = 6. C Next, we prove the auxiliary inequality
3
, x2 2 + x2 2 2 + 2x2 , 2 x2 , x4 2 2 2x2 x 2
2
, x2 2 + x2
0. Since
2 2
p
2
;
5
, 2x2 + 2 x2 , x4
g0x = 1 , x2 + x , cos x ; g00 = 0 ; 24 00 x = ,x + x + sin x ; g g00 0 = 0 ; 6 g30 = 0 ; g3x = ,1 + x2 + cos x ; 4x = x , sin x ; g g40 = 0 ; and 5x = 1 , cos x 0 : g Thus, g 4x increases, so that g 4x g 40 = 0. Further, g 3x increases, so that g 3x g 30 = 0. And so on. Thus, g x increases, so that g x g 0 = 0, as claimed. D Because of Cp and since sin x 0 since x 2 0; =2, inequality 1 , with = opt = 6 that is, x2 6 , x2 6 + x2 sin2 x will follow from 3 x5
2 0 : x26 , x2 , 6 + x2 x , x + 120 6
2 4 3 2
is valid for all x 0. x Indeed, let g x = x , x6 + 120 , sin x. Then g 0 = 0. Furthermore, we have
3 5
x sin x x , x + 120 6
2
x6 ,x6 , 34x4 + 400x2 , 960 := x6 hx 0 : 14400 14400 , But, h0 x = 6x5 , 136x3 + 800x = 2x x4 , 68x2 + 400 , , 2x h3x4 , 69x2 + 399 = 6x x4 , 23x2 + 133 , i 0; = 6x x2 , 23 2 + 3 2 4 that is, hx increases as x 0 increases. Since 1:6 and h1:6 ,142, 2
the proof is complete.
This inequality is equivalent to
509
2259. 1997: 301 Proposed by Paul Yiu, Florida Atlantic University, Boca Raton, Florida, USA. Let X , Y , Z , be the projections of the incentre of 4ABC onto the sides BC , CA, AB respectively. Let X 0 , Y 0 , Z 0 , be the points on the incircle diametrically opposite to X , Y , Z , respectively. Show that the lines AX 0 , BY 0, CZ 0, are concurrent. I. Solution by Christopher J. Bradley, Clifton College, Bristol, UK. The areal coordinates of the incentre I are a+1+c a; b; c, and those of b X are 21a 0;a + b , c; a , b + c. , ,! ! , Since XI = IX 0 , the areal coordinates of X 0 are twice those of I minus those of X , and hence are
The equation of AX 0 is thus
2a ; c , a + bc + a , b ; b , a + cb + a , c
: a+b+c 2aa + b + c 2aa + b + c
yb + a , c = zc + a , b : The equations of BY 0 and CZ 0 are obtained by cyclic changes, and all are concurrent at the point P with coordinates 1 b + c , a; c + a , b; a + b , c : a+b+c
II. Solution by Florian Herzig, student, Cambridge, UK. This problem is very similar to problem 2250 1997:245, 1998:372 . I will use the following known theorem a proof can be found in 1 : In an inscriptable quadrilateral the incentre and 1 the midpoints of the diagonals are collinear. First let A0 , B 0 , C 0 be the midpoints of AX , BY and CZ . Because of 1, in the limiting case of a triangle, the lines A0 K , B 0 L, C 0 M pass through I . Since XA0 : XA = XI : XX 0 = 1 : 2, the lines AX 0 and KI are parallel. Let G be the centre of gravity of 4ABC and also of 4KLM . A dilatation with centre G and factor ,2 maps 4KLM onto 4ABC . If P is the image of I under that mapping, then AX 0 , BY 0 , CZ 0 pass through P since the line KI is mapped onto the line AP but also onto the line AX 0 as AX 0 k KI . Moreover, because of the dilatation, P lies on IG and
IG : GP = 1 : 2 :
Reference 1 H. Dorrie, Triumph der Mathematik, Physica, Wurzburg 1958. 100 Great Problems in Elementary Mathematics , Dover, N.Y. 1965.
510
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CON AMORE PROBLEM GROUP, Royal Danish School of Educational Studies, Copenhagen, Denmark; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY
LEVERSHA, St. Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; JOHN VLACHAKIS, Athens, Greece; and the proposer. Bradley commented that if J is the centre of mass of the triangle thought of as a uniform wire framework, then P lies on the extension of IJ and is such that IJ = JP . The point P is the Nagel point. Also, G lies on this line, and IG : GJ = 2 : 1. Further, G and P are the internal and external centres of similitude of the incircles of triangles ABC and the median triangle with vertices at the midpoints of the sides of triangle ABC . Lambrou, who gave an indirect proof, gave a generalization and noted that it is the converse of a result due to Rabinowotz problem 1353, Mathematics Magazine, 65 1992, p. 59. Let X , Y , Z be the points of contact of the incircle of triangle ABC with the sides BC , CA, AB respectively. Let I 0 be any point within the incircle not necessarily the incentre. Let XI 0 , Y I 0 , ZI 0 cut the incircle again at X 0 , Y 0 , Z 0 respectively. Then AX 0 , BY 0 , CZ 0 are concurrent. Seimiya also noted that the point of concurrency is the Nagel point of 4ABC .
1997: 364 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a right angled triangle with the right angle at A. Points D and E are on sides AB and AC respectively, such that DEkBC . Points F and G are the feet of the perpendiculars from D and E to BC respectively. Let I , I1 , I2 , I3 be the incentres of 4ABC , 4ADE , 4BDF , 4CEG respectively. Let P be the point such that I2 P kI1 I3 , and I3 P kI1 I2 . Prove that the segment IP is bisected by the line BC . Solution by Florian Herzig, student, Cambridge, UK. AE Let = AD = AC : Also de ne x; x1 ; x2 ; x3 ; y to be the lengths of AB perpendiculars of I; I1 ; I2 ; I3 ; P to BC respectively. It is su cient to prove that x = y . First, see, for example, Roger A. Johnson, Modern Geometry 1929, p. 189, Then, by similarity, and so,
2264.
x = a +bc=2c=2 = a +bc+ c : b+ b DF = b BD = 1 , bc ; a a
x1 = DF + x = 1 , bc + a +bc + c : a b
511 Moreover,
bc2 b2 c x2 = BD x = 1 , aa + b + c ; x3 = 1 , aa + b + c : BC By de nition of P , y = x1 , x2 , x3 c = 1 , bc + a +bc+ c , 1 , abcb ++c a b a+b = x
after an easy calculation, and that is what we wanted to prove. Editor's note: To see that y = x1 , x2 , x3 ; let Q = I1 P ^ I2 I3 , R = I1P ^ BC , and z be the length of the perpendicular from Q to BC . Then x2 + x3 = 2z and so x1 = y + 2z = y + x2 + x3 .
y z I1R = I1Q + QR = QP + QR = RP + 2QR = I1R x + 2I1R x ;
1 1
Solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
2267. 1997: 364 Proposed by Clark Kimberling, University of Evansville, Evansville, IN, USA and Peter Y , Ball State University, Muncie, IN, USA. In the plane of 4ABC , let F be the Fermat point and F 0 its isogonal conjugate. Prove that the circles through F 0 centred at A, B and C meet pairwise in the vertices of an equilateral triangle having centre F . Solution by Toshio Seimiya, Kawasaki, Japan. Equilateral triangles BCD, CAE , ABF are erected externally on the sides of 4ABC . Then AD, BE , CF are concurrent at the Fermat point F , and AFB = BFC = CFA = 120 see, for example, H.S.M. Coxeter, Introduction to Geometry 1961, x1.8. Let the circles through F 0 centred at B and C meet at P . Since 0 = BP and CF 0 = CP , P is the re ection of F 0 across BC . SimiBF larly Q is the re ection of F 0 across AC , and R0 is the re ection of F 0 across AB, where the circles through F 0 centred at A and C meet at Q and those centred at A and B meet at R.
512 Since CAQ = F 0 AC = BAF , we get
FAR = FAC + CAQ = FAC + BAF = BAC :
Similarly we have RAF = BAC , so that FAQ = RAF . Since AQ = AF 0 = AR, AF is the perpendicular bisector of QR. Similarly CF is the perpendicular bisector of PQ. Thus we have FR = FQ = FP , so that F is the circumcentre of 4PQR. Since AF ? QR and CF ? PQ, and AFC = 120 , we have PQR = 60 . Similarly, we have PRQ = 60 , so that 4PQR is equilateral.
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, and
MAR IA ASCENSION LOPEZ CHAMORRO, I.B. Leopoldo Cano, Valladolid, Spain 2 solutions; FLORIAN HERZIG, student, Cambridge, UK; MICHAEL LAMBROU, University of Crete, Crete, Greece; and the proposer.
2269. 1997: 365 Proposed by Cristobal S anchez Rubio, I.B.
Penyagolosa, Castellon, Spain.
Let OABC be a given parallelogram with AOB = 2 0; =2 .
A. Prove that there is a square inscribable in OABC if and only if
OA sin , cos OB sin + cos
and
sin , cos OB sin + cos : OA
B. Let the area of the inscribed square be Ss and the area of the given parallelogram be Sp . Prove that
2Ss = tan2 ,OA2 + OB 2 , 2Sp :
Editor's comment. All those who sent in submissions on this problem noted that the proposer had mislabeled the parallelogram OABC instead of OACB .
513 Solution by Hans Engelhaupt, Franz Ludwig Gymnasium, Bamberg, Germany.
B S M
R
C
O
P
E
QA
D
Suppose that the diagonals of parallelogram OACB meet at M . The centre of the square must be M . If OB and are xed, then CD = OB sin is xed. If the square rotates about M with one vertex on the line OA, then the other vertices lie on the lines QR, RS and SP . Thus, there is an inscribed square if and only if the segment OB meets the segment SP . Therefore OA OB cos + OB sin from B = S and A = Q, and OB sin , OB cos OA from O = P and C = R. The same must hold if we interchange the roles of A and B . So part A follows immediately.
S H
B M
R
C
O
P
G E
A Q
D
Let OB = b, OP = x, HP = y , PG = z . Then y = x tan and z = b sin , y. Therefore OF = 1 a + b cos and x = OF , PF = 2 1 , b sin , giving 4Ss = 4y2 + z2 and Sp = ab sin . 2 a + b cos
514 It follows that
4Ss = 4y2 + 4z2 = a + b cos , b sin 2 tan2 + 4b sin , x tan 2 = a + b cos , b sin 2 tan2 + 2b sin , a + b cos , b sin tan 2 , = tan2 2a + b cos , b sin 2 + 4b2 cos2 ,4ba + b cos , b sin cos :
Therefore
2Ss = tan2
,a + b cos , b sin a , b cos , b sin +2b2 cos2 , = tan2 ,a , b sin 2 , b cos 2 + 2b2 cos2 = tan2 ,a2 + b2 , 2ab sin = tan2 OA2 + OB2 , 2Sp ;
as required.
Also solved by FLORIAN HERZIG, student, Cambridge, UK; and the proposer. Walther Janous, Ursulinengymnasium, Innsbruck, Austria and Michael Lambrou, University of Crete, Crete, Greece commented on the error in notation. Janous gave a condition for the problem to be true.
2276. 1997: 430 Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Quadrilateral ABCD is cyclic with circumcircle ,0; R. Show that the ninepoint Feuerbach circles of 4BCD, 4CDA, 4DAB and 4ABC have a point in common, and characterize that point. Solution by Francisco Bellot Rosado, I.B. Emilio Ferrari, and Mar
a Ascension Lopez Chamorro, I.B. Leopoldo Cano, Valladolid, Spain and a
summary of their comments. This problem is a classical result about cyclic quadrangles. We will present a solution, then discuss some references. Let HA ; BB ; HC ; HD be the respective orthocentres of the triangles BCD; CDA; DAB , and ABC , and let NA ; NB ; NC ; ND be their ninepoint centres. Theorem 1. For any cyclic quadrangle, the centres of the ninepoint circles of the four triangles formed, taking the vertices of the quadrangle three at a time, form a homothetic quadrangle. Proof. The ninepoint centres NA ; NB ; NC ; ND are the midpoints of the segments OHA ; OHB ; OHC ; OHD ; therefore the quadrangles NA NB NC ND and HAHB HC HD are homothetic with ratio 1 . But 2
515
HA HB HC HD is also homothetic and oppositely congruent 1 ABCD. to Hence NA NB NC ND and ABCD are homothetic with ratio , 2 . As a consequence, the circumcentre M of NA NB NC ND is the midpoint of the segment joining O to the circumcentre of HA HB HC HD , while the circumradius is R . In other words, M is the centre of symmetry of 2 ABCD and HAHB HC HD the common midpoint of AHA, etc. Since our four ninepoint circles all have radius R , they must all pass through M , which 2
is the desired result. Comments. It seems that this result is from Lemoine, 1869, Nouvells Annals de Math
matiques pp. 174 and 317. M is sometimes called the e Mathot point, after Jules Mathot, Mathesis, 1901, p. 25. In 1 M is called the anticentre of the cyclic quadrangle ABCD. Here is a summary of some of the properties of this point: Theorem 2. If ABCD is cyclic, the ninepoint circles of the triangles BCD, CDA, DAB, ABC , HB HC HD ; HC HD HA, HD HAHB ; HA HB HC , and the Simson lines of these triangles with respect to the fourth vertex of each quadrangle all contain the Mathot point of ABCD, as do the lines perpendicular to a side of ABCD that pass through the midpoint of the opposite side. Convenient references that present a solution to our problem include 1 Nathan Altshiller Court, College Geometry. Barnes and Noble, 1965 theorem 263, p. 132.
2 Roger A. Johnson, Advanced Euclidean Geometry. Dover, 1960 pp. 209
and 243. Part of Theorem 1 was used in the 1984 Balkan Math. Olympiad. See 1985: 243 for related references.
Also solved by NIKOLAOS DERGIADES, Thessaloniki, Greece; FLORIAN HERZIG, student, Cambridge, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete,
Greece; GERRY LEVERSHA, St. Paul's School, London, England; ISTVAN REIMAN, Budapest, Hungary 2 solutions; TOSHIO SEIMIYA, Kawasaki, Japan; and the proposer.
2279. 1997: 431 Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. With the usual notation for a triangle, prove that
sin3 A cos B cos C = 4sr4 2R2 , s2 + 2R + r2 : R cyclic
X
,
516 Solution by HeinzJurgen Sei ert, Berlin, Germany. On 1997: 447 448 it was shown by Herzig that
X
sin3 A cos B cos C =
Y
sin A
X
cos2 A :
Using see, for example, 1996: 130
Y
and
sin A = 2sr2 R
X
the desired identity easily follows.
2 2 X cos2 A = 3 , sin2 A = 3 , s , 4Rr , r 2R2 6R2 , s2 + 4Rr + r2 ; = 2R2
Also solved by HAYO AHLBURG, Benidorm, Spain; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; GORAN CONAR, student, Gymnasium Varadin, Varadin, Croatia; NIKOLAOS DERGIADES, Thessaloniki, Greece; z z FLORIAN HERZIG, student, Cambridge, UK; RICHARD I. HESS, Rancho Palos Verdes, California, USA; MURRAY S. KLAMKIN, University of Alberta, Edmonton,
Alberta; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; SAI C. KWOK, San Diego, California, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; VEDULA N. MURTY, Maharanipeta, India; ISAO NAOI, SekiShi, Gifu, Japan; BOB PRIELIPP, University of Wisconsin Oshkosh, Wisconsin, USA; D.J. SMEENK, Zaltbommel, the Netherlands; PANOS E. TSAOUSSOGLOU, Athens, Greece; PAUL YIU, Florida Atlantic University, Boca Raton, Florida, USA; and the proposer. Ahlburg and Prielipp gave the same solution as Sei ert. Several other solvers also used the identity from 1997: 447 448 .
1997: 431 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a triangle with incentre I . Let D be the second intersection of AI with the circumcircle of 4ABC . Let X , Y be the feet of the perpendiculars from I to BD, CD respectively. 1 Suppose that IX + IY = 2 AD. Find BAC . Combination of solutions by Michael Lambrou, University of Crete, Crete, Greece and by Florian Herzig, student, Perchtoldsdorf, Austria. We show that
2280.
IX + IY = sin AAD : 1 5 The given condition then implies that sin A = , so that A = or A = . 2 6 6
517
IBC + CBD = B + CAD 2 B + A = ABI + BAI = BID : = 2 2 Thus IY = ID sin CDI = BD sin CDA. Let the diameter of the circumcircle of 4ABC be 1 so that a chord equals the sine of either angle it IBD =
subtends on the circumference, in which case, Similarly,
Note rst that 4IBD is isosceles with ID = BD:
IY = BD AC :
Thus by Ptolemy's Theorem
IX = CD AB :
IX + IY = BD AC + CD AB = BC AD = sin AAD :
Also solved by MIGUEL AMENGUAL COVAS, Cala Figuera, Mallorca, Spain; FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; SABIN CAUTIS, student, Earl Haig Secondary School, North York, Ontario; NIKOLAOS DERGIADES, Thessaloniki, Greece; GERRY LEVERSHA, St. Paul's School, London, England;VEDULA N. MURTY, Visakhapatnam, India; VICTOR OXMAN, University
of Haifa, Haifa, Israel; ISTVAN REIMAN, Budapest, Hungary; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer.
1997: 431 Proposed by Toshio Seimiya, Kawasaki, Japan. ABC is a triangle, and D is a point on the side BC produced beyond C , such that AC = CD. Let P be the second intersection of the circumcircle of 4ACD with the circle on diameter BC . Let E be the intersection of BP with AC , and let F be the intersection of CP with AB . Prove that D, E , F , are collinear. Solution by Nikoloas Dergiades, Thessaloniki, Greece with his notation modi ed to make use of directed line segments. Let G be the point where AP intersects BD. The convex quadrilateral APCD is cyclic and 4ACD is isosceles so GPC = CDA cyclic = CAD isosceles = CPD cyclic: Hence PC is the bisector of GPD; since PB ? PC , PB is the exterior bisector of the same angle. It follows that
2281.
GC = BG ; CD BD
or
BG = , BD : GC DC
518 Ceva's Theorem applied to 4ABC gives
BG CE AF = 1 : GC EA FB
Thus
BD CE AF = ,1 ; DC EA FB which, by Menelaus's Theorem, means that D; E; F are collinear.
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; FLORIAN HERZIG, student, Cambridge, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; D.J. SMEENK, Zaltbommel, the Netherlands 2 solutions; and the proposer.
2282. 1997: 431 Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. A line, `, intersects the sides BC , CA, AB , of 4ABC at D, E , F respectively such that D is the midpoint of EF . Determine the minimum value of jEF j and express its length as elements of 4ABC . Solution by Florian Herzig, student, Perchtoldsdorf, Austria. Let M be the midpoint of BC . Drop perpendiculars from E , F onto BC having lengths h1, h2 respectively. Observe that h1 = h2 by similarity, as ED = DF . Hence the triangles BMF , CME are equal in area. This implies that ABC = AFME , using the common notation for areas. Therefore
1 ABC = AFME = 1 AM EF sin AM; EF 2 AM EF 2
and so
ABC EF 2 AM with equality if and only if EF ? AM . Equality is always possible by a continuity argument: Consider all lines ` perpendicular to AM ; the lines passing through B and C yield the extreme ratios FD : DE = 0 or in nity,
whence there is a desired line in between. Ed: Dou gave three constructions for the points E 0 , F 0 that provide the minimum value of jEF j. Here is one of them: De ne B 0 as the point where the perpendicular from B to AM meets the side AC ; and D0 to be where the line joining A to the midpoint of BB 0 meets BC ; the perpendicular from D0 to AM meets AB and AC in the desired points F 0 and E0.
519 Remark: The lines EF envelope a parabola see 1991: 97 that touches BC in M; AC in Y and AB in Z where AC = CY , AB = BZ . Indeed this problem shows that in this case the shortest segment of a tangent intercepted by AB and AC comes from the tangent perpendicular to the axis of the parabola. I will sketch a proof: Let the directrix of the parabola be d and the focus be P . Drop perpendiculars from Y; Z on d which have feet K; L. Then, as AY; AZ are tangents to the parabola, AY; AZ are perpendicular bisectors of PK; PL using the tangent property for parabolae. Hence A is the circumcentre of 4PLK and so the perpendicular from A to d is actually the bisector of the lines KY; LZ ; that is, it passes through M . Thus AM is perpendicular to the directrix; that is, it is parallel to the axis of the parabola.
Also solved by NIKOLAOS DERGIADES, Thessaloniki, Greece; JORDI DOU, Barcelona, Spain; RUSSELL EULER and JAWAD SADEK, NW Missouri State University, Maryville, Missouri, USA; WALTHER JANOUS, Ursulinengymnasium, Inns
bruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; GERRY LEVERSHA, St. Paul's School, London, England: TOSHIO SEIMIYA, Kawasaki, Japan; and the proposer. Dou also proved the parabola property of the minimum EF . Several readers interpreted the question as calling for the minimum value of EF in terms of the side lengths of 4ABC :
pa + b + c,a + b + ca , b + ca + b , c : min jEF j = p2b2 + 2c2 , a2
2283. 1997: 432 Proposed by Waldemar Pompe, student, University of Warsaw, Poland. You are given triangle ABC with C = 60 . Suppose that E is an interior point of line segment AC such that CE BC . Suppose that D is an interior point of line segment BC such that
AE = BC , 1 : BD CE Suppose that AD and BE intersect in P , and the circumcircles of AEP and BDP intersect in P and Q. Prove that QE k BC .
Solution by Michael Lambrou, University of Crete, Crete, Greece. De ne F to be the point on AC such that CBF = 60 , and R to be the point on BF such that ER is parallel to BC . We shall show that the circumcircles of AEP , BDP intersect at R. In other words R and Q coincide and the required conclusion QE jjBC will follow.
520 If the projections of R and E on BC are R1 and E1 respectively, then BR1 = CE1 = 1 CE as C = 60 , so that RE = R1 E1 = 2 BC , BR1 , CE1 = BC , CE. Hence from the stated condition we have
so that
AE BC BD = CE , 1 ; = RE CE
AE CE = BD RE : 1 Suppose that the circumcircle of RAC cuts the extension of RE at G. As RG, AC are intersecting chords we have AE CE = EG RE . Comparing with 1 we see EG = BD. Using this we see that triangles BRD, ECG are congruent: they have BD = EG; BR = EC clear and B = 60 = ECB = CEG alternate on parallels. Thus EGC = RDB and thus quadrilateral RGCD is cyclic. In other words C , G, A, R and D are all on one circle. Hence RAD = RCD same arc = CBE symmetry = BER parallel lines; that is, RAP = REP , showing that quadrilateral RAEP is cyclic, as
required. It remains to verify that quadrilateral BDPR is cyclic:
BPD =
= = =
APE ARE ACG BRD
as triangle RAE is cyclic as quadrilateral RACG is cyclic as 4BDR ECG: =
This completes the proof.
AE BC BD = CE , 2 cos C : Note that our featured solution accommodates this generalization: de ne F to be the point on AC for which CBF = C .
Also solved by NIKOLAOS DERGIADES, Thessaloniki, Greece; FLORIAN HERZIG, student, Cambridge, UK; TOSHIO SEIMIYA, Kawasaki, Japan; D.J. SMEENK, Zaltbommel, the Netherlands; and the proposer. Pompe shows more generally that QE jjBC still holds if the given condition were
521 1997: 432 Proposed by Toshio Seimiya, Kawasaki, Japan. ABCD is a rhombus with A = 60 . Suppose that E, F , are points on the sides AB , AD, respectively, and that CE , CF , meet BD Ed: not BC2as was 2originally printed in error at P , Q respectively. Suppose that BE + DF = EF 2. Prove that BP 2 + DQ2 = PQ2. Solution by Walther Janous, Ursulinengymnasium, Innsbruck, Austria. Apparently, there is a misprint in the published condition: ... meet BC at ..." should read ... meet BD at ...". We solve the corrected version. Without loss of generality, let AB = 1, and put EB = x, and FD = y . Then DB = 1, AE = 1 , x, and AF = 1 , y . By applying the Cosine Law to 4AEF EAF = 60 we transform the condition x2 + y 2 = EF 2 to
2284.
x2 + y2 = 1 , x2 + 1 , y2 , 2 1 1 , x1 , y; 2
1 , x , y , xy = 0;
which simpli es to and so,
Furthermore, BP is the anglebisector of EBC in 4EBC , so, by a wellknown formula,
y = 1,x : 1+x
1
BC cos x BP = 2EBEB + BC 60 = x + 1 : y Similarly, DQ = y+1 . Using 1, DQ = 1,x . Now, 2
Finally,
x 1 2 PQ = DB , BP , DQ = 1 , x + 1 , 1 , x = 21+ xx : 2 + BP 2 + DQ2 =
x
2 1 , x
2 x+1 + 2 2+ 4 1 2 2 = 1 + 2x x2x = 41+ xx2 = PQ2 ; 41 + +
which completes the proof.
522
Also solved by FRANCISCO BELLOT ROSADO, I.B. Emilio Ferrari, Valladolid, Spain; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; GORAN CONAR, student, Gymnasium Varadin, Varadin, Croatia; NICOLAOS z z DERGIADES, Thessaloniki, Greece; FLORIAN HERZIG, student, Cambridge, UK;
VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; ANGEL JOVAL ROQUET, I.E.S. Joan Brudieu, La Seu d'Urgell, Spain; D.J. SMEENK, Zaltbommel, the Netherlands; PANOS E. TSAOUSSOGLOU, Athens, Greece; and the proposer. Most of the submitted solutions are similar to the above. All solvers have noticed the misprint and solved the corrected problem.
2285. 1997: 432 Proposed by Richard I. Hess, Rancho Palos Verdes, California, USA.
An isosceles right triangle can be 100 covered by two congruent tiles. Design a connected tile so that two of them maximally cover a nonisosceles right triangle. The two tiles must be identical in size and shape and may be turned over so that one is the mirror image of the other. They must not overlap each other or the border of the triangle. What coverage is achieved for a 30 60 90 right triangle? Editorial comments by Bill Sands. Two readers sent in solutions to this problem, but both seem to have assumed that the tile must be a triangle, so the amount of the right triangle their solutions cover is quite a bit less than optimal. The proposer's solution is the best of the three received, although it has not been proved optimal either, and this in fact may be quite di cult to do! The proposer's solution also had an error at one point, which slightly a ected the end results. So below is a corrected version of his solution. The proposer has independently veri ed the numerical results given below. Let the given right triangle have angle 45 , with the side adjacent to this angle having length 1, so the other side has length tan and the hypotenuse has length sec . The area of the triangle is tan =2. The proposer gives two methods for partially tiling this triangle with two congruent tiles, each method better than the other for certain values of .
523
Method A.
tan
csc x pp pp pp pxppp p pxp cotpp p p p 
x
tan 1
From the gure, x + tan + x cot = 1, giving The uncovered area is
x = 1 , tan : 1 + cot
x2 cot = 1 , tan 2 cot ; 2 21 + cot 2
so the fraction of uncovered area is
, tan
2 Fa = 1 + tan : 1 Note that this method is only valid for 45 , and for = 45 gives Fa = 0, the perfect tiling of the isosceles right triangle.
Method B.
A3 ! ppp p ! v
pp pp p
6
w pp . ppp pp p pp p p v. A2 p ? p p pp pp p p
1
. A1 p p pp . pp p .

. = 2
524 Let = 2 . From the gure one can derive, for arbitrary v between 0 and tan =1 + cos , this formula was given incorrectly by the proposer , and
2 A1 = sin w cot , 1 ; 2
sin2 w = vcos ++ sin cos
2 A2 = v sin ; 4
A3 = sin w , v cos 2tan , v , w ; where A1 ; A2 ; A3 are the uncovered areas of the triangle as shown. For a given , and hence , vary v so that A = A1 + A2 + A3 is minimized. After much calculation, one obtains: the minimum value of A occurs at
whence
v = cos3 sin 2 cos 2 , 1 ; + cos
cos w = cos2sin+ cos , 1 ; and the minimum value of A is 3 3 cos2 , 4 A = 2sin 2 cos 1cos3 + 2cos 2 , 1 2 : cos , cos , 1
Thus the uncovered fraction is
2 2 , 4 , Fb = sincos3 cos 2 cos2cos, 12 1 : 3 +
Fb = 20p6 + 15p3 , 22p2 + 74 :019915536; 15 3 + 36 that is, just under 2 of the 30 60 90 right triangle remains uncovered. Can
anyone do better? One can also calculate that Method A and Method B give the same uncovered area when satis es the remarkably simple equation
When = 30 = 15 , we get
p
4 sin cos4 = 1; namely for 17:648 , or 35:3 . Method B gives the better result when 35:3 , and Method A when 35:3 45 . In particular,
Method B is the better way for the 306090 triangle.
525 If anyone can nd a reasonably short derivation of these relations, let us know! And of course the entire problem is still open, in the sense that no proof of minimality has been given, including for the 30 60 90 triangle. Maybe some reader can nd a third method that beats both of the proposer's methods for some values of .
2287. 1997: 501 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. Let G denote the point of intersection of the medians, and I denote the point of intersection of the internal angle bisectors of a triangle. Using only an unmarked straightedge, construct H , the point of intersection of the altitudes. Solution  see problem 2234 1997:168, 1998: 247
New solutions were sent in by FLORIAN HERZIG, student, Cambridge, UK; and MICHAEL LAMBROU, University of Crete, Crete, Greece.
2288. 1997: 501 Proposed by Victor Oxman, University of Haifa, Haifa, Israel. In the plane are a circle without centre and ve points A, B , C , D, E, on it such that arc AB = arc BC and arc CD = arc DE. Using only an unmarked straightedge, construct the midpoint of arc AE . Solution  see problem 2251 1997: 300, 1998: 373
New solutions were sent in by NIKOLAOS DERGIADES, Thessaloniki, Greece; CYRUS HSIA, student, University of Toronto, Toronto, Ontario; and GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria.
USA.
2289?.
1997: 501 Proposed by Clark Kimberling, Evansville, IN,
Use any sequence, fck g, of 0's and 1's to de ne a repetitionresistant sequence s = fsk g inductively as follows: 1. s1 = c1 , s2 = 1 , s1 ; 2. for n 2, let L = maxfi 1 : sm,i+2 ; : : : ; sm; sm+1 = sn,i+2 ; : : : ; sn ; 0 for some m ng; L0 = maxfi 1 : sm,i+2 ; : : : ; sm; sm+1 = sn,i+2 ; : : : ; sn ; 1 for some m ng: so that L is the maximal length of the tailsequence of s1; s2 ; : : : ; sn ; 0 that already occurs in s1 ; s2 ; : : : ; sn , and similarly for L0 , and
526
8 0 if L L0 ; sn+1 = : 1 if L L00 ; cn if L = L :
For example, if ci = 0 for all i, then
s = 0; 1; 0; 0; 0; 1; 1; 0; 1; 0; 1; 1; 1; 0; 0; 1; 0; 0; 1; 1; 1; 1; 0; 1; 1; 0; 0; 0; 0; 0; 1; 0; 1; 0; : : :
Prove or disprove that s contains every binary word. No solutions have been received  the problem remains open.
2290. 1997: 501 Proposed by Panos E. Tsaoussoglou, Athens, Greece. For x; y;z 0, prove that
,x + yy + zz + x2 xyz2x + y + z2y + z + x2z + x + y :
Solution by Florian Herzig, student, Cambridge, UK. First note that we can assume that x, y , z are greater than zero since 1 1 1 otherwise the inequality becomes trivial. Now let a = x ; b = y and c = z . The inequality is therefore equivalent to
a + b2b + c2 c + a2 bc + mca + mab + m ; where m = bc + ca + ab. Since a + ba + c = a2 + m, etc., this is
equivalent to
a2 + mb2 + mc2 + m bc + mca + mab + m ;
which is a consequence of Cauchy's Inequality:
a2 + mb2 + m ab + m2
the three such expressions are then multiplied together. Equality therefore holds if and only if a = b = c; that is, x = y = z or two of the x, y , z are zero.
Also solved by MICHEL BATAILLE, Rouen, France; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE CHRONIS, Athens, Greece; NIKOLAOS DERGIADES, Thessaloniki, Greece; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MURRAY S. KLAMKIN, University of Alberta, Edmonton, Alberta; MICHAEL LAMBROU, University of Crete, Crete, Greece 3 solutions; KEEWAI LAU, Hong
527
Kong; VEDULA N. MURTY, Dover, PA, USA; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; HEINZJURGEN SEIFFERT, Berlin, Germany; DIGBY SMITH, Mount Royal College, Calgary, Alberta; GEORGE TSAPAKIDIS, Agrinio, Greece; G. TSINTSIFAS, Thessaloniki, Greece; and the proposer.
1997: 501 Proposed by K.R.S. Sastry, Dodballapur, India. Let a, b, c denote the side lengths of a Pythagorean triangle. Suppose that each side length is the sum of two positive integer squares. Prove that 360jabc. Solution by Florian Herzig, student, Cambridge, UK. More generally we show that this is even true for all right triangles with each side being the sum of any two integer squares. We use the following wellknown criterion ? throughout the solution: a positive integer n is the sum of two integer squares if and only if all prime factors p with p 3 mod 4 are contained in n to an even power. We may assume that a, b, c have no common factor since if d 1 is the highest common factor of them and a = a1 d, etc., then by ? a1 , etc. are again sums of two squares d has to contain each prime of the form 4k + 3 to an even power. Hence we can write
2291.
a = m2 , n2 b = 2mn c = m2 + n2 where m n are relatively prime and m 6 n mod 2. For the power of 3 in abc we know that a2 + b2 = c2 and since 2 1 mod 3 for x not divisible by 3, it follows that at least one of a, x b, c is divisible by 3. By ? this implies that this one is even divisible by 9, whence 9jabc. For the power of 2 in abc notice that m even, n odd is impossible since it would then follow that a 3 mod 4. Thus m is odd and n is even. As b=2 is the sum of two integer squares and m, n have no common factor, it follows that m, n are each the sum of two integer squares. The same follows for m , n and m + n as a = m , nm + n is the sum of two integer squares. Hence m and m , n are of the form 4k + 1 as both are odd. But this implies that n is divisible by 4 and so 8jabc. Finally as x2 1 or 4 mod 5 for x not divisible by 5, we conclude since a2 + b2 = c2 that at least one of a, b, c is divisible by 5 and hence also abc. This shows that 360jabc.
Moreover this is the best possible, even for side lengths being the sum of two positive integer squares: just consider a; b;c = 225; 272; 353, 153; 104; 185, and 65; 72; 97.
Also solved by RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL
528
LAMBROU, University of Crete, Crete, Greece; JOEL SCHLOSBERG, student, Robert Louis Stevenson School, New York, NY, USA; HEINZJURGEN SEIFFERT, Berlin, Germany; KENNETH M. WILKE, Topeka, Kansas, USA; and the proposer. There were three incorrect and two incomplete solutions submitted.
1997: 502 Proposed by K.R.S. Sastry, Dodballapur, India. A convex quadrilateral Q has integer values for its angles, measured in degrees, and the size of one angle is equal to the product of the sizes of the other three. Show that Q is either a parallelogram or an isosceles trapezium. Solution by Florian Herzig, student, Cambridge, UK slightly edited. Let the angles be x, y , z , and xyz in degrees, which are positive integers. Then x + y + z + xyz = 360, and since the quadrilateral is convex, xyz 180. Without loss of generality, x y z. Hence
2292.
3z x + y + z = 360 , xyz
and so z 60. But then 180 leaves us two cases to check.
180 ;
xyz
60xy, which gives xy 2, and 180.
If x = 1 and y = 2, then z = 119, which contradicts xyz
If x = y = 1 we obtain z = 179 and the last angle is also 179 degrees. Depending on whether the angles of one degree are opposite or adjacent, Q is a parallelogram or isosceles trapezium.
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; NIKOLAOS DERGIADES, Thessaloniki, Greece; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; RICHARD I. HESS, Rancho Palos Verdes, California, USA; CYRUS HSIA, student, University of Toronto, Toronto, Ontario;
VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; VICTOR OXMAN, University of Haifa, Haifa, Israel; D.J. SMEENK, Zaltbommel, the Netherlands; KENNETH M. WILKE, Topeka, Kansas, USA; and the proposer.
529
2293. 1997: 502 Proposed by Claus Mazanti Sorensen, student, Aarhus University, Aarhus, Denmark. A sequence, fxn g, of positive integers has the properties:
1. for all n 1, we have xn,1 nxn ; 2. for arbitrarily large n, we have x1 x2 : : : xn,1 nxn ; 3. there are only nitely many n dividing x1 x2 : : : xn,1 . Prove that
1 X ,1k is irrational. k=1 xk k!
Solution by Michael Lambrou, University of Crete, Crete, Greece. We shall show the required irrationality without using conditions 1 and 3. We argue by contradiction. Suppose condition 2 is valid for all n n0 for some xed n0 2 N, and suppose on the contrary that
1 X ,1k p = q k=1 xk k!
with p; q 2 N. By multiplying both terms of the fraction p=q by a positive integer, if necessary, we may assume that q n0. In particular we have by condition 2 that
x1x2 : : : xq q + 1xq+1 1 and, for all k q + 1, since xi 1 and 0 xk,1 x1 x2 : : : xk,1 kxk that 0 xk,1 kxk . So for k q + 1 we have xk,1 k , 1!
Thus the terms of
1 X ,1k k=1 xk k!
1
1 : xkk!
are decreasing in absolute value. So by a wellknown estimate of alternating series see, for example, G.H. Hardy, A Course of Pure Mathematics, Cambridge University Press, Tenth Edition, 1952, p. 377 we have
0
1 , xq+21 + 2! xq+1q + 1! q 1 X ,1k 1 ,1q+1 xq+1q + 1! ; k=q+1 xk k!
530 and so
! q p , X ,1k 1 0 q k=1 xk k! xq+1q + 1! : Multiplying across by q !x1x2 : : : xq we obtain " q k q+1 q !x1x2 : : : xq p , X ,1 q !x1x2 : : : xq 0 ,1 q k=1 xkk! x1x2 : : : xq x q + 1 :
,1q+1
q+1
The middle term is an integer as a sum of integers but by inequality 1 it is strictly between 0 and 1. This is absurd, and the required irrationality follows. Also solved by FLORIAN HERZIG, student, Cambridge, UK; and the proposer. Herzig also avoided use of condition 3.
2294. 1997: 502 Proposed by Zun Shan and Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. For the annual SinoJapanese Go" tournament, each country sends a team of seven players, Ci 's and Ji 's, respectively. All players of each country are of di erent ranks strengths, so that
C1 C2 : : : C7
and
J1 J2 : : : J7:
Each match is determined by one game only, with no tie. The winner then takes on the next higher ranked player of the opponent country. The tournament continues until all the seven players of one country are eliminated, and the other country is then declared the winner. For those who are not familiar with the ancient Chinese Chess" game of Go", a better and perhaps more descriptive translation would be the surrounding chess". a What is the total number of possible sequences of outcomes if each country sends in n players? b? What is the answer to the question in part a if there are three countries participating with n players each, and the rule of the tournament is modi ed as follows: The rst match is between the weakest players of two countries determined by lot, and the winner of each match then plays the weakest player of the third country who has not been eliminated if there are any left. The tournament continues until all the players of two countries are eliminated.
531 Solution to a by Walther Janous, Ursulinengymnasium, Innsbruck, Austria and the proposers. Let Tn denote the number sought. Suppose, for some k = 1, 2, : : : , n, the winning country has lost k , 1 of its players before the tournament is over; that is, the kth player of the winning country defeats the strongest player of the losing country. Since each match eliminates exactly one player, the total number of matches before the last game is n , 1+ k , 1, or n + k , 2, and among these n + k , 2 matches, k , 1 of them resulted in the elimination of a player from the winning country. , Hence,for each k, the total number of k such sequences of outcomes is Wk = n+,,2 , and so k 1
n X n + k , 2
Tn = 2 Wk = 2 k,1 k=1 k n , 1
=1
n + 1
2n , 2
n + = 2 0 + 2 + :::+ 2n , 1
21n 2n , 1
2n
n , 1 = 2 n,1 = n n,1 = n : n X r X n + k
n + r + 1
= : r k=0 k
Ed: The summation formula used above is a special case of the wellknown identity This identity can readily be proved by induction. See, for example, p. 207 of Applied Combinatorics 2nd Edition, John Wiley and Sons, Inc., 1984, by Alan Tucker. Alternate solution to a by the proposers. We represent the 2n players by a row of 2n white marbles. Choose any n of the marbles and colour them black. Let the white black marbles represent the players from China Japan, and so they are listed from left to right in order of their elimination, so that the colour of the last marble indicates the winning country. Clearly, there is a onetoone correspondence between all such colourings and all the possible sequences of outcomes; for example, the colouring depicted below corresponds to the following sequence of outcomes: C1 defeats J1 and J2, but loses to J3; C2 defeats J3 and J4, but loses to J5; C3 defeats J5, but loses to J6 ; C4 loses to J6; C5 defeats J6, but loses to J7 ; nally, C6 defeats J7 .
w
w
g
w
w
g
w
g
g
w
g
w
g
g
532 Therefore Tn =
open.
No other solutions were received to either a or b? . Hence, b? remains
2n
n as in solution I.
If one lets f n denote the number sought, then using bruteforce computations on a tree diagram, the proposers found that f 1 = 12 and f 2 = 84. They feel that, for general n, this is a di cult problem.
2295. 1997: 503 Proposed by D.J. Smeenk, Zaltbommel, the Netherlands. Find three positive integers a, b, c, in arithmetic progression with positive common di erence, such that a + b, b + c, c + a, are all perfect squares. Solution by Murray S. Klamkin, University of Alberta, Edmonton, Alberta. We take the arithmetic progression as x , y , x, and x + y . Then we must have
2x , y = u2; 2x = v 2; 2x + y = w2 so that x = w2 + u2 =4, y = w2 , u2 =2, and u2 + w2 = 2v 2.
The problem of nding three squares in arithmetic progression is wellknown and is obtained easily from the identity
1 + ip + iq2 = p2 , q 2 , 2pq + ip2 , q 2 + 2pq : 2p2 + q22 = 2pq + p2 , q 22 + 2pq + p2 , q 22:
On taking the square of the absolute values of each side, we have Hence we can take
u = 2pq + q2 , p2; w = 2pq + p2 , q2; and v = p2 + q2 with p q. In order to have x y , we must have 3u2 w2 or p2+q 2 2 8pq p2,q 2. This can be satis ed for an in nite number of pairs p; q in which p is close to q and not too small. For example, if we choose p = 10 and q = 9, we get x = 1992 + 1612 =4 and y = 1992 , 1612=2. Since the Diophantine equation is homogeneous, we can multiply x and y by 4, to give x = 65;522 and y = 27;360, and then 2x , y = 3222, 2x = 3622, and 2x + y = 3982.
Also solved by CHARLES ASHBACHER, Cedar Rapids, Iowa, USA; SAM BAETHGE, Nordheim, Texas, USA; NIELS BEJLEGAARD, Stavanger, Norway; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; THEODORE CHRONIS, student, Aristotle University of Thessaloniki, Greece; CHARLES R. DIMINNIE, Angelo State University, San Angelo, TX, USA; FLORIAN HERZIG, student, Cambridge, UK; RICHARD I. HESS, Rancho Palos Verdes, California, USA; WALTHER
JANOUS, Ursulinengymnasium, Innsbruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete,
533
Crete, Greece; GERRY LEVERSHA, St. Paul's School, London, England; KATHLEEN E. LEWIS, SUNY Oswego, Oswego, NY, USA; DAVID E. MANES, SUNY at Oneonta, Oneonta, NY, USA; GOTTFRIED PERZ, Pestalozzigymnasium, Graz, Austria; HEINZJURGEN SEIFFERT, Berlin, Germany; DIGBY SMITH, Mount Royal College, Calgary, Alberta; KENNETH M. WILKE, Topeka, Kansas, USA; and the proposer. There was one incorrect solution submitted. Hess and Wilke both determine constraints on the ratio p=q which will guarantee that x y above: or
Many solvers also pointed out that to get integers for x and y we need to have p and q having the same parity, which explains why Klamkin needed to multiply his values by 4 to get integers.
p q p q
2 , 3 + 2 2 , 3 1:303225373 2 + 3 + 2 2 + 3 7:595754113 :
p p
q q
p p
2296. 1997: 503 Proposed by Vedula N. Murty, Visakhapatnam, India. x 2x2 for 0 x 1. Show that sin2 2
2
Hence or otherwise, deduce that
1+x
Solution by Florian Herzig, student, Cambridge, UK. From the in nite product formula for cos x Ed.: See, for example, 1.431, 3 on p. 45 of Tables of Integrals, Series and Products 5th edition, by I.S. Gradshteyn and I.M. Ryzhik, Academic Press, 1994 , we get
sin x x1 , x
4 for 0 x 1.
cos 2
Since 0
x
=
1 Y
1 , x2 1 + x2 1 for 0 x 1, we have
1 x2 2 x cos 2 1 , x22 1 , x2 ; + sin
2
k=0
1 , 2k + 12 1 , x2
x2
for 0 x 1 :
1
or, equivalently,
2x2 : 2 1 + x2 From 1, we deduce that cosy 1 , 4y 2 for 0 y 1 . Hence, for 2 0 x 1 and y = 1 , x, we get 2 2 sin x 1 , 1 , 2x2 = 4x1 , x :
x
534
x Dividing by x1 , x and noting that f x = sin,x = f 1 , x, we have x1 proved the right side of the double inequalities. Note that equality holds when x = 1 . Ed.: Hence, the right inequality, as given, was incorrect. 2 To prove the left inequality, note rst that y = cost is concave on 1 0; 1 . Since the chord joining 0; 1 to 2 ; 0 has equation y = 1 , 2t, we 2 1 have, by integrating from 0 to x 0 x 1 that sinx x1 , x, 2 x or sin,x . Again, since f x = f ,x, the inequality holds for all x1 x 2 0; 1.
Also solved by PAUL BRACKEN, CRM, Universit
de Montr
al, Montr
al, e e e Qu
bec; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; NIKOLAOS e DERGIADES, Thessaloniki, Greece; WALTHER JANOUS, Ursulinengymnasium, Inns
bruck, Austria; VACLAV KONECNY, Ferris State University, Big Rapids, Michigan, USA; MICHAEL LAMBROU, University of Crete, Crete, Greece; HEINZJURGEN SEIFFERT, Berlin, Germany; and the proposer. There were also one incomplete and one incorrect solution. The in nite product formula was also used by Kone cny, Lambrou and Sei ert.
Besides Herzig, Bradley, Lambrou and Sei ert also pointed out the error mentioned in the solution above. Herzig remarked that the same problem appeared on p. 277 of G.H. Hardy's A Course of Pure Mathematics, Cambridge University Press, and that it was also used in the Mathematical Tripos 1993 for part IA Cambridge exams after the rst year, paper 3, problem 7D. Janous asked the question:
Determine the set of all real numbers such that sin x 1 2x x 2 + holds for all x 2 0; 1 .
Lambrou strengthened the rst inequality to
sin2 x 2
20x2 91 + x2 2x2 1 + x2
if 0
1 x 2;
Both Kone cny and Sei ert pointed out that the left hand side of the double inequal
itiers is weaker than the following known inequality:
sin2 x x 2
if 1 x 2
1:
1 , x2 sinx 1 + x2 x
for all x 6= 0 :
See CRUX 1993: 433 . According to Klamkin and Manes, this inequality is not as easy to prove".
535
2297. 1997: 503 Proposed by Bill Sands, University of Calgary, Calgary, Alberta. Given is a circle of radius 1, centred at the origin. Starting from the point P0 = ,1; ,1, draw an in nite polygonal path P0 P1 P2 P3 : : : going counterclockwise around the circle, where each Pi Pi+1 is a line segment tangent to the circle at a point Qi , such that jPi Qi j = 2jQi Pi+1 j. Does this path intersect the line y = x other than at the point ,1; 1? Solution by Richard I. Hess, Rancho Palos Verdes, California, USA. The total length of the in nite B polygon P0 P1 P2 P3 : : : is q
q1
P0
qP1 qP qQ
q
Aq
1
Q0
q
1 2
P1
q
1 2
q Q
2 1 4 1 2 4 1
The total length of the arc from A to B is 3. Therefore P0P1P2 : : : will never reach B. By calculation the limit point P1 has angle 1 154:76 degrees.
1 1 1 1+ 2 + 1 + 1 + 1 + 8 + 8 + = 3 : 2 4 4
Also solved by CHRISTOPHER J. BRADLEY, Clifton College, Bristol, UK; NIKOLAOS DERGIADES, Thessaloniki, Greece; FLORIAN HERZIG, student, Cambridge, UK; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; VICTOR OXMAN, University of Haifa, Haifa, Israel; and the proposer. Janous and Oxman also gave the simple solution above.
2298. 1997: 503 Proposed by Bill Sands, University of Calgary, Calgary, Alberta. The Tickle Me" Feather Company ships its feathers in boxes which cannot contain more than 1 kg of feathers each. The company has on hand a number of assorted feathers, each of which weighs at most one gram, and whose total weight is 1000001=1001 kg. Show that the company can ship all the feathers using only 1000 boxes. Solution by Kathleen E. Lewis, SUNY Oswego, Oswego, New York, USA. Before packing the feathers in the boxes, rst sort them by weight, from heaviest to lightest. Now put 1000 feathers in each box, beginning with the heaviest. Thus, the 1000000 heaviest feathers are now in the boxes with no box containing more than 1 kg of feathers. For each remaining feather beginning still with the heaviest, put it in the rst box that it will t" in; that is, the rst which it will not put over weight capacity. Continue this process until all feathers are packed or until you reach a feather that will not t anywhere.
536 Suppose you nd a feather that does not t. Then in each box, adding that feather would push the weight over 1 kg. Since there are at least 1000 feathers already in each box, which are each the same weight as, or heavier than, the new feather that will not t, adding this feather would increase the weight by a factor of at most 1001=1000. Thus the weight of each box without this feather must be strictly greater than 1000=1001 kg. If we add the weight of this feather to that of the feathers in any one box, it exceeds 1 kg. The total weight of the feathers in the other 999 boxes is more than 999 1000=1001 kg. Putting these weights together, we get a value strictly greater than the 1000001=1001 kg which is supposed to be the total weight of all the feathers. Therefore this situation cannot happen, so it must be possible to pack all the feathers.
Also solved by NIKOLAOS DERGIADES, Thessaloniki, Greece; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; MICHAEL LAMBROU, University of Crete, Crete, Greece; DIGBY SMITH, Mount Royal College, Calgary, Alberta; and the proposer. Janous and the proposer note the more general result: let a 1 and let n be a positive integer, and let fw1 ; w2 ; : : : g be positive real numbers satisfying
of each set is at most 1. Readers may like to prove this. The given problem is the case a = n = 1000. The proposer also notes that the result is best possible in two ways: 1 1 if there are 1000001 feathers each of weight 1001 + 1000 kg, for su ciently small, then the total weight of the feathers is just slightly greater than 1000001=1001 kg, but if only 1000 boxes are used, some box must contain at least 1001 feathers which would weigh altogether more than 1 kg; 1 if there are 999001 feathers each of weight 1000 + kg, for su ciently small, then each feather weighs just slightly more than 1 gm, and the total weight of the feathers is still at most 1000001=1001 kg, but if only 1000 boxes are used, some box must contain at least 1000 feathers which would weigh altogether more than 1 kg.
+ wi na+ 11 : a Then the wi 's can be partitioned into at most n sets so that the sum 1 max wi a
and
X
Founding Editors R
dacteursfondateurs: L
opold Sauv
& Frederick G.B. Maskell e e e Editors emeriti R
dacteuremeriti: G.W. Sands, R.E. Woodrow, Bruce L.R. Shawyer e Founding Editors R
dacteursfondateurs: Patrick Surry & Ravi Vakil e Editors emeriti R
dacteursemeriti: Philip Jong, Je Higham, e J.P. Grossman, Andre Chang, Naoki Sato, Cyrus Hsia
Crux Mathematicorum
Mathematical Mayhem
537
YEAR END FINALE
Again, a year has own by! This year was owing smoothly until I had a three month sickleave break. Our Associate Editor, CLAYTON HALFYARD, completed issue 5, and did issues 6 and 7. He did a splendid job. In particular, Memorial University student PAUL MARSHALL did almost all the diagrams for issues 6 and 7. He too did a splendid job. I have resumed responsibilities with this issue. Thanks to all for their patience and assistance. The online version of CRUX with MAYHEM continues to attract attention. There is a new look" now. We recommend it highly to you. Thanks are due to LOKI JORGENSON, NATHALIE SINCLAIR, FREDERIC TESSIER, and the rest of the team at SFU who are responsible for this. There are many people that I wish to thank most sincerely for particular contributions. Again, rst and foremost is BILL SANDS. Bill is of such value to me and to the continuance of CRUX with MAYHEM. As well, I thank most sincerely, CATHY BAKER, ILIYA BLUSKOV, ROLAND EDDY, CHRIS FISHER, BILL SANDS, JIM TOTTEN, and EDWARD WANG, for their regular yeoman service in assessing the solutions; DENIS HANSON, DIETER RUOFF, CHRIS FISHER, RICHARD MCINTOSH, DAIHACHIRO SATO, DOUG FARENICK, HARLEY WESTON, for ensuring that we have qual
ity articles; TED LEWIS, ANDY LIU, MARIA FALK de LOSADA, BILL SANDS, JIM TIMOURIAN, for ensuring that we have quality book reviews, ROBERT WOODROW, who carries the heavy load of two corners, and RICHARD GUY for sage advice whenever necessary. The editors of the MAYHEM section, NAOKI SATO, CYRUS HSIA, ADRIAN CHAN, RICHARD HOSHINO, DAVID SAVITT and WAI LING YEE, all do a sterling job. I also thank two of our regulars who assist the editorial board with proof reading; THEODORE CHRONIS and WALDEMAR POMPE. The quality of these people is a vital part of what makes CRUX with MAYHEM what it is. Thank you one and all. As well, I would like to give special thanks to our Associate Editor, CLAYTON HALFYARD, for guiding issues 5, 6 and 7 to fruition, and for keeping me from printing too many typographical and mathematical errors; and to my colleagues, PETER BOOTH, RICHARD CHARRON, ROLAND EDDY, EDGAR GOODAIRE, ERIC JESPERS, MIKE PARMENTER, DONALD RIDEOUT, NABIL SHALABY, in the Department of Mathematics and Statistics at Memorial University, and to JOHN GRANT MCLOUGHLIN, Faculty of Education, Memorial University, for their occasional sage advice. I have also been helped by some Memorial University students, MIKE GILLARD, DON HENDER, PAUL MARSHALL, JON MAUGER, SHANNON SULLIVAN, TREVOR RODGERS, as well as a WISE Summer student, SHAWNA LEE.
538 The staff of the Department of Mathematics and Statistics at Memorial University deserve special mention for their excellent work and support: ROS ENGLISH, MENIE KAVANAGH, WANDA HEATH, and LEONCE MORRISSEY; as well as the computer and networking expertise of RANDY BOUZANE, DAVID VINCENT and PAUL DROVER. A Also the LTEX expertise of JOANNE LONGWORTH at the University of Calgary, ELLEN WILSON at Mount Allison University, and the MAYHEM sta , is much appreciated. GRAHAM WRIGHT, the Managing Editor, continues to be a tower of strength and support. Graham has kept so much on the right track. He is a pleasure to work with. The CMS's TEX Editor, MICHAEL DOOB has been very helpful in ensuring that the printed master copies are up to the standard required for the U of T Press who continue to print a ne product. Finally, I would like to express real and heartfelt thanks to the Head of my Department, HERBERT GASKILL, and to the Acting Dean of Science of Memorial University, WILLIE DAVIDSON. Without their support and understanding, I would not be able to do the job of EditorinChief. Last but not least, I send my thanks to you, the readers of CRUX with MAYHEM. Without you, CRUX with MAYHEM would not be what it is. Keep those contributions and letters coming in. We need your ARTICLES, PROPOSALS and SOLUTIONS to keep CRUX with MAYHEM alive and well. I do enjoy knowing you all. In a letter from Ron Weedon, Reading, UK, concerning the reference to F. G.M., the author of the valuable 1912 book entitled Exercises de G
om
trie, we learned that the author's full name is F. GabrielMarie which e e members of the Editorial Board had not known. The letter made us wonder if there is a copy of the F. G.M. book anywhere in Canada. Let us know if you have a copy. Apparently, the University of Toronto does not have a copy, and the copy in the British Library is damaged.
?
u u
CRUX
u
u
MAYHEM
u
539
INDEX TO VOLUME 24, 1998
Crux Articles
Cyclic ratio sums and products Branko Grunbaum ......................................................... 20 One Problem Six Solutions Georg Gunther ............................................................. 81 Sum of powers of a nite sequence: a geometric approach William O.J. Moser ....................................................... 145 A Note on Special Numerals in Arbitrary Bases Glenn Appleby, Peter Hilton and Jean Pedersen ......................... 210 Pythagoras Strikes Again! K.R.S. Sastry ............................................................. 276 The Academy Corner Bruce Shawyer February No. 16 .......................................................... 1 March No. 17 ......................................................... 65 April No. 18 ....................................................... 129 May No. 19 ....................................................... 193 September No. 20 The Bernoulli Trials 1998 Questions Ian Goulden and Christopher Small ............... 257 December No. 21 The Bernoulli Trials 1998 Answers Ian Goulden and Christopher Small ............... 449 The Olympiad Corner R.E. Woodrow February No. 187 ........................................................ 4 March No. 188 ....................................................... 68 April No. 189 ...................................................... 131 May No. 190 ...................................................... 196 September No. 191 ...................................................... 261 October No. 192 ...................................................... 322 November No. 193 ...................................................... 385 December No. 194 ...................................................... 452 The Skoliad Corner R.E. Woodrow February No. 27 ........................................................ 26 March No. 28 ........................................................ 88 April No. 29 ....................................................... 148 May No. 30 ....................................................... 215 September No. 31 ....................................................... 281 October No. 32 ....................................................... 334 November No. 33 ....................................................... 398 December No. 34 ....................................................... 477
Miscellaneous
Canadian Mathematical Society Award for Contributions to Mathematical Education Eric Muller, Chair, CMS Education Committee ........................... 18 Letter From the Editors .................................................... 321 Year End Finale ............................................................. 537
540
Book Reviews Andy Liu MiniReviews Update Reviewed by Andy Liu ..................................................... 12 From Erdos to Kiev: Problems of Olympiad Caliber
by Ross Honsberger Reviewed by Bill Sands .................................................... 78
Vita Mathematica Mathematical Challenge & More Mathematical Challenges Models that Work
Edited by Ronald Calinger Reviewed by Maria de Losada ........................................... 143 by Tony Gardiner Reviewed by Ted Lewis ................................................... 208 by Alan Tucker Reviewed by Jim Timourian .............................................. 275
Juegos y acertijos para la ense~ anza de las Matem
ticas n a Dissections: Plane and Fancy The Last Recreations Problems
by Bernardo Recam
n Santos a Reviewed by Francisco Bellot Rosado .................................... 333 by Greg N. Frederickson Reviewed by Andy Liu .................................................... 396 by Martin Gardiner Reviewed by Andy Liu .................................................... 475 February 2301 2313 ..................................................... 45 March 2314 2325 .................................................... 107 April 2306, 2326 2337 ............................................. 175 May 2338 2350 .................................................... 234 September 2326, 2329, 2351 2363 ....................................... 301 October 2364 2375 .................................................... 363 November 2374, 2376 2387 ............................................. 424 December 2388 2400.................................................... 503 February 2145, 2153, 2167, 2181, 2198 2207 ........................... 48 March 2208 2220 .................................................... 110 April 2219 2222, 2224 2226, 2229 2230 .......................... 178 May 2223, 2227 2228, 2231 2239 ................................. 237 September 2015, 220A, 2206, 2240 2246, 2248 .......................... 305 October 2247, 2249 2256, 2258 ....................................... 366 November 1637, 2257, 2260 2263, 2265 2266, 2268, 2270 2275, 2277 2278 ................................. 427 December 2090, 2259, 2264, 2267, 2269, 2276, 2279 2285, 22872298 ................................. 507
Solutions
541
Mathematical Mayhem
Mayhem Articles
February March April May September October November December
................................................................. 30 ................................................................. 92 ............................................................... 158 ............................................................... 220 ............................................................... 286 ............................................................... 342 ................................................................ 406 ............................................................... 485
Shreds and Slices
Powers of Two Wai Ling Yee ............................................................... 33 The Cantor Set and Cantor Function Naoki Sato ................................................................. 37 The Order of a Zero Naoki Sato ................................................................ 101 Tips on Inequalities Naoki Sato ................................................................ 161 Riveting Properties of Pascal's Triangle Richard Hoshino .......................................................... 168 The Pentagon Theorem Hiroshi Kotera ............................................................ 291 A Combinatorial Triad Cyrus Hsia ................................................................ 296 Pushing the Envelope Naoki Sato ................................................................ 357 Bogus Arguments and Arcane Identities Ravi Vakil ................................................................. 413 The Fibonacci Sequence Wai Ling Yee .............................................................. 415 Dividing Points Equally Cyrus Hsia ................................................................ 499 February ................................................................. 30 March ................................................................. 92 May ............................................................... 220 September ............................................................... 286 November ................................................................ 406 December ............................................................... 494 The K Method ............................................................... 30 The Fibonacci Triangle ....................................................... 92 From the Archives .......................................................... 220 Primitive Roots and Quadratic Residues, part 2 ........................... 286 An Algebraic Relation with a Geometric Twist ............................. 406 IMO Report Adrian Chan ................................................... 412 1998 1999 Olympiad Correspondence Problems E.J. Barbeau ............. 495
Miscellaneous
542
Mayhem Problems
High School Problems
February March April May September October November December
................................................................. 41 ................................................................. 94 ............................................................... 158 ............................................................... 222 ............................................................... 288 ............................................................... 342 ................................................................ 410 ............................................................... 485
High School Solutions
February H223 H226 ................................................... 42 April H237 H240 .................................................. 159 September H241 H244 .................................................. 289 November H245 H248 .................................................. 410 March May October December H217 H221 H224 H229 H218 ................................................... 95 H224 .................................................. 222 H228 .................................................. 342 H232................................................... 485
Advanced Problems
Advanced Solutions
February A209 A212 .................................................... 43 April A213 216 .................................................... 159 September A217 A220 ................................................... 289 November A221 A224 ................................................... 411 March May October December A193 A197 A201 A205 A196 .................................................... 98 A200 ................................................... 226 A204 ................................................... 349 A208.................................................... 489
Challenge Board Problems
Challenge Board Solutions
February C75 C76 ....................................................... 43 April C77 C78 ...................................................... 160 September C79 C80 ...................................................... 290 November C81 C82 ...................................................... 411 March May October December C73 ........................................................... 100 C74 ........................................................... 228 C75 C76 ...................................................... 353 C77 C78 ...................................................... 492 231 232 298 300 421 422 497 498
Contests
Swedish Mathematics Olympiad 1986 Qualifying Round .................. J.I.R. McKnight Problems Contest 1982 ................................... Swedish Mathematics Olympiad 1987 Qualifying Round .................. J.I.R. McKnight Problems Contest 1983 ................................... J.I.R. McKnight Problems Contest 1984 ................................... Swedish Mathematics Olympiad 1988 Qualifying Round .................. J.I.R. McKnight Problems Contest 1985 ................................... Swedish Mathematics Olympiad 1989 Qualifying Round ..................
543 Proposers, solvers and commentators in the PROBLEMS and SOLUTIONS sections for 1998 are:
Gerald Allen: 2207, 2208, 2228, 2249, 2256, 2260, 2273, 2275 Hayo Ahlburg: 2249, 2268, 2279 Miguel Amengual Covas: 2209, 2235, 2237, 2238, 2244, 2249,
2253, 2254, 2258, 2259, 2263, 2280 Claudio Arconcher: 2209, 2262 Sefket Arslanagi
: 2181, 2204, 2205, 2207, 2218, 2221, 2232, c 2237, 2238, 2249 Charles Ashbacher: 2219, 2232, 2238, 2245, 2295 Sam Baethge: 2221, 2225, 2226, 2237, 2238, 2245, 2249, 2258, 2295 Edward J. Barbeau: 2219, 2219, 2278 Michel Bataille: 2207, 2232, 2249, 2290 Frank P. Battles: 2238, 2275 F.R. Baudert: 2271 Niels Bejlegaard: 2236, 2239, 2248, 2251, 2258, 2295 Francisco Bellot Rosado: 2181, 2203, 2204, 2209, 2221, 2224, 2226, 2235, 2237, 2239, 2240, 2241, 2242, 2246, 2249, 2250, 2252, 2253, 2254, 2258, 2259, 2263, 2264, 2266, 2267, 2276, 2279, 2280, 2281, 2284 Adrian Birka: 2229, 2235, 2239, 2249 Mansur Boase: 2199, 2209, 2211, 2218, 2219, 2220, 2221, 2222, 2224, 2225, 2226, 2227, 2229, 2235, 2236, 2237, 2238, 2239, 2248, 2254 Paul Bracken: 2206, 2245, 2256, 2258, 2260, 2268, 2296 Christopher J. Bradley: 2181, 2199, 2200, 2201, 2203, 2204, 2205, 2206, 2207, 2208, 2209, 2211, 2212, 2214, 2215, 2217, 2218, 2219, 2221, 2222, 2224, 2225, 2226, 2227, 2229, 2231, 2235, 2236, 2237, 2238, 2239, 2240, 2241, 2242, 2243, 2244, 2245, 2246, 2247, 2249, 2250, 2251, 2252, 2253, 2254, 2255, 2256, 2257, 2258, 2259, 2260, 2261, 2262, 2263, 2264, 2265, 2266, 2268, 2270, 2271, 2273, 2275, 2290, 2292, 2295, 2296, 2297 Jeremy T. Bradley: 2200 Adam Brown: 2181 James T. Bruening: 2245 Miguel Angel Cabez on Ochoa: 2204, 2205, 2207, 2212, 2221,
2238, 2245, 2258, 2273 Sabin Cautis: 2280 Adrian Chan: 2204, 2209, 2218, 2221, 2224, 2226, 2229, 2232, 2235, 2239, 2249 Ji Chen: 220A, 2015 Denise Cheung: 2237 Jimmy Chui: 2229, 2235, 2237, 2239, 2240, 2249 Theodore Chronis: 2181, 2215, 2218, 2245, 2248, 2249, 2253, 2254, 2255, 2256, 2258, 2260, 2271, 2274, 2275, 2279, 2284, 2290, 2295, , , 2443 Goran Conar: 2181, 2204, 2212, 2218, 2221, 2245, 2249, 2253, 2256, 2279, 2284 Curtis Cooper: 2225 Filip Crnogorac: 2235 Tim Cross: 2273 Luz M. DeAlba: 2256, 2275 PaulOlivier Dehaye: 2199 Nikolaos Dergiades: 2268, 2276, 2278, 2279, 2280, 2281, 2282, 2283, 2284, 2288, 2290, 2292, 2296, 2297, 2298 Charles R. Diminnie: 2207, 2208, 2218, 2219, 2225, 2226, 2228, 2244, 2245, 2249, 2256, 2260, 2268, 2273, 2275, 2295 Angel Dorito: 2261 David Doster: 2199, 2207, 2217, 2221, 2222, 2227, 2232, 2237, 2245, 2248, 2249, 2256, 2268, 2275, 2203, 2209, 2220, 2234, 2235, 2237, 2244, 2255, 2282 Mayumi Dubree: 2238 Keith Ekblaw: 2207, 2211, 2212, 2238, 2260, 2261, 2275 Hans Engelhaupt: 2251, 2256, 2258, 2269, 2271, 2273, 2275 Russell Euler: 2254, 2256, 2260, 2268, 2275, 2282 Noel Evans: 2256, 2273 C. FestraetsHamoir: 2201, 2203, 2204, 2208, 2209 F.J. Flanigan: 2181, 2199, 2275, 2443 Je rey K. Floyd: 2238
2207, 2208, 2209, 2211, 2212, 2213, 2214, 2215, 2217, 2218, 2219, 2219, 2220, 2221, 2224, 2225, 2226, 2229, 2230, 2231, 2232, 2235, 2237, 2238, 2239, 2245, 2247, 2248, 2251, 2253, 2254, 2256, 2258, 2259, 2260, 2261, 2262, 2263, 2264, 2265, 2266, 2267, 2269, 2271, 2272, 2273, 2274, 2275, 2276, 2277, 2279, 2280, 2281, 2282, 2283, 2284, 2287, 2290, 2291, 2292, 2293, 2295, 2296, 2297 Richard I. Hess: 2181, 2198, 2199, 2200, 2207, 2208, 2211, 2212, 2217, 2218, 2219, 2220, 2221, 2222, 2225, 2227, 2229, 2231, 2233, 2237, 2238, 2244, 2245, 2247, 2248, 2256, 2258, 2260, 2261, 2262, 2273, 2275, 2279, 2290, 2291, 2292, 2295, 2297 John G. Heuver: 2226, 2231, 2235, 2238 Joe Howard: 2206, 2214, 2218, 2232, 2256, 2260, 2275 Cyrus Hsia: 2199, 2209, 2212, 2218, 2219, 2221, 2225, 2288, 2292 Robert B. Israel: 2208, 2211, 2213, 2217, 2218, 2221, 2222, 2225, 2226, 2227 Walther Janous: 2015, 2090, 2181, 2198, 2199, 2201, 2202, 2203, 2204, 2205, 2205, 2206, 2206, 2207, 2208, 2209, 2212, 2214, 2215, 2217, 2218, 2219, 2220, 2221, 2222, 2224, 2225, 2227, 2229, 2231, 2232, 2233, 2235, 2236, 2237, 2238, 2239, 2240, 2243. 2244, 2245, 2247, 2248, 2249, 2250, 2252, 2253, 2254, 2255, 2256, 2258, 2260, 2261, 2262, 2263, 2264, 2266, 2268, 2269, 2271, 2273, 2274, 2275, 2276, 2278, 2278, 2279, 2281, 2282, 2284, 2290, 2291, 2292, 2294, 2295, 2296, 2297, 2298 D. Kipp Johnson: 2181, 2226, 2229, 2232, 2237, 2238, 2254, 2255, 2258 Dag Jonsson: 2254 Angel Joval Roquet: 2258 Geo rey A. Kandall: 2254 Murray S. Klamkin: 2145, 2181, 2199, 2279, 2290, 2295 Clark Kimberling: 2267, 2289 Kenneth Kam Chiu Ko: 2225, 2229, 2239 V
clav Kone n
: 2181, 2199, 2200, 2203, 2204, 2205, 2206, 2209, a cy 2212, 2214, 2215, 2217, 2218, 2219, 2222, 2224, 2225, 2236, 2237, 2238, 2240, 2242, 2245, 2249, 2251, 2253, 2254, 2256, 2258, 2261, 2266, 2271, 2274, 2275, 2278, 2279, 2282, 2284, 2292, 2295, 2296 Edward J. Koslowska: 2211 Sai C. Kwok: 2279 Michael Lambrou: 2181, 2198, 2199, 2201, 2202, 2203, 2204, 2205, 2205, 2206, 2207, 2208, 2209, 2211, 2212, 2213, 2214, 2215, 2215, 2217, 2218, 2219, 2220, 2221, 2222, 2224, 2225, 2226, 2227, 2228, 2229, 2231, 2232, 2233, 2235, 2236, 2237, 2238, 2239, 2240, 2241, 2242, 2243, 2244, 2245, 2246, 2247, 2248, 2250, 2251, 2253, 2254, 2255, 2256, 2257, 2258, 2259, 2260, 2261, 2262, 2263, 2265, 2266, 2267, 2268, 2269, 2271, 2273, 2274, 2275, 2276, 2278, 2278, 2279, 2280, 2281, 2283, 2284, 2287, 2290, 2291, 2292, 2293, 2295, 2296, 2297, 2298 KeeWai Lau: 220A, 2204, 2204, 2205, 2206, 2214, 2218, 2221, 2222, 2227, 2229, 2237, 2239, 2244, 2245, 2248, 2249, 2254, 2255, 2256, 2258, 2262, 2263, 2268, 2271, 2273, 2275, 2290 James Lee: 2235 Gerry Leversha: 2201, 2202, 2203, 2204, 2205, 2207, 2208, 2209, 2211, 2212, 2215, 2217, 2218, 2219, 2221, 2222, 2223, 2224, 2225, 2226, 2231, 2232, 2235, 2237, 2238, 2239, 2240, 2243, 2245, 2248, 2251, 2252, 2253, 2254, 2258, 2259, 2260, 2261, 2262, 2263, 2266, 2273, 2275, 2276, 2278, 2279, 2280, 2282, 2284, 2292, 2295 Kathleen E. Lewis: 2211, 2225, 2238, 2295, 2298
Ian June L. Garces: 2212, 2224, 2226 Robert Geretschlager: 2145, 2199, 2211, 2212 Shawn Godin: 2222, 2248, 2251 Joaqu
n Gomez Rey: 2210, 2220, 2223, 2227, 2245, 2277, 2278
Douglass Grant: 2015 Herbert Gulicher: 2231 David Hankin: 2226 Stergiou Haralampos: 2244 Yeo Keng Hee: 2203, 2214, 2217, 2221, 2225 G.P. Henderson: 2233, 2274 Florian Herzig: 2181, 2199, 2201, 2202, 2203, 2204, 2205, 2206,
544
Alan Ling: 2239, 2245, 2248, 2249 Mar
a Ascensi on Lopez Chamorro:
2226, 2253, 2265, 2267, 2276 Nick Lord: 2256, 2260 David E. Manes: 2295 2204, 2221, 2224,
HeinzJurgen Sei ert:
Pavlos Maragoudakis: 2253, 2254, 2260 Anne Martin: 2232 Giovanni Mazzarello: 2212, 2224, 2226 J.A. McCallum: 2222, 2238, 2273 Phil McCartney: 2256, 2260 Sean McIlroy: 2211, 2229, 2232 Can Anh Minh: 2218, 2232 William Moser: 2212 Vedula N. Murty: 2198, 2206, 2245, 2249, 2253, 2260, 2275,
2279, 2280, 2290, 2296 Grady Mydlak: 2238 Isao Naoi: 2279 David Nicholson: 2235 Victor Oxman: 2213, 2218, 2234, 2236, 2240, 2244, 2245, 2249, 2251, 2254, 2255, 2256, 2258, 2260, 2262, 2263, 2268, 2280, 2287, 2288, 2292, 2297 Michael Parmenter: 2181, 2260, 2275 M. Perisastry: 2256, 2275 Gottfried Perz: 2199, 2209, 2211, 2212, 2221, 2225, 2226, 2232, 2235, 2237, 2240, 2244, 2284, 2288, 2290, 2295 Waldemar Pompe: 1637, 2015, 2224, 2228, 2230, 2238, 2257, 2258, 2265, 2266, 2283 Luis A. Ponce: 2254, 2258 Bob Prielipp: 2218, 2221, 2245, 2248, 2249, 2279 Efstratios Rappos: 2222 Istv
n Reiman: 2201, 2203, 2204, 2209, 2218, 2221, 2224, 2230, a 2231, 2236, 2237, 2252, 2253, 2258, 2259, 2276, 2280 Renato Alberto Rodrigues: 2256 JuanBosco Romero M
rquez: 2218, 2262, 2268 a Angel Joval Roquet: 2284 Jawad Sadek: 2254, 2256, 2260, 2268, 2275, 2282 Rose Marie Saenz: 2211 Bill Sands: 2015, 2207, 2211, 2217, 2285, 2297, 2298 Jeevan Sandhya: 2244 Crist obal S
nchez Rubio: 2269
a Christos Saragiotis: 2238, 2245, 2232 K.R.S. Sastry: 2226, 2237, 2242, 2244, 2249, 2252, 2291, 2292 Joel Schlosberg: 2211, 2217, 2225, 2229, 2238, 2245, 2248, 2273, 2278, 2291, 2443 Robert P. Sealy: 2232, 2245 Harry Sedinger: 2181
2181, 2198, 2199, 2202, 2204, 2205, 2206, 2206, 2207, 2211, 2212, 2214, 2221, 2222, 2227, 2229, 2232, 2233, 2236, 2237, 2238, 2240, 2245, 2248, 2256, 2258, 2260, 2262, 2268, 2273, 2275, 2278, 2279, 2290, 2291, 2295, 2296 Toshio Seimiya: 2201, 2203, 2209, 2224, 2230, 2231, 2234, 2235, 2236, 2237, 2240, 2241, 2244, 2246, 2250, 2251, 2252, 2253, 2254, 2255, 2257, 2258, 2259, 2262, 2263, 2264, 2265, 2266, 2267, 2276, 2280, 2281, 2282, 2282, 2283, 2284 Reza Shahidi: 2226, 2232, 2238, 2245, 2275 Zun Shan: 2202, 2204, 2206, 2207, 2208, 2211, 2212, 2215, 2217, 2218, 2219, 2221, 2225, 2245, 2294 ShiChang Shi: 2015 Trey Smith: 2207, 2228, 2244, 2249, 2256, 2260, 2273, 2275 D.J. Smeenk: 2181, 2201, 2203, 2209, 2221, 2224, 2226, 2231, 2232, 2235, 2236, 2237, 2238, 2240, 2241, 2242, 2244, 2245, 2246, 2250, 2251, 2253, 2254, 2257, 2258, 2259, 2263, 2264, 2265, 2266, 2268, 2270, 2275, 2276, 2279, 2280, 2281, 2283, 2284, 2292, 2295 Digby Smith: 2181, 2208, 2212, 2218, 2219, 2232, 2238, 2245, 2247, 2249, 2256, 2275, 2290, 2295, 2298 Christopher So: 2235, 2239, 2249 Claus Mazanti Sorensen: 2293 David R. Stone: 2181, 2199, 2207, 2211, 2221, 2225, 2226, 2228, 2232, 2238, 2243, 2245, 2247, 2248, 2249, 2256, 2260, 2268, 2273, 2275, 2278 Peter Tingley: 2272 Panos E. Tsaoussoglou: 2198, 2199, 2204, 2205, 2208, 2219, 2221, 2226, 2231, 2236, 2238, 2245, 2249, 2251, 2253, 2254, 2255, 2256, 2258, 2260, 2279, 2284, 2290 George Tsapakidis: 2262, 2275, 2290, 1637 G. Tsintsifas: 2290 Enrique Valeriano: 2254, 2255, 2258 Meletis D. Vasiliou: 2237 David Vella: 2249 John Vlachakis: 2254, 2258, 2259, 2260, 2262, 2266, 2275 Edward T.H. Wang: 2181, 2202, 2204, 2206, 2207, 2208, 2211, 2212, 2215, 2217, 2218, 2219, 2221, 2225, 2232, 2239, 2245, 2249, 2260, 2294 Kenneth M. Wilke: 2181, 2225, 2291, 2292, 2295 Yeo Keng Hee: 2209, 2215, 2218 Karen Yeats: 2235, 2244 Paul Yiu: 2181, 2226, 2244, 2254, 2258, 2258, 2259, 2273, 2279 Peter Y : 2267 Vrej Zarikian: 2226, 2228, 2247, 2256, 2443 Roger Zarnowski: 2207, 2228, 2249, 2256, 2260, 2268, 2273, 2275
Con Amore Problem Group:
2227, 2238, 2239, 2242, 2244, 2252, 2254, 2255, 2256, 2257, 2258, 2259, 2261, 2268, 2275, 2443
Skidmore College Problem Group: 2238 University of Arizona Problem Solving Lab: 2256
Proposers, solvers and commentators in the MAYHEM PROBLEMS and SOLUTIONS sections for 1997 are:
Mohammed Aassila: A218, A219, A223, C79 Miguel Carri on Alvarez: A202, A204, H225, H226
Evan Borenstein: H222 Adrian Chan: H224 Keon Choi: H229, H230, H231 Lino Demasi: H229, H230, H231 Deepee Khosla: A194, A195, A196 Christopher Long: C78 Vedula N. Murty: H228
Katya Permiakova: H229 Waldemar Pompe: A220, A224 Joel Schlosberg: H217, H222, H224 D.J. Smeenk: A195, A197, A207 Matt Szczesny: C73 Alexandre Trichtchenko: A202, A204, A207, A208, A217, H240 Hoe Teck Wee: H224 Wendy Yu: H229, H230, H231