G. L. Alexanderson - The William Lowell Putnam Mathematical Competition. Problems and Solutions - 1965-1984.-Mathematical Association of America (1985) PDF

THE WILLIAM LOWELL PUTNAM
MATHEMATICAL COMPETITION
PROBLEMS AND SOLUTIONS: 1965-1984
Gerald L. Alexanderson
University of Santa Clara
Leonard F. Klosinski
University of Santa Clara
Loren C. Larson
St. Olaf College
The
WILLIAM LOWELL PUTNAM
PROBLEMS AND SOLUTIONS:
1965-1984
Edited by
Loren C. Larson
Published and distributed by

The Mathematical Association of America
C 1985 by
The Mathematical Association of America (Incorporated)

Library of Congress Catalog CardNumber 85-062263
ISBN 0-88385-441-4
Printed in the United States of America
Current printing (last digit):
10 9 8 7 6 5 4 3 2
PREFACE
Let us make clear from the start that we have not tried with this collection to imitate the
scholarly and extensive treatment of the first twenty-five contests by Gleason, Greenwood, and
Kelly (The William Lowell Putnam MathematicalCompetition/Problemsand Solutions: 1938-1964.
Washington: MAA, 1980). That splendid volume shows the years of work spent in following up
on problems, compiling better solutions, and tracing effects of some of the problems in subse-
quent work.
We have done none of that here. We have compiled material essentially already available in the
American MathematicalMonthly and Mathematics Magazine, correcting in several cases solutions
where errors had crept in. The present volume is mainly an attempt to put together in convenient
form existing material. A volume comparable to the Gleason, Greenwood, Kelly book will have to
wait for another time. We hope that in the meantime the present collection will benefit students
interested in preparing for the Competition, faculty who wish to organize problem seminars, or
any others just interested in problems.
For information about the history of the Putnam Competition, we refer the reader to the
excellent articles by Garrett Birkhoff and L. E. Bush in the earlier collection. These articles also
appeared in the Monthly in 1965. We are happy to have in the present collection a further bit of
information about the origins of the Competition, an essay on the first contest by Herbert
Robbins as told to Alan Tucker.
We have summarized lists of winning teams and individual participants; more extensive
information on winners and teams appears in annual reports in the Monthly.
Our work would have been much more difficult had we not had the reports of the Competition
carefully prepared by former directors of the Competition, James H. McKay (Oakland University)
and Abraham P. Hillman (University of New Mexico). We wish especially to thank them for their
many contributions over the years and specifically for their excellent reports. They are largely
responsible for the presentation of solutions that have appeared in the Monthly during their
directorships, though, of course, they had the benefit of having the solutions given them by
members of the Questions Committees over those years. And, of course, had the members of the
Questions Committee not provided the questions (and in many cases solutions) there would have
been no Competition. We therefore wish to thank the members of the Questions Committee:
H. S. M. Coxeter (University of Toronto), Adriano M. Garsia (California Institute of Technology),
Robert E. Greenwood (University of Texas, Austin), Nicholas D. Kazarinoff (University of
Michigan, Ann Arbor), Leo Moser (University of Alberta), Albert Wilansky (Lehigh University),
Warren S. Loud (University of Minnesota, Minneapolis), Murray S. Klamkin (Ford Scientific
Laboratories), Nathan S. Mendelsohn (University of Manitoba), Donald J. Newman (Yeshiva
University), J. Ian Richards (University of Minnesota, Minneapolis), Gulbank D. Chakerian
(University of California, Davis), Joseph D. E. Konhauser (Macalester College), Richard J.
Bumby (Rutgers University, New Brunswick), Lawrence A. Zalcman (University of Maryland,
College Park), Edward J. Barbeau, Jr. (University of Toronto), Kenneth B. Stolarsky (University
of Illinois, Urbana-Champaign), Joel H. Spencer (State University of New York, Stony Brook),
William J. Firey (Oregon State University), Douglas A. Hensley (Texas A & M University),
vii
viii THE WILLIAM LOWELL PUTNAM MATHEMATICAL COMPETITION
Melvin Hochster (University of Michigan, Ann Arbor), Bruce A. Reznick (University of Illinois,
Urbana-Champaign), and Richard P. Stanley (Massachusetts Institute of Technology).
We would further like to thank Alan Tucker, Chairman of the Publications Committee of the
MAA, A. B. Willcox, Executive Director, and Beverly Joy Ruedi of the Editorial Office of the
MAA.
Loren C. Larson
March, 1985
CONTENTS
PAGE
DEDICATION ....................................................... v
PREFACE ......................................................... Vii
RECOLLECTIONS OF THE FIRST PUTNAM EXAMINATION BY HERBERT ROBBINS ................ xi
LIST OF PROBLEMS ......................................................................... 3
SOLUTIONS TO THE PROBLEMS IN THE VARIOUS COMPETITIONS
Twenty-sixth ....................................................... 47
Twenty-seventh ....................................................... 51
Twenty-eighth ....................................................... 55
Twenty-ninth ................................................... 59
Thirtieth .......................................................... 62
Thirty-first ....................................................... 66
Thirty-second ....................................................... 70
Thirty-third ....................................................... 76
Thirty-fourth ....................................................... 83
Thirty-fifth ....................................................... 87
Thirty-sixth ....................................................... 91
Thirty-seventh ....................................................... 95
Thirty-eighth ....................................................... 99
Thirty-ninth .................................................... 103
Fortieth........................................................ 109
Forty-first ....................................................... 113
Forty-second ................................................... 117
Forty-third ....................................................... 122
Forty-fourth ....................................................... 127
Forty-fifth ..................................................... 132
APPENDICES
Winning Teams ....................................................... 137

Winning Individuals ................................................... 141
INDEX OF PROBLEMS ................................................ 145

RECOLLECTIONS OF THE FIRST PUTNAM EXAMINATION
HERBERT ROBBINS
as told to Alan Tucker
The well-known story of the origin of the initial 1933 Putnam contest in mathematics is as
follows (I believe this story to be mostly true). During half-time of the 1931 Harvard-Army
football game, President A. Lawrence Lowell said to the Commandant of the U.S. Military
Academy that while Army was showing that it could trounce Harvard in football, Harvard would
just as easily win any contest of a more academic nature. The Commandant took President Lowell
up on his challenge and it was decided to have a mathematics contest between the two schools. I
would guess that the field of mathematics was chosen because it is a subject that was studied at
both West Point and Harvard (all cadets, then as now, took two years of math) and because a
relative of President Lowell, George Putnam, was an amateur mathematician who was involved in
arrangements for the contest and got it named after his relative William Lowell Putnam.
I came to Harvard in the fall of 1931 with what I thought was some knowledge of the
humanities, but no mathematics beyond quadratic equations. To address this deficiency, I enrolled
in Math A, Analytic Geometry and Calculus. The texts were Analytic Geometry by Osgood and
Graustein and Calculus by Osgood. These books were casual in their treatment of real numbers
and limits but had challenging problems that assumed a good knowledge of physics. The
best-prepared students in Math A were put in the section taught by the eminent Professor Julian
Lowell Coolidge, and the rest were taught by junior instructors. I was put in the section taught by
Sumner B. Myers. I skipped a good many of the classes (being occupied with certain extracurricu-
lar interests most of the year) but did well in the tests.
At the end of the course in May, 1932, I was invited to be part of the team that would represent
Harvard in the mathematics competition with West Point the following spring. Since cadets took
two years of mathematics, the Harvard team was restricted to students who would be completing
their second year of mathematics at the time of the test. Selection for this team led me to continue
my study of mathematics for a second year.
The Harvard Mathematics Department assigned Professor Marston Morse to coach the team,
and we met with him about four times during the fall and winter. It was assumed that our
Harvard intellects would easily carry the day, and our meetings with Morse were spent in general
conversation rather than problem-solving. However, these sessions were very important to me, for
I became impressed by Morse as a person and by the incomprehensibility of the mathematics he
spoke of.
I had many conversations with Morse in addition to the Putnam sessions. Morse was in a low
mood then because his wife had recently left him to marry Professor William Fogg Osgood, a
distinguished Harvard mathematician with a long white beard who was 28 years Morse's senior
(this scandal forced Osgood to leave Harvard). However, my association with Morse was not
enough to persuade me to pick mathematics as my major just yet; I was experimenting then with
majors in the sciences and philosophy.
One spring weekend the Harvard team traveled to West Point for the competition. There was a
morning and afternoon session, and the problems were rather cut and dried, technical integrations
and the like, with little call for originality. The highlight of the weekend for me was a date
Saturday night in New York City with a girl I had met the previous summer.
Back at Harvard we found out, to our shame, that we had lost the competition to Army. I was
told that I had done well on the exam (we never saw our exam books) and decided to be a math
xi
xii THE WILLIAM LOWELL PUTNAM MATHEMATICAL COMPETITION
major. I never would have studied more than a year of mathematics, much less have become for a
time a mathematician, were it not for my experience with the Putnam competition.
The next year Morse left Harvard to go to the Institute for Advanced Study. He told me to
continue my studies, to get a PhD in math at Harvard, and then to get in touch with him. I had no
further contact with Morse until five years later when I defended my thesis at Harvard and sent
him a telegram: "Have PhD in mathematics." His response was equally brief: "You are my
assistant starting September 1."
It would seem that the reason I finally became a math major was that most Harvard
mathematics professors were rather pompous know-it-alls and that I wanted to show them that
any reasonably bright person could do mathematics. Unfortunately, I won the battle but lost the
war.
Herbert Robbins was the subject of an interview in the January 1984 College Mathematics Journal. He was co-author
with Richard Courant of What is Mathematics? and is Higgins Professor of Mathematical Statistics at Columbia
University.
PROBLEMS
THE TWENTY-SIXTH WILLIAM LOWELL PUTNAM
November 20, 1965
A-i. Let ABC be a triangle with angle A <angle C<90'<angle B. Consider the bisectors of the
external angles at A and B, each measured from the vertex to the opposite side (extended).
Suppose both of these line-segments are equal to AB. Compute the angle A.
A-2. Show that, for any positive integer n,
V('-)/2]
tO
pn-2r ( \)
n Or)
On-1) n
1 (2n-2
where [x] means the greatest integer not exceeding x, and () is the binomial coefficient
"n choose r," with the convention (")=i.
A-3. Show that, for any sequence at, a2, -* - of real numbers, the two conditions
(A) lim ) + e(is) + . + e(C(,)

nemo n
and
(B) lir 0e(^) + e) + . + e(i-.'
n--e ns
are equivalent.
A-4. At a party, assume that no boy dances with every girl but each girl dances with at least one
boy. Prove that there are two couples gb and g'b' which dance whereas b does not dance with
g' nor does g dance with b'.
A-5. In how many ways can the integers from 1 to n be ordered subject to the condition that,
except for the first integer on the left, every integer differs by 1 from some integer to the
left of it?
A-6. In the plane with orthogonal Cartesian coordinates x and y, prove that the line whose
equation is ux+vy= 1 will be tangent to the curve x"+ym = 1 (where m>1) if and only if
ut +v" = I and m- 1 +n± = 1.
B-1. Evaluate
lim I
n-X fo
A
o~
fo~
cos' 2n
(xl + X2 + x) dxldx2 . dx..
B-2. In a round-robin tournament with n players Pi, P2 , * , Pt (where n>l), each player
plays one game with each of the other players and the rules are such that no ties can occur.
Let wAand 1, be the number of games won and lost, respectively, by P.. Show that
E = E 4.
r.-l -1
3
4 THE WILLIAM LOWELL PUTNAM MATHEMATICAL COMPETITION
B-3. Prove that there are exactly three right-angled triangles whose sides are integers while the
area is numerically equal to twice the perimeter.
B-4. Consider the function
AX ()O) + ("2) x+(n) X+,

f (x, a) = -
(1i) + (3)x (5)x2

where n is a positive integer. Expressf(x, n+l) rationally in terms off(x, n) and x. Hence,
or otherwise, evaluate lim-,,-f(x, n) for suitable fixed values of x. (The symbols ( ) represent
the binomial coefficients.)
B-5. Consider collections of unordered pairs of V different objects a, b, c, , k. Three pairs
such as bc, ca, ab are said to form a triangle. Prove that, if 4E - V2 , it is possible to choose E
pairs so that no triangle is formed.
B-6. If A, B, C, D are four distinct points such that every circle through A and B intersects
(or coincides with) every circle through Cand D, prove that the four points are either collinear
(all of one line) or concyclic (all on one circle).
THE TWENTY-SEVENTH WILLIAM LOWELL PUTNAM
November 19, 1966
A-1. Letf(n) be the sum of the first n terms of the sequence 0, 1, 1, 2, 2, 3, 3, 4, , where
the nth term is given by
1 n/2 if n is even,
1(n-1)/2 if n is odd.
Show that if x and y are positive integers and x>y then xy=f(x+y) -f(x-y).
A-2. Let a, b, c be the lengths of the sides of a triangle, let p= (a+&+c)/2,and r be the radius
of the inscribed cricle. Show that
1 1 1 1
(p-a)' (p-b)2 I+ (p-c)2 r
A-3. Let 0<xi<1 andx,+il=x,,(l-x),n=1, 2, 3, *- Showthat
lim nx, = 1.
A-4. Prove that after deleting the perfect squares from the list of positive integers the number
we find in the nth position is equal to n + {V/n), where I v/nI denotes the integer closest to -/n.
A-5. Let C denote the family of continuous functions on the real axis. Let T be a mapping
of C into C which has the following properties:
I. T is linear, i.e. T(ci*i +C jP)
2 = cl Tiki +c2 Tik2, for cl and c2 real and ^61
and V2 in C.
2. T is local, i.e. if Vitb
_2 in some interval I then also Tsb MTV2 holds in I.
Show that T must necessarily be of the form T#,(x) =f(x)V&(x) where f(x) is a suitable continuous
function.
A-6. Justify the statement that
3 =-I+2V4+3+ 1+5V1T,7
B-1. Let a convex polygon P be contained in a square of side one. Show that the sum of the
squares of the sides of P is less than or equal to 4.
B-2. Prove that among any ten consecutive integers at least one is relatively prime to each
of the others.
B-3. Show that if the series
.1
is convergent, where pi, P2, P3 , P., are positive real numbers, then the series
5
i ~n2 P
.- I (Pi+ P2 + * *+ P.)I
is also convergent.
B-4. Let O<al<a 2 < . ..<am"+l be mn+1 integers. Prove that you can select either m+1
of them no one of which divides any other, or n+1 of them each dividing the following one.
B-5. Given n ( 2 3) distinct points in the plane, no three of which are on the same straight line,
prove that there exists a simple closed polygon with these points as vertices.
B-6. Show that all solutions of the differential equation y"+ely= 0 remain bounded as x- oo.
THE TWENTY-EIGHTH WILLIAM LOWELL PUTNAM
December 2, 1967
A-1. Letf(x) =aisin x+a 2 sin 2x+ * * +asin

± nx, where a,, a2, , an are real numbers and
I
where n is a positive integer. Given that If(x) | |sin x I for all real x, prove that
jao+2a2 + - . +na,,I !1.
A-2. Define So to be 1. For n>1, let S. be the number of nXn matrices whose elements are
nonnegative integers with the property that aij= aj, (ij = 1, 2, * * *, n) and where Fs'-Iai.
=1,(j=1,2, - ,n).Prove
(a) Sn+i = S. + nS,- 1 ,
(b) j S. - = exp(x + x'/2), where exp(x) = eX.

,-0 n
A-3. Consider polynomial forms ax2 -bx+c with integer coefficients which have two distinct
zeros in the open interval O<x<1. Exhibit with a proof the least positive integer value of a for
which such a polynomial exists.
A-4. Show that if X> I there does not exist a real-valued function u such that for all x in the
closed interval O•x -1, u(x) =l+Xf|' u(y) u(y-x) dy.
A-5. Show that in a convex region in the plane whose boundary contains at most a finite num-
ber of straight line segments and whose area is greater than 7r/4 there is at least one pair of points a
unit distance apart.
A-6. Given real numbers {as and { bi), (i = 1, 2, 3, 4), such that alb2 -a 2 b, p6O. Consider the
set of all solutions (xI, X2, X3, X4) of the simultaneous equations
aix, + a2x2 + asxa + a4x4 = 0 and bixi + b2x2 + b x, + b 4x 4 0,.
for which no xi (i = 1, 2, 3, 4) is zero. Each such solution generates a 4-tuple of plus and minus signs
(signum xi, signum X2 , signum x3, signum X 4 ).
(a) Determine, with a proof, the maximum number of distinct 4-tuples possible.
(b) Investigate necessary and sufficient conditions on the real numbers Ias and Ibi such
that the above maximum number of 4-tuples is attained.
B-1. Let (ABCDEF) be a hexagon inscribed in a circle of radius r. Show that if ABf=D
=2FS=r, then the midpoints of BC, D!, FA are the vertices of an equilateral triangle.
B-2. Let 0•-p-l and 0:5r<1 and consider the identities

(a) (px + (1 - p)y)2 = Ax' + Bxy + Cy',
(b) (px + (1 - p)y) (rx + (1-r)y) = aT2 + flxy + yy.
Show that (with respect to p and r)
(a) maxIA, B, C} 2 4/9,
(b) max{Ir, A,y} 2 4/9.
7
B-3. If f and g are continuous and periodic functions with period 1 on the real line, then
lin,..,, flf(x) g(nx) dx = (fJf(x)dx) (f0 g(x)dx).
B-4. (a) A certain locker room contains n lockers numbered 1, 2, 3, , n and all are origi-
nally locked. An attendant performs a sequence of operations T., T2, * * *, T. whereby with the
operation Tk, 1 - k < n, the condition of being locked or unlocked is changed for all those lockers and
only those lockers whose numbers are multiples of k. After all the n operations have been performed
it is observed that all lockers whose numbers are perfect squares (and only those lockers) are now
open or unlocked. Prove this mathematically.
(b) Investigate in a meaningful mathematical way a procedure or set of operations similar to
those above which will produce the set of cubes, or the set of numbers of the form 2m', or the set of
numbers of the form m 2 +l, or some nontrivial similar set of your own selection.
B-5. Show that the sum of the first n terms in the binomial expansion of (2-1)-" is i, where n
is a positive integer.
B-6. Letf beareal-valued function having partial derivatives and which is defined for xt+y' S 1
and is such that tf (x, y) | - 1. Show that there exists a point (xo, ye) in the interior of the unit circle
such that
(xo, yo)) + (af (0o,yo)) S 16.

THE TWENTY-NINTH WILLIAM LOWELL PUTNAM
December 7, 1968
A-1. Prove
22 x4(1- dx.
7 = J 1+ dx.
A-2. Given integers a, b, e, c, d, andf with ad bc, and given a real number e>0, show that
there exist rational numbers r and s for which
0< Ira+sb-el <.,
0< Irc+sd-fI <4.
A-3. Prove that a list can be made of all the subsets of a finite set in such a way that (i) the
empty set is first in the list, (ii) each subset occurs exactly once, (iii) each subset in the list is ob-
tained either by adding one element to the preceding subset or by deleting one element of the
preceding subset.
A-4. Given n points on the sphere I (x, y, z) :x 2+y2 +z2 = 1 }, demonstrate that the sum of the
squares of the distances between them does not exceed a2.
A-5. Let V be the collection of all quadratic polynomials P with real coefficients such that
IP(x) | S 1 for all x on the closed interval [0, I ]. Determine
sup II P'(0) j P G V}.
A-6. Determine all polynomials of the form Z0axl-i with ai l(O-ign, 1 n < x) such
that each has only real zeros.
B-1. The temperatures in Chicago and Detroit are x0 and y', respectively. These temperatures
are not assumed to be independent; namely, we are given:
(i) P(xo=70'), the probability that the temperature in Chicago is 70°,
(ii) P(y'=700), and
(iii) P(max(x0 , yo) = 70°).
Determine P(min(x0 , y0)= 70 0).
B-2. A is a subset of a finite group G (with group operation called multiplication), and A- con-
tains more than one half of the elements of G. Prove that each element of G is the product of two
elements of A.
B-3. Assume that a 600 angle cannot be trisected with ruler and compass alone. Prove that if
n is a positive multiple of 3, then no angle of 360/n degrees can be trisected with ruler and compass
alone.
B-4. Show that if f is real-valued and continuous on (-so, 0o) and f X f (x)dx exists, then
f f (x--) dx = f_()dx.
B-5. Let p be a prime number. Let J be the set of all 2X2 matrices (d) whose entries are
chosen from 10, 1, 2, * , p-I }and satisfy the conditions a+daI (mod p), ad-bc-0 (mod p).
9
Determine how many members J has.

B-6. A set of real numbers is called compact if it is closed and bounded. Show that there does
not exist a sequence I K. }I = of compact sets of rational numbers such that each compact set of ra-
tionals is contained in at least one K..
THE THIRTIETH WILLIAM LOWELL PUTNAM
December 6, 1969
A-1. Letf(x, y) be a polynomial with real coefficients in the real variables x and y defined over
the entire x-y plane. What are the possibilities for the range of f(x, y)?
A-2. Let D. be the determinant of order n of which the element in the ith row and the ith
column is the absolute value of the difference of i andj. Show that D. is equal to
(-1)" (n - 1)2-2.
A-3. Let P be a non-self-intersecting closed polygon with n sides. Let its vertices be Pi,
P2 , * - , PF,. Let m other points, Q1, Q2, * * ,, Q, interior to P be given. Let the figure be tri-
angulated. This means that certain pairs of the (n+m) points Pi, * * *, Quad are connected by line
segments such that (i) the resulting figure consists exclusively of a set T of triangles, (ii) if two
different triangles in T have more than a vertex in common then they have exactly a side in
common, and (iii) the set of vertices of the triangles in T is precisely the set of (n+m) points
PI, - * * , Q.,,. How many triangles in T?
A-4. Show that
xtdx = E )+Inn
(The integrand is taken to be 1 at x =0.)

A-5. Let u(t) be a continuous function in the system of differential equations
dx dy
- 2y +u(t), - =- 2x +u(t).
Tt di
Show that, regardless of the choice of u (), the solution of the system which satisfies x =xo, y =yo
at t=0 will never pass through (0, 0) unless xo=yo. When xo=yo, show that, for any positive value
to of t, it is possible to choose u(t) so the solution is at (0, 0) when t =t o.
A-6. Let a sequence {x.} be given, and let y,.=x.,_+2x., n =2, 3, 4, *- . Suppose that the
sequence {y,} converges. Prove that the sequence {x"} also converges.
B-1. Let n be a positive integer such that n+l is divisible by 24. Prove that the sum of all
the divisors of n isdivisible by 24.
B-2. Show that a finite group can not be the union of two of its proper subgroups. Does the
statement remain true if "two' is replaced by "three'?
B-3. The terms of a sequence T. satisfy
T.T,+, = n (n = 1, 2, 3) and lim -= 1.

no- T.+,
Show that irT =2.
B-4. Show that any curve of unit length can be covered by a closed rectangle of area 1/4.
11
B-5. Let a,<a2<as< * be an increasing sequence of positive integers. Let the series
E 1/a,.
nIl
be convergent. For any number x, let k(x) be the number of the a.'s which do not exceed x. Show
that limurn,, k(x)/x =0.
B-6. Let A and B be matrices of size 3 X2 and 2 X3 respectively. Suppose that their product
in the order AB is given by
=[ 8 2 -2-
AB = 2 5 4 .
-- 2 4 5-
Show that the product BA is given by
BA = [9Lo 01.
9J
THE THIRTY-FIRST WILLIAM LOWELL PUTNAM
December 5, 1970
A-1. Show that the power series for the function

escosbx (a.>Ob>O)
in powers of x has either no zero coefficients or infinitely many zero coefficients.
A-2. Consider the locus given by the real polynomial equation
Ax' + Bxy + Cy' + Dx2 + Ex'y + Fxy' + Gy' = 0,
where B'-4AC<O. Prove that there is a positive number a such that there are no points of the
locus in the punctured disk
0 < x' + y' <82.
A-3. Find the length of the longest sequence of equal nonzero digits in which an integral
square can terminate (in base 10) and find the smallest square which terminates in such a sequence.
A-4. Given a sequence txnd n = 1, 2, * , such that limit,, {xs-x,.s J =0.
Prove that
limit = 0.
nor n
A-5. Determine the radius of the largest circle which can lie on the ellipsoid
x2 Y2 Z
- 2 + -- + 1 (a > b > c).
a b2 C
A-6. Three numbers are chosen independently at random, one from each of the three intervals
[0, Li] (i=1, 2, 3). If the distribution of each random number is uniform with respect to length in
the interval it is chosen from, determine the expected value of the smallest of the three numbers
chosen.
B-1. Evaluate
1 2.
limit-jI (n' + it) V.
"-.- .- 1
B-2. The time-varying temperature of a certain body is given by a polynomial in the time of
degree at most three. Show that the average temperature of the body between 9 A.m. and 3 P.M.
can always be found by taking the average of the temperatures at two fixed times, which are inde-
pendent of which polynomial occurs. Also, show that these two times are 10:16 A.M. and 1:44 P.M.
to the nearest minute.
B-3. A closed subset S of RI lies in a <x <b. Show that its projection on the y-axis is closed.
B-4. An automobile starts from rest and ends at rest, traversing a distance of one mile in one
minute, along a straight road. If a governor prevents the speed of the car from exceeding ninety
miles per hour, show that at some time of the traverse the acceleration or deceleration of the
car was at least 6.6 ft./sec.'
13
B-5. Let u, denote the "ramp" function

-n for x <-n,
u.(x) = x for-n < x5 n,
n for x > n,
and let F denote a real function of a real variable. Show that F is continuous if and only if u. o F
is continuous for all n. (Note: (un o F)(x) =u, [F(x) ].)
B-6. A quadrilateral which can be inscribed in a circle is said to be inscribable or cyclic. A
quadrilateral which can be circumscribed to a circle is said to be circumscribable. Show that if a
circumscribable quadrilateral of sides a, b, c, d has area A = \Iabcd, then it is also inscribable.
THE THIRTY-SECOND WILLIAM LOWELL PUTNAM
December 4, 1971
A-I. Let there be given nine lattice points (points with integral coordinates) in three dimen-
sional Euclidean space. Show that there is a lattice point on the interior of one of the
line segments joining two of these points.
A-2. Determine all polynomials P(x) such that P(x2 + 1) = (p(X)) 2 + I and P(O) = 0.
A-3. The three vertices of a triangle of sides a, b, and c are lattice points and lie on a circle
of radius R. Show that abc 2 2R. (Lattice points are points in the Euclidean plane with
integral coordinates.)
A-4. Show that for 0 < 6 < I the expression (x + y)l (X2 -(2- e)xy + y 2 ) is a polynomial
with positive coefficients for n sufficiently large and integral. For a = .002 find the
smallest admissible value of n.
A-5. A game of solitaire is played as follows. After each play, according to the outcome,
the player receives either a or b points (a and b are positive integers with a greater than
b), and his score accumulates from play to play. It has been noticed that there are thirty-
five non-attainable scores and that one of these is 58. Find a and b.
A-6. Let c be a real number such that nc is an integer for every positive integer n. Show that
c is a non-negative integer.
B-i. Let S be a set and let Obe a binary operation on S satisfying the two laws
x x x for all xin S, and
(x y) z = (y . z) o x for all x, y, z in S.
Show that . is associative and commutative.
B-2. Let F (x) be a real valued function defined for all real x except for x = 0 and x = I and
satisfying the functional equation F(x)+F{(x- )/x} = I + x. Find all functions F(x)
satisfying these conditions.
B-3. Two cars travel around a track at equal and constant speeds, each completing a lap
every hour. From a common starting point, the first starts at time t = 0 and the second
at an arbitrary later time t = T > 0. Prove that there is a total period of exactly one
hour during the motion in which the first has completed twice as many laps as the
second.
15
B4. A "spherical ellipse" with foci A, B on a given sphere is defined as the set of all points P
on the sphere such that PA + PB = constant. Here PA denotes the shortest distance
on the sphere between P and A. Determine the entire class of real spherical ellipses which
are circles.
B-5. Show that the graphs in the x-y plane of all solutions of the system of differential
equations
x" + y' + 6x = O,y"-x' + 6y=O ('= d/dt)
which satisfy x' (0) = y' (0) = 0 are hypocycloids, and find the radius of the fixed
circle and the two possible values of the radius of the rolling circle for each such solution.
(A hypocycloid is the path described by a fixed point on the circumference of a circle
which rolls on the inside of a given fixed circle.)
B-6. Let a (x) be the greatest odd divisor of the positive integer x. Show that
I 6(n)/n -2x/3
IX= < 1, for all positive integers x.
THE THIRTY-THIRD WILLIAM LOWELL PUTNAM
December 2, 1972
A-i. Show that there are no four consecutive binomial coefficients ( (rn+ n),)'(r+ 2)' (r+ )
(n, r integers > 0 and r + 3 < n) which are in arithmetic progression.
A-2. Let S be a set and let * be a binary operation on S satisfying the laws
x*(x*y)=y for all x,y in S,
(y * x) * x = y for all x, y in S.
Show that * is commutative but not necessarily associative.
A-3. If for a sequence xl, X2, x3 , ***,limn o0 (xi +x 2 + ... + x")/n exists, call this limit the C-limit
of the sequence. A functions (x) from [0, 1I to the reals is called a supercontinuous function
on the interval [0, 1] if the C-limit exists for the sequence f (xi)f(xz),f(x3 ),- .. whenever
the C-limit exists for the sequence xi, X 2 , X3 . Find all supercontinuous functions
on [0, 1].
A-4. Of all ellipses inscribed in a square, show that the circle has the maximum perimeter.
A-5. Show that if n is an integer greater than 1, then n does not divide 2n - 1.
A-6. Letf(x) be an integrable function in 0 - x 2"n( + 1) in a set of
positive measure.
B-i. Show that the power series representation for the series E' (x (x-1)2n)In! cannot have
three consecutive zero coefficients.
B-2. A particle moving on a straight line starts from rest and attains a velocity vo after traversing
a distance so. If the motion is such that the acceleration was never increasing, find the
maximum time for the traverse.
B-3. Let A and B be two elements in a group such that ABA = BA 2B, A 3 = I and B 2 .- 1
= I
for some positive integer n. Prove B = 1.
BA4. Let n be an integer greater than 1. Show that there exists a polynomial P(x, y, z) with integral
coefficients such that x = P(xn, xn + 1, x + Xn + 2 ).
B-5. If the opposite angles of a skew (non-planar) quadrilateral are equal in pairs, prove that the
opposite sides are equal in pairs.
17
B-6. Let n, < n2 < n3 < * < nk be a set of positive integers. Prove that the polynomial I + ze
+ Zn2 + + z'k has no roots inside the circle jzj < (15 - 1)/2.
THE THIRTY-FOURTH WILLIAM LOWELL PUTNAM
December 1, 1973
A-l. (a) Let ABC be any triangle. LetX, Y, Z be points on the sides BC, CA, AB respectively.
Suppose the distances BX < XC, C'Y YA, AZ < ZB (see Figure 1). Show that the
area of the triangle X YZ is - (1/4) (area of triangle ABC).
(b) Let ABC be any triangle, and let X, Y, Z be points on the sides BC, CA, ABrespectively
(but without any assumption about the ratios of the distances BX/XC, etc.; see Figures I
and 2). Using (a) or by any other method, show: One of the three corner triangles AZY,
BXZ, CYX has an area < area of triangle XYZ.
A A
X I -
FIG. I FIG. 2
A-2. Consider an infinite series whose nth term is + (1/n), the ± signs being determined according
to a pattern that repeats periodically in blocks of eight. [There are 28 possible patterns of
which two examples are:
++-- ++
The first example would generate the series
I + (1/2) - (1/3) - (1/4) - (1/5) - (1/6) + (1/7) + (1/8)
+ (1/9) + (1/10) - (1/11) - (1/12) - -]..1
(a) Show that a sufficient condition for the series to be conditionally convergent is that there
be four " +" signs and four "-" signs in the block of eight.
(b) Is this sufficient condition also necessary?

[Here "convergent" means "convergent to a finite limit.")
19
A-3. Let n be a fixed positive integer and let b(n) be the minimum value of
n
k + k
k
as k is allowed to range through all positive integers. Prove that b(n) and V'4n + I have the
same integer part. [The "integer part" of a real number is the greatest integer which does not
exceed it, e.g. for n it is 3, for /21 it is 4, for 5 it is 5, etc.]
2
A-4. How many zeros does the functionf(x) = -1 - x have on the real line? [By a "zero"
of a functions, we mean a value x 0 in the domain of f (here the set of all real numbers) such
thatf(xo) = 0.]
A-5. A particle moves in 3-space according to the equations:
dx dy dz
- = yz, - = zx, dt = xy.
dt d'd ' di'
[Here x(t), At), z(t) are real-valued functions of the real variable t.] Show that:
(a) If two of x(O), y(O), z(0) equal zero, then the particle never moves.
(b) If x(O) = y(O) = 1, z(0) = 0, then the solution is:
x = sec t, y = sec t, z = tan t;
whereas if x(0) = y(O) = 1, z(0) = - 1, then
x = 1/(t + 1), y = lI(t + 1), z =-1/(t + 1).
(c) If at least two of the values x(O), y(O), z(0) are different from zero, then either the particle
moves to infinity at some finite time in the future, or it came from infinity at some finite
time in the past. [A point (x, y, z) in 3-space "moves to infinity" if its distance from the
origin approaches infinity.]
A-6. Prove that it is impossible for seven distinct straight lines to be situated in the euclidean plane
so as to have at least six points where exactly three of these lines intersect and at least four
points where exactly two of these lines intersect.
B-i. Let al, a2, ... a, + Xbe a set of integers such that, if any one of them is removed, the remaining
ones can be divided into two sets of n integers with equal sums. Prove al = a2 = ... = a2n + X-
B-2. Let z = x + iy be a complex number with x and y rational and with I z I. Show that the
number Iz2n - 1 I is rational for every integer n.
B-3. Consider an integer p > 1 with the property that the polynomial x 2 - x + p takes prime
values for all integers x in the range 0 s; x < p. (Examples: p = 5 and p = 41 have this
property.) Show that there is exactly one triple of integers a, b, c satisfying the conditions:
b 2 - 4ac = I - 4p,
0 < a : c,
- a < b < a.
PROBLEMS: THE THIRTY-FOURTH COMPETITION 21
B-4. (a) On [0, 1], let f have a continuous derivative satisfying 0 < f'(x) :5 1. Also suppose that
f(O) = 0. Prove that
fJ f(x)dx] 2 fJ If(X)]3dX.
[Hint: Replace the inequality by one involving the inverse function tog.]
(b) Show an example in which equality occurs.
B-5. (a) Let z be a solution of the quadratic equation
aZ2 +bz + c =0
and let n be a positive integer. Show that z can be expressed as a rational function of
zn', a, b, c.
(b) Using (a) or by any other means, express x as a rational function of x 3 and x + (I/x).
(Display your answer explicitly in a clearly visible form.)
[By a rational function of several variables, we mean a quotient of polynomials in those
variables, the polynomials having rational numbers as coefficients, and the denominator
being not identically zero. Thus to obtain x as a rational function of u = x2 and v -
x + (I/x), we could write x = (a + I)/v.]
B-6. On the domain 0 00! 2n:

(a) Prove that sin 2 o* sin (20) takes its maximum at n/3 and 47r/3 (and hence its minimum
at 2n/3 and 5nt/3).
(b) Show that
I sin20(sjn 3(20) sin 3(40)... sin 3(2- l@0))sin (2n0)

takes its maximum at 0 = n/3. (The maximum may also be attained at other points.)
(c) Derive the inequality:
sin 2 8 *sin 2 (20) - sin 2 (40) ... sin2(2n 9 ) : (3/4)5.

THE THIRTY-FIFTH WILLIAM LOWELL PUTNAM
December 7, 1974
A-I. Call a set of positive integers "conspiratorial" if no three of them are pairwise relatively prime. (A set
of integers is "pairwise relatively prime" if no pair of them has a common divisor greater than 1.)What
is the largest number of elements in any "conspiratorial" subset of the integers I through 16?
A-2. A circle stands in a plane perpendicular to the ground and a point A lies in this plane exterior to the
circle and higher than its bottom. A particle starting from rest at A slides without friction down an
inclined straight line until it reaches the circle. Which straight line allows descent in the shortest time?
(Assume that the force of gravity isconstant over the region involved, there are no relativistic effects,
etc.]
path of
descent
A Isltart) The starting point A and the circle are
fixed; the stopping point B is allowed
* B slOP)
I to vary over the circle.
Note. The answer may be given in any form which specifies the line of descent in an unambiguous manner; it
is not required to find the coordinates of the point B.
A-3. A well-known theorem asserts that a prime p > 2 can be written as the sum of two perfect squares
(p = mn+ n', with m and n integers) if and only if p I (mod 4). Assuming this result, find which
primes p > 2 can be written in each of the following forms, using (not necessarily positive) integers x
and y:
(a) x 2+ 16y 2;
(b) 4x2 + 4xy + 5y2.
A-4. An unbiased coin is tossed n times. What is the expected value of I H T-Il, where H is the number of
heads and T is the number of tails? In other words, evaluate in closed form:
I^- E , (. -2k) (n)
(In this problem, "closed form" means a form not involving a series. The given series can be reduced
to a single term involving only binomial coefficients, rational functions of n and 2^, and the greatest
integer function [xl.)
A-5. Consider the two mutually tangent parabolas y = x' and y - x2. [These have foci at (0, 1/4) and
(0, -1/4), and directrices y = -1/4 and y = 1/4, respectively.) The upper parabola rolls without
slipping around the fixed lower parabola. Find the locus of the focus of the moving parabola.
22
PROBLEMS: THE THIRTY-FIFTH COMPETITION 23
\ / f the rolling parabola
\ (0, 1/4), the original position

of the moving focus
the fixed parabola
A-6. It is well known that the value of the polynomial (x + 1)(x + 2)*.. (x + n) is exactly divisible by n for
every integer x. Given n, let k = k(n) be the minimal degree of any monic integral polynomial
f(x)=xa+axai'+ .+a,
(with integer coefficients and leading coefficient 1) such that the value of f(x) is exactly divisible by n
for every integer x.
Find the relationship between n and k = k(n). In particular, find the value of k corresponding to
n = I 000 000.
B-I. Which configurations of five (not necessarily distinct) points pi, -,p, on the circle x'+ y'= I
maximize the sum of the ten distances
, d(p,,p)?
[Here d(p,q) denotes the straight line distance between p and q.]
B-2. Let y(x) be a continuously differentiable real-valued function of a real variable x. Show that if
(y')'+ y'- as x - + , then y(x) and y'(x)-O as x - +a.
B-3. Prove that if a is a real number such that
cos 7ra = 1/3,
then a is irrational. (The angle nra is in radians.)

B-4. In the standard definition, a real -valued function of two real variables g: Rt'.R is continuous if, for
every point (x.,y.)ER' and every e>0, there is a corresponding 8>0 such that
l(x - x.)' + (y .- Y)a]/ C' implies Ig(x, y)-g(x 0, YO)I < e.
By contrast, f: R '- R' is said to be continuous in each variable separately if, for each fixed value
Yoof y, the function f(x, yo) is continuous in the usual sense as a function of x, and similarly &fxo,y) is
continuous as a function of y for each fixed x0.
Let f: R'- R' be continuous in each variable separately. Show that there exists a sequence of
continuous functions go: R-. R' such that
f(x, y) = tim g.(x, y) for all (x, y) E R'.
B-5. Show that I +(n/l!)+(n'/2!)+ *-- -+(n^/n!)> e/2 for every integer n -0.
REMARKS You may assume as known Taylor's remainder formula:
k'
x2 n |(x - t)- ' dt,
as well as the fact that
n =foe 'dt.
B-H. For a set with n elements, how many subsets are there whose cardinality (the number of elements in
the subset) is respectively = 0 (mod 3). I (mod 3), = 2(mod 3)? In other words, calculate
* X (n) for i = 0,1,2.

Your result should be strong enough to permit direct evaluation of the numbers s,.. and to show clearly
the relationship of so. and s*,, and s2,, to each other for all positive integers n. In particular, show the
relationships among these three sums for n = 1000. (An illustration of the definition of s,, is
-() + (6) + (6) 22.1

THE THIRTY-SIXTH WILLIAM LOWELL PUTNAM
December 6, 1975
A-i. Supposing that an integer n is the sum of two triangular numbers,

a'+a b'+b
2 2'
write 4n + I as the sum of two squares, 4n + 1 = x2 + y', and show how x and y can be expressed in terms of a
and b.
Show that, conversely, if 4n + 1 = x2 + y', then n is the sum of two triangular numbers.
[Of course, a, b, x, y are understood to be integers.]
A-2. For which ordered pairs of real numbers b, c do both roots of the quadratic equation
z'+ bz + c = 0
lie inside the unit disk {Iz I< 1) in the complex plane?
Draw a reasonably accurate picture (i.e., 'graph') of the region in the real bc-plane for which the above
condition holds. Identify precisely the boundary curves of this region.
A-3. Let a, b, c be constants with 0 < a < b < c. At what points of the set
{x' + yb + z' = 1, x - 0, y2 0, z2 0)
in three-dimensional space R' does the function f(x, y, z) = x' + y' + z' assume its maximum and minimum
values?
A-4. Let n = 2m, where m is an odd integer greater than 1. Let 0 = e 2 '". Express (I - 5)-' explicitly as a
polynomial in 0,
a*k+a. ,e_-'+***+a,6+aa,
with integer coefficients a,.
[Note that 6 is a primitive n-th root of unity, and thus it satisfies all of the identities which hold for such
roots.]
A-5. On some interval I of the real line, let y,(x) and y,(x) be linearly independent solutions of the differential
equation
y = f(x)y,
where f(x) is a continuous real-valued function. Suppose that y,(x) >0 and y2(x) >0 on 1. Show that there
exists a positive constant c such that, on I, the function
z(x) = c Vy,(x)y 2 (x)
satisfies the equation
z'+-3 f(x)z.
State clearly the manner in which c depends on y,(x) and y,(x).

A-6. Let P., P,, P, be the vertices of an acute-angled triangle situated in three-dimensional space. Show that it
is always possible to locate two additional points P. and P, in such a way that no three of the points are
collinear and so that the line through any two of the five points is perpendicular to the plane determined by the
other three.
25
In writing your answer, state clearly the locations at which you place the points P. and P,.
B-i. In the additive group of ordered pairs of integers (m, n) [with addition defined componentwise:
(m, n)+(m', n')= (m + m', n + n')] consider the subgroup H generated by the three elements
(3, 8), (4, - 1), (5,4).
Then H has another set of generators of the form
(l,b), (0,a)
for some integers a, b with a > 0. Find a.
[Elements g,,..., g&are said to generate a subgroup H if (i) each g, E H, and (ii) every h E H can be
written as a sum h = ng, + * + nm& where the n, are integers (and where, for example, 3g, - 2g, means
g, + g g,
+ - g2 -
B-2. In three-dimensional Euclidean space, define a slab to be the open set of points lying between two
parallel planes. The distance between the planes is called the thickness of the slab. Given an infinite sequence
S,, S, ... of slabs of thicknesses d,, d, .. , respectively, such that I,-L, di converges, prove that there is some
point in the space which is not contained in any of the slabs.
B-3. Let sk(a,,.. ., a")denote the k-th elementary symmetric function of a,, . . ., a.. With k held fixed, find the
supremum (or least upper bound) Mk of
sa(a, . a.)[s,(a,,..., a)]
for arbitrary n Ž: k and arbitrary n-tuples a,. a. of positive real numbers.

[Thesymmetric function se(a,,..., a.) is thesumofall k-fold productsofthe variables a,, a.. Thus, for
example:
s,(a,... ) a+ a+ + a;
s3(a,, a2, a3, a.) = a,aa 3 + a,a2 a, + a,aa, + a~a~a,.

It should be remarked that the supremum M, is never attained; it is approached arbitrarily closely when, for
fixed k, the number n of variables increases without bound, and the values a, >0 are suitably chosen.]
B-4. Does there exist a subset B of the unit circle x2 + y' = I such that (i) B is topologically closed, and (ii) B
contains exactly one point from each pair of diametrically opposite points on the circle?
[A set B is topologically closed if it contains the limit of every convergent sequence of points in B.]
B-5. Let f,(x) = e' and fh,,(x) = xf,(x) for n = 0,1, 2_.. Show that
f'(l)=e,.
- n!
B-6. Show that if s. = I + 2 + 3 + ** ' + 1/n, then
(a) n(n + 1)"' <in + s, for n >1, and
(b) (n - I)n "' < n - s. for n > 2.
THE THIRTY-SEVENTH WILLIAM LOWELL PUTNAM
December 4, 1976
A-I. P is an interior point of the angle whose sides are the rays OA and OB. Locate X on OA and Y on OB so
that the line segment XY contains P and so that the product of distances (PX)(PY) is a minimum.
A-2. Let P(x,y)= x'y +xy' and Q(x,y)= x'+xy +y'. For n 1,2,3,..., let F.(x,y) =(x + y)' -x" -y'
and G.(x,y)-(x +y)' +x" +y'. One observes that G,=2Q, F,-3P, G,= 2Q0, F,=5PQ, =-2Q0+3P'.
Prove that, in fact, for each n either F. or G, is expressible as a polynomial in P and Q with integer coefficients.
A-3. Find all integral solutions of the equation
Ip' -q'I=1
where p and q are prime numbers and r and s are positive integers larger than unity. Prove that there are no other
solutions.
A-4. Let r bearoot of P(x)= x3+ax'+bx - I = 0and r + I bea root of y%+cy'+dy + I =0, where a, b, c,
and d are integers. Also let P(x) be irreducible over the rational numbers. Express another root s of P(x) = 0 as a
function of r which does not explicitly involve a, b, c, or d.
A-5. In the (x, y)-plane, if R is the set of points inside and on a convex polygon, let D(x, y) be the distance
from (x, y) to the nearest point of R. (a) Show that there exist constants a, b, and c, independent of R, such that
f f
J e-e''-"dx dy = a + bL + cA,
where L is the perimeter of R and A is the area of R. (b) Find the values of a, b, and c.
A-6. Suppose f(x) is a twice continuously differentiable real valued function defined for all real numbers x
and satisfying |f(x)! <1 for all x and (f(0))' + (f'(0)) =4. Prove that there exists a real number x, such that
f (x.) +f "(X.) = 0.
B-I. Evaluate
him I Ink] -2 [n)
and express your answer in the form log a - b, with a and b positive integers.
Here [x] is defined to be the integer such that [x] 'x <[x]+ I and logx is the logarithm of x to base e.
B-2. Suppose that G is a group generated by elements A and B, that is, every element of G can be written as
a finite "word" A "B'"A "' - *B" , where n,, .. , n, are any integers, and A" = B" = I as usual. Also, suppose that
A=B'= ABA 'B =I, A'#1, and Bjil.
(a) How many elements of G are of the form C' with C in G?
(b) Write each such square as a word in A and B.
B-3. Suppose that we have n events A,. A,, each of which has probability at least I -a of occurring,
where a < 1/4. Further suppose that A, and A, are mutually independent if I i - j I > 1, although Ai and Ai, may
be dependent. Assume as known that the recurrence u = u, - au, -, u,, = 1, u, = I - a, defines positive real
numbers u, for k =0,1,... Show that the probability of all of A,,..., A. occurring is at least u,.
B-4. For a point P on an ellipse, let d be the distance from the center of the ellipse to the line tangent to the
ellipse at P. Prove that (PF,)(PF,)d' is constant as P varies on the ellipse, where PF, and PF, are the distances
from P to the foci F, and F, of the ellipse.
27
B-5. Evaluate
E (_ ), (n) (.-k)".
= k
B-6, As usual, let e(N) denote the sum of all the (positive integral) divisors of N. (Included among these
divisors are I and N itself.) For example, if p is a prime, then a(p) = p + I. Motivated by the notion of a "perfect"
number, a positive integer N is called "quasiperfect" if o,(N) = 2N + 1. Prove that every quasiperfect number is
the square of an odd integer.
THE THIRTY-EIGHTH WILLIAM LOWELL PUTNAM
December 3, 1977
Problem A-1
Consider all lines which meet the graph of
y -2x 4 +7x 3 +3x-5
in four distinct points, say (xj,yj), i - 1, 2,3,4. Show that
x 1 +X 2
+X 3
+X 4
4
is independent of the line and find its value.
Problem A-2
Determine all solutions in real numbers xy,z,w of the system
x+y+z-w,
I+ll1I I
x y z w
Problem A-3
Let u,f, and g be functions, defined for all real numbers x, such that
u(x+l)+u(x-l) -f(x) and u(x+4)+u(x-4)_g(x).
2 2
Determine u(x) in terms of f and g.
Problem A-4
For 0<x < 1, express
W xr
n-O -x
as a rational function of x.
Problem A-5
Prove that
(pb)(a)(modp)
for all integers p,a, and b with p a prime, p >0, and a >bO.
Notation: (in) denotes the binomial coefficient ml
Problem A-6
Let f(x,y) be a continuous function on the square
S-{(x~y):O <x< ,O <y 1).
For each point (a,b) in the interior of S. let S(,ab) be the largest square that is contained in S, is centered at (a,b),
and has sides parallel to those of S. If the double integral fff(x,y)dxdy is zero when taken over each square
29
S(., by must J(x,y) be identically zero on S?
Problem B-aI
Evaluate the infinite product
ri n3-I
3
n-2 l + I
Problem B-2
Given a convex quadrilateral ABCD and a point 0 not in the plane of ABCD, locate point A' on line OA, point B'
on line OB, point C' on line OC, and point D' on line OD so that A'B'C'D' is a parallelogram.
Problem B-3
An (ordered) triple (x1 ,x2 ,x 3 ) of positive irrationalnumbers with x1 +x 2 +x 3 -I is called "balanced" if each
xi < 1/2. If a triple is not balanced, say if xj > 1/2, one performs the following "balancing act"
B(x 1 , x 2 , x 3) -(Xl X2, X3)
where x,- 2x, if i:j and xj- 2xj - 1. If the new triple is not balanced, one performs the balancing act on it. Does
continuation of this process always lead to a balanced triple after a finite number of performances of the
balancing act?
Problem B-4
Let C be a continuous closed curve in the plane which does not cross itself and let Q be a point inside C. Show
that there exist points PI and P2 on C such that Q is the midpoint of the line segment PIP2 .
Problem B-5
Suppose that a,, a2,...,a, are real (n> 1) and
n I ~ \2
A+ < 22a,
Prove that A <2aaj for I < i Q < n.
Problem B-6
Let H be a subgroup with h elements in a group G. Suppose that G has an element a such that for all x in H,
(xa)3 _ 1, the identity. In G, let P be the subset of all products x1ax2 a*** xa, with n a positive integer and the x, in
H.
(a) Show that P is a finite set.
(b) Show that, in fact, P has no more than 3h2 elements.
THE THIRTY-NINTH WILLIAM LOWELL PUTNAM
December 2, 1978
Problem A-1
Let A be any set of 20 distinct integers chosen from the arithmetic progression 1, 4, 7,..,,100. Prove that there
must be two distinct integers in A whose sum is 104.
Problem A-2
Let a,bp ,p2 ... 'p,, be real numbers with a #b. Define f(x)-(p 1 - XXP2 - xXp 3 - x) ... (p,,-x). Show that
p a a a ... a a
b P2 a a ... a a
b b p3 a ... a a
- bf(a)-aj(b)
detb b b p4 ... a a
b-a
b b b b * p,,.- a
b b b b *.. b p.
Problem A-3
2 3 4 5 6
Letp(x)-2+4x+3x +5x +3x +4x +2x . For k with O<k<5, define
D k
k' lP(X)
For which k is Ik smallest?

Problem A4
A "bypass" operation on a set S is a mapping from S XS to S with the property
B(B(w, x), B(y,z)) - B(w,z) for all w, x, y, z in S.
(a) Prove that B(a,b)-c implies B(c,c)-c when B is a bypass.

(b) Prove that B(a,b)-c implies B(a,x)-B(c,x) for aU x in S when B is a bypass.
(c) Construct a table for a bypass operation B on a finite set S with the following three properties:
(i) B(x,x)-x for all x in S.
(ii) There exist d and e in S with B(d,e)-dF e.
(iii) There exist f and g in S with B(fg)#f.
Problem A-5
Let O<x,< ifor i-1,2,...,n and set
XI + X2 + + X.
n
Prove that
1 sinx, sinxy
i-l Xi X
31
Problem A-6
3 2
Let n distinct points in the plane be given. Prove that fewer than 2n " pairs of them are unit distance apart.
Problem B-1
Find the area of a convex octagon that is inscribed in a circle and has four consecutive sides of length 3 units and
the remaining four sides of length 2 units. Give the answer in the form r+ sVr with r, s, and t positive integers.
Problem B-2
Express
i-I m-I 2
m n+mn +2mn
2
as a rational number.
Problem B-3
The sequence {Q.(x)) of polynomials is defined by
Q1(x) I + x, Q2(x) - I + 2x,

and, for m > 1, by
Q2-+ 1(x) -Q2m(X) + (m + I)XQ2. - I(x),

Q2. + 2(X) -Q2,. + I(x) + (m + I0XQ2,f(4)
Let xn be the largest real solution of Q.(x)-O. Prove that (x,,) is an increasing sequence and that lim-,,x,,-0.
Problem B34
Prove that for every real number N, the equation
X1+ X2+ X3 -XX2X3 + XX2X4 + XIX3X4 + X2X3X4
has a solution for which xI, x2 , X3 , X4 are all integers larger than N.
Problem 3-5
Find the largest A for which there exists a polynomial
4 2
P(x)-Ax + Bx3 + Cx + Dx + E,
with real coefficients, which satisfies

0<P(x)<l for -IcxIl.
Problem 3-6
Letp and n be positive integers. Suppose that the numbers ck(h-1,2,...,n;k-1,2,...,ph)satisfy 0<cck^ 1.
Prove that
(w h c i )2over ah k
where each summation is over all admissible ordered pairs (hok).

THE FORTIETH WILLIAM LOWELL PUTNAM
December 1,1979
Problem A-1
Find positive integers n and al, a 2 ... , a. such that
a, +a 2 + +a.- 1979
and the product ala2 a. is as large as possible.
Problem A-2
Establish necessary and sufficient conditions on the constant k for the existence of a continuous real valued
function f(x) satisfying J(f(x)) - kx' for all real x.
Problem A-3
Let xI, X 2 , X 3 ,... be a sequence of nonzero real numbers satisfying
x" - 2X.- Iforjs-3,4,5,....
n2X.-2-X.- I
Establish necessary and sufficient conditions on x, and x2 for x., to be an integer for infinitely many values of n.
Problem A-4
Let A be a set of 2n points in the plane, no three of which are collinear. Suppose that n of them are colored red
and the remaining n blue. Prove or disprove: there are n closed straight line segments, no two with a point in
common, such that the endpoints of each segment are points of A having different colors.
Problem A-5
Denote by [xl the greatest integer less than or equal to x and by S(x) the sequence [x, [2xx],3x.... Prove that
there are distinct real solutions a and 8 of the equation x 3 - 1Ox 2 +29x-2S-O such that infinitely many positive
integers appear both in S(a) and in S(p).
Problem A-6
Let O<pi < I for i-1,2,..., n. Show that
n I I8~l.I I-l
i-I l| <8n(1+ + 5+ + 2n-Ix)
for some x satisfying 0 < x < I.

Problem B-1
Prove or disprove: there is at least one straight line normal to the graph of y - cosh x at a point (a,cosh a) and also
normal to the graph of y -sinhx at a point (c,sinbc).
[At a point on a graph, the normal line is the perpendicular to the tangent at that point. Also, coshx-(ex+
e -')/2 and sinh x-(e - e )/2.
Problem B-2
Let 0<a<b. Evaluate l 1/l
Ii {| [bx+a(l-~x)]tdx}/
[The final answer should not involve any operations other than addition, subtraction, multiplication, division, and
exponentiation.]
33
Problem B-3
Let F be a finite field having an odd number m of elements. Let p(x) be an irreducible (i.e., nonfactorabit)
polynomial over F of the form
2
x +bx+c, b, cEF.
For how many elements k in F is p(x)+k irreducible over F?
Problem B-4
(a) Find a solution that is not identically zero, of the homogeneous linear differential equation
2 2
(3x +x- I)y"-(9x +9x-2)y'+(18x+3)y-0.
Intelligent guessing of the form of a solution may be helpful.
(b) Lety=f(x) be the solution of the nonhomogeneous differential equation

2 2
(3x +x- I)y"-(9x +9x-2)y'+(18x+3)y-6(6x+ 1)
that hasf(0)-I and (,f(- I) - 2)(f(l) - 6)- I. Find integers a, b, c such that (f(- 2) -a)Xf(2) - b) c.
Problem B-5
In the plane, let C be a closed convex set that contains (0,0) but no other point with integer coordinates. Suppose
that A(C), the area of C, is equally distributed among the four quadrants. Prove that A(C) C 4.
Problem B-6
For k - IX 2,..., n let z,-xk +1yk, where the xk and yk are real and i IVI
. Let r be the absolute value of the
real part of
Prove that r <IxI I+ i x21+ *** +Ix,,.

THE FORTY-FIRST WILLIAM LOWELL PUTNAM
December 6, 1980
Problem A-l
Let band cbe fixed real numbersand let the ten points(jy.v).j= 1.2....,10, lieon theparabolay = X2 + bx + c.
For j = 1.2.... 9, let 1, be the point of intersection of the tangents to the given parabola at (j, y,) and
(j + I, ,+ I ). Determine the polynomial function Y = g( x) of least degree whose graph passes through all nine
points I,.
Problem A-2
Let r and s be positive integers. Derive a formula for the number of ordered quadruples (a, b, c, d) of positive
integers such that
3'. 7'= lcm[a, b, c] =lcm[a, b, d] =lcmja, c,d] lcm[b, c,dl.
The answer should be a function of r and s.
(Note that I cm [x, v.:] denotes the least common multiple of x, y, z.)
Problem A-3
Evaluate
2 dx
I + (tan x)
Problem A-4
(a) Prove that there exist integers a, b, c. not all zero and each of absolute value less than one million, such that
la + be2 +c cy- 1I0 - " -
(b) Let a, b, c be integers, not all zero and each of absolute value less than one million. Prove that
la + b1 +c31 > lo-21,
Problem A-5
Let P(t) be a nonconstant polynomial with real coefficients. Prove that the system of simultaneous equations
0 =IP(t)sintdtf
= P(i)costdt
has only finitely many real solutions x.

Problem A-6
Let C be the class of all real valued continuously differentiable functions/ on the interval 0 c x < I with ,(0) -0
and f(l) = 1. Determine the largest real number u such that
for af'(l)l-f(x)dx
for all f in C.
35
Problem B-I
For which real numbers c is (et+ e -)/2 C e"' for all real x?
Problem B-Z
Let S be the solid in three-dimensional space consisting of all points (x, v. z) satisfying the following system of six
simultaneous conditions:
x0, v -. z'0,
x +y + z < 11,
4
2x + V + 3: < 36,
2x + 3z < 24.
(a) Determine the number tv of vertices of S.
(b) Determine the number e of edges of S.
(c) Sketch in the bc-plane the set of points (b, c) such that (2,5,4) is one of the points (x, v, z) at which the linear
function bx + cy + z assumes its maximum value on S.
Problem B-3
2
Forwhich real numbers does thesequencedefinedby the initialconditionuo0 = and therecursion u,, 1 = 2u- n
have u,, > 0 for all n ; 0?
(Express the answer in the simplest form.)
Problem B-4
Let Al, A 2. A*,

A1066 be subsets of a finite set X such that IAil > 'I XI for I < i e 1066 Prove there exist ten
elements xI,..., x1o of X such that every A, contains at least one of xI,.. , xlO.
(Here I SI means the number of elements in the set S.)
Problem B-5
For each t ; 0. let S, be the set of all nonnegative, increasing, convex, continuous, real-valued functions f(x) defined
on the closed interval [0, II for which
f(l) - 2f(2/3) +f(l/3) ;- t[f(2/3) - 2f(1/3) +f(0)].
Develop necessary and sufficient conditions on t for S, to be closed under multiplication.
(This closure means that, if the functions f(x) and g(x) are in S., so is their product f(x) g(x). A function f(x) is
convex if and only iff(su + (I -s)v) < sf(u) + (I -s)f(v) whenever 0C s < 1.)
Problem B-6
An infinite array of rational numbers G(d, n) is defined for integers d and n with I A: d C n as follows:
G(l,n)--, G(d,n)-- = , G(d-1,i-I) for d> 1.
For I < d < p and p prime, prove that G(d, p) is expressible as a quotient sIt of integers s and t with t not an
integral multiple of p.
(For example, G(3, 5) = 7/4 with the denominator 4 not a multiple of 5.)
THE FORTY-SECOND WILLIAM LOWELL PUTNAM
December 5, 1981
Problem A-i
Let E(n) denote the largest integer k such that 5k is an integral divisor of the product 1'2233 n'. Calculate
tim E(n)
Problem A-2
Two distinct squares of the 8 by 8 chessboard C are said to be adjacent if they have a vertex or side in common.
Also, g is called a C-gap if for every numbering of the squares of C with all the integers 1,2,..., 64 there exist two
adjacent squares whose numbers differ by at least g. Determine the largest C-gap g.
Problem A-3
Find
lim [e fJe - end d_ ]
or show that the limit does not exist.

Problem A-4
A point P moves inside a unit square in a straight line at unit speed. When it meets a corner it escapes. When it
meets an edge its line of motion is reflected so that the angle of incidence equals the angle of reflection.
Let N( T) be the number of starting directions from a fixed interior point POfor which P escapes within T units
of time Find the least constant a for which constants b and c exist such that
N(T) < aT2 + hT + c
for all T > 0 and all initial points PO.
Problem A-5
Let P(x) be a polynomial with real coefficients and form the polynomial
2 2
Q(x) = ( l) p(X)p'(x) + x([P(x)] + [P'(x)] ).
Given that the equation P(x) = 0 has n distinct real roots exceeding 1, prove or disprove that the equation
Q( v) = 0 has at least 2n - I distinct real roots.
Problem A-6
Suppose that each of the vertices of AABC is a lattice point in the (x, y)-plane and that there is exactly one lattice
point P in the interior of the triangle. The line AP is extended to meet BC at E. Determine the largest possible value
for the ratio of lengths of segments
API
IPEI
[A lattice point is a point whose coordinates x and v are integers.]
37
Problem B-I
Find
4 2 2
lirn (5h l8hi k + 5k4
Problem B-2
Determine the minimum value of
(r +) (S
(r
)2 +
) (s
(~ 1)+(
I
)
for all real numbers r, s t with I < r < s < t < 4.
Problem B-3
2
Prove that there are infinitely many positive integers n with the property that if p is a pnme divisor of n + 3, thenp
2 2
is also a divisor of k + 3 for some integer k with k < n.
Problem B4
Let V be a set of 5 by 7 matrices, with real entries and with the property that rA + sB E V whenever A, B (E V and
r and s are scalars (i.e., real numbers). Prove or disprove the following assertion: If V contains matnces of ranks 0, 1,
2, 4, and 5, then it also contains a matrix of rank 3.
[The rank of a nonzero matrix M is the largest k such that the entnes of some k rows and some k columns form a
k by k matrix with a nonzero determinant.1
Problem B-5
Let B(n) be the number of ones in the base two expression for the positive integer n. For example. B(6) = B(1102)
2 and B(15) = B(l 11 12) = 4. Determine whether or not
Y-( fn(n + 1))
is a rational number. Here exp(x) denotes en.
Problem B-6
Let C be a fixed unit circle in the Cartesian plane. For any convex polygon P each of whose sides is tangent to C. let
N( P, A, k) be the number of points common to P and the unit circle with center at (h, k). Let Hf(P) be the region
of all points (x, y) for which N(P, x, y) 2 I and F(P) be the area of H(P). Find the smallest number u with
F(P)JJN(Px, Y)dxdy'< u
for all polygons P. where the double integral is taken over H(P).
THE FORTY-THIRD WILLIAM LOWELL PUTNAM
December 4, 1982
Problem A-I
Let V be the region in the Cartesian plane consisting of all points (x. v) satisfying the simultaneous conditions
xjI<y < Ixj + 3 and v< 4.
Find the centroid (x, y) of V.
Problem A-2
For positive real x, let
B,(x) = 1x + 2' + 3' + + nx
Prove or disprove the convergence of
E B(log.2)
.- 2 (nlog2 n)2
Problem A-3
Evaluate
foOArctan(vx) - Arctan x dx.
C x
Problem A4
Assume that the system of simultaneous differential equations
yV =
y -Z 3.Z ,
-v3 V3
with the initial conditions y(O) = 1, z(O) - 0 has a unique solutiony =f(x), z = g(x) defined for all real. Prove
that there exists a positive constant L such that for all real x.
f (x + L)-f (x), g(x + L) - g(x).
Problem A-5
Let a, b, c, and d be positive integers and
ac
hd'
Given that a + c 6 1982 and r > 0, prove that
>19833
Problem A-6
Let a be a bijection of the positive integers, that is, a one-to-one function from (1,2,3.... onto itself. Let
X,, x 2 , X3,... be a sequence of real numbers with the following three properties:
39
(i) Ix.1 is a strictly decreasing function of n;

(ii) [a(n) - nj Ix.1 0 as n - o;
(iii) lim_-Y- Ix' 1.
Prove or disprove that these conditions imply that
lim E X.k)- 1
Problem B-l
Let M be the midpoint of side BC of a general AA BC. Using the smallest possible n, describe a method for cutting
AAMB into n triangles which can be reassembled to form a triangle congruent to AAMC.
Problem B-2
Let A(x, y) denote the number of points (m, n) in the plane with integer coordinates m and n satisfying
ml + n' < x2 + Y.2 Let g = Ek'- e52. Express
f Af(x )e id d
as a polynomial in g
Problem B-3
Let p,, be the probability that c + d is a perfect square when the integers c and d are selected independently at
random from the set (1, 2, 3,.... n}. Show that lim,- .parn) exists and express this limit in the form r(v/; - t).
where s and t are integers and r is a rational number.
Problem B4
Let nl, n2 - . - n, be distinct integers such that
(n, + k)(n2 + k) (n, + k)
is an integral multiple of njn 2 n, for every integer k. For each of the following assertions, give a proof or a
counterexample:
(a) In lI I for some i.
(b) If further all n, are positive, then
(n,, n2,., n,) (12,.. s).
Problem B-5
For each x > e' define a sequence S, = u0, u:. u2 ... recursively as follows: uo = e. while for n > 0. u,, | is the
logarithm of A to the base u,. Prove that St converges to a number g( x) and that the function g defined in this way
is continuous for v > en.
Problem B-6
Let K(x. v. :) denote the area of a triangle whose sides have lengths x. v, and :. For any two tangles with sides
a, b. c and a'. h', c'. respectively, prove that
VK(a, b. c) + f-K(a', b, c') < VK(a + a'. b + b'.c + c')

and determine the cases of equality.
THE FORTY-FOURTH WILLIAM LOWELL PUTNAM
December 3, 1983
Problem A-1
How many positive integers n are there such that n is an exact divisor of at least one of the numbers 104°, 2030?
Problem A-2
The hands of an accurate clock have lengths 3 and 4. Find the distance between the tips of the hands when that
distance is increasing most rapidly.
Problem A-3
Let p be in the set (3, 5.7.11,... ) of odd primes and let

F(n) - I + 2n + 3n 2 + +(p -+ )n 2
Prove that if a and b are distinct integers in (0,1, 2,..., p - 1) then F(a) and F(b) are not congruent modulo p,
that is, F(a) - F(b) is not exactly divisible by p.
Problem A4
Let k be a positive integer and let m 6k - 1. Let

S() 2k-1
S(m) j=l
E (-1)J+'(3jm 1).
For example with k - 3,
S(17) (17) (17) +( 17) (17) 1(7)
ProvethatS(m)isneverzero. [AsuMual,(,) - !( - !]
Problem A-5
Prove or disprove that there exists a positive real number u such that [ 0l - n is an even integer for all positive
integers n.
Here [x] denotes the greatest integer less than or equal to x.
Problem A-6
Let exp(t) denote e' and
3
F(x) - xpx | x| exp(u 3 + v3) dvdu.
Find limx-<,F(x) or prove that it does not exist.
41
42 THE WILl IAM LOWELL PUTNAM MATHEMATICAL COMPETITION
Problem B-I
Let v be a vertex (comer) of a cube C with edges of length 4. Let S be the largest sphere that can be inscribed in
C. Let R be the region consisting of all points p between S and C such that p is closer to a}than to any other
vertex of the cube. Find the volume of R.
Problem B-2
For positive integers n, let C(n) be the number of representations of n as a sum of nonincreasing powers of 2,
where no power can be used more than three times. For example, C(8) - 5 since the representations for 8 are:
8, 4 + 4, 4 + 2 + 2, 4 + 2 + 1 + 1, and 2 + 2 + 2 + 1 + 1.
Prove or disprove that there is a polynomial P(x) such that C(n) - [P(n)] for all positive integers n; here Jul
denotes the greatest integer less than or equal to u.
Problem B-3
Assume that the differential equation

y "' + p(x)y" + q(x)y' + r(x)y -0
has solutions yl(x), y 2 (x), and y 3 (x) on the whole real line such that
yR(x) +y2(x) +Y3(X)-1

for all real x. Let
2 2
f(x) - (yff(X)) +(y2;(x)) +(y3(X)) 2 .
Find constants A and B such that f(x) is a solution to the differential equation
y' + Ap(x)y - Br(x).
Problem 84
Let f(n) = n + [v;;1 where [x] is the largest integer less than or equal to x. Prove that, for every positive integer
m, the sequence
m,f(m),f(f(m))f(f(f(m)))
contains at least one square of an integer.
Problem B-S
Let lull denote the distance from the real number u to the nearest integer. (For example, 1j2.811
- .2 - 113.211.)
For positive integers n, let
a,,--ijiflhax.
Determine lim,,-,,a. You may assume the identity
z 2 4 4 6 6 8 '8'' ' 2'
Problem B-6
Let k be a positive integer, let m - 2k + 1, and let r * 1 be a complex root of z' - 1 - 0. Prove that there
exist polynomials P(z) and Q(z) with integer coefficients such that
(P(r)) +(Q(r)) - -1.

THE FORTY-FIFTH WILLIAM LOWELL PUTNAM
December 1, 1984
Problem A-I
Let A be a solid a x b x c rectangular brick in three dimensions, where a, b, c > 0. Let B be the set of all
points which are a distance at most one from some point of A (in particular, B contains A). Express the volume of
B as a polynomial in a, b, and c.
Problem A-2
Express E-k- 1(6k/(3"+ - 2+ I )(3A - 2k)) as a rational number.
Problem A-3
Let n be a positive integer. Let a, b. x be real numbers, with a s h. and let M,, denote the 2n X 2n matrix
whose (i, j) entry m,, is given by
if ij.
mj= a if i * j and i + j is even,
Xb if i * j and i + j is odd.
2
Thus, for example, 2 = (h x h . Express lim,- det M,,I(x - a) " 2 as a polynomial in a, b, and n,
ba b x
where det M,, denotes the eterminant of M.
Problem A-4
A convex pentagon P = ABCDE, with vertices labeled consecutively, is inscribed in a circle of radius 1. Find the
maximum area of P subject to the condition that the chords AC and BD be perpendicular.
Problem A-5
Let R be the region consisting of all triples (x, v. z) of nonnegative real numbers satisfying x + v + z < 1. Let
W= -x y - z. Express the value of the triple integral
1 9 4
Jtx y Z Iw dx dv d-
R
in the form a!b!c-!d!/n!, where a, b. c, d, and n are positive integers.

Problem A-6. Let n be a positive integer, and let f(n) denote the last nonzero digit in the decimal expansion of n!.
For instance, f(5) = 2.
(a) Show that if a, a ... , a, are distinct nonnegative integers, then f(5"8- + 52++ + 5" ) depends only on
the sum a + a2 +-' +a .
(b) Assuming part (a), we can define
g(s) |(5"1 + 5"' + ... .4 5'
where s = ai + a 2 + * -ak. Find the least positive integer p for which

g(s) = g(s + p), for all s > 1,
or else show that no such p exists.
43
Problem B-I
Let n be a positive integer, and define
f(n) = I! + 2! +-
Find polynomials P(x) and Q(x) such that
2) = P(n )f(n + L) + Q(n )f(n)
+-(n
for all n > 1.
Problem B-2. Find the minimum value of
(u-v)2 +(- 9)2
for 0 0.
Problem B-3
Prove or disprove the following statement: If F is a finite set with two or more elements, then there exists a
binary operation * on F such that for all x, v, z in F.
(i) x * z = y z implies x - y (right cancellation holds),

and
(ii) x *( v * z) (x * y) * z (no case of associativity holds).
Problem B4
Find, with proof, all real-valued functions y = g(x) defined and continuous on [0, so), positive on (0, x). such
that for all x > 0 the y-coordinate of the centroid of the region
Rx - {( S. t)10 4%s S<x, 0 < t _<g(s)}

is the same as the average value of g on [0, x].
Problem B-5
For each nonnegative integer k, let d(k) denote the number of I's in the binary expansion of k (for example.
d(O) = 0 and d(5) - 2). Let m be a positive integer. Express
2^-
E ( l)d(k)k=
k-0
in the form (- 1)'maf')(g(nm))!, where a is an integer and f and g are polynomials.
Problem B-6
A sequence of convex polygons { P, }, n > 0, is defined inductively as follows. Po is an equilateral triangle with
sides of length 1. Once P, has been determined, its sides are trisected; the vertices of P.,01 are the interior trisection
points of the sides of P,. Thus, P, l is obtained by cutting corners off P., and P, has 3 - 2" sides. ( P, is a regular
hexagon with sides of length 1/3.)
Express lim,, -_Area(P,) in the form Fa/b, where a and b are positive integers.
SOLUTIONS
THE TWENTY-SIXTH WILLIAM LOWELL PUTNAM
November 20, 1965
A-1. Suppose the bisector of the exterior angle at A intersects line BC at X

and the bisector of the exterior angle at B meets the line A C at Y. The assump-
tion that C is between B and X contradicts the fact that Z B > L C so we may
assume that B is between X and C. Similarly, we conclude that C is between
A and Y because Z A < Z C.
If Z is a point on line AB with B between A and Z, we have from triangle
ABY that ZZBY=2A. Hence, ZBXA=ZABX= ZZBC=2ZZBY=4A,
and the angle sum of triangle ABX is 90°-4A+8A. Thus, A=12'.
A-2. Substituting s=n-r in the given summation reveals that twice this
sum is equal to:
n n- 2r(n)}
r-O{ n (r)}
(I1)-2 ( = E (( -4 + 4)(
(2fl) - E (is) (4 (n -)2
n C- r r- I
. 2nA 2 - 1) + 4(2n - 2) (2n 2-
2 28
\-1nn n- I n \n -2
= {2n(2n- 1) n-1} 2n-2) 2 (2n-2)
n2 n n- I n n- 1
Comment: This solution assumes the well-known identities
( 2) (n) and E (kIn) ( ) = ( + n)
which may be proved by comparing coefficients in the expansion of
(1 + x)" (1 + X)n = (1 + X)"+.
A-3. That (A) implies (B) follows from the fact that subsequences of a con
vergent sequence converge to the limit of the sequence. We simplify the nota-
47
48 IHE WILLIAM LOWELL PUTNAM MATHEMATICAL COMPETITION
tion by setting cr=expia, and S(t)=cI+c2 + * * +ct.

± Note that Icj =1 and
|S(t+k)-S(t) <•k. Suppose now that (B) holds and write m=n2 +k, where
O<k•2n.
S(m) S(n2) S(m) S(n2 ) I + IS(n 2 ) S(n12 )l
mk+ 2(1 -
m 1
1) k
+ n2
mn=
mn m
2k <4n
m 112 m) m m 12
We conclude that linim..(S(m)/m-S(n2)/n2) =O or that S(m)l/m converges to a.
A-4. Let b be a boy who dances with a maximal number of girls (i.e., there
may be another boy who dances with the same number of girls, but none dances
with a greater number). Let g' be a girl with whom b does not dance, and b'
a boy with whom g' dances. Among the partners of b, there must be at least one
girl g who does not dance with b' (for otherwise b' would have more partners
than b). The couples gb and g'b' solve the problem.
A-5. On the basis of the first few cases we conjecture the answer is 21-1 and
proceed by induction.
We first show (also by induction) that an n-arrangement ends in 1 or n.
Note that when n is deleted from an n-arrangement, the result is an (n-i)-
arrangement. If an n-arrangement does not end in 1 or n, deletion of n produces
an (n-1)-arrangement ending (by induction) in (n-1). This implies the n-
arrangement ended in n because n cannot precede (n-1).
For any n-arrangement (a,, a2, - *, a.) there is another n-arrangement
(a{, a2., . ., an.'), where al' = n+ 1 -ai. If one of these ends in n, the other ends
in 1 and consequently exactly half of the n-arrangements end in n.
All of the n-arrangements which end in n can be obtained by adjoining an n
to the end of all (n-1)-arrangements, and by induction there are 2n-2 of these.
Hence, there are 2-1 n-arrangements.
A-6. The problem is not well set being true only under rather heavy restric-
tions on the x, y, u and v. For example, all is in order if they are nonnegative.
However, if m is rational with odd numerator and odd denominator there are
tangent lines to the curve for which u"+Vn> 1, while if m is rational with odd
numerator and even denominator then n is rational with odd numerator and
odd denominator and there are solutions (u, v) of u"±+v = 1 such that the line
ux+vy= 1 is not tangent to the curve xm+ym= 1.
Let (x0,y0) be a point on the curve xm+ym= 1. The tangent to this curve is
x0 n-x+yn-'y= 1. If this line is ux+vy = 1, then u =x0T-' and v =yn' with both
u and v nonnegative. The relation I/m+1/n=1 gives m/(m-1) =n and we ob-
tain u + v=-X4+y'=1.
Conversely, let m-'+n-'= 1 and let u and v be nonnegative and such that
Uin+vn= 1. Define x0 and yo by the equations X0=Un Im and yo=vnlIm. Then x0 and
y, are nonnegative and x'±+y,=un+v-=1. Thus, (x0, y,) is on the curve
SOLUTIONS: THE TWENTY-SIXTH COMPETITION 49
x-+ym = 1 and the line ux+vy = 1 is the tangent to the curve by the calculation
above.
In this solution we use the fact that for nonnegative a and positive r and s
(a,) = are.
B-1. The change of variables xk-4l -xi yields
f **
*' OS+ {-(C+X 2 * + Xn)} dXldX2 *.. dx
o o o2n
= f... f1 sin 2 (xI + x2 + + x.) dxldxg ... dxo.
Each of these expressions, being equal to half their sum, must equal J. The
limit is also i.
B-2. Clearly Wr+1,=n-1 for r=1, 2, , n and 1 a,= ET l,. Hence,
r = h (w°- 1r)(w, + I) = (n 1) I (wI- lr) = (n - 1)-0 = 0.

1 1 1 1
B-3. All Pythagorean triples can be obtained from x=X(p2 -q 2 ), y=22Xpq,

z=X(p2 +q 2 ) where O<q<p, (p, q)=l and ppq mod 2, X being any natural
number.
The problem requires that 2xy=2(x+y+z). This condition can be written
X2(p 2 -q 2 )(pq)=2X(p 2 -q 2 +2pq+p2 +q 2 ) or simply X(p-q)q=4. Since p-q is
odd it follows that p -q = 1 and the only possibilities for q are 1, 2, 4.
If q=1, p=2, X=4, x=12, y=16, z=20.
If q=2, p=3, X=2, x=10, y=24, z=26.
If q=4, p=5, X=1, x=9, y=40, z=41.
B-4. Since
nnf(X, + I it) + X
( )='nf(X,
+ 1) = X ) + 1
If x is such that f(x, n) converges when n tends to infinity, the limit F(x) must
satisfy F(x) = (F(x) +x)/(F(x) + 1), F2(x) =x. The convergence to \/x is obvious
when x = 0 or 1. To show this convergence for any positive x we first note that
(1 + V/x)" + (1 - V/x)^
f(x, n) = \/x (1 + -\VX)n (1 - \/x) -
When 0<x<1, write a=(1-\x)/(1+v\x); then 0<a<1 and
f(x, n) = \/x -4 N/x.

1 - an
When x>1, write b=(x/x-1)/(Vx+1); then O<b<l and
f(x, n) = v'x + ( \/X.

1 - ba
The limit fails to exist for negative values of x; but for all other complex num-
bers the limit exists and is that square root of x which lies in the right half plane.
B-5. Divide the objects into two subsets {a,,a .}
,a, and {Ibb ,2 . . . obn}a
where m+n= V. Then the mn pairs (a3 , bk), where j= 1, 2, * , m and k= 1, 2,
, n, obviously contain no triangles. If V is even, take m=n= V/2, and if
V is odd, take m = (V+ 1)/2, n = (IV-1)/2. Then mn - V 2 /4- E.
B-6. Suppose A, B, C, D are neither concyclic nor collinear. Then p, the
perpendicular bisector of segment AB, cannot coincide with q, the perpendicular
bisector of segment CD. If the lines p and q intersect, their common point is the
center of two concentric circles, one through A and B, the other through C and
D. If instead p and q are parallel, so also are the lines AB and CD. Consider
points P and Q, on p and q respectively, midway between the parallel lines AB
and CD. Clearly, the circles ABP and CDQ have no common point.
THE TWENTY-SEVENTH WILLIAM LOWELL PUTNAM
November 19, 1966
A-i It is easily verified by induction that

f(n) = n2/4 when n is even,
A(n2 - 1)/4 when n is odd.
Therefore, since x+y and x-y always have the same parity, in any case we
must have
J(x + y)-f(x - y) = (X + v) 2 - (X - y) y
A-2 The area of the given triangle can be calculated in two ways,
A =pr and A = v'p(p-a)(p-b)(p-c). Squaring and equating we get p2r2
=p(p-a)(p-b)(p-c). Setting x=1/(p-a), y=1/(p-b), z=1/(p-c) we can
write this equation in the form
1 1 1 1
-= pxyz= xyz X I+ -.
Thus we need only show that yz+xz+xy<x2 +y 2 +z2 . However this follows
from the trivial inequalities y 2 +z 2 >-2yz, x 2+z 2 >2xz, x 2 +y 2 >2xy.
A-3 Multiplying the defining relation by (n+1) we get
2
(1) (n + 1)x 3 +1 = nx. + xn - (n + 1)(xn) = nX. + X,[1 - (n + 1)xn].
To prove that nx. is increasing, we need to show that 1 - (n +1 )x,, Ž0. From
the graph of x(1 -x) we note that X2 < ' and that xn <•a < I implies xn+ <•a(l -a).
So by induction,
(n + 1)x. - (n + 1) -(-) = 1--< 1.

n n n-
Furthermore, nx,<(n+1)x,<1 and so nx, is bounded above by 1. Thus nx"

converges to a limit a with O<nx,<a•1. Now summing (1) from 2 to n we
obtain
51
(2) (n + l)xn+i
= 2x2 + x2(1 - 3X2) + X3(1 - 4x3) + ±+xn[1 - (n + 1)x.].
If az1 then [1-(n+l)xI> (1-a)/2 for all large n and thus (2) would show
that EX, is convergent. However nx,>xl and so x,>xi E(1/n).
A-4 To prove the formula by induction, it suffices to show that the dif-
ference A =n±+ {Ivn --(n-1 + { \/n-1I ) = 1 or 2, with the value 2 occurring
}
if and only if the number n+ {-Vn- 1 is a perfect square. For convenience, let
{/n-1} =q. Then of course q- 1 < /n-1 <q+ or better q 2 -q+ I < n-l
2
<q2+q+'. This gives q2+5/4<n+{ /n-1} <(q+1) +±. Therefore the
number n + {-/n-l is a perfect } square if and only if n = (q +1) 2
-q. However,
then and only then V/n>q+12> /n-1. In other words then and only then
\/Vn } - { /Vn-1I } = 1, because this difference is never greater than 1.
A-5 We shall show that TVt(x) =f(x)+P(x) for all ACC, where f is the
image under T of the function 1 which sends x into 1. For each x0 and 1k, define
l' by
IV() {+(x) if x < XI,

= (x0 ) if x > x0.
Since T is local we must have T{'(x) = T4(x) for all x <x0. On the other hand,
for x>xo, T#I'(x) = t'(xo)T1 (x) =#,(xo) .f(x). By continuity of Tik', T4!(xo) =ip(xo)
*f(xo).
Comment: The condition (1) is needed only in the case where C2 =0. Also the interval I in
condition (2) should be required to have positive measures since the problem is trivial if I can be
taken as a point.
A-6 We understand the statement to mean that
3= lim + 2/
±/1 +3A+(-1) 1.
We see that
3 = 1 24= a1+ 2v /1 + 2A/1 +32
3
= /1 + 2A/1 + A/1 + 4A/36
This leads us to conjecture the relation

3 =
I1+2A/+3A/1+ +± +n /n ) for all n - 1.
Proceeding by induction we verify that (n+2)2 =n 2 +4n+4=1+(n+l)(n+3,
=1+(n+1) (n+3)/ This given, we must have
3 - 1/1 + 2AV/; + 3AI/ a/1- + (n - 1) ;
To set an inequality in the other direction; observe that for any a> 1
V TTa< a vrl/Tn.
SOLUTIONS: THE TWENTY-SEVENTH COMPETITION 53
A repetition of this inequality gives then

3 - (n + 2)a.I+2V4+3A/1+. *
V1+(f )1+ where a= 21-n.
B-i Let Pi, P 2 , , P. be the vertices of P. Let P 1i, P2 ,, * Put
be the projections of PI, P2 , * , Pit upon one of the sides of the squares, and
let Pi", P 2 ", * * *, Pn" be the projections of Pi, P2, , Pit upon a side that is
orthogonal to the previous one. Since P is convex, the first side will be covered
at most twice by the segments PI P 2' .. YPn', P n' P'. We thus deduce
the inequality PI'P2 '2 + * +P,'P1 " <2. Similarly P1 P2 " 2
< 2. Adding these two inequalities and using the Pythagorean theorem the asser-
tion follows.
B-2 Any common factor of two of such numbers would have to be
divisible by 2, 3, 5 or 7. So it is sufficient to prove that among any ten consecu-
tive integers there is at least one that is not divisible by 2, 3, 5 or 7. We get such
an integer by elimination as follows. Strike out those divisible by 3. There may
be either 3 or 4 of them. Among these there is either at least one or two respec-
tively that are divisible also by 2. Thus if we strike off also those that are
divisible by 2 we will have eliminated at most seven of the integers. Note that by
so doing we have stricken off at least one number divisible by five. Thus we are
left with three integers only two of which can be divisible by 5 or 7.
B-3 Set q.=Pl+p2 + +Pd(qo=O). We are led to estimate SN
.
= Ad I(n/q,,)1(q.-q,) in terms of T= , 1/p12. Note that
1 N 1 N n2 N n
SNV :5 + E7 (q - q.-I) -+ E -2 E X2
PI n-2 qnqn-1 PI n-2 qn-I n-2 qn
1 N-1 (n +1 ) i n2 5 + n N 1
PI n-I qn n=2 q. pi n-2 qn n_2 qn
By Schwarz's inequality,
(N n2 N n2 1
n-2 qn) .- 2 q2n n n ItPitP

and thus
SN • - + 2 /SNT + T.
This quadratic inequality implies that VYSN VT++

/ 2T-+5/pi.
B-4 Let, for each •i~mn+1, ni denote the length of the longest
chain, starting with ai and each dividing the following one, that we can select
out of ai, a,+ 1 , * * *, a. 1+,. If no ni is greater than n then there are at least
m+1 ni's that are the same. However, the integers ai corresponding to these
nit's cannot divide each other, because ai aj implies that ni Ž nj+ 1.
B-5 Let these points be denoted by P1 , P2 , * - , P.. To every permuta-

tion (al, a , o-) of (1, 2, 3, * , n) we associate a closed polygon, namely
P,,P,, * * P,,Po, This way we obtain (n - 1)! distinct closed polygons some
of which may have selfintersections. We claim that anyone of these polygons
whose length is the shortest possible is simple. By the hypothesis that no three
Pi's are on the same line, a selfintersection occurs if and only if two segments
say P,,Pa, and P,.P..,, cross each other. However, then the closed polygon
PO2. P.-1PomPO,PP_,-l * Pa,,iP,2 would have shorter length. Thus
there can't be a cross if the length of P, *.PI.P.,is shortest possible.
Alternate solution: Take two points P1 and P2 such that all the other points
P3 , P4 , - - * , P. are on the same side of the line connecting P1 and P2 . Each
point Pi, i>2, determines an angle As between P1 P2 and P1 Pi, with O<Oi<ir.
By hypothesis, Oiz~j if i-j. Let (i 3, i4,*- *, in) be the permutation of
(3, 4, *, n) such that &3,<0i,< .<i. Then PIP2PP 2 ,* Pi.P1 is a
closed simple polygon.
B-6 Multiplying the equation by e-xy' and integrating from 0 to T we
get, after rearranging terms,
y2 (T) + 2 j eay'y" dx = y2(O).
Note then that for some 0 << T we must have
W +o
2f ZY"
ery dx =2 r YIYI dx {yl(r) 2- {y'(0) }2.
We then obtain yl(T) + I yl(r) }2=y2(0) + {y'(0) }2

THE TWENTY-EIGHTH WILLIAM LOWELL PUTNAM
December 2, 1967
A-1
la, +2a2* = Jf'(O)f limn(X) -- f(O)

=- 11. em
, f(x) I sin x . f(X)I
=im sin x =l x A • 1.
X-.O Sinlxi- X z-'OlSinxlX
A-2 S. is the number of symmetric n X n permutation matrices (a permu-

tation matrix has exactly one 1 in each row and column with 0's elsewhere).
Let the 1 in the first row be in the kth column. If k = 1, then there are S.-, ways
to complete the matrix. If kPI1 then alk=akl=1 and the deletion of the 1st
and kth rows and columns leaves a symmetric (n-2)X(n-2) permutation
matrix. Consequently S= Sni + (n-1) Sn,2.
For part (b), let
F(x) = a{5x} F'(x) = E2{5 x }
wS Xn-I xn-1
j=
- 1+ (n 1)Sn-2 (n-i
n=1
= A{Sn-
n,=o S
(n
Xn!1
+
x
{nn- 2
n2 (1-)
}
(n TJI
F(x) + xF(x).
Hence F'(x)/F(x) = 1 +x. Integration and use of F(O) = = 1 yields F(x)

=exp(x+x 2 /2). Now the series for F(x) is uniformly convergent for all x, so all
the operations are legal.
Comment: Sn is the number of permutations lr, with 7r2= 1, in the symmetric group on n sym-
bols. Some contestants used this observation together with known formulas for the number of
permutations with a given cycle structure to check part (b) by multiplying the series for exp(x)
and exp(x 2 /2).
A-3 Let f(x) =ax2 -bx+c=a(x-r)(x-s). Then f(0) f(1) =a 2 r(r-1)
.s(s-1). The graph of r(r-1) shows that O<r<1 implies O<r(r-1)<4,with
equality if and only if r = 2. Similarly, 0 <s(s-1) - L. Since rids, r(r-l )s(s-1)
<1/16 and 0<f(O).f(1)<a 2 /16. The coefficients a, b, c are integers and thus
55
1 f(0) -f(1). Consequently a 2 > 16, i.e. a>5.

The discriminant b 2 -4ac shows that the minimum possible value for b is 5.
Furthermore, 5x 2 -5x+l has two distinct roots between 0 and 1.
A-4 Assuming that there is a solution u, then integrating with respect to
x from 0 to 1, one obtains fJu(x)dx =f 1 dx+Xff l4fu(y)u(y-x)dy jdx. In the
iterated integral, one can interchange the order of integration, and letting
fu(x)dx=a, one gets a=1+Xf Ju(y)fu(y-x)dxjdy. Now, holding y fixed,
let z=y-x to get a=1+XfS{u(y)fOu(z)dz}dy. Set f(y) =fu(z)dz. Then a=1
+Xfof'(y)f(y)dy=l+X {2Xf2(1)-ljf2(0)} =1+X- 2ca, orXa 2 -2a+2=0. The dis-
criminant of this quadratic shows that if X> i then the roots are imaginary.
A-5 Let the maximum diameter be 2d and assume d<1. Take such a
diameter as the x-axis with the origin at the mid-point. Since this is a maximum
diameter the region is bounded between the lines x= -d and x=d. The upper
and lower boundaries of the region are functions, because of convexity. Denote
them byf(x) and -g(x), where f and g are nonnegative for -d -x-d. Calculat-
ing the distance between (x, f(x)) and (-x, -g(x)) shows that f(x)+g(-x)
<-\/(1 +4x2 ), for -d -x -d. Area =f-d {f(x) +g(-x) dx<f-dV/(1 l+4x2 )dx
<fY12,2V(1 +4X )dx = ,'r This contradiction proves that d- 2 and so there
2
must be at least two points a unit distance apart.

Comment: The requirement that the boundary contain at most a finite number of straight line
segments was extraneous.
A-6 Solving the given equations in terms of x3 and X4, leads to the
equivalent system: x1 =A 1 x3+Blx4 , x 2 =A 2x 3 +B2x4 , X3=xa, X4 =X4 , where
A1 = (a2b3 - a3b2)/(aib2 - a2b1 ), etc.
Each point in the X3, x 4-plane corresponds uniquely to a solution (xI, x2,
Signum x1 is positive for (X3, X4) on one side of Ax3±+BIx4 =0 and nega-
X3, X4 ).
tive on the other side. Similarly for signum X2, signum X3, and signum x4, using
A2x3+B 2x3=0, X3=0 and X4 =0, respectively. These four lines through the
origin, in general, divide the plane into eight regions, each having a different
4-tuple of signum values. Hence the maximum number of distinct 4-tuples is
eight.
The maximum number of eight will occur if and only if there are actually
four distinct lines. This is equivalent to the conditions A 1 00, A 27#0, B 1 O,
B2Fs0 and AiB 2 -A 2 B#00 or, simply, aibj-ajbi3#0 for i, j=1, 2, 3, 4 and i<j.
B-1 Consider the figure in the complex plane with the center of the circle
at the origin. We can take A, B, C, D, E, F as complex numbers of absolute
value r. Furthermore B = Aw, D = Cw and F=Eco, where w = cos(7r/3) +i sin (ir/3).
Since 3 =1 and w71, w2 -±+1=0. The mid-points of BC, DE and FA are
P=1(Aw+C), Q=I(Cw+E) and R=-(Ew+A). If the segment from Q to R is
rotated through 7r/3 about Q, then R is carried into Q+w(R-Q), which equals
P. Thus P, Q, R are vertices of an equilateral triangle.
B-2 For part (a) one has immediately that A =p2 , B=2p(1-p), and
SOLUTIONS: THE TWENTY-EIGHTH COMPETITION 57
C= (1-p) 2 . The result follows by examination of the graphs for A, B and C on

O-p•1.
For part (b), a=pr, f3=p(l-r)+r(l-p), y=(l-p)(l-r). Consider the

region R in the p, r-plane defined by Op<1 and O<r•1. We will show that
there is no point in R with a<4/9, 13<4/9 and -y< 4 /9 . If a< 4 /9 and y<4/9
then (p, r) is between the hyperbolas pr=4/9 and (1-p)(I-r) -4/9. These
hyperbolas have vertices in R at (2/3, 2/3) and (1/3, 1/3), respectively. The
symmetry about (2, 2) suggests setting p'=p-2 and r'=r-1. Then 0=2
-2p'r' and thus 13<4/9 if and only if p'r'> 1/36. Note that the vertices for the
hyperbola pr'= 1/36 are at (p, r) =(1/3, 1/3) and (2/3, 2/3). By looking at
asymptotes, we see graphically that the region 13<4/9 does not overlap the
region in R where a<4/9 and yy<4 / 9 .
B-3 Sincef is uniformly continuous, for any e>O there is an integer n

such that lx-yl <1/n implies |f(x)-f(y)j <e.
I -1 m.+l/n n-1 'm+l/n
f(x)g(nx)dx
J on-0
= Ef m/n
f(x)g(nx)dx = Er
m-0 m/n
f(m/n)g(nx)dx
rn- m+l/n
r-I
m=0 m/n
(f (x) - f(m/n))g(nx)dx.
The first term equals -o(l/n)f (m/n)f lg(t)dt and becomes (fo'ff(x)dx) (fg(x)dx)
as n-ioo. Furthermore,
mr+l/n r+l/n
m
Jrn/n {f(x) -f(rn/n) g(nx)dx < rn f(x) - f(m/n) I g(nx) I dx
<
r +l/n
rn I|e g(Tx) I dx < e/n f
1
I g(t) I dt.
Thus the absolute value of the second term is less than or equal to ef'J g(t)I d
which becomes 0 as e -O.
B-4 Locker m, 1 <m -n, will be unlocked after the n operations are
performed if and only if m has an odd number of positive divisors. If in=pq
* .* r8, then the number of divisors of m is (a+l)(/3+1) . . . (7+l), which is
odd if and only if a, 13, . *, y are all even. This is equivalent to the condition
that m is a perfect square.
For part (b), the set of numbers of the form 2m2 are obtained by having Tk
change lockers whose numbers are multiples of 2k. The set m 2 +1 results from
T.k changing locker i if i-1 is a multiple of k, with the stipulation that locker
number 1 is changed only by T1.
B-5 Let A. be the sum of the first n terms in the binomial expansion of
(2 t1)-n.
An=
n-I
ieO
n~i-1
) 2-n-i 22+ £
i=l/
/-1
n+i-2\
(.) +
/,n+i-2Y\
. 2
2-
8 -2 (n+i-2\ /2n-3\ 2-n-,/Ij-

=2-n+ E ) 2-n-i+ ) 2-2n+1-2-n} + . 2n
i-o_ n-1 I ~
/2n-3\ /2n-2
= 2-n+lAn-l+ 2-2n+I - 2-n+ I An- 2-2n
2
=IAn-I+IA+2 n {2n-3\ (2n-3) _ (2n-3)} 'Ani+'A.
=~+2 A 2 +2
n-1
| ) n-1 n- 2 )t= Aff
A2n
s
Thus A.=An-1, but Al=2-'=2 and so An = 2 for all positive integers n.
Alternate solution: Consider a random walk starting at (0, 0), such that if
one is at (x, y) the probability of moving to (x + 1, y) is 2 and the probability
of moving to (x, y+1) is I. Let Sn be the square with vertices at (0, 0), (n, 0),
(n, n), (0, n). By symmetry, the probability R2 (n) of first touching Sn at (n, i),
0<i<n, equals the probability T 1(n) of first touching S. at (i, n), 0<i<n,
and hence En`O Ri(n) = , Ti(n)= z. Furthermore, Ri(n) = (f+t1-)(1)n+i.
B-6 Consider the function g whose values are defined by g(x, y) =f(x, y)
+2(x2 +y 2 ). On the circumference of the unit circle, g(x, y) - 1, and at the origin,
g(O, 0); 1. Hence, either g(x, y) =constant and f(x, y) =constant-2(x 2 +y 2 ),
or there is a minimum value for g(x, y) at some interior point. In the case
g(x, y) =constant, the result is immediate. Otherwise, let (xo, yo) be the coordi-
nates of a point where g(x, y) has a minimum. Then
dg ag
- =-= 0 at (xo, yo)
dx ay
and -(X 0 , yo) < 41 xo Of (x0 , y0 ) < 4 1 yo I-

I Ox ay
Thus the conclusion follows.
THE TWENTY-NINTH WILLIAM LOWELL PUTNAM
December 7, 1968
A-1 The standard approach, from elementary calculus, applies. By

division, rewrite the integrand as a polynomial plus a rational function with
numerator of degree less than 2. The solution follows easily.
A-2 The easy solution is obtained by selecting a rational number p
with O<p<e and solving the linear system
ar + bs = e + p
cr + ds =f + p.
The solution for r and s exist, since adobc, and are rational numbers which
satisfy the given inequalities.
A-3 The proof is by induction. For a singleton set {1 the list is 0,
{ 1}. Thus the result is true for singleton sets. Suppose the result is true for all
sets with n-l members. Let S= {1, 2, 3, * *, n } and T = {1, 2, 3, * - *, n. }.
Let To, T 1, * * *, Tt (t = 2n-1-1) be the list of subsets of T satisfying the require-
ments. Then the desired list of subsets of S are SO, Si, - * * , S. (s=28-1) where
Si=Ti, for O<i<t, and S1=TtY{n1, Se+i=Tt-'J{n}, * * *, S= {,n.
Comments: This problem is equivalent to finding a Hamiltonian circuit on an n-cube.
A-4 The n points can be represented by vectors v; (i= 1, * - *, n) with
Ivi|= 1. Expanding "the sum of the squares of the distances between them" in
the case n = 3 suggests the following general identities:
E I Vi - V12 = E ,i- Vj).(Vi -j)
15 i<j~gn Is ;<ji n
= (n 1) A vi vi-2 E vi vi
1jisg 1si<j;Sn
= n vEvi - vi vi + 2 E vs vi
1siiln 15~ ig n lii<i~sn
n2 - ( V ( E v).
T s tion I i if o
Thus the result follows and, in addition, equality exists if and only if E15isn Vi = °-
59
A-5 (t1) Let f(x) =ax2 +bx+c be an arbitrary quadratic polynomial. Then
f(0) =c,f(') =la+±b+c, andf(l) =a+b+c.f'(0)=b=4f(6)-3f(0)-f(l). Using
the given conditions, P'(0)I <41P(2)1 +31P(0)I +IP(1)f <8. Furthermore,
P(x) =8x 2 -8x+1 satisfies the given conditions and has IP'(0)I =8.
A-6 (0) The desired polynomials with ao = -1 are the negative of those with
ao = 1, so consider ao = 1. The sum of the squares of the zeros of xn +axn-l+**
+an is a2-2a 2 . The product of the squares of these zeros is a . If all the zeros are
real, we can apply the arithmetic-geometric mean inequality to obtain
al-2a2 2 1/n
n
with equality only if the zeros are numerically equal. In our case this inequality
becomes (1 ± 2)/n > 1 or n < 3. Note that n > 1 implies a2 =-I and n = 3 implies
all zeros are ± 1. Thus the list of polynomials is:
2
(X-), ±(X + 1), ±(X2 + X- 1), ±(X - X-l)
3 2 3 2
(X + X - X- 1), ± (X - X - X + 1).
B-1 Denote the four events x° =700, y =700, max(x°, y°) =70°,
min(x°, y°)=7 0 ° by A, B, C, D, respectively. Then AUB=CUD, and
ACn B = CrD.HenceP(A) +P(B) = P(A UJB) +P(A(nTB) = P(CUJD)
+P(CCnD) =P(C)+P(D) and P(min(x°, y°) = 700) =P(x° = 700) +P(y°= 700)
-P (max (x°, y°) = 70°).
B-2 Let g be any element of G. The set {ga-'| aCA } has the same
number of elements as A. If these two sets are disjoint, their union would contain
more elements than G. Thus there exist a,, a 2CA sUCch that al=gai-' and
g = ala2.
Alternate Solution: Let G have n elements, A have -melements, and consider
the multiplication table of G. AnI element g in G must appear exactly once in
each row and column of the mIultiplication table. It appears at most 2(n-rm)
times outside the table for A and n times in the table for G. Thus it appears at
least n-2 (n-m) = 2n -n times in the table for A, and we are given that 2m > n.
B-3 We need to make use of the following facts about fields and con-
strtictibility: (1) If Q is the field of rational numbers, the degree of Q extended
by cos(360°/k), where k is a positive integer, is k(k), where k is the Euler func-
tion. (2) If K, L, M are fields with KCLCM and [L:K] < -, [M:L] < - then
[AT:KI = [M:L]. [L:KI. (3) Given cos(360°/k), theen cos(360°/3k) is con-
structible if and only if
s 30) °Q(cos 3600)
i- a po3k :Q 2.
is a power of 2.
SOLUTIONS: THE TWENTY-NINTH COMPETITION 61
Consequently, [Q(cos 360'/3k):Q(cos 360'/k) I .qS(k) =k(3k). Now

[3+(3a-~1), if a.> I
0(3a) = 31'.2 =
{2,
3
if a = 1
and, by the multiplicative property of the Euler function,
0(3k) = {3+(k),
2+(k),
if 31 k
if 31k.
Therefore an angle of size 360'/k is trisectible if and only if 31k.
B-4 The graph of y =x - 1/x suggests splitting the integral into the form
f f(x - l/x)dx = lim

-~ 2I'.~
f(x-l/x)dx + lim
JaJ~~
-
b-.0 _f~x-
f(x -lx)dx
lxd
rl rd
+ lim f f(x - l/x)dx + lim ff(x - l/x)dx
and making the change of variables x = § [y - Vy2 + 4], in the first two integrals,
and the change of variables x = [y+
y v/y2+4], in the second two integrals. Since
both of these functions of y have continuous first derivatives on the intervals
involved, the change of variables is valid. After the changes of variable, we have
four improper integrals. The convergence of each of these integrals is established
by a corollary of the Dirichlet Test (Advanced Calculus, R. C. Buck, McGraw-
Hill, p. 143). Thus it is permissable to rewrite the first and third of these im-
proper integrals as a single integral by adding the integrands, since they have
the same limits from - o to 0. The result is fJ' °,f(y)dy. Likewise, the other two
integrals combine to give fJ;f(y)dy. In this combining, there is a canceling of a
term involving y/v/y2 +4 because it appears once with a plus sign and once
with a minus sign. We have shown both the convergence of f'. 0f(x-l/x)dx
and the desired equality.
B-5 If a = 0 then d =1, and if a = I then d = 0. In either case bc =O and
b or c is 0, while the other is arbitrary. There are 2p -1 distinct solutions to
bc = 0 and thus the case a = 0 or a = 1 accounts for a total of 4p -2 solutions.
If a #0 or 1, then d is uniquely determined and bc ad 0 (mod p) implies that
for each b #0, there is a unique c, since the integers mod p form a field. Hence for
each a in this case, there are p - 1 solutions. The total number of solutions is
4p-2+(pp-2)(p-1) =p 2 +p.
B-6 Let {Kn} be any sequence of compact sets of rational numbers.
For each in, there is a rational rfEEKf, with 0<rr,<1/n. Otherwise, it would be
that K,, contained all rationals in [0, l/n], and hence some irrationals (since K.
is closed). Let S= {0, r,, r2, * . Then S is compact and not included in
any K,,.
THE THIRTIETH WILLIAM LOWELL PUTNAM
December 6, 1969
A-i The continuity of f(x, y) implies that the range is connected (i.e.,
if a, b are in the range and a <c <b then c is in the range). If the range is bounded
above and below, then the polynomial f(x, kx) is a constant for each value of
k and thus f(x, y) is the constant f(O, 0). Thus the only possibilities are: (i) a
single point; (ii) a semi-infinite interval with end-point; (iii) a semi-infinite
interval without end-point; and (iv) all real numbers.
Examples are easily given for (i), (ii) and (iv). An example for (iii) is harder
to find. One way is to have each cross-section of the surface (for fixed y) be a
parabola with a minimum which decreases asymptotically toward some con-
stant as y approaches ± oo. A suitable example is (xy-1)2 +x 2 .
A-2 Subtract the first column from every other column. Then add the
first row to every other row. The last row now has all zeros except for (n-1)
in the first column. D. is (-i)n- 1 (n-1) times the minor formed by deleting
the first column and last row from the transformed determinant. This minor has
only zeros below the main diagonal and thus is equal to the product of its
diagonal elements. Hence the minor has value 2n-2 and Dn = (-1)n-(n-1)2 2.
Alternate Solution: From the bottom row of Dn+i, subtract 1/(n-1) times
the first row and n/(n-1) times the nth row. This shows that Dn+1
=-[2n/(n-1)iDn, for n>1. The result follows easily by iteration and the
observation that D2 = -1.
A-3 Let t be the number of triangles. The sum of all the angles is 7rt
(since it is 7r for each triangle) and it is also 27rm+(n-2)7r.
Alternate Solution: Let t be the number of triangles. In Euler's formula

V-E+F=2, F=t+i, and V=n+m. Since every edge is on two faces, 2E
= 3t +n. Substitution leads directly to the answer t = 2m +n-2.
Comment: It should have been stated in the problem that the interior of the polygon is tri-
angulated. If any of the additional line segments are outside of the polygon, the answer is different.
A-4 A reasonable way to get a series (other than using Riemann sums,
which apparently doesn't work) is to write the integrand as a power of e and
62
SOLUTIONS: THE THIRTIETH COMPETITION 63
use the series expansion for e. Then uniform convergence can be applied to
interchange integration and summation, and show that
f'xxdx= x(og x)"dx.
Let F(m, k) =f 1 xm(log x)kdx. Integration by parts shows, if applied to a typical

term in the summation, why we are interested in F(m, k) and also shows that
F(m, k)=-k/(m+1)F(m, k-1) for m>ŽO and k-1. As a result, F(m, m)
= (- 1)-m!(m+1)--F(m, 0) = (- )mm(m+l)-^-l. To get the given formula in
the problem, replace m+l by n and adjust the limits on the summation ac-
cordingly.
A-5 Subtracting the two equations eliminates u(t) and provides the
simpler equation d(x-y)ldt =2(x-y), which has thesolution x-y = (xo-yo)e2 t .
If xowyo, the right hand side is never zero and so x=y=O can never occur.
For the second part, xo =yo. In this case, x(t) =y(t) and so every solution
is a parametrization of the line x =y. We can attempt to get a solution of the
form x=xo-at, y=yo-at. This will be a solution if u(t)=2(xo-at)-a. By
takinga =xo/to, x =y =0 att =to.
A-6 Let y=lim,,-.y. and set x=y/3. We will show that x=lim.-.
0 xn.
For any e> 0 there is an N such that for all n > N, I y,,- <e/2.
e./2 > I y. - 9|= I x.-1 + 2x. - 3x I = I 2(x. - x)+ (x.-I - x)|

2t 21| xn - x | - I Xn-I - XI -
This may be rewritten as IxX.-xj <E/4±+2x.--x, which can be iterated to
give
|x+-x|< e./4 2-i) + 2- (m+l) |X -X <- /-24 2-(ml
By taking m large enough, 2-(m+l)l X, - I <e/2. Thus for all sufficiently large
k, |Xk-X1 <e.
B-1 The condition 24|n+1 is equivalent to n= -1 (mod 3) and
n- -1 (mod 8). Letd be a divisor of n, then d=1 or 2 (mod 3) and dl1, 3, 5 or
7 (mod 8). Since d (n/d) = n -1 (mod 3) or (mod 8), the only possibilities are:
d 1, n/d 2 (mod 3) or vice versa
d 1, n/d 7 (mod 8) or vice versa
d = 3, n/d 5 (mod 8) or vice versa.
In every case, d+n/d=-O (mod 3) and (mod 8). Thus d+n/d is a multiple of
24. Note that d pn/d and thus no divisor is used twice in the pairing, so the sum
of all the divisors is a multiple of 24.
B-2 The number of elements in a subgroup is a divisor of the order of

the group. Thus a proper subgroup can have no more than half of all the ele-
ments. Two subgroups always have the identity in common and hence their
union cannot be the entire group.
An example for the second part of the problem is the Klein group, which has
an identity and three elements x, y, z of order two. The product of any two
distinct elements from {xI y, zI is the third. This group is the union of three
proper subgroups.
Alternate Solution: Let G = HJK, with H and K proper subgroups. There
exists kCK with kXH. None of the elements in kH are in H and so kHCK.
Hence HCk-1 K =K and K =HUK = G, a contradiction.
B-3 The first relation implies that
(/- 1) (1n-3).* 3 1
n= 3- if n is even,
(n-2)(n-4) . .. 2 T
(n-1)( -3) ... 2
T. T1 if it is odd.
(n - 2)(n -4)... 1
If n is odd,
T. 2 2 4 4 6 (n- 1)(n-)
= (T 2.--.-. ...--
T+l . 1 3 3 5 5 (n-2) n
The Wallis product is 7r/2 =2 1j 4' - . After an even number of factors the
partial product is less than 7r/2 and after an odd number of factors the partial
product is greater than 7r/2. Thus for the case when n is odd, Tn/T,+±<j7rTj.
A similar calculation shows that, when n is even, Tn/Tn+i<2/1irT7. Since the
limit of Tn/T,+1 = 1, 1 is less than or equal to both 7rTr, and its reciprocal. This
implies that 7rT' = 2.
B-4 Place the curve so that its endpoints lie on the x-axis. Then take
the smallest rectangle with sides parallel to the axes which covers the curve.
Let its horizontal and vertical dimensions be a and b respectively. Let Po and P6
be the endpoints of the curve, and let P1 , P2 , P3 , and P4 be the points on the
curve, in the order named, which lie one on each of the four sides of the rect-
angle. Draw the broken line PoPIP2 P3 P 4P6 . This line has length at most one.
The horizontal components of the segments of this broken line add up at least
to a, since one of the vertices of the broken line lies on the left end of the rec-
tangle and one on the right end. The vertical segments add to at least 2b since
we start and finish on the x-axis and go to both the top and bottom sides. This
implies that the total length of the broken line is at least (a2 +4b2 ) I/2.
We now have that a and b both lie between 0 and 1 and that a 2 +4b 2 •1.
Under these conditions the product ab is a maximum for a = 2-%,12, b = -1-/2 and
so the maximum of ab is 4. Thus the area of the rectangle we have constructed
is at most 4-.
B-5 The following proof shows it is not necessary to stipulate that
SOLUTIONS: THE THIRTIETH COMPETITION 65
the a. be integers. Suppose for some e> 0 there are xj- with k (x,) /xi 2 e. Note
that if 1 <n •k(xi), then (because the a. increase) ant Sat(.,) f xi and 1/ant 1/xj.
Now for any positive integer N,
0 k(zi) k(xi) - N
E l/a. 2 sup Il/a, 2 sup ' sup(e-N/x,) = e.
ft-N J n-A J Xi i
But this contradicts the convergence of D'_j/a., which implies
lim F 1/aft = 0.
N-s n-N
B-6 Observe that ABAB =9AB. AB isof rank two so A is onto and B is
one-to-one. Hence there exist matrices A' and B' such that A'A =I=BB',
where I is the 2X2 identity matrix. ThenA'(ABAB)B'=BA =91.
Alternate Solution: (AB)2=9AB. The rank of BA is greater than or equal
to the rank of A(BA)B, which is 2. Thus BA is nonsingular. But (BA)8
=B(AB) 2A B(9AB)A =9(BA)2 and the result follows since BA has an inverse.
THE THIRTY-FIRST WILLIAM LOWELL PUTNAM
December 5, 1970
A-1 Note that ea. cos bx is the real part of e(a+ib)x. Thus the power
series is
e- cos bx = E Re{(a + ib)i-

n-O n!
In this form, it is easily seen that if xn has a zero coefficient, then xkn has a zero
coefficient for every odd value of k.
A-2 Let (x, y) = (r cos 0, r sin 0), r>0, be a point of the locus. Then
A cos 2 0 + B sin 0 cos 0 + C sin2 0
( D cos's + E cos2' sin 0 + F cos 0 sin2 0 + G sin' l
The denominator of (1) is less than or equal to I Di + I El + I FI + I GI, whereas
the numerator has a positive minimum
N IA + CI - V(A - C)2 + Bj2
2
2
since B <4AC. Therefore
N
I DI + I El + I F| + |G|
and there ar'e no points of the locus within O<r<8.
Alternate Solution: Set H(x, y) equal to the polynomial on the left hand side
of the given equation. The standard theory for maxima or minima of functions
of two variables can be used together with the condition B 2 < 4A C to show that
H(x, y) has a local maximum or a local minimum at (0, 0).
A-3 If x is an integer then x 2 =O, 1, 4, 6 or 9 (mod 10). The case x 2 =0
(mod 10) is eliminated by the statement of the problem. If x2 =11, 55 or 99
(mod 100), then x 2 =3 (mod 4) which is impossible. Similarly, x2 -66 (mod 100)
implies x 2 e2 (mod 4) which is also impossible. Therefore x2 =44 (mod 100).
If x 2-- 4444 (mod 10,000), then x 2 = 12 (mod 16), but a simple check shows that
66
SOLUTIONS: THE THIRTY-FIRST COMPETITION 67
this is impossible. Finally note that (38)2 = 1444.

A-4 For e>0, let N be sufficiently large so that IXn.X--2I <e for all
n - N. Note that for any n > N,
Xn X- = (Xn - Xn-2) - (X.-l.- Xn-S) + (Xn- 2 - Xn- 3 ) '
± (XN+l - XN-1) + (XN - XN-1)-

1
Thus X.-XR-11 <(n-N)e+jxN-xN..I and limn-* (x,,-x- 1)/n=0.
A-5 Since parallel cross sections of the ellipsoid are always similar
ellipses, any circular cross section can be increased in size by taking a parallel
cutting plane passing through the center. Every plane through (0, 0, 0) which
makes a circular cross section must intersect the y-z plane. But this means
that a diameter of the circular cross section must be a diameter of the ellipse
x=0, y 2 /b 2+z 2 /c 2 = 1. Hence the radius of the circle is at most b. Similar reason-
ing with the x-y plane shows that the radius of the circle is at least b, so that
any circular cross section formed by a plane through (0, 0, 0) must have radius
b, and this will be the required maximum radius. To show that circular cross
sections of radius b actually exist, consider all planes through the y-axis. It can
be verified that the two planes given by a2 (b 2 -c 2 )z2 =c2 (a2 -b 2 )x 2 give circular
cross sections of radius b.
A-6 Let x be selected from [0, Li], y from [O, L2], z from [0, Ls], and
assume La - L2 - L1 . Let X = min (x, y, z).
LIL2LsE[X = i Lff X dzdy dx
= fLI L {f zdz+ Lmdz} dydx, where = min(x,y),
= JL f L2 - 1p2} dy dx
0 0
= fL, { f (L~y - 4y2)dy + f (Lax - JxI)dy dx
12 1 14
= *-- =- LL 2 Ls - -L 1 (L2+L 3) +-L.
2 6 12
Alternate Solution: For 0 a L,
P(X ! a) =P(x 5 a) + P(y ! a) + P(z ! a) - P(x : a)P(y a)
-P(x : a)P(z S a) - P(y 6 a)P(z 5 a)
+P(x - a)P(y S a)P(z - a)
a a a / a2 a+ a2 as
L1 L2 Ls LIL2 L2Ls LL 1 / L1L2 L,

The answer follows easily from the formula

CL dP(X - a)
E[X] a=Ja da.
Jo da
B-1 Let
1 2n
an - (n 2 + i2) ln
n 4-
Then
1 2n i2 \
log an = log 1 + -2,
and
2
limit log an J log(1 + x 2 )dx = 2 log 5-4 + 2 arctan 2.
2
B-2 Let P(t) =a+±bt +ct+d. The equation
2T f P(t)dt= 2 {P(tU) + P(t2 )}
is satisfied for all values of a, b, c, and d if and only if t2 =-t = ± T/l,/3. If

T=3 hrs, T/V 3% 1 hr, 43.92 min. Therefore, in the case considered, the critical
times are 1 hour 44 minutes each side of noon.
B-3 Let yn-*y with (Xn, y.)ES for all n. The Bolzano-Weierstrass
Theorem implies that a subsequence xk() -- x. Then 3k() -y and since S is
closed, (x, y) ES. Thus y is in the projection of S on the y-axis.
B-4 Converting units to feet and seconds, we have 0 <v(t) <132 for
all te [0, 60]. Suppose | v'(t)J <6.6 for all tE [0, 60]. Then v(t) =-f v'<6.6t, and
v(t) =f, -v'<6.6(60-t) for all tE [0, 60]. Thus
5280 = J 60
v(t)d1 <
s
J o 60
min{6.6t, 6.6(60 - 1), 132}dt.
This last integral is the area under a trapezoid and equals the value 5280, which
is a contradiction.
B-5 Clearly un is continuous. So, if F is continuous, then u. o F is
the composition of continuous functions and hence is continuous. Conversely,
suppose un o F is continuous for all n. To prove F is continuous it is enough to
show F'-[(a, b) ] is open for every bounded interval (a, b). Let n> max(I al , b).
Then unj' [(a, b) ] = (a, b) so
F 1 [(a, b)] = F1 [u2n{ (a, b) }] = (u. o F)'[(a, b)],
which is an open set by the continuity of u, o F.
SOLUTIONS: THE THIRTY-FIRST COMPETITION 69
B-6 Since the quadrilateral is circumscribable, a+c=b+d.Let k be the

length of a diagonal and angles a and / selected so that k 2 =a2 +b 2 -2ab cos a
=c 2 +d 2 -2cd cos,. If we subtract (a-b)2 = (c-d) 2 , we obtain
(1) 2ab(1 -cos a) = 2cd(1 - cosB).
From A = lab sin a+ cd sin / = V/abcd,
4A2 = 4abcd = a 2b2(1 - cos 2 a) + c2d2(1 - cos 2 , ) + 2abcd sin a sin (.

Using (1) twice on the right hand side,
4abcd = ab(l + cosa)cd(l- cos,3) + cd(l + cosB)ab(1 - cosa)
+ 2abcd sin a sin ,B.
On simplifying, 4=2-2 cos(a+#3), which implies that a+/3=ir and so the
quadrilateral is cyclic.
THE THIRTY-SECOND WILLIAM LOWELL PUTNAM
December 4, 1971
A-I The set of all lattice points can be divided into eight classes according
to the parities of the coordinates, namely, (odd, odd, odd), (odd, odd, even), etc.
With nine lattice points some two, say P and Q, belong to the same class. The mid-
point of the segment PQ is a lattice point.
A-2 P(O) = 0, P(1) = [p(o)]2 + 1 = 1, P(2) = [p(1)] 2 + 1 = 2, P(5)
= [P(2)] 2 + 1 = 5, p(52 + 1) = [P(5)] 2 + 1 = 26, etc. Thus the polynomial P(x)
agrees with x for more values than the degree of P(x), so P(x) - x.
A-3 For a triangle with sides a, b, c, area = A and circumradius = R we have
abc = 4 RA. But if the vertices are lattice points the determinant formula (or Pick's
Theorem or direct calculation) for the area shows that 2A is an integer. H.-nce
2A - 1, so that abc > 2R. To obtain the formula abc = 4RA note that if a is the
angle opposite side a, then side a subtends an angle 2ac at the center and a = 2R sin xt,
A=Ibc sin a.
2 2
A-4 In the expansion of (x + y)n(x - (2 -s)xy + y ) the coefficient of
xk+ yn+-k is
(k -1-2
n (2~
- )(n ) (k + (k + 1
( n) -k+ + T--k (2 - s)
Now for fixed n consider the expression
= n(k)k+1 +n k (2-
If k is taken to be a continuous positive variable

0'(k) - (n + I) {(k + 1)
2
-(n-k + 1)2 }
2
(n-k + 1) (k + 1)2
2
Hence 0Y(k) = 0 at k = n/2 and it follows easily that q5(k) is minimum at k =n/ .
70
SOLUTIONS: THE THIRTY-SECOND COMPETITION 71
We needn't consider end point minima since it easily follows that for n > 2 the
polynomial has its first two and last two coefficients positive. We may also note that
if the two mid-terms in the expansion are non-positive for a given odd value of n then
for the next larger value of n the mid-term remains non-positive. Hence if the mid-
coefficients become positive, the first value of n for which this occurs is odd. Now if
n is odd and k = J(n + 1) then 0(k) = -+ -I + c, and +(k) > 0 for n >--3.
If e = .002, n > 1997 and n is odd. Hence the minimum n for which all terms are
positive is 1999.
A-5 The attainable scores are those non-negative integers expressible in the
form xa + yb with x and y non-negative integers. If a and b are not relatively prime
there are infinitely many non-attainable scores. Hence (a, b) = 1. It will be shown
that the number of non-attainable scores is I (a - 1) (b -1).
If m is an attainable score, the line ax + by = m passes through at least one
lattice point in the closed first quadrant. Because a and b are relatively prime, the
lattice points on a line ax + by = m are at a horizontal distance of b. The first-
quadrant segment of ax + by = m has a horizontal projection of mr/a and thus every
score m - ab is attainable. Every non-attainable score must satisfy 0 - m < ab.
if 0 < m < ab, the lirst-quadrant segment of the line ax + by - m has a horizon-
tal projection less than b, and so contains at most one lattice point. Thus there is a
one-to-one correspondence between lattice points (x,y) with 0 < ax + by < ab in the
first quadrant and attainable scores with 0 • m < ab. The closed rectangle 0 < x - b,
0 - y - a contains (a + 1)(b + 1) lattice points, so the number of lattice points in
the first quadrant with 0 < ax + by < ab is I (a + 1) (b + 1) - 1. This is the number
of attainable scores with 0 < m < ab. Hence the number of non-attainable scores in
this range (which is all of them) is ab - J(a + 1)(b + 1) + 1 = i(a - 1)(b - 1).
In our given example 70 = (a - 1)(b - 1) = 1(70) = 2(35)=5(14) = 7(10). The
conditions a > b, (a,b) = 1 yield two possibilities a = 71, b = 2 and a = 11, b = 8.
Since 58 = 71(0) + 2(29), the first of these alternatives is eliminated. The line lix + 8y
= 58 passes through (6, - 1) and (-2,10) and thus does not pass through a lattice
point in the first quadrant. The unique solution is a = 11, b = 8.
A-6 The case n = 2 shows that c is non-negative. If the ordinary mean value
theorem is applied to xc on the interval [u, u + 1] there is a 4 with u < 4 < u + 1 such
that c tC = (u + 1)' - uc. For any positive integer u the right hand side is a positive
integer. Now, in the case 0 < c < 1, u could be taken large enough so uc- I < 1/c and
so c Cc-1 < 1. Thus the mean value theorem for the first derivative eliminates all c
with 0 < c < 1.
There is an extension of the mean value theorem which states that iff (x) is k-times
differentiable in [a, b] then there is a X,a < 4 < b, such that hkf(k)(4) = Akf (a), where
h = -b- and Ak is the k-th difference for intervals spaced h apart. Take k as the
unique integer such that k - 1 < c < k and apply this extension of the mean value
theorem on the interval [u,u + k]. There is a 4 with u < 4 < u + k such that
c(c - 1)(c -2) ... (c - k + 1) c-Ck = Akf(u)
The right hand side is an integer, and by taking u sufficiently large ¢-' becomes
sufficiently small so that the left hand side, though non-negative, is less than I. Hence
c(c - 1)(c - 2) ... (c - k + 1) = Oand so c = k-1.
B-1 Using the given laws we have
x oy = (x oy) o (x oy)= [(x o y) o x] oy =[(y a x) x] oy
ox) oy] oy (x cy) oy = (yoy) ox =yox.
-[(x
From this commutative law we obtain

(x oy)o z = (y a z) ox = x o(yo z).
B-2 In the given functional equation
(1) F(x) + F= 1+ X
we substitute X for x, obtaining

X
(2) (I x) F(I) 2 x--
Also in (1), we substitute for x and obtain
(3) F 1 ) + F(x) = X 2
Adding (1) and (3) and subtracting (2) gives
2F(x) = 1 + x +X 2 -2-1 =X X,
(4) X -1I X(x -l
X3 X2 _ I
(x) =2x(x-1)
That F(x), defined in (4), does satisfy the given functional equation is easily verified.
Therefore (4) is the only solution of the problem.
B-3 At tim- t, car 1 has co00fl ted [t] laps and car 2 has co.n?lated
[t - T] laps. The problem is to find values of t - T for which [t] = 2[t - T].
Let T = k + 6, where 0 < 6 < 1, k an integer. Consider any integral interval
[m, m + 1] and let m < t < m + 1. Then t = m + e, where 0 < a < 1. Then the
equation to be solved becomes
[t]= m = 2[t - T] = 2[m + - (k + 6)] = 2[m - k+ -6].
Thus m = 2(m-k), if £ > 6 and m = 2(m -k-1), if £ < 6. If I > £ > A, then
m = 2k and the equation is satisfied during [2k + 6, 2k + 1], which has length
1 -6.
If 0 <£ < 6, then m = 2k + 2 and the equation is satisfied during [2k + 2,
2k + 2 + 6] which has length 6. Therefore the total length is I - 6 + 6 = 1.
Comment: The problem should have been more explicit by stating "after the
start of the second car" instead of "during the motion ". The solution is given for
this interpretation, whereas, if t < T, [t - T] is negative but the second car would
have completed zero laps.
B-4 We take the radius of the sphere as unity and denote the constant sum
PA + PB by 2a. To avoid trivial and degenerate cases we assume that 0 < AB < 7i
and that AB < 2a < 27r - AB.
The case 2a > it can be reduced to the case 2a < 7. For, if A' and B' are the
points diametrically opposite to A and B then PA + PB = 2a if and only if
PA' + PB' = 2t - 2a; that is, the spherical ellipses PA + PB = 2a and PA'
+ P B' = 27t - 2a are identical. Since min (2a, 27r - 2a) - 7r, we may assume without
loss of generality that 2a - 7r.
Let A and B lie on the equator. There are two points V,and V2 (the "vertices") on
the equator which lie on the spherical ellipse. Obviously, V1 V2 = 2a. The "center"
of the spherical ellipse (common midpoint of the arcs AB and V1 V2 ) will be denoted
by C.
l V1 V2
FIG. I FIG. 2
We first treat the case 2a < 7r and show that in this case the spherical ellipse
cannot be a circle. Assume it were a circle; call it r (see Figure 1). r would have to
be symmetric with respect to the equatorial plane, thus lie in a plane perpendicular
to the equatorial plane. r would also have to pass through the vertices. Therefore its
spherical diameter would be V1 V2 = 2a and its spherical radius would be equal
to a. The spherical center of r would be C, the center of the ellipse. Let M be one of
the two points on F which lie half-way between the two vertices. Then, sinceMis
supposed to be a point on the spherical ellipse, 2a = MA + MB > 2MC = 2a (note
that MAC is a right spherical triangle with the right angle at C and with side
MC = a < 1ir). Contradiction shows that the only possible spherical ellipses which
are circles must occur when 2a = 7r.
In case 2a = 7r, V1 and V2 are diametrically opposite points on the equator. We
shall show that the great circle r through the vertices and perpendicular to the
equatorial plane is identical with the spherical ellipse PA + PB = it. To see this, let
B* be the reflection of B about the plane of I. B* is on the equator diametrically
opposite to A (see Fig. 2). Let P be an arbitrary point on the sphere, and draw the
great circle through A, P and B*. Then PA + P = i*t. Hence, PA + PB= it if
and only if PBt = PBJ*, that is, if and only if P is on F. This shows that F is the
spherical ellipse PA + PB = it, as stated above.
Thus the only circles on the sphere that are spherical ellipses are the great circles.
For any given great circle F the foci can be any two points A and B which lie on the
same great circle perpendicular to I, on the same side of I and at equal distances
from f. The equation of any such spherical ellipse is PA + PB = 7r.
B-5 We put z = x + iy. Then both differential equations can be combined
into one, namely
(1) z" -iz' + 6z =O.
This is a standard linear equation of the second order with constant coefficients and
has the general solution
3it
z(t) = cle~i + C2 e--2it
The initial conditions imply z'(O)= 0 or 3ic, - 2ic2 = 0. We may set cl = 2A and
c2 = 3A, where A is any complex number. The general solution of the given system is
(2) z(t) = 2Ae3 't + 3Ae 2 ".
If A = R e"a, then a rotation of axes through the angle ae produces

(3) Z(t) = 2Re3i' + 3Re-2 ,'
or in rectangular form
X(t) = 2Rcos(3t) + 3Rcos(2t)
Y(t) = 2Rsin(3t) - 3Rsin(2t).
This is the standard form for a hypocycloid when the radius of the rolling circle is 3R
and the fixed circle is of radius 5R. On time reversal it becomes the standard equations
of a hypocycloid with radius of the rolling circle of 2R and the radius of the fixed
circle of SR.
B-6 Set
S X) 3 (n)
( nI
= n
Note that 6(2m + 1) = 2m + 1, 6(2m) = 6(m) and that S(2x + 1) = S(2x) + 1.
Dividing the summation for S(2x) into even and odd values of the index produces
the following relation:
S(2x) = 36(2m) + I - 1)= jS(x) + x.
m Im m=1 2m - 1
If we denote S(x) - 2- by F(x), the above relations translate into
F(2x) = jF(x), and F(2x + 1) = F(2x) + i.

Now induction can be used to show that 0 < F(x) < i, for all positive integers x.
This result is sharper than that requested.
THE THIRTY-THIRD WILLIAM LOWELL PUTNAM
December 2, 1972
A-1 For a given n and r, in order for the first three binomial coefficients
to be in arithmetic progression, we must have
(1)2 n nn
()(r + 1) (r ) (r + 2)
or equivalently
(2 2 = -r + +2
The condition that the last three given binomial coefficients are in arithmetic
progression is found from (1) by replacing r by r + 1. Consequently both r and r + 1
must satisfy equation (2) if all four terms are in arithmetic progression.
Note that the two terms in equation (2) are interchanged it r is replaced by
n - r - 2. Thus the quadratic equation (2) has roots
r, r + 1; n - r - 3, n - r - 2.
Since (2) can have only two roots, r = n - r - 3 and n = 2r + 3. The four binomial
coefficients must be
(2r + 3) (2r + 3) (2r +3 (2r + 3)
( r J. ( r + 1), ( r + 2), ( r + 3)
which are the four middle terms. They cannot be in arithmetic prog ession since
binomial coefficients increase to the middle term(s) and then decrease.
A-2 Label the given laws (1) and (2), respectively.
1. We first show that
(3) (x*y,*x=y.
This follows from (x * y) * x = (x * y) [(x * y) * y] = y. (First apply (2) with x and
y interchanged; then apply (1) with x replaced by x*y.)
We now obtain
76
SOLUTIONS: THE THIRTY-THIRD COMPETITION 77
(4) y*x = [(x*y)*x]*x = x*y.

(First apply (3); then apply (2) wth y replaced by x*y.) This proves that * is
commutative.
11. Let S be the set of all integers. Define x *y =-x - y. Then
(5) x*(y*z) = -x + y + z; (x*y)*z =x + y - z.
It follows from (5) that, in the first place, (1) and (2) hold and, secondly, * fails to be
associative: simply choose x # z in (5).
Alternate So'ution, Part I (suggested by Martin Davis):
Wrnte the equation x * y = z as P(x, y, z). Then law (l) may be written "If x * y = z
then x*z = Y" or
(6) P(x, y, z) implies P(x, y, z).
Similarly, the law (2) may be written
(7) P(y, x, z) implies P(z, x, y).
These two implications, (6) and (7), show that the permutations (23) and (13) on
the location of the variables in P(x, y, z) are permitted. Since (13), (23) generate the
symmetric group S3, we find (12) is also permitted.
Thus, P(x, y, z) implies P(y, x, z, or x * y = z implies y * x = z, which means
x*y = y*x.
A-3 A function is "supercontinuous" if and only if it is affine,f(x) = Ax + B.

The sufficiency is trivial (and was worth 1 point in the grading). For the necessity:
First we note that it is not assumed that f(C-limit) = C-limit (f) (otherwise the
solution could be materially simplified). The essential steps are to show, that if f is
supercontinuous, then (l)f is continuous, and (2)f((a + b)/2) = (f(a +f(b))/2 for
all a, b. These two statements imply that f is affine. The proofs of (1) and (2) are
similar; we give (2) (which is the harder). Set c = (a + b) /2, and supposef(c) ¢ (f(a)
+ f(b))/2. Imagine any sequence of integers NAwhich "grows very rapidly"; say let
N,+1 exceed 2NI'. Then construct a sequence of points {x,} as follows: Break the
sequence into blocks, alternating between
and {x,} = a, b, a, b, a, b, ..
{xn} = C, C,C, C,CC,, *
the ab pattern holding for N 2,- 1 < n < N2 j, and the c pattern holding for
N 2 j < n < N 2i+ 1. Then {x,} has the C limit c, but the averages of {f(xj)} oscillate
(because the lengths of the blocks N. < n < Ni + I increase very fast, and f(c) ¢ the
average off(a) and f(b)). Thus the C-limit of {/(x,)} does not exist, a contradiction.
A-4 Let the square of sidelength 2R have the vertices (± R 12,0) and
(0, + R\ 2). The ellipse
(I) a22 + b
2
Pb = I
with 0 < b - a < R1/2 has the line x + y = RV/2 as a tangent if and only if the
quadratic equation x2 la2 + (RV2 - x) 2 lb 2 =1 has a double root. It can be verified
that its discriminant vanishes if and only if a2 + b2 = 2R2 . As a varies from R to
R Vj2 and b varies from R to 0, the curve (1) varies from the circle of radius R through
all the non-circular ellipses inscribed in the square to the degenerate "flat" ellipse
lying on the x-axis.
Let 4L denote the length of the ellipse x = a cos t, y = b sin t, 0 < t - 27r. Then
L= f [a 2 sin2 t + b2 cos 2t]+dt = f [la2(1 - cos 2t) + lb2 (1 + cos 2t)]Idt
= f [R2 - C2 cos 2t]+dt,
where c2 = a2 - b2 . The last integral we split into one from 0 to 7r/4 and one from
xz/4 to 7r /2, and in the latter we substitute t = 7T/2 - i', obtaining
(2) L = |{[R2 - ic2cos2t]i + [R2 + -c 2 cos2t]1 }dt.
Note that cos 2t > 0 for 0 • t < 7r/4.

Now the function f(u) = (p - u)+ + (p + u)+ decreases in the interval 0 < u < p,
because 2J'(u) = - (p - u)- + (p + u)-+ < 0 for 0 < u < p. Thus the integral in
(2) as a function of c has its largest value when c = 0, that is, for the inscribed circle.
To show that an ellipse inscribed in the square must have its axes along the
diagonals of the square, we choose the square as having sides u ± R and v-+ R
and the ellipse as having the equation
Au 2 + Buy + Cv 2 + Du + Ev + F 0,
where
(1) 4AC-B 2 > 0.
Taking the "highest,'' "lowest," "rightest," and "leftest" points on the ellipse, we
see that all four sides of the square must be tangents to the ellipse.
The line u = R is a tangent if and only if the equation Cv 2 + (BR + E)v
+ (Ar 2 + Dr + F) = 0 has a double root or
(2) (BR + E)2 -4C(AR2 + DR + F) = 0.
The corresponding conditions for u = - R, v = R and v = - R are

(3) (-BR+E)2 -4C(AR 2 -DR+F)=0,
(4) (BR + D) 2 - 4A(CR 2 + ER + F) = 0,
2
(5) (- BR + D)2 -4A(CR -ER + F) =O,
respectively. Subtract (2) from (3) and divide by 4R; this gives
(6) 2CD-BE =O.

Similarly, from (4) and (5),
(7) -BD + 2AE =O.

By (6), (7) and (1), D = E = 0. Therefore (2) and (4) become
B2 R2 - 4ACR 2
- 4CF = 0, B2 R 2 - 4ACR 2 - 4AF = 0,
respectively. Since F # 0, we have A = C; this means that the ellipse has its axes
along the lines u + v = 0.
A-5 Assume that it divides 2" - I for some n > 1. Since 2' - I is odd, n is
odd. Let p be the smallest prime factor of n. By Euler's Theorem, 20(P) 1I (mod p),
because p is odd. If A is the smallest positive integer such that 2A - 1 (mod p) then A
divides +(p) = p - 1. Consequently A has a smaller prime divisor than p. But 2" - 1
(mod p) and so . also divides n. This means that n has a smaller prime divisor than p
Contradiction.
A-6 The conditions imply fC(x - )"f(x)dx = 1. Suppose lf(x) I < 2'(n + 1)
except for a set of measure 0.
Then 1 = Jf,(x - D-)'(x)dx < 2"(n + 1) PI I x -1 dx = 1, a contradiction.
B-I For the proposed solution the problem could have been stated in the
more general form: The series expansion about any point for exp(P(x)), if P(x) is a
cubic polynomial, will not have three consecutive zero coefficients.
If J(x) = exp(P(x)), where P(x) is a cubic polynomial, then f'=f. P' and
f" =f'P' +f P". In general for k > 2,
()f'k + 1) =f'k). P, + ( )k
P P, + ( 2 )k 2 P ,.
1
It follows from (1): if, at some (real or complex) point xo,f(k -2) (x 0 ) =f(k- )(xo)
= f(k) (x,) = 0, then also f(k+I'(x 0 ) = 0. By the same argument, f (u(xo) = 0 for
= k + 2, k + 3, so that f(x) would reduce to a polynomial. This is evidently
impossible.
Alternate Solution: In the given form of the problem it can be shown that no
coefficient of xk is zero. The product x'(I x)2" has a non-zero coefficient for xk if
0 - k - n < 2n or, equivalently, k/3 • n < k. This coefficient is the integer
( 1 )k (2n)
which we denote by a(n, k). The coefficient of xk in the given series is

Ck= k a(n, k)
n=[k/3]+ 1 n
Multiplying through this summation by (k - 1)! will convert each term, except the
last term, to an integer. The last term becomes 1 /k. Since (k - 1)! times Ck is not an
integer for k > 1 and C, = CO = 1, there are no zero coefficients in the expansion of
the given series in powers of x.
B-2 We take vo as positive (see Comment) and consider the graph of v as a

function of t (see Figure 1). From the given data we know that the curve starts at the
origin and is concave downward since the acceleration a = dv/dt does not increase.
FIG. I
Let to be the time of the traverse. Then v(to) = vo. The distance so is represented by the
area bounded by the curve v = v(t), the t-axis, and the line t = to. The area of the
right triangle with vertices at (0, 0), (to, 0) and (to, vo) has area less than or equal to s.
Thus i vo to < so or
to< 2so
V0
Equality is possible and gives the maximum value of to (for given so and vz,) when the
graph of v(t) is the straight line v(t) = (vo /to)t = (v 1/2so)t.
Comment: If vo is zero or negative, there is no maximum time to for the traverse.
In the case vo = 0 the equation of motion
S = so[3(t/to) 2 - 3(t Ito) 3 ], 0- t ' to
satisfies the conditions of the problem for any to > 0.

B-3 From ABA = BA 2 B = BA'-B, we have
AB 2 = ABA -A- 1 B = BA-'BA-'B = BA-1 *ABA = B2A.
By induction, AB 2 ' = B2FA so that AB = AB2" = B2 'A = BA. Since A and B
commute, ABA = BA 2 B implies A 2B = A2 B2, or B = B2 , or B = 1.
Alternate Solution: It can be shown that A and B commute by expressing each as

powers of the same group element. Because A 3 = 1 it is tempting to multiply ABA
= BA 2B on the right by A2 and then on the left by BA 2 to get B2 = (BA 2) 3. Set
X = BA2 and use B" = B to obtain
(1) B = X3 .
From X = BA 2, we get XA = B, A = X'B, or
(2) A =X3-
The conclusion that B = I is as before.

B-4 Let x t + In+2. We construct a polynomial P(x, y, z)
= tn, y = t" I, z =
with integral coefficients such that P(x, y, z) = t. We have
z = t + tn+2,
Zy = tn+ 2 + t2n+3,
2 4
Alzy = t2n+3 + t3n+ ,
. . . . . . . . . .
2
Zn-2 = tn -n-I + tn2
Multiply the above equations alternately by + 1 and - 1 and add:

z[j _ y + y 2 + (-
2 2
1)n- yn- ] = t + (_ 1),-t_2 = t + (- 1)Ox-.
Hence, if we define
P(x, y, x) = 2 - 1y + (1- -I Xn'
Then P(t", t"+l 1 t + tn+

I2) = t. = t.
B-5 For the skew quadrilateral ABCD, let AB = a, BC = b, CD =c,

DA = d, AC = x, BD = y. None of these lengths can be zero. By the law of cosines:
a2 + b2 - x 2
C2 + d2 - x 2
ab cd
2
or (ab - cd)x2 = (bc - ad) (ac - bd). Similarly, (ad - bc)y = (cd - ab)(ac - bd).
CASE 1: ab - cd = 0.
Then, ad - bc = 0 and a = c, b = d.
CASE 2: ab-cd # 0.
Then, bc - ad $ 0, ac - bd # 0 and X 2 y 2 = (ac - bd)2 . Consequently,
ac = xy + bd or bd = ac + xy.
By Ptolemy's Theorem (in space), ABCD must be concyclic which violates the skew
condition.
Alternate Solution: If AC = BC and AD = BD, the conclusion that AC = BD
and BC = AD is obvious (see Figure 2) so assume AC # BC. With this assumption
we first show BD = AC. If BD 0 AC there exists a unique point D* in the plane of
AADB with BD* = AC, AD* = CB. / AD*B = /ACB = 0. From A's ADE and
BD*E it follows that L DAE = L D*BE.
D*
-/ -
FIG. 2
From the congruent A's CD*A and CD*B it follows that LCAD* =KCBD*.
These angle equalities prove that the trihedral angles A - CDD* and B - CDD* are
congruent. Hence the angle which CA makes with the plane ADD* is equal to the angle
CB makes with the plane BDD* (which is the plane ADD*). If H is the foot of the
altitude from C to this plane the A's CHA and CHB are congruent right triangles.
That is AC = CB. This is a contradiction and BD = AC.
Interchanging the roles of B and A in the above shows that AD = BC.
B-6 Let P(z) denote the given polynomial. The power series expansion of
1/(1 - z) - 2 P(z) has coefficients + 1 with leading coefficient -1. Hence,
(1) 1 +1 - 2P(z) <•IZ I+IZ12 + = z1

1 -z 1 ZIz
Also,
12P(z)j -Ž 1+ 1- 1+-1 z-2P(z)
1+ 1 = 21
Z III ZIZ 12
The latter term is positive for I z I < (,5 - 1)/2.

THE THIRTY-FOURTH WILLIAM LOWELL PUTNAM
December 1, 1973
A-1. (a) If X, Y, Z are at the midpoints of the sides, the area of AXYZ is one
fourth of the area of AABC. Also, as long as BX < XC, CY < lA and AZ < ZB,
moving one of X, Y, Z to the midpoint of its side, while leaving the other two fixed,
does not increase the area of AXYZ since the altitude to the fixed base of AXYZ
decreases or remains constant.
(b) Under the hypothesis of (a) the three corner triangles have no more than
three fourths of the total area and so one of them must have smaller area than
AXYZ. All other cases are similar to the one in which XC < BX and CZ < YA.
Then consideration of the altitudes to base XY shows that ACYX has smaller
area than AX YZ.
A-2. The ideas in both parts are similar and the answer in (b) is "Yes." Let un be
the nth term lln
/- and Sn = u + -. + u. Since un 0 as n -o, {Sn} will con-
verge if and only if {S8 m} does. Using the facts that
1 _ I k
n n+k n(n+k)'
that 1(l/n2 ) converges, and that X(1/n) diverges, one shows that with four "+"
signs and four "- " signs in each block, {S8m} converges as the term-by-term sum of
four convergent sequences while an imbalance of signs makes {S 8ml divergent as
the sum of a convergent and a divergent sequence.
A-3. Let c(n) = ./4n + 1 and let [x] denote the greatest integer in x; then we
wish to show that [b(n)] = [c(n)]. Let k(n) be a value of k that minimizes k + (n/k).
Then
b(n - 1) - k(n) + {(n - 1)/k(n)} < k(n) + {n/k(n)} = b(n),
i.e., b(n - 1) < b(n). Let rn be a positive integer. Then
b(m 2 ) = 2m, b(m 2 + m) = 2m + 1. (1)
It follows from formulas (I) and the strictly increasing nature of b(n) that
83
[b(n)] = 2m for m 2
- n < m
2
+ m, [b(n)] = 2m + 1 for m
2
+ m (11)
- n < (M + 1)2.
On the other hand, c(n) is also an increasing function and

c(m 2 -1) = 4m2 - 3 < 2m, c(in2 ) = v/4tn2 + 1 > 2m, c(m
2
+ m)
= 14112 + 4m + 1 = 2m + 1.
These facts show that (II) remains true when [c(n)] is substituted for [b(n)].
A-4. Three; at 0, 1, and some x > 1. The first two are clear and the other
followsfromf(4) <0 and f(5)> 0orfromf'(1) <0 whilef(x) - + ooasx - + a0.
There are no more zeros since four zeros of f would imply a zero of f"' using an
extension of Rolle's Theorem; but f'(x) = (log2) 3 2x # 0 for all x.
A-5. (a) If two of x(0), y(O), z(0) vanish, then x'(0) = y'(0) = z'(0) = 0, and
the uniqueness theorem applies (the equations are clearly "Lipschitz").
(b) Clear (this was intended as a hint for part (c)).
(c) Now write the equations in the symmetric form:
xx' = 31y' = zz' = xyz.
2
Thus X -C = Y2 - C2 = Z2 - C3 with constant c,. Say without loss of generality
that c1 - C2 - C3 , and then set C3 = 0. Thus z2 < y2 - x2, and:
z 2 =X 22 - C1 = y2 - C2 , Ci 2 0;
dz
dti= ± 1(z2 + Cl)(Z2 + C2).
Now let time t move in the direction which makes I z I increase (this depends on the
sign of z and on the + sign in the square root).
For simplicity assume that z(0) - 0, and that the sign on the square root is +
then let time move positively. Since
t = dz/ /(Z2 + Cl)(Z2 + C2)
and the z-integral converges, a finite amount of time suffices to push z out to infinity.
A-6. Any two distinct lines in the plane meet in at most one point. There are
altogether ( 2) = 21 pairs of lines. A triple intersection accounts for 3 of these
pairs of lines, and a simple intersection accounts for 1.
Finally, 6 3 + 4 1 = 22 > 21.
B-1. Since the sum of the 2n integers remaining is always even, no matter which
of the ai is taken away, all of the ai must have the same parity. Now a similar argument
SOLUTIONS: THE THIRTY-FOURTH COMPETITION 85
shows that they are all congruent (mod 4); for the property held by the aj is shared
by the integers a1/2 or (a,- 1)/2 (depending on whether the ai are all even or all odd).
Continuing in this manner, all of the ai are congruent (mod2 k) for every k. This is 2
possible for integers only if they are equal.

B-2. Let z = e@' and z' = w
u + iv (with u and v real). Then | 2n - 1
=
2
=IV - i I = [(U2 - V2 - = 2 1v |, using u2 + v2 = 1. [I z2 n -1
2
1)2 + (2uv) ]1/
2
= 2 1sin nO I is also easily shown geometrically using an isosceles triangle.] Hence it

suffices to show that v = sin nO is rational when x = cos 0 and y = sin 0 are rational.
For n - 0, this follows from (x + iy)f = u + iv or by mathematical induction using
the addition formulas for the sine and cosine. Then the case n < 0 follows using
sin( - x) = sin a.
B-3. One triple (a, b, c) satisfying the conditions is (1, - lp); it remains to
show that this is the only solution. Clearly b must be odd since b2 =1 (mod 4).
Also b2 = ( - b)2 , so write I b I = 2x-1. Then b2 - 4ac = 1 - 4 p gives
x2 -x+p = ac.
If 0 • x < p, the hypothesis tells us that ac is prime; then 0 < a ! c implies that
a = 1, it follows from - a 2, one sees
that x < p using
3a 2 = 4a 2 - a2 4ac -b = 4p - I,
bI < a •< V(4p -- 0)/3,

x = (I b I + 1)/2 <Vp3 + (1/2) < p.
B-4. We give two solutions; the first does not use the hint and the second does.
THEOREM. If ] is continuous on [0,1], f(O) = 0 and 0 < f '(x) f, [f(X)] 3
dx
unless, identically on [0, 1], either f(x) = x or f(x) = 0.

Proof. Define G(t) = 2 JIt(x)dx - [f(t)]2 for t e [0, 1]. Then G(0) = 0 and
G'(t) = 2f(t)[ I-f'(t)] > 0, so that G(t) > 0 and consequently J(t)G(t) > 0.
2
Now define F(t) = [fJf(x)dx] - fo[f(x)] 3 dx for te [0,1]. Then F(0) = 0
and F'(t) = f(t)G(t) > 0 so that F(t) - 0 and in particular F(l) - 0.
Equality is possible only if f(t)G(t) = F'(t) = 0 for all t, which implies that,
for some K, f = 0 on [0,K] and G' = 0, with f > 0, on (K,1). We then have
f' = 1 on (K, 1), which is admissible only if K = 0 or K = 1, since otherwise
f'(K) is simultaneously defined and undefined.
The unique answer to (b) is f(x) = x. The following is an outline of a proof

of (a) using the hint. Let f(1) = c. The hypothesis implies that f has an inverse g
with g'(y) - 1 on 0 < y - c. Let
A =Jo f(x)dx] and B = o [f(x)] 3 dx.

Then
A = 14 yg'(y)dyl = Xf yg'(y)zg'(z)dzdy = 2 fX yg'(y)zg'(z)dydz
using the symmetry of the integrand about the line y = z. Now g'(y) > 1 implies
A> f Zg'(Z) f2yd)y dz = f z 3g'(z)dz = B.
B-5. (a) Let r - b/a and s = - c/a. Let polynomials p,, and q, in r and s
be defined by the initial conditions po = 0, pi = 1, q0 = 1, and q1 = 0 and the
recursion formulas P. = rP- 1 + Sp.- 2 and q. = rq.-.I + sq.- 2 for n > 1. Using
zn = rz"-I + SZ,- 2 and mathematical induction, one proves that zn = pnz+ q.
and that all the coefficients in p"(r, s) are positive. Then multiplying numerator and
denominator of the right hand side of z = [z" - qn(- b/a, - c/a)]Ip"(- b/a, - c/a)
by the proper power of a leads to z = F(z", a, b, c)/G(a, b, c), where F and G are
polynomials with integer coefficients. Since all the coefficients in p"(r, s) are positive,
the same is true of G(a, b, c). Therefore G(a, b, c) is not identically zero and FIG is
the desired rational function.
(b) Let v = x + (1/x). Then x2 - Vx + 1 = 0. Using (a) with z replaced by x,
one finds that x 3 = p3X + q 3 with p3 = V 2 -1 and q 3 -v. Then
x = (x 3 - q 3 )/p 3 = (X 3 +V)/(V 2 -1).
B-6. (a) Simple calculus.
(b) By induction: The case n = 1 is just (a).
Now the ratio of the expression for n+ I to the expression for n is equal to:
Isin 2 0 *sin 2` ° |.
Since 0 = 7r/3 gives 2`0 =-27r/3 or 47r/3 (mod 27t), this ratio is maximized at 0 = z/3,
and by induction, then, the whole expression is maximized.
(c) Set 0 = 7r/3, and observe that the expression in part (b) is then exactly equal
to (3/4) 3n/2; its 2/3 power is thus equal to (3/4)". That is the maximum; in general
the 2/3 power of the expression in (b) is - (3/4)n. To get from that to the expression
in (c), we would increase the powers of the end factors sin 0 and sin 2"0; this can only
decrease the product, since I sin a | < 1.
THE THIRTY-FIFTH WILLIAM LOWELL PUTNAM
December 7, 1974
A-I.
A conspiratorial subset (CS) of {l,2,-. ,16} has at most two numbers from the pairwise
relatively prime set {1,2,3,5,7, 11, 13} and so has at most 16-(7-2)= II numbers. But
{2, 3, 4, 6, 8, 9, 10, 12, 14, 15. 16}
is a CS with II elements; hence the answer is II.
A-2.
Let C be the otherpoint of intersection of line AB with the circle and let 6 be the inclination of
AB. Let AB= b and AC= c. The square of the time of descent is proportional to b/sin 6 and hence
to l/(c sin 6), since it is well known that bc is constant with respect to 6. The time is minimized by
maximizing c sin 6; this is done by choosing C as the bottom of the circle.
A-3.
If p I (mod 4). either (A): p 31 (mod 8) or (B): p = 5 (mod 8). We show that (A) and (B) are
necessary and sufficient for (a) and (b), respectively. If p = m 2 + n2 and p is odd, one can let m be
oddand n be even. Then = m2 +4v 2 with m2' 1 (mod8). With (A), v iseven and p = m 2 + 16 w2 .
Conversely, p = m2 + 16w 2 implies p 3 1 (mod 8). With (B), v is odd, m = 2u + v for some
integer u, and p =(2u + v)2 +4v 2 =4u2 +4uv +5v2 . Conversely, p =4u2 +4uv +5v2 with p odd
implies p = (2u + v)2 + 4v2 with v odd and hence p 3 5 (mod 8).
A-4.
The answer is
2F-7 ([(nn-1)/21)
since
(n -2k)Q)= E {(n -k)(n) -k(n)

kin/2 k cn/2k k
= n2t(k-I n(n I = k)f/2t(nk ) (n -l)}
= nt(n-l)~/2])
A-5.
Let F be the fixed focus, M be the moving focus, and T be the (varying) point of mutual
87
tangency. The reflecting property of parabolas tells us that the tangent line at T makes equal angles
with FT and with a vertical line. This and congruence of the two parabolas imply that MT is vertical
and that the segments FT and MT are equal. Now M must be on the horizontal fixed directrix
y = 1/4 by the focus-directrix definition of a parabola.
A-6.
Let p(k,x) be the monic polynomial (x + l)(x + 2).. (x + k) and let m be an integer. Then
p(k,m) is exactly divisible by k! since the absolute value of the quotient is a binomial coefficient
(even when m is negative). Hence, if n | k! there is a monic integral polynomial f(x) of degree k with
n Jf(m) for all integers m. Conversely, the condition n Ik! is necessary since the k-th difference k!
of a monic integral polynomial of degree k is divisible by any common divisor of all the values f(m).
In particular, k(10') =k(56 26 ) =25 since the smallest s with 561 s! is s = 25.
B-I.
Since the pi need not be distinct, the sum is a continuous function on the compact set
C x C x C x C x C, where C is the circle. Hence maxima exist. One proves that the maximum
occurs when the pi are the vertices of a regular pentagon by showing that this configuration
simultaneously maximizes both of the sums:
S = d(p, p2 ) + d(p2 ,p3 ) + d(p3 , p) + d(p4 , p,) + d(p5 ,p1 ),
T = d(p, pI) + d(p2 , p4 ) + d(p3 ,p5 ) + d(p4 , p,) + d(p5 ,p 2 ).
For S or T. one can fix four of the points; then the varying part of the sum is of the form.
D = d(p,a)+d(p,b), with a and b fixed.
Using the Law of Sines, one shows that D is a constant times sin a + sin 3 where a = 4 pab,
3 = 4 pba, and a + / is constant. Then it is easy to show that D is not a maximum unless p is
symmetrically situated with respect to a and b.
B-2.
If y '(x,) - 0 for a sequence {X, } approaching + o, the hypothesis insures that y(x, ) 0. Since
these x. may include any relative maxima and minima, this case must have y(x) - 0 as x - + .
Then one also has y'(x) - 0 as x -. + o.
In the remaining case, there is an x0 such that for x > x0 one has y' # 0 and so (y')2 >0. We
restrict ourselves to the x's with x > xo and consider two subcases:
(a) y' >0. If y is unbounded above, so are y' and (y')2 + y3 . This contradicts the hypothesis
(y') 2 +y- *0 as x-- + . If y is bounded above, it approaches a finite limit. Then y3 , (y')2 and y'
approach limits. Since y is bounded, the limit for y' must be 0. Then y also has 0 as its limit.
(b) y' < 0. There is no problem unless y is unbounded below. Then we may assume that y < 0
and compare v to a solution of the differential equation
y' =-(I/2)1 yI"', yY< 0
Every solution diverges to - in a finite interval, hence so does y(x); this contradicts the
hypothesis that y is defined and smooth for all large x.
B-3.
If a = ris with r and s integers and s > 0, then cos(n7ra) takes on at most 2s distinct values for
integral choices of n. When cos ira = 1/3, the formula cos 20 = 2 cos 2 0- I and mathematical
induction can be used to show that
cos(2'ra) = t/32 - [m = 1,2, 3,']
SOLUTIONS: THE THIRTY-FIFTH COMPETITION 89
with t an integer not divisible by 3, and hence that these cosines form an infinite set of distinct
values. Thus a is irrational.
B-4.
For each n, we construct the function gn(x, y) as follows: First divide the xy- plane into vertical
strips of width l/n separated by the lines {x = m/n}, m an integer. Now set g,(x, y) = f(x, y) along
each vertical line x = mrin, and interpolate linearly (holding y fixed and letting x vary) in between.
Then g,(x, y) is continuous because f(x0, y) is continuous in y; g.(x, y) - f(x, y) because f(x, yo) is
continuous in x.
REMARKS. This result has two interesting consequences for functions which are continuous in
each variable separately:
(i) Such functions are Borel measurable.
(ii) They are continuous (in the usual sense) except on a set of points of the first Baire category.
(In particular, there is no function which is continuous in each variable separately and yet
discontinuous at every point.)
B-5.
We want to show that
e - (n - t)&e'dt >
or, equivalently, that
n ! > 2e-' |' (n - t)'e'dt,
1'e-'dt > 2e (n - t)re'dt.

Letting u = n - t, this can be transformed into
f t'e -'dt > 2fu' e ~du,

which is equivalent to
f une- du > f u'e- du.
Let f(u) = u'e-'. Then it suffices to show that

f(n + h) - f(n - h) for O- h - n.
This is equivalent to
(n + h) e > (n - e),
n In(n + h) - h > nIn(n - h) + h.
Let g(h) = n ln(n.+ h) - n In(n - h) - 2h. Then g(0) = 0 and
2
dg = f + -2 = 22n 22>
dh n +h - h n -h2
for 0 < h < n. Hence g(h) > 0 for 0 < h < n. The desired result follows.
B.-6.
Let n - r (mod 6) with r in {0, 1,2,3,4,5). Then the pattern is
r 0 1 2 3 4 5
So.. a+l b c d-I e f
s a b c+l d e f
52,n a b- I c d e+l f
This is easily proved by mathematical induction using the formulas
Si, = Si-I-I + S;-1. [Here 0 - I =2 (mod 3).1
These formulas follow immediately from the rule
(k) = (n - l) +( - )
The sums may be computed readily using the above patterns and
S0O, + SI.. + S2.. 2".
For n = 1000, r = 4 and
5° 100 = 5 1.1000= S 2 .1000- = (2'°- 1)/3.
THE THIRTY-SIXTH WILLIAM LOWELL PUTNAM
December 6, 19'75
A-1.
Let n = [(a 2 + a)/21 + [(b 2 + b)/2J, with a and b integers. Then

4n + 1 = 2a 2 + 2a + 2b 2 + 2b + 1 = (a + b + 1)2+ (a - b)2 .
Conversely, let 4n + 1 =x 2 + y2 , with x and y integers. Then exactly one of x and y is odd and so
a = (x + y - 1)/2 and b = (x - y - 1)/2 are integers. One easily verifies that
[(a 2 + a)12] + [(b2 + b)/2] = (x 2 + y2 - 1)/4 = n.
A-2.
The desired region is the inside of the triangle with vertices (0, - 1), (2,1), (- 2, 1). The boundary
segments lie on the lines
L: c =1, L2 : c - b + =0, L3 : c + b + 1 =0.
2
To see this, we let f(z) = z + bz + c and denote its zeros by r and s. Then - b = r + s and c = rs.
Also
(r + )(s + 1)=rs + r + s + 1 = c - b + 1= f(- 1),
(r - )(s - )=rs - r - s + = c + b + f(l).
On or. below L2 , at least one zero is real and not greater than - 1; this follows either from (r + 1)-
(s + 1)c0 or from f( - 1) ' 0 and the fact that the graph of y = f(x), for x real, is an upward opening
parabola. Similarly, on or below L3 one zero is real and at least 1. On or above L, at least one zero
has absolute value greater than or equal to 1. Hence the desired points (b, c) must be inside the
described triangle.
Conversely, if (b,c) is inside the triangle, Icl< l and so Irl< I orIsI<1 orboth. If thezeros are
complex, they are conjugates and Ir I = Is 1;then I r =I Is I < 1 follows from I c I < 1. If the zeros are
real, Ic I 0 and (r - 1)
(s - 1) = f(1) > 0 imply that the other zero is also in (- 1, 1).
For full credit, the region had to be depicted.
A-3.
Let h(x) = x' - xb and k(z) = z' - zb. The desired points also give the maximum and minimum of
the function
g(x, z) = (x' + y' + z )-(xb + yb + zb) = h(x)+ k(z)
on the domain obtained by projection of the solid domain on the xz-plane. For all points under
91
consideration, both x and z are in [0,1]. Examining its derivative, one sees that h(x) increases from O
at x = 0 to a maximum at x, = (alb)/(b-" and then decreases to O at x = 1. (This uses the hypothesis
0< a <b.) Similarly, k(z) decreases from 0 at z= 0 to a minimum at z0 = (b^c)"1 -b and then
increases to O at z = 1. Since (1, zo) and (xo, 1) are not in the domain of g(x, z), the function f achieves
its maximum only at (x, y, z) = (xo, [1- x 0) and achieves its minimum only at (0, [1- zo']lb, zo).
0Ii/b,
A-4.
Let n = 4k + 2 with k >0. Then
0 = o1 - i = o4k,2- 1 = (o2k.1 - 1)(02k+1 + 1),
0 = (02k- 1)(6 + 1)(O2k - o2k I+ 2 - - +1).
Since 6 is a primitive nth root of unity with n > 2k + 1 and n > 2,

(0 - 1) (6 + 1)/ 0.
Hence
(A) 62k - o2k -I1+2k-2.. .+ 2_ 0 + 1 =0,
1= _ 0o2+ 03_- * . * 2k = (1 -e)(0+ + 6+l+ **+ 2k-1),
(1 _ 0)-1 = + 03+ .+ 6k-' [where 2k - I = (n - 4)/2].

Another solution is (1 - 6) = I + 62 + 04 + * + 02k as one sees from (A).
A-S.
The answer for c is V/2Iw, where w is the wronskian yy2- y2y. (and will be seen below to be
constant).
Let C2 = 2k. Then z2 /2 = kyIy2 . Differentiating twice, one has
zz'= k(yly2+ y2y1), zz+ (z')2 = k(yty2'+ y2Yt+ 2yfy2).
Since y I = fy and Y2'= fY2, this implies
zz" + (z')2 = 2k(fyy 2 + yy 2') = f(2ky.y 2)+2ky~y2 = fz2 + 2ky y2.
Now
2 4 2
z z +(ZZ1) = fz + 2kz yy,
z 3z"+ k2(yly,+ y 2y92 = fz4+ 4k2(YsY2Yi'Yi)
z3z" + k 2(yy2 - y 2y )2 = fz4,

3 4 2 2 2 4 2
(1) z z _fz = - k (y y2-Y2Y) - k W = C W /4
2
Since w'=jlyy2- y 2 y;)' = yy'2'- y 2 y' = y#(fy2 ) - y 2 (fY,) = 0, w is a constant. Solving C'w /4= 1for c
gives c = C2w; for this c, (1) impliesiz"- fz = - z or z"+ z- = fz.
A-6.
Let A denote the line through the desired points P 4 and P5. Let ir be the plane of P., P2, and P3
and let H be the intersection of A with 7r.
Let v. be the vector HP, and Iv. I be its magnitude. We wish to have the dot product
(I) d= Ph Pk. PiPi, = (Vk - vh ) ' (Vj - vA) = Vs . Vj - Vk . V - Uh ' Vj + Vh ' Vv
zero for all choices of h, k, i, j as distinct indices in {1, 2,3,4, 5}.

Since A is to be perpendicular to i7, we must have
SOLUTIONS: THE THIRTY-SIXTH COMPETITION 93
(2) vh,-v=0 for hE{4,5} and iE{1,2,3}.

If h, k e {4, 5} and i, j e {1, 2,3}, (2) implies that the dot product d of (1) is zero. If h E {4, 5} and
k, i, j E {1,2,3}, (2) implies that the d of (1) becomes
(3) d = Vk ' V, - Vk *vi =VI *(v, - vi) = HPk Pi Pi.
Clearly the d of (3) are zero simultaneously if and only if H is the orthocenter (i.e., intersection of
altitudes) of APIP2P3. With this choice of H, the vanishing of the d of (3) implies
(4) v2 *V3 =v * 3 = v V2.
Now let h, i E {4,51 and k, j e {1, 2,3}. Then (2) implies

(5) d =Vk ' Vj + V4c V 5.
Assuming (4), one sees that all the d's of (5) will be zero if V4' = - v2. The hypothesis that
V
AP1 P2 P3 is acute-angled tells us that H is inside the triangle. Then at least one (actually, all) of the
angles 4PHP2 , 4P 2 HP3 , 4P3 HP, must be obtuse and so the equal dot products of (4) must be
negative. Hence v4 * v5 must be positive; this means that P4 and P5 must be on the same half-line of A
determined by H.
Now the location of P 4 and P5 can be given. Let H be the orthocenter of APP2P3 and JLbe either
half-line perpendicular to plane PP 2P3 at H. Then P 4 may be any point on p. such that IV 4 1 is neither
zero nor (- vI *v2)" and P 5 must be the unique point on a with lvsI = - yr v*/I v.4 Then each d of
(1) is zero and no three of the Pi are collinear.
B-1.
The answer is a = 7. Also one must have b = 5(mod 7).
Proof: The subgroup H must contain 4(3,8) - 3(4, - 1) = (0,35), 4(5,4) - 5(4, - 1) = (0,21), and then
2(0,21) - (0,35) = (0,7). Now (0,7) and (1, b) will generate H iff (1,b) is in H and there exist integers u,
v, and w such that
(3, 8) = 3(1, b) + u(O, 7), (4, - 1) = 4(1, b) + v (O,7), (5, 4) = 5(1, b) + w (0,7).
These hold iff 8 = 3b + 7u, - 1 = 4b + 7v, and 4 = 5b + 7w. With b = 5 + 7k, k any integer, the desired
coefficients u, v, and w exist in the form u = -1- 3k, v = - 3 - 4k, w = - 3-5k. It now suffices to
let k = 0 and to note that (1,S) = (4, - 1) - (3,8) + 2(0,7) is in H.
B-2.
Let Ei, = d and let S be a sphere of radius r > d/2. The area of S contained in slab Si is at most
21rd,. It follows that the area of S contained in the union of the slabs Si is at most 21rd < 47rr = (area of
S). Hence there are points of S that are not in any of the slabs.
The problem may also be done using volumes of intersection of the slabs with an appropriately
chosen sphere.
B-3.
In the expansion of s = (a, + a2 + * + a, )k, every term of sk appears with k! as coefficient and
the other coefficients are nonnegative. Hence sk/sl -' I/k!
If we let each ai = 1,
A = (kn / n( 1)... (nk )= I (1-n) 2) .. (I k )'
which approaches 1/k! as k is held fixed and n goes to infinity. These facts show that the supremum
A4 is 1/k!.
B4.
No. Since the mapping with (x, y)-* (- x, - y) is a homeomorphism of the unit circle on itself, the
complement - B of such a subset B would also be closed. Thus the existence of such a B would make
C the union - B U B of disjoint nonempty closed subsets; this would contradict the fact that C is
connected.
B-5.
Since fo(x) = Ek-.ox k/k!, one easily shows by mathematical induction that fa(x) = YZ-o(knxk/k!).
Then, since all terms are positive, one has
f(l)E k" I ± k-= ek= et
n-0 n! n._ k-6 k! n! kto ! n-6 n! kOk!
B-6.
Both parts are done easily using the Inequality on the Means. For (a), one has
n+s, (1+1)+(1+2_)+__ _ +(l+(_/n))
n n2 >V( + 2). . . (I + (I/n)) = 32- 3 ... (n +I)ln

(n + I)
and so n + s, > n(n + 1)1/n.
For (b), one has
n-Sn ( 2-')+( -3)+ (+(I-(t/n)) 2___________3____ (n_-_1)_ n
n -I n- I 1/n2/(3I
= n-s-s>n
and so n- Sn >(n - )n v"c")
THE THIRTY-SEVENTH WILLIAM LOWELL PUTNAM
December 4, 1976
A-i.
Let A be the angle bisector of 4AOB and A be the perpendicular to /. through P. Then the
intersections of A with OA and OB are chosen as X and Y respectively.
This construction makes OX = OY and there is a circle F tangent to OA at X and to OB at Y.Let
X, Y, be any other segment containing P with X, on OA and Y. on OB. Let X2 and Y2 be the
intersections of X, Y. with F. A theorem of Euclidean geometry states that (PX) (PY) = (PX2 ) (PY2 ).
Clearly (PX2 )(PY 2 ) is less than (PX,)(PY,). Hence (PX)(PY) is a minimum.
One can also locate X and Y by saying that (ir -4AOB)/2 should be chosen as the measure of
4 OXP or 4 OYP.
A-2.
One easily verifies that
(x + y) = (x + y)f 2Q + (x + y)'-'P,
2 3
x +y =(x, -yI )O-(x, -yI )P.
Subtracting or adding corresponding sides gives
(R) F. = QF, n+ PG. -, G. = QG-2 + PF.,.
The desired results now follow by strong mathematical induction using the given results for G2, F3,
G4. F,, and G6 and (R).
A-3.
We show that the only solutions are given by 3' - 2' = 1,i.e.. (p, r,q, s) = (3,2.2,3) or (2,3.3.2).
Clearly either p or q is 2. Suppose q = 2. Then p is an odd prime with p' ± I = 2'. If r is odd,
(p' + I)I(p + 1) is the odd integer p`-' p- +p' p' -'+.. + 1. which is greater than I since
r > I; this contradicts the fact that 2' has no such factor.
Now we try r as an even integer 2t. Then p' + I = 2' leads to
2' = (p) 2
+ I = (2n + 1)2+1 = 4n 2 +4n + 2,
which is impossible since 4 2' for s > I and 4. ?(4n 2 + 4n + 2).
Also r = 2r and p' - I = 2 ' leads to (p')- I = (2n + 1)' - = 4n'+ 4n = 4n(n + 1 =2'. Since
either n or n + I is odd. this is only possible for n = 1, s = 3, p = 3, and r = 2.
A-4.
We show that one answer is s = - I/(r t 1) and another answer is s - (r + I)/r = - I - (I/r).
Since P(x) is irreducible, so is M(x)= P(x - I). Hence M(x) is the onlv monic cubic over the
rationals with r + I as a zero. i.e., M(x) =x' + cx: + dx + 1.If the zeros of P are r, s, and t, the zeros of
95
M are r + 1.s + 1. and t + 1. Now the coefficients - I and I of x' in P and M, respectively, tell us that
rst = I and (r + 1) (s +
-)(t + 1)= -1. Then
st = 1 s + t = (S +I) (t + 1)- st-I = - - r+

r ri-I r r(r+ 1)
Hence s is either root of
r2 +-3r
XI+ r t +1I - = O.
r(Ir+1 r
Using the quadratic formula, one finds that s is - 1I(r + 1) or - (r + I)/r.
A-5.
It is shown below that a = 27r, b = 1, and c = I. We use I4S] to denote the integral of e over
a region S. Since D(x. y) = 0 on R, I[RI = A. Now let a be a side of R. s be the length of a, and S(a)
be the half strip consisting of the points of the plane having a point on a as the nearest point of R.
Changing to (u. v)-coordinates with u measured parallel to a and v measured perpendicular to a, one
finds that IES(a)I = fJf,) e-'dvdu = s. The sum Y, of these integrals for all the sides of R is L.
If v is a vertex of R, the points with v as the nearest point of R lie in the inside T(v) of an angle
bounded by the rays from v perpendicular to the edges meeting at v; let a = a(v) be the measure of
this angle. Using polar coordinates. one has
I[T(v)] =L L reC'drdO = a.
The sum Y2 of the I[T(v)J for all vertices v of R is 2fr. Now the original double integral equals
,+1, +A =27r+L +A. Hence a = 2 ir and b = I = c.
A-6.
Let G(x)= V(x)f+ [f'(x)] 2 and H(x) = f(x)+f"(x)Since H is continuous, it suffices to show
that H changes sign. We assume that either H(x)>() for all x or H(x)<0 for all x and obtain a
contradiction.
Since f(0)! - I and G(0) =4. either f'(0) > V3 or f'(O) ' - V3. We deal with the case in which
H(x)> 0 for all x and f'(0)> \ 3; the other cases are similar.
Assume that the set S of positive x with f'(x) V\3 and continuity of f'(x) imply g >(0. Now f'(x)- O and H(x)' O for
0' x - g lead to
G(g)=4+ 24gf'( x) [f( x) +f"(x)] dx >-4.
Since If(g)[ l. this implies f'(g)> V3. Then continuity of f'(x) tells us that there is an a >0 such
that f'(x)> I for 0() x < g + a. This contradicts the definition of g and hence S is empty. Now
f'(x)- I for all x and this implies that f(x) is unbounded, contradicting ,f(x)l - 1.This contradiction
means that H(x) must change sign and so H(x,,) = ( for some real x,,.
Alternately. we use the Mean Value Theorem to deduce the existence of a and b with
-2<a <0<h --2 and
'l)(0)- ,f ?< f() s-If( --2)c I + I
andsimilarly !f'(b) .I. Then G(a)(=) 1(ao) [f'(a)I- I +1= 2 andalso G(b)<2. Since G(0)= 4,
G(x) attains its maximum on a ' x ' b at an interior point x, and hence G'(x,,) = f'(x,,)H(x,) = 0.
But f'(x,,) k 0 since otherwise If(x,,)'= G(x,,) - 4 and 'f(x,,)> 1. Thus H(x,,) =0.
SOLUTIONS: THE THIRTY-SEVENTH COMPETITION 97
B-1.
It is shown below that a = 4 and b = 1. Let f(x) = [2/xl - 2[llxJ. Then the desired limit L equals
f(.f(x)dx. For n = 1,2,...,f(x)=O on 2/(2n + 1)<x •'1/n andf(x)= Ion 1/(n + 1)<x -•2/(2n + 1).
Hence
= (3 4-) + 2_2-) + * = -1+22(- 2 +3- )
1 Fdx
=-- +2j dx I-+21n2=n4 -1.
B-2.
Theanswersare(a)8; (b) 1,A2BB 2 ,B3 ,B4 ,B',B6 . SinceB = (B4 )2,B 3 (B5 )2, B 5 = (B6)2 , the
elements in the answer to (b) are all squares in G. They are distinct since B has order 7 and A has
order 4. To show that there are no other squares, we first note that ABA 'B = I implies AR = B-'A.
Then
AB 2 =(B 'A )B = B '(AB) = B -'(B -'A )B 'A.
Similarly AB" = B "A for the other n's in {O. l, . .., 6} and so for all integers n.. With this, one obtains
(P) (B'Ai)(B"Ak)= B"A with u = i+(- IYh, v =j+k.
Thus the set S of elements of the form BA' is closed under multiplication. S is finite since i and j
may be restricted to 0 - i ' 6 and 0 - j3. Hence S is a group and so S = G. It then follows from (P)
that the squares in G are the B"A° with u =i[ +(- IY I and v =2j. If j is odd, u =0 and
v = 2(mod 4). If j is even, v - 0 (mod 4). Thus there are no squares other than those listed above.
B-3.
The statement to be proved is false for n ' 5 unless the hypothesis is strengthened to state that Ai
is independent of the conjunction of A, A2,... Ai 2 for 3 ' i -' n.
The following counterexample with n = 5 was furnished by Professor David M. Bloom of
Brooklyn College. Let h = 33/37 and k = 1/(64+ h). Let P(A,) be the sum of the numbers in the
second row of the following table for which Ai appears in the heading:
AIA 2 A 4 A5 j A4 j A1A 3 A 4 A5 AIA2 A3A 4 A5 A1A 2 A3A 4 A 2 A3 A 4 A5
12k 13k i 6k I 7k I 12k I 6k I
AA 3 A 5 A1A 2 A3A5 A 2 A 3 A5 A3 A 4 A 5
3k I 9k I 3k I 3k
Then each P(A,) is 49k and P(Ai AAJ)= 37k for all i, j with Ii- j > 1. Since (49k)2 = 37k, the
original independence hypothesis holds. Also, P(Ai) = 1- a, where a = (15 + h)k < 1/4. However,
for any a '•1/4, we have u5 Ž 7/64 and P(A, AA 2 A A 3 AA 4 AA 5)= 7k < 7/64.
B-4.
We let P = (x, y) and the ellipse have the equation b2x 2 + a2 y2 = a2 b2 , with a > b >0. Then
F,=(-c,0) and F, = (c, 0) with c 2 - a 2 - b2 . Let r, = PF, and r 2 = PF2. Then r, + r2 = 2a and
[(r, + r2 ) 2 - r'2- r']
r, r2 = (12)
2
=(')[4a2 -(x +C) 2 -y 2
-(x _C) y2 ]
=2a2 -x 2 2 2
=a 2 +b 2
-y -c x2 y 2
A point (u, v) on the tangent to the ellipse at P satisfies
a2 b2
Putting this in the form u cos 0 + v sin 0 = d, one finds that

1 a'b 4
(xa 2 )2 +(y/b 2 )2 =b 4 x2 + a4 y2
But bx 2+a 4y = b2(a2 b2 - a2 y 2) +a2 (a2b2 b 2 x2) = a2 b2(a2 +b2 x2 - y

2
) =a 2b2 rr. Hence
d'rr, = a 4 b4rr2la2b 2rr,=a2 bW,a constant.
B-5.
The sum is n! since it is an nth difference of a monic polynomial, x", of degree n.
B-6.
'.. plj where a and the P, are nonnegative integers and the p. are distinct odd
Let N = 2'pi'p '2
primes. Then
a,(N) = o(2a )or(pe,) .. c0(pk
Since cr(N) = 2N + I is odd, it follows that o-(p,3 ') is odd, I ' i -' k. But
(P i I p pj 2j p j
is odd if and only if Pi is even; for if 1 were odd, the right hand side would be the sum of an even
number of odd numbers and hence even. It follows that the odd part of N must be a square, so that we
may write
(I) N=2'M2, a>0.
where M is odd. It remains to show that a = 0.
Since N is quasiperfect, cr(N) = 2"'M2 + 1, while from (1) we deduce cr(N) = o(2")O(M 2 ) -
(2atg- 1)o-(M 2 ). Hence 2"+'M2 + I = (2"+'- 1)ca(M 2 ) so that

(2) M2 + I =0(mod 2 +' - 1).
If a > 0, 2"a - I -- 3 (mod 4). Consequently 2" - I has a prime divisor p - 3 (mod 4). Equation (2)
implies
(3) M2 + I - 0 (mod p).
But since - I is a quadratic non-residue modulo p whenever p -3 (mod4), (3) is impossible. Thus,
a = 0.
THE THIRTY-EIGHTH WILLIAM LOWELL PUTNAM
December 3, 1977
A-I.
A line meeting the graph in four points has an equation y = mx + b. Then the xi are the roots of
2x 4 + 7x3 + (3- m)x-(5 + b) = O,
their sum is - 7/2, and their arithmetic mean ()x,)/4 is - 7/8, which is independent of the line.
A-2.
We show that w must equal one of x, y, z and that the remaining two unknowns must be
negatives of each other. Let s = x +y and p = xy. Then the given equations imply that w - z = s
and that
sx+y I + 1 I z-w - s
p xy y x W z zw zw
Then s/p=s/(-zw) implies that either s=0 or -zw=p. If s=0, then y=-x and w=z. If
- zw =p = xy, then - z and w are the roots of the quadratic equation T2 - sT+p = 0, which has
x andy as its roots; this case thus leads to either w=x and -z=y or w=y and -z=x.
A-2. Alternate solution

Substitute w from the first equation into the second. The resulting expression factors into
(x + y)(x + z)(y + z) = 0. In this form we see that two of the numbers x, y, z must be negatives
of each other, and the third number must equal w (by the first equation).
A-3.
We show that there are an infinite number of expressions for u(x) in terms of f and g; some of
the simpler ones are:
u(x) =g(x) -f(x + 3) +A(x + 1) +A(x - )-f(x-3)
=-g(x + 2) +f(x + 5)-f(x + 3) +f(x+ I) +f(x-- )
=g(x+4)-f(x+7) +f(x+5)
-f(x +3)+fA(x+ 1).
1
Let E be the shift operator on functions A defined by EA(x) = A(x + 1). Then (E + E - )u(x) =
4 4 2 8
2f(x) and (E +E- )u(x)=2g(x) are given. Thus (E +l)u(x)=2Ef(x) and (E +l)u(x)-
2E 4 g(x). Motivated by the fact that E 2 + I and E 8 + I are relatively prime polynomials in E, one
finds that
I= (EB+ 1)-2 (E6-E4+E2-1)(E2+ 1)
4 2
u(x)= 2(E+ l)u(x)-j (E6 -E + F 2 -)(E + I)u(x)
99
u(x) = E 4 g(x) - (E 6 -E 4
+ E 2 - l)Ef(x),
u(x)= E 4 g(x)+(- E 7 + E 5 - E 3 + E)f(x),
u(x)=g(x+4)-f(x+7) +f(x+5)-f(x+3)+f(x+ 1).

Other expressions are obtained using
g(y)= -g(y -2) +f(y+ 3) +f(y -5)
A4. =-g(y+2)+A(y+5)+Ay -3).
2
n-O ] X n ( __X I X )
| -X - X2--1 I|-X I-X
since jxj < 1.

A-5.
It is well known that (P )-0 mod p for i = 1,2, .,p - or equivalently that in Z,[x] one has
(I + x)' = I + xP, where Z, is the field of the integers modulo p. Thus in Zp[x],
(pa ) k =( + X)P =[(I + X) ] =[ + X] ,2 ja)),
Since coefficients of like powers must be congruent modulo p in the equality
kO ( k)X Y. j
in Zf[xl, one sees that
( pb)-(ab)(modp)
for b=0,I.a.
A-6.
For (a,b) in S, let I(a,b) be ff f(xy) dxdy over the rectangle 0 <x <a, 0 <Y <6b. Also let (a,b)
define inductively a sequence (a,,b,) using aj a,b =b, a,+,=a,-b. and b,,+=b. when
0<b,,6a., and a., l= a. and b,+, =b,,-aa when 0<a.<bb. Then the hypothesis implies that
I(ab)=I(a,,,b.)for all n. Since f is bounded on S and lim,_.a. ==0=lim,-b,,, it follows that
I(a,b)=O for all (a,b) in S.
If f(x,y) is not zero for all (x,y) in S, then f must be positive (or negative) in some rectangle
Rs{(x,y):csxsd,h sy <k) and hence I= fRff(x,y)dxdy must be positive (or negative).
But this contradicts
I = I(h, k) - I(h, d) - I(c, k) + I(c, d) = 0.
Thus f is identically zero on S.
B-i.
n3- 1 23-l 33-l k 3 -lI

n-2 n3 + I 23+1 33+1 k3+1I
SOLUTIONS: THE THIRTY-EIGHTH COMPETITION 101
[ 1 7 2-13 3 21 (k-l)(k2 2+k+1)

[-xk
3 34-7 5-13 (k+I)(k -k+l)
2 k2 +k+l1 2
=l~k.3 k(k+ 1) J 3
B-2.
Let O' be any point different from 0 on the line of intersection of planes A OC and BOD, e.g.,
O' may be the intersection of lines AC and BD. Let A' be the intersection of line OA with the
line through O' and parallel to OC. Let C' be the intersection of line OC with the line through
O' which is parallel to OA. Then OA'O'C' is a parallelogram and its diagonals 00' and A'C'
bisect each other at a point M. Choosing B' and D' in the same way, one obtains a
parallelogram OB'O'D' whose diagonals 00' and B'D' also bisect each other at the midpoint
M of segment 00'. Hence segments A'C' and B'D' bisect each other (at M) and A'B'C'D' is a
parallelogram. (The parallelogram is not unique.)
B-3.
Let xi=, ~ with aE{0,l}, be the binary expansion of xi. The triple is balanced if
la2-j,
al1 =a 2 l=a 3 l=0. Otherwise, a,1 =l for exactly one i and the balancing act produces xi'=
E:y laJ+j2-'. An unbalanced triple that remains unbalanced after any finite number of
balancing acts is constructed by choosing the au, so that exactly one of a1 j, a2 j, a3j equals I for
each j while taking care that no one of sequences a,1 , a 2, ... repeats in blocks, i.e., that each xi is
irrational. One such solution has
aj= I if and only ifje (1,9,25,49,... },
a2= 1 if and only ifjE{4,16,36,64,...),
a3j=l ifandonlyifje(2,3,5,6,...}.
B-4.
We can assume that Q=0, the origin. Let - C be the image of C under the reflection P-a- P.
-C is again a continuous closed curve surrounding 0 and C n - C 0 since they have the
same diameter and both surround 0 (hence neither can be exterior to the other). Let P, E C n -
C. Then there exists P2 E C such that P, =-P 2 . These are the two desired points.
B-5.
From the Cauchy-Schwarz Inequality, one has
2 2
[(a,+a 2 )+a 3 +a 4 + ++a.j<[12+12+..- +12[(a,+a 2 ) +a2 + +a.]
or
(2a,)2 <(n-l)[(Xaj2)+2aa 2 ] or [l/(n-l)](1ai)2 <(Xa,2 )+2aa 2 .
Using the hypothesis, one then has
A < -(Ea?2)+ 1 (2 a) 2 < -(1a,2)+(1a, 2 )+2ala 2 =2ala2 .
Similarly, A < 2ajaj for 1 i <j <n.

B-6.
Clearly I E H. Also x E H implies x -'E H. Then the hypothesis implies that a - = a2 and that
xaxaxa = 1= x -'ax -'ax 'la when x E H. Thus one easily shows that
(i) axa = x -'a2x -', (ii) a 2xa2 =-x -'ax -'.
Let
A={xay:x,yEH}, B={xa2 y:x,yEH},
C={xa2 ya:x,yEH}, and Q=AuBUC.
Each of A, B, C has at most h2 elements; hence Q has at most 3h2 elements. Thus it suffices to
prove that xjax2 a*** xa E Q when each xi E H. We do this by induction on n.
For n=l, one sees that xla=x 1 a -lEAQ. Now let x 1 ,x 2 .... xk+,IEH; x~ax 2a...xka
qxk+ = z, and qza =p. Inductively, we assume q EQ and seek to show p EQ. The assumption
implies that q is in A, B,or C. If q= xay eA, then
p = (xay)za = xa(yz)a = x(yz) a2 (yz) 'EB C Q,
using (i). If q = xa2y E B, then p = xa2 yza E C C Q. If q = xa2 ya E C, then
p = xa2 y(aza) = xa2y(z -'a 2z -I) = x[ a 2 (yZ - I)a
2
]z-I
=x(yz -) 'a(yz') 1z 'EACQ,

using (i) and (ii).
THE THIRTY-NINTH WILLIAM LOWELL PUTNAM
December 2, 1978
A-I.
Each of the twenty integers of A must be in one of the eighteen disjoint sets
{1}, {52}, {4,100}, {7,97}, {10,94},...,{49,55}.
Hence some (at least two) of the pairs {4, 1001,..., {49, 55} must have two integers from A. But
the sum for each of these pairs is 104.
A-2.
Let M, be the matrix obtained by subtracting I from each entry of the given matrix and let GQ)
be the determinant of M,. By subtracting the entries of any row from the corresponding entries
of each other row, one sees that G(t) is linear in t. Then one notes that G(a) =f(a) and
G(b)-f(b) using the fact that they are determinants of triangular matrices. Then linear
interpolation shows that the desired determinant G(O) is
[bG(a) - aG(b)]/(b - a) = [bf(a) - af(b)]/(b - a).

A solution can be given using induction. For the inductive step, subtract the second column from
the first and expand by cofactors down the first column. Apply the inductive assumption to the
two resulting determinants using the functions F(x) = (P2 - X)(p 3 - x) (Pk - x) and G(x)
= (a -x)( p3 - x) ... (Pk - x) respectively. Simplify the resulting expression, noting that G(a)
= 0 and (p- a)F(a)= f (a).
A-3.
Since the integral converges for -I <k <5, one can consider Ik to be defined on this open
interval. Letting x = I/ t, one finds that
0 C k t -dt 0X t4-kdt
Ik = --
t-6pt Jo1)4-
=f
Then
Ik (Ik+I4-k)/2 [(Xk+ X4 -)/ 2]dx > x

rXo I2dx
4
since (xk+X4-I)/2> xkx k =X2 by the Arithmetic Mean-Geometric Mean Inequality.
Thus Ik is smallest for k = 2.
103
A4.
(a) The defining property with [w,x,y,z]=[a,b,a,b] and the hypothesis B(a,b)-c give us
B(c,c)=B(B(a,b),B(a,b))=B(a,b)=c.
(b) The defining property with [w,x,y,z]=[a,b,x,x] and B(a,b)=c give

B(c, B(x, x)) = B(B(a, b), B(x, x)) = B(a,x).
Then using the result in (a) and [w,x,y,zI=[c,c,x,xj, one has
B(c, B(x,x)) = B(B(c,c), B(x,x)) = B(c,x).
Together, these show that B(a, b) = c implies B(a, x) = B(c, x) for all x in S.
(c) An easy way to obtain a bypass with property (i) is to let S be a Cartesian product I X J and
to define the operation B by
B((ij), (h,k)) = (ik).
Properties (ii) and (iii) will hold if I and J, respectively, have more than one element. Except for
notation, every bypass is obtained this way.
Tables with S= {a,bc,d} and with S= (u,v,w,x,y,z) follow:
aorc bord uorx vory worz
aorb a b uor vorw u v w
cord c d xoryorz x y z
A-5.
Let g(x) = ln[(sin x)/x] = ln(sin x) - In x. Then
2
g"(x)=-csc x+ 12 = 12 1 for O<x <si
x2 X sin 2 X <0 fr<<
since x > sinx for x > 0. Thus the graph of g(x) is concave down and hence
~~Xi
-n g(xi) < g i-n g(x),
or X g(x;) < ng(x). Since e' is an increasing function, this implies
fl sinX, = exg(i) < g(e (= smx)

i-I XiX
A-6.
For a set (p, .pn} of points in the plane, let e* be the number of pj one unit from pi. Then
E = (e, + . + e,)/2 is the number of pairs with unit distance. Let C, be the circle with center at
p and radius 1. Each pair of circles has at most 2 intersections, so the C1 intersect in at most
2( 2 ) = n(n - 1) points. It suffices to treat the case in which each e, > 1.
The point p occurs (2 ) times as an intersection of Cj. Hence

2
(A) n(n-1) > (2)= e,(ei-1)/2 > (1/2) .
1(e,-)
In (A) and what follows, all sums are over i= 1,2,...,n. Using the Cauchy-Schwarz Inequality
SOLUTIONS: THE THIRTY-NINTH COMPETITION ]05
and (A) one has

[ 2 (ej-)]2 < I][(e*- ])2 ] <n2n(n-1) < W.
2
Hence X(e, -1) < V2 n3/ and so
E = (I e,)/2 < (n + V2 n3 / 2 )/2 < 2n3 /2 .
Let f (n) be the maximum number of pairs with unit distance apart from a set of n points in a
plane. Then f(l) = 0, f (2) = 1, f(3) = 3, and f(4) = 5.
Suppose we have an array of n > 2 points which realizes f(n) pairs unit distance apart. Suppose
each point has at least k points (in the set) unit distance away and that one of them, namely x(,
has exactly k points unit distance away. Let these points be x,1,..., Xk.
For i = 0,1. k, let C, be the circle with radius 1 and center x,. For i > 0, C, goes through x(
and at most two other x;. Therefore, there are at least k -3 points of the array other than
xo,..., x, on C,. Any one of these k - 3 points can appear on at most one other C, since there are
only two unit circles through x 0 and another point. Hence
k( 3) k( k -1)
2 2
Thus k is such that the triangular number k(k -1)/2 is less than n; we let the largest such
integer k be k,,.
The array with x0 removed has at most f(n- 1) pairs unit distance apart; therefore f(n) <
f(n - 1)+ k,,. By repeating this argument we find that f(n) < k2 + k3 + * + k,,.
The definition of kn implies that k. = t for (2) < n (t1). From this it follows that k,, <1
+ r2( n -1) . Hence
f(n)<k 2 +k,+ - +kn
• n -1 +2( F +V
V+ + )n-
• n -1+ vJ dx
= n -1+22(n 3/2 -1)/3
< n +.95n3/2 < 2n3/2

B-i.
The area is the same as for an octagon inscribed in a circle and with sides alternately 3 units and
2 units in length. For such an octagon, all angles measure 3ir/4 and one can augment the
octagon into a square with sides of length 3 + 2 V2 by properly placing a V2, V2, 2 isosceles
right triangle on each of the sides of length 2. Hence the desired area is
(3 + 2V )2 - (40 *V2 /2) = 13 + 12V .
A second solution follows. Let r be the radius of the circle and let a and i8 be half of the
central angles for the chords of lengths 3 and 2, respectively. Then 8a+8,8=27r and so
,8= (X/4)-a. Also
I
-32r=sma, -=sin,8=sin
r
.7T \
S-aJf
cosa-sina
2I ,
2 2r I cosa-sina cota-I
3 3 r V sina V/2
Now cota=(3+2V2 )/3=[(3+2V2 )/2]/(3/2) and hence the distance from the center of the
circle to a chord of length 3 is h3 = (3 + 2 V )/2. Similarly the distance to a chord of length 2 is
h2 -(2 + 3 V2_ )/2. Finally, the desired area is
4(3h3 +2h2 )/2=(9+6V )+(4+6V'1 )=13+12V'L
B-2.
Let S be the desired sum. Then
S = 0, I 2 (l-
2n rn- n+ [m m+n+2 n
Hence
"-I(n n 2+) . [ 2I n+ 2 k k_~+ I k+n

2=
n0 I - n I ) lirnI
I +2)
ll-+ I
unI [i(i+.
(h h+2) i 2
+~.+~~+ h)] ) ) ~
+ h(h+ I +I I+ I+( 2 h+)h+
- 6+3+2 + 12+6+4+3
I +
+1+-+..
1 1 \ +1( 1 -+--+
I
63 + 2 12 45+ 3644 I3 46
11I 25 1II ( II1 7
6 +24+ 32 I 34) 2
Thus S =7/4.
B-3.
Clearly, xI = -1, X 2 An easy induction shows that each Q. is positive for x > 0. Hence
x, < O, if Q. has zeros.
Assume inductively that xI <x 2 < <X2 -,i<X 2 r. Then Q2,-

0 (x)>0 for X>X2--l. In
particular, Q2. - I(X2-) > 0. Hence
Q2n+ (X 2
-)= Q 2
.(X 2
-)+(m+ l)X2mQ2r-l(x 2
.)
=(m+1)x rnQ
2 2 -,I(X 2-)<0.
SOLUTIONS: THE THIRTY-NINTH COMPETITION 107
This implies thai Q 2 ,+1 (x)=O for some x>x2 ., i.e., x2 +,1 >x2,. Similarly, one shows that
X2m+2 >X 2 .+1
Let a = - I/(m + 1). Using the given recursive definition of the Q.(x), one finds that
Q 2 .+ 2 (a) =Q 2 +1 (a) -Q 2 ,(a)= - Q2 _1 (a).
Hence at least one of Q2 .+ 2(a) and Q2,- I(a) is nonpositive. Thus either X2m+2 >a or x2 ,- 1 >a.
But each of these implies that both X2 ,, 2 > - I/(m + I) and X2 .+3> -I/(m+ 1). It follows that
-2/n x <0 for all n and then that limx. =0.
B-4.
Clearly (1, 1, 1, 1) is a solution. Thinking of xI, X2 , X 3 as fixed, the equation is quadratic in X4

and one sees that the x4 of a solution can be replaced by x4= xIx 2 + xIx3 + x 2 x 3 - x 4 to obtain a
new solution when X47'X 4. Also, the xi may be permuted arbitrarily since the equation is
symmetric in the x;. Thus we may assume that x 4 <m=min(x1 ,x2 ,x 3 ). Also assume that each
x > 1. Then x4' > 3m2 -m >m. This implies that one can start with the solution (1, 1, 1, 1) and
through repeated use of the procedures stated above obtain a solution with each x, an integer
greater than N.
B-5.
The solution is very easy if one knows that the Chebyshev polynomial C(x) = 8x 4 -8x 2 + 1=
cos(4Arccosx) has the largest leading coefficient of all fourth degree polynomials f(x) satisfying
-I < f(x) < I for - I <x < 1; then one lets P(x) = [C(x) + 1]/2 and has 4 as the largest A.
Without this information, one can use various substitutions to change the problem into
equivalent ones of maximizing A in simpler functions satisfying conditions over intervals. With
Q(x) - [P(x) + P(- x)]/2, the condition becomes
0< Q(x)=Ax4 + Cx2 +E 1 over -I x~ 1.

Letting x 2 -y, this becomes
OR(y)=Ay2 +Cy+E<l over O<yl.
Letting y =(z + 1)/2 and S(z) = R [(z + 1)/2], one has
0<S(z)=(A/4)z2 +Fz+G'l over -I<zdl.
With T(z) - [S(z) + S( - z)]/2, one obtains
0<(A/4)z2 +G<l over -I z<1.
2
Finally, letting z w, it becomes
0<(A/4)W+Gd over O<w<l.
Now it is clear that the maximum A is 4 and that this maximum is achieved with G = 0, i.e., with
T(z) -z 2 R(y) = (2y - 1)2, Q(x) = 4x 4 -4x2 + 1.
B-6.
2
Let ah ( Cck)/h. Clearly, O ah < p. We now prove that (-
I= Iah) '2p2'_,(hah), which is
equivalent to the assertion of the problem, by induction on n.
For n = 1, one has a2 < pal < 2pal as required. Suppose the inequality established for n = m.
Then
m+la,2= m h2+ m
( E- ah= h 2a.+1 E_ a,+a.+l
(h-I / h-I h-I
9 2p i (hah)+2am,+lpm+2pa.+,
h-I
r m ,n+
42p (m + I)am+I+ j (ha,) 2pI I (hah),
h-a . h-e
as desired.
THE FORTIETH WILLIAM LOWELL PUTNAM
December 1, 1979
A-l.
We see that n = 660 and that all but one of the ai equal 3 and the exceptional a, is a 2 as follows.
No a, can be greater than 4 since one could increase the product by replacing 5 by 2 *3, 6 by 3 -3,
7 by 3 4, etc. There cannot be both a 2 and a 4 or three 2's among the a, since 2 4 < 3 3 and
2 -2 < 3*3. Also, there cannot be two 4's since 4 4 < 2 3 3. Clearly, no a, is a 1. Hence the a,
are 3's except possibly for a 4 or for a 2 or for two 2's. Since 1979-3-659+2, the only exception
is a 2 and n=660.
A-2.
The condition is k > 0. If k > 0, one sees that f(x) = satisfies f(f(x)) kx9 . For the
9
converse, we note that f(f(x)) = kx for all real x with k #0 implies that f takes on all real values
since kx9 does and implies that f is one-to-one since f(a) =f(b) leads to ka'=f(f(a)) = kb9 and
hence a = b. But a continuous one-to-one function f from the real numbers R onto itself must be
strictly monotonic. Also, if f is monotonic, either always increasing or always decreasing, f(f(x))
will always be increasing and so cannot equal kx 9 if k <0.
A-3.
The condition will be seen to be that x,=x,=m for some integer m. Let r,,=Il/x,,. Then
rn=(2x.-2-xn-,)/x.-2xn-1 =2rn-1 -rn-2 and the rn form an arithmetic progression. If xn is a
nonzero integer when n is in an infinite set S, the r. for n in S satisfy - I < r, < I and all but a
finite number of the other r. are also in this interval due to being nested among rn with n in S;
this can only happen if the r. are all equal since the terms of an arithmetic progression are
unbounded if the common difference rt,, - r, is not zero. Equality of the r, implies that
xI= x2 = m, an integer. Clearly, this condition is also sufficient.
Alternatively, let the r, form the arithmetic progression defined above. If xi and x, are integers
with i=#j, then r, and r. and the common difference (r, - rj)/(i -j) are rational. It follows that r1
and r2 are rational and hence that r, =a/q and r2 =(a+d)/q with a, d, and q integers. Then
x,,= l/r. =q/[a+(n- I)d]. Since q has only a finite number of integral divisors, x. can be an
integer for an infinite set of n's only if d=O. This gives the same condition as in the first
solution.
An easy induction argument shows that
xI X2
(n -1)xI -(nx-2)x
(x -x 2 )n +(2x2 x) ' n= 3,4,5
109
In this form we see that x,, will be an integer for infinitely many values of n if and only if
xl = x2 = m for some integer m.
A-4.
There are a finite number (actually n!) of ways of pairing each of the red points with a blue
point in a 1-to-I way. Hence, there exists a pairing for which the sum of the lengths of the
segments joining paired points is minimal. We now show that for such a pairing no two of the n
segments intersect.
Let red points R and R' be paired with B and B', respectively, and assume that segments RB
and R'B' intersect. The triangle inequality implies that the sum of the lengths of these segments
exceeds the sum of the lengths of segments RB' and R'B. Then interchanging B and B' would
give us a new pairing with a smaller sum of segment lengths. This contradiction proves the
existence of a pairing with nonintersecting segments.
A-5.
29
Let f(x) = X- lOx 2 + x -25. Then the table
x 1 2 3 5 6
f(x) -5 1 - I -5 5
shows thatf(x)=O has three real solutions a, b, c with I <a<2,2<b<3,5<c<6. The number
of integers that the set (1,2,..., n} has in common with S(a), S(b), and S(c) is [n/a], tn/b],
and [n/c], respectively. Since
+a+ >- 3 +-
one sees that
lim +0
n a [b ][c
and hence that an infinite number of positive integers appear in more than one of S(a), S(b),
S(c). This implies that some pair of these sets must have an infinite intersection.
A-6.
For k=0, I__.2n- I let Ik be the open interval (k/2n,[k+ 1]/2n). Among the 2n intervals 'k
there exist n not containing any of the pAand we place an x; at the center of each of these n
intervals. Let Jx.-pA =d, and
( 3 5 2 n- I)
For fixed i, the d,, satisfy d,, > 1/4n, at most two of them do not satisfy dj > 3/4n, at most four
do not satisfy dij > 5/4n, etc. Hence
n I 4n
zl -d-, <2Y 1+2h =B.
(This inequality can be improved.) Thus we have
nI(nI) I)
j- l - i j- j- li
So clearly there is a value of j for which 17-, (l/dj;) < B and the rj for such aj can serve as the
desired x.
SOLUTIONS: THE FORTIETH COMPETITION III
B-1.
We assume that there is such a common normal and obtain a contradiction. This assumption
implies
a-c
a sc -= cosh c = sinh a. (I)
cosh a -sinh c
Since cosh x >0 for all real x and sinh x >0 only for x >0, (I) implies a >0. Using the fact that
sinh x < cosh x for all x and (1), one obtains
sinh c < cosh c = sinh a < cosh a.
This, a>O, and the fact that coshx increases for x>O imply that c<a. Thus the leftmost
expression in (I) is negative and cannot equal coshc. This contradiction shows that no common
normal exists.
B-2.
Let u = bx + a(l - x); then the definite integral becomes
l(t) = I- I u'du (lb+ )(-a )

W-aa (I +t)(b -a)~
Using standard calculus methods for evaluating limits of indeterminate expressions, one finds
that
[J(t)]"'/ el(bb/aa)l/(b ) as t -- 0.
B-3.
Let r = (m - 1)/2. We show that q(x) =p(x) + k is irreducible over F for r elements k of F. Since
m is odd, the characteristic of F is not 2, 1 + I =2#0, 2 -lb is an element h of F, the 2r+ I
elements of F can be expressed in the form O,f,, -f .... f,, -f,, and {Of .f, 2 } is the set of the
r+ I distinct squares in F. Now
q(x) (x + h) 2 _(h2'- c-k)
is irreducible over F if and only if it has no zero in F, i.e., if and only if h2 - c - k is not one of
the r+ I squares f2 in F. Hence k must be one of the r elements left when the r+ I elements of
the form h 2 - c -f2 are removed from the 2r + I elements of F.
B-3. Alternate solution
The polynomial x2 + bx + d is reducible if and only if there are elements s and t in F such that
st= d
s+t=-b,
or, equivalently, if and only if there is an element s in F such that - s(s + b) = d. For s E F,
define f(s) = s(s + b). We have just seen that the number of reducible polynomials of the form
x2 + bx + d is equal to the number of elements in the image of f. Notice that f(s) = f(t) if and
only if either t = s or t = - s -b. Because F is not of characteristic 2, s = - s - b for only one s
in F. It follows that the image of F has 1 + (m -1)/2 distinct elements, and therefore the number
of irreducible polynomials of the form x2 + bx + c + k is m -(1 +(m -1)/2) = (m - 1)/2.
B4.
(a) Trial of e- shows that y = e 3. satisfies the homogeneous equation. Trial of a polynomial
xd+... shows that d must be 2 and then trial of x 2 +px+q shows thaty=x 2 +x is a solution.
Any linear combination he 3, + k(x 2 + x), with at least one of the constants h and k not zero, is
an answer.
(b) It is easy to see that y = 2 satisfies the nonhomogeneous equation and hence f(x) is of the
form 2+he3 ,+k(x 2 +x). Nowf(O)= l gives us2+h= I orh= -1. Then [f(- I)-21[f(l)-61= I
leads to
-e- 3 (2+2k-e3 -6)=l, (2k-4)e 3 =O, k=2.
Hence f(x)=2 - e3, + 2(x2 +x), f(-2)-6-e -6 , f(2) = 14-e6 . Therefore we let a-6, b - 14, and
c= 1.
We note that if one stops guessing after obtaining an answer g(x) to (a), the standard
substitutions y = g(x)z followed by z' = w will reduce the nonhomogeneous equation to a linear
equation which can be solved by a well-known method.
B-5.
A support line for C is a straight line touching C such that one side of the line has no points of
C. There is a support line containing (0, 1); let its slope be m. If m > 1/2, the part of the area of
C in the fourth quadrant is no more than I and we are done. Similarly, if m < - 1/2. So we
assume that - 1/2 <m < 1/2 and assume the analogous facts for support lines containing (1,0),
(0, - 1), and (- 1,0). At least one of the angles of the quadrilateral formed by these four support
lines is not acute; we may take this angle a to be at a vertex (hk) in the first quadrant. Then
a > r/2 implies that h + k < 2, and this in turn implies that the area of C in the first quadrant
does not exceed 1. Hence A(C) < 4.
B-6.
Let X=(x_,...,x.) and Y=(y ..... y.). Also let a+bi be either square root of Zl2+ +Z, Then
ab=X-Y=xIy+ +xy,, and
2 2 2
a2- b = IIXIl - 11y I =I(X2+ +X2)-(y2+ +y.2)
The Cauchy-Schwarz inequality tells us that IX YI < IlXII t Y11 and hence lal *IbI <
llXlj-11 Y1. Therefore, the assumption that IaI>lIXii would imply that IbI<IIYII. This and
a2 =_1XI1 2-11Y1y2 +b 2 would yield a2 <IIXII and thus the contradiction IaI<IIXII. Hence the
assumption is false and r =aI <II X II. Since lIXI12 < (IxI+ * + Ix1) 2, this implies the desired
,r<Ix'l+ + Ix,,l.
THE FORTY-FIRST WILLIAM LOWELL PUTNAM
December 6, 1980
A-1.
We show that g(x) = X 2 + bx + c - (1/4). The equation of the tangent to the given parabola at
Pj = (j, yj) is easily seen to be y = L>, where L, = (2j + b)x-j2 + C. Solving y = Lj and
y = Lj+, simultaneously, one finds that x = (2j + 1)/2 and soj = (2x - 1)/2 at Ij. Substituting
this expression forj into L, gives the g(x) above.
A-2.
We show that the number is (I + 4r + 6r2 )(l + 4s + 6s2 ). Each of a, b, c, d must be of the
form 3m7' with m in 0,1,. . r} and n in {0,1I. s). Also m must be r for at least two of the
four numbers, and n must be s for at least two of the four numbers. There is one way to have
m = r for all four numbers, 4r ways to have one m in (0,1, ..., r-I ) and the other three equal
to r, and (4 )r2 = 6r 2 ways to have two of the mar's in (0, 1_. r - I) and the other two equal to
r. Thus there are I + 4r + 6r2 choices of allowable m's and, similarly, I + 4s + 6s2 choices of
allowable n's.
A-3.
Let I be the given definite integral and F2 = r. We show that I = 7r/4. Using x = (iT/2) - u, one
has
I JO1 odu f/2 tan' u du
4/2 I + COt' U JO, tan' u + I
Hence
r0/ I + tan'x
21=I~ /J I+ , dxJ=I dx-= r/2 and I = r/4.
A-4.
(a). LetSbe thesetof the 10"' real numbersr + sFJ + tF3 witheachofr,s, tin (0, I,. 106- I}
and let d = (I + vf + J) 106 . Then each x in S is in the interval 0 < x < d. This interval is
partitioned into 1018 - I "small" intervals (k - I)e < x < ke with e = d/(10' 8 - 1) and k taking
on the values 1,2 . , 10'8 - 1. By the pigeonhole principle, two of the 10 8 numbers of S must be
in the same small interval and their difference a + bFy + c/3 gives the desired a, b, c since
c < 10-.
(b) Let F. = a + bF2 + cF3 and F2 , F3 , F4 be the other numbers of the form a ± b; -_clr.
Using the irrationality of F2 and F3 and the fact that a, b, c are not all zero, one easily shows
that no F, is zero. (The demonstration of this was Problem A- I of the 15th Competition, held on
March 5, 1955. For the proof, see page 402 of The William Lowell Putnam Mathematical
Competition Problems and Solutions: 1938-1964, published by the MAA, or see this MONTHLY, 62
113
(1955) 561.) One also sees readily that the product P = FF 2 F3 F4 is an integer. Hence IP 1.
ThenIF. l/lFF 3 F41 > 10- 2 1 since IF, < 7 and thus I/I F,I > O 7 for each i.
A-5.
Let Q= -P P" + p ~V- Using repeated integrations by parts, the equations of the given
system become
J P(t)sin t dt =-Q(x)cos x + Q'(x)sin x + Q(O)= 0,
0
I xP(t)costdt = Q(x)sinx + Q'(x)cosx - Q'(0) = 0.
These imply that

Q(x) = Q'(O)sinx + Q(O)cosx. (E)
Siaice P and, hence, Q are polynomials of positive degree and the right side of (E) is bounded,
equation ( E) has all of its solutions in some interval Ix I - M. In such an interval, P(x)sin x has
only finitely many zeros and fox P(t)sin t dt = 0 has at most one more zero by Rolle's Theorem.
Q.E.D.
A-6.
We show that u = I/e. Sincef'-f = (fe -x)'ex and ex I for x 2 0,
f'I f -fIdx= 'I(fe -)'eildxj '(fe -)'dx= [fe X] l/e.

To see that I/e is the largest lower bound, we use functions f,(x) defined by
fa(x)=(eao-/a)x for 0<x a, f.(x)=ex-' for a x l.
Let m = ea- '/a.Then
IIt(x) -fa(x )l dx = al - mxldx = m a - 2 ea-'
As a - 0, this expression approaches I/e. The function f,(x) does not have a continuous
derivative, but one can smooth out the corner, keeping the change in the integral as small as one
wishes, and thus show that no number greater than I/e can be an upper bound.
B-i.
The inequality holds if and only if c 2 1/2. For c ; 1/2,
ex+ e a X2 X2"lA2/2cx
2 2~-!<1
' '! e <ex
forallxsince(2n)!22"n! forn=0,1,....
Conversely, if the inequality holds for all x, then
2 2
ecx'-H(e'+e-x) (I +cx + )-(I + Jx + ,) I
Ohm 2 .lim -C--
x-o xx-
x x2 2
and soc > 1/2.
B-2.
(a) v = 7. The seven vertices are VO= (,0, 0), VI = (11, 0, 0), V2 = (0, 9, 0), V3 (0, 0, 8), V4 =
(0, 3, 8), V5 = (9,0, 2), and V6 = (4, 7, 0).
(b) e = I1. The eleven edges are VOV
1, VOV 2, VOV 3 , V1V5, VV6 , V2V4, V2V6, V3V4, V3V5 , V4 V5, and
V4 V6 .
SOLUTIONS: THE FORTY-FIRST COMPETITION 115
.y
(c) The desired (b, c) are those with b + c = 2 and 2/3 < b < 1. Let L(x, y, z) = bx + cy + z.
Since L is linear and (2,5,4) is on edge V4 V6 , the maximum of L on S must be assumed at V4 and
at V6 and the conditions on b and c are obtained from L(O,3,8) = L(4,7,0) a L(x, y, z), with
(x, y, z) ranging over the other five vertices.
B-3.
We show that un > 0 for all n > 0 if and only if a > 3. Let Au. = u - u.. Then the recursion
(i.e., difference equation) takes the form (I - A)u. = n2 . Since n2 is a polynomial, a particular
solution is
u = (I -A)-n2z= (I + A+A2 + *--)n 2
=n 2
+ (2n+ 1)+2=n2 +2n+3.
(This is easily verified by substitution.) The complete solution is un = n2 + 2n + 3 + k-2", since
v. = k 2' is the solution of the associated homogeneous difference equation v,,, - 2v. = 0. The
desired solution with uo = a is u, = n2 + 2n + 3 + (a- 3)2". Since limn-.[2"/(n2 + 2n + 3)]
= + oo, un will be negative for large enough n if a - 3 < 0. Conversely, if a - 3 a 0, it is clear
that each u.> 0.
Alternatively, one sees that uo = a and u, = 2a and one can prove by mathematical induction
that
n-I
=2"2 - 2"--kk2 for n 2.
k=-
2
Hence un > O for n 00 if and only if a > In-' 2 -kk and this holds if and only if a a L,
wher
LS 2-l1-k V. Let D mendd.Then for Ix I < 1,
k-O
(I -X Iy = Xk
k=O
D(l -x)1 = (I -x)- = i kx k |

k=l
D(l - x) 2= 2(1 - x)-3 = 2 k(k - l)xk-2

k=2
Let g(x) = 2x 3 ( - x)-3 + X

2
(l -X)-2. Then L = g(l/2) = 3 and the answer is all a 3.
B4.
The result we are asked to prove is clearly not true if IXi < 10. Hence we assume that IXI a 10 or
that the Ai are distinct, which implies that IXI > 10.
LetX {xl,..., x,}, with m = IXI, andlet n, be the numberofj such that, is in A1. LetN
be the number of ordered pairs (i, j) such that x, is in Ai. Then
N=n 1 +n 2 + ---+n,= [AIl + IA2 1 + +IA ,661
1 > 1066(m/2)=533m.
Hence one of the ni, say n1, exceeds 533.
Let B,..., B, be those sets Ai not containing xl and Y = {X2 , x 3 ,... I Xml. Then s = 1066-
n I 532 and each IBj > IYJ/2. We can assume that x2 is in at least as many Bj as any other x,
and let C, ... , C, be the Bj not containing x2 . As before, one can show that t < 265.
We continue in this way. The 4th sequence of sets D,_., D. win number no more than 132.
The numbers of sets in the 5th through 10th sequences will number no more than 65, 32, 15, 7, 3,
and 1, respectively. Thus we obtain the desired elements xl,..., x, 0 unless X has fewer than 10
elements.
B-5.
The answer is I > t (or 0 < t 1). The product fg of two nonnegative increasing continuous
real-valued functions has the same properties. Using the fact that 0 < a < c and 0 - b < d imply
ad + bc < cb + cd, one shows thatfg is convex when and g are convex. The functionf(x) = x is
in S, for all t. If S, is closed under multiplication, x2 is in S, and so 2/9 = I - 2(4/9) + (1/9) >
t[4/9 - 2(1/9)] = 2t/9 or I t.
The following argument proves the converse. Let t c*[0,1]. For a real valued function h defined
on [0,1], let E(h) = [h(l)- 2h(2/3)+ h(1/3)]- t[h(2/3)- 2h(1/3)+ h(0)]. Suppose that f and g
are in St, so E(f)>0 and E(g)>0. Then E(fg)=g(2/3)E(f)+f(l/3)E(g)+
[f(l)- f(1/3)][g(1)- g(2/3)]- t[f(1/3)- f(O)][g(2/3)- g(O)]. By convexity, f(1)- f(l1/3)>
2[f(1/3)-f(O)], and g(l)-g( 2 / 3 ) - 2[g(2/ 3 )-g(O)]. If t < 1, this implies E(fg)> O 0sofg is in
S..
B-6.
Let Fd(x) =1=d G(d, n)x¶. Then F1(x) = 1"=Ix'ln and F(x) = 2=ox". One sees that
Fd(x) = dFd-,(x)F^(x) by finding the coefficients of x"-' on both sides and using nG(d, n) =
dy4'=d G(d - 1, i - 1). Then an induction gives us Fd(x) = [F,(X)]d. Now, for I < d < p, the
coefficient G(d,p) of xP in Fd(x) is the coefficient of xP in [Y. p-d+1xt/n]d, and hence
G(d, p) = s/t with s and t integers and t a product of primes less than p.
THE FORTY-SECOND WILLIAM LOWELL PUTNAM
December 5, 1981
A-i.
We show that the limit is 1/8. Let T(m) = I + 2 + + m = m(m + 1)/2, [x] denote the
greatest integer in x, h = [log 5 n], and e, be the fractional part (n/5') - [n/5'] for I < i • h.
Then
E(n) = 5T([n/51) + 52T([n/52]) + + 5hT([n/5h])
2E(n) = 5([n/5]2 + [n/51) + 52([n/52] + [n/52]) + * + 5h([n/5h]2 + [n/5"])
=n2 _
2eln
2 5e 2 + n-e + -+ 5. n2
- _
2ehn
+ + e
2 +n
- e
i
5( 522 - 5 5+) ) k52h h 5 h

E~n)
I I )+ h eA +2eh +e
n2 2 c2 hI nn
+3 +(e-e 1 ++ . +5 n~~h
5(e'2 - e,) + 5h(e' - eJ
+ h
2n2
Since 5h • n < 5h+ I and 0 e, < 1, one sees that h/n-0 and E(n)/n 2
< _ 1/8 as n - o.
A-2.
For any numbering, one can go from the square numbered I to the square numbered 64 in 7 or
fewer steps, in each step going to an adjacent square; thus (64- 1)/7 = 9 is a C-gap. It is the
largest C-gap since with coordinates (a, b), I < a < 8 and I < b < 8, for the squares we can
number (a, b) with 8(a - 1) + b and thus find that no number greater than 9 is a C-gap.
A-3.
Let G(t) be the double integral. Then
lim [G(t)/e'] = lim [G'(t)/e']
I-00 1-00
by L'Hbpital's Rule. One finds that
G'(t)=le-dx+fe et dy= 2 e e dx.

foXx-t Joy Y-t Jo x -I
Then using e' = e'[I + (x - t) + (x - t)2/2! + ], one sees that e-'G'(t) - oo as t -p
since for sufficiently large t,
G'(t)ffex II d Jr
f e dy>f} eI d>(I-e ')logt.
2e' f x - t
117
A-4.
Set up coordinates so that a vertex of the given unit square is (0, 0) and two sides of the square are
on the axes. Using the reflection properties, one can see that P escapes within T units of time if
and only if the (infinite) ray from PO, with the direction of the first segment of the path, goes
through a lattice point (point with integer coordinates) within T units of distance from PO. Thus
N(T) is at most the number L(T) of lattice points in the circle with center at PO and radius T.
Tiling the plane with unit squares having centers at the lattice points and considering areas, one
sees that
N(T) < L(T) 7T
r[T + (r2 /2)] 2
2
Hence there is an upper bound for N(T) of the form 7T + bT + c, with b and c fixed. When just
one coordinate of PO is irrational,
N(T) = L(T) 2 vr[T- (V2/2)] 2 .
This lower bound for N(T) exceeds aT 2 + bT + c for sufficiently large T if a < 7r; hence so is the
desired a.
A-5.
We show that Q(x) has at least 2n - I real zeros. One finds that Q(x) = F(x)G(x), where
F(x) = P'(x) + xP(x) = e _2e12 x2P(x)]', G(x) = xP'(x) + P(x) = [xP(x)]'.
We can assume that P(x) has exactly n zeros a, exceeding I with I < a < a 2 <.. < a. It
follows from Rolle's Theorem that F(x) has n - I zeros b, and G(x) has n zeros c, with
I < a, < b, < a2 < b2 < ... < bn -l < antI < cl < a, < C2 < a2 < .''' < c, < a..
If b, $ c, , for all i, the b's and c's are 2n - I distinct zeros of Q(x). So we assume that
bi = c.+I = r for some i. Then
P'(r) + rP(r) = 0 rP'(r) + P(r)
and so (r2 _I)P(r) = 0.Since r = bi > 1, P(r) = 0. Since a: < r < a,+,, this contradicts the
fact that the a's are all the zeros exceeding I of P(x). Hence Q(x) has at least 2n - I distinct
real zeros.
A-6.
Treating each point X of the plane as the vector A X with initial point at A and final point at X, let
L = (B + C)/2,M= C/2, and N = B/2
(be the midpoints of sides BC, AC, and AB). Also let
S = (2L + M)/3 = (B + C + M)/3, T= (2L + N)/3
=(B+C+N)/3,Q=2P- B, and R=3P-B-C.
Clearly Q and R are lattice points. Also Q $ P and R $ P since Q = P implies P = B and R = P
implies that P is the point L on side BC. Hence Q is not inside AABC and this implies that P is
not inside A NBL since the linear transformation f with f(X) = 2 X - B translates a doubled
ANBL (and its inside) onto AABC (and its inside). Similarly, P is not inside AMCL. Using the
mapping g(X) = 3X - B - C and the fact that R is not inside ALMN, one finds that P is not
inside ALST. Since the distance from A to line ST is 5 times the distance between lines ST and
BC, it follows that IAP1/IPE1 • 5. This upper bound 5 is seen to be the maximum by
considering the example with A = (0, 0), B = (0, 2), and C = (3, 0) in which P = (1, 1) = T is the
only lattice point inside AABC and IATI/ITEI = 5.
SOLUTIONS: THE FORTY-SECOND COMPETITION 119
B-I.
Let Sk(n) = lk + 2 k + * + n k. Using standard methods of calculus texts one finds that
S 2 (n) = (n 3 /3) + (n 2 /2) + an
and
S4 (n) 2 (n 5 /5) + (n 4 /2) + bn3 + cn
2
+ dn,
with a, b, c, d constants. Then the double sum is

18[S 2 (n)]2 = (2n 6 + 5n' + - - -) (2n6 + 6n5 + 5
lOnS 4 (n) - . ) = nn +
and the desired limit is -1.

B-2.
First we let 0 < a < b and seek the x that minimizes
f~)= - I -- I on a < x < b.
Let x/a = z and b/a = c. Then
f(x) = g(z) = (Z 1)2 + (c- 1)

Now g'(z) = 0 implies
3 2 2
Z4 -z + cz- C2 =(Z c)(z -Z + C) = 0;
the only positive solution is z = vA. Since 0 < a < b, c > 1, r > 1, and
g(l) = g(c) = (C 1)2 = ( _ )2(r + 1)2 2( - 1)2 g()

Hence the minimum of g(z) on I < z < c occurs at z = r. It follows that the minimum for f(x)
on a < x < b occurs at x = afb/a = Va-b. Then the minimum for the given function of r, s, t
occurs with r = V_, t = 4 = 2r, and s = Vr= rvr. These imply that r = C2 , s = 2, t = 2- .
Thus the desired minimum value is 4(2 - 1)2 = 12 - 8vr.
B-3.
As m ranges through all nonnegative integers,
n = (M 2 + m + 2)(m 2 + m + 3) + 3
takes on an infinite set of positive integral values. Let f(x) = x2 + 3. Examination of (f(m)} =
3, 4, 7, 12, 17,28, 39, 52, 67, 84, ... leads one to conjecture that
/(x)f(x + 1) =f[x(x + 1) + 3] =f(X2 + x + 3).
This is easily proved. Using this property and the above relation between m and n, one has
2
(n) =f(m2 + m + 2)f(m2 + m + 3) =f(m + m + 2)f(m)f(m + 1).
Thus pjf(n) with p prime implies that p1f(k) with k equal to rn, m + 1, or m2 + m + 2. Since
each of these possibilities for k satisfies k2 < n, the desired result follows.
B-4.
Let M = M(a, b, c) denote the 5 by 7 matrix (a,_) with
a,, a, a12 = a 33 = a44 = a55 = b, a16 = a2 7 = c,
and a,, 0 in all other cases. Then the set V of all such M (with a, b, c arbitrary real numbers) is
closed under linear combinations. Also, M(O,0,0), M(l,0,0), M(O,0, 1), M(O, 1,0), and M(l, 1,0)
have ranks 0, 1, 2, 4, and 5, respectively. But no M in V has rank 3 since b * 0 implies that the
rank is 4 or 5 and b = 0 forces the rank to be 0, 1, or 2.
B-5.
If n has d digits in base 2, 2 d I < n and so
5 d 5 I + log2 n. B(n)
This readily implies that Er,-.[B(n)/n(n + I)] converges to a real number S. Hence the
manipulations below with convergent series are allowable in the two solutions which follow.
Each n is uniquely expressible as n0 + 2n, + 22 n2 + with each n, in (0, 1) (and with
n, = 0 for all but a finite set of i). Since
I + 2 + 22 + . + 2'- ' = 2 - 1,
one sees that n, = I if and only if n is of the form k + 2' + 2'+j with k in {0,,..., 2' - 1) andj
in (0, 1, 2, ... ). Thus
x 1 ) n,
S n-I n (n+ l),z0"

X x 2'- 1 __
t-Gj-O k-0 (k + 2' + 2 ''j)(1 + k + 2' + 2'+j).

Using l/s(s + I) = I/s -/(s + 1), the innermost sum telescopes and
i oj-0 O[2 + 2j) 2i(2 +2j)] ,-02J ( ) J
Since it is well known that I - 2 + - + = In 2,
S = (( 2- ')n2
) = 21n2 = ln4
and eS is the rational number 4.

Alternatively, we note that B(2m) = B(m), B(2m + 1) = I + B(2m) = I + B(m).
Then
X B(n) _ B(2mn+1)
1 + B(2m)
n= I (n + I) ,.n0 (2m + 1)(2m + 2) r- 2m(2m + 1)
x0 I + B(m) .0 B(m)
m-C (2m + 1)(2m + 2) + M,--, 2m(2m + 1)
=m-
- (2
+
)(
1)(2mI+2
+2)
+ E
m-
B(m)[
2m(2m + 1)
+
(2m + 1)(2m + 2)1
I2+I 00 B(m) _=1n2+S
2 m (m +1 ) 2
Hence S/2 = In 2, S = In4, and exp(S) is the rational number 4.
B-6.
Let L - L( P) be the perimeter of P. One sees that H( P) consists of the region bounded by P. the
regions bounded by rectangles whose bases are the sides of P and whose altitudes equal 1, and
sectors of unit circles which can be put together to form one unit circle. Hence
F(P) = (L/2) + L + 1r = 1T+ 3L/2.
If A and B are two consecutive vertices of P, the contribution of side AB to the double integral
SOLUTIONS: THE FORTY-SECOND COMPETITION 121
I is double the area of the region (of the figure) bounded by the unit semicircles with centers at A
and B and segments CD and EF such that A BCD and A BEF are rectangles and IA D I = I = IA Fl.
D C
FiG 1.
One doubles this area because there is a symmetric region bounded by CD, EF, and the other
halves of the unit circles centered at A and B. The overlap of the two regions counts twice. By
Cavalieri's slicing principle, this contribution of side AB to I is 4 times the length of AB. Hence
I = 4L and
I _ 4L 8
F(P) 1r + 3L/2 3 + (21T/L)'
One can make L arbitrarily large (e.g., by letting P be a triangle with two angles arbitrarily close
to right angles). Hence the desired least upper bound is 8/3.
THE FORTY-THIRD WILLIAM LOWELL PUTNAM
December 4, 1982
A-1.
Let T consist of the points inside or on the triangle with vertices at (0, 3), ( -1,4), (1,4) and let U
be the set of points inside or on the triangle with vertices at (0, 0), ( - 4,4), (4,4). Then T and V
overlap only on boundary points and their union is U. The centroids of T and U are (0, 11/3) and
(0,8/3), respectively. The areas of T, V, and U are 1, 15, and 16, respectively. Using weighted
averages with the areas as weights, one has
I 0 + 15x = 16 0, I -'3+ 15y = 16
It follows that x = 0, y = 13/5.
A-2.
Since x = log.2 > 0, Bn(x)= Ix + 2x + + n' n*n and
5 2
B.(log,,2) n _nll g 2
2
(nlog2 n) (nlog2n) n(1og2n)
As LE [2/n(log n) converges by the integral test, the given series converges by the compari-
son test.
A-3.
r Arctan( vx ) -Arctan v- I rt x d
j dx Arctan( ux) dx
2 dudx =ffA Xdxdu

1 + (.Xu) I 4) 1 + (xu)
- du = -ln 7r.
.1U 2 2
Let
f(a) Arctan(ax) -Arctanx dx.

o X
Then
f'a
fo xl+a
(
x2x dx= aa J dy
+ l ia
a 2'
122
SOLUTIONS: THE FORTY-THIRD COMPETITION 123
Therefore f(a) = (77/2)]n a + C. But C = 0 since f(l) = 0. Hence f ( o) = (7r/2)ln 7T.
A-4.
The differential equations imply that
y3y, + z3z zy'-y'z' = 0
and hence that y + z is constant. This and the initial conditions give y 4
4 4
+ z4 = 1. Thinking of x
as a time variable and (y, z) as the coordinates of a point in a plane, this point moves on the
curve y 4 + z4 = I with speed
[(Y)2 + (Z,)2] 2 = (Z, + Y )
4 4
At any time, either or z is at least 2 and so the speed is at least ([(1)1/4]6)1/2. Hence the point
will go completely around the finite curve in some time L. As the speed depends only on y and z
(and not on x), the motion is periodic with period L.
A-5.
We are given that
a c bd-ad- bc
r= - - = bd > 0.
Thus bd - ad - bc is a positive integer and so r > I/bd. We may assume without loss of
generality that b < d. If b < d < 1983, r > 1983-2 > 1983 -3. Since a + c < 1982, if 1983 < b <
d, one has
a c 1982 1 1
1983 1983 1983 1983 3
The remaining case is that with b < 1983 < d. Then the d that minimizes r for fixed a, b, c is
I + [bc/(b - a)], where [x] is the greatest integer in x. This d is at most 1983b since b - a > I
and c < 1982 and thus
r 1 I
,Q 1983b 2 l983V
Hence we have the desired inequality in all cases.
A-6.
We disprove the assertion. Let y, = I /(n + I)ln(n + 1). Then E' (I- I)"+*y, converges to some
g > 0 since y. - 0 as n - x and y, > Y2 > *. Letix (- l)y/g. Then conditions (i) and
(iii) are satisfied. Let a0 , a,,... be positive integers to be made more definite later. Let bo = 0 and
bi+, = bi + 4a, for i = 0, 1.. . The bisection a is defined as follows:
u(n) = 2n - I-b, for bi < n < bi + 2ai,
a(n) = 2n - bi+, for bi + 2a, < n < b1+ 1.
Then 0 < a(n) < 2n and hence la(n) - nl < n. Thus
lo(n)o~n)-
- nl Ix.I <<
n - jx~ n
(n + I)ln(n + I)
which implies condition (ii). Let C(n) = (nxi and D(n) = £". jx.(,w Then
I a I a,
(A) D(b, + 2a,) - C(b, + 2a,) = E Yb,+2j + - E Ybi+2a,+2k-1
Since y2 + y 4 + Y6 + * diverges to + so by the integral test, the a, can be chosen large enough
so that the first sum in (A) exceeds I for each i. Then D(n) > I + C(n) for an unbounded
sequence of n's. Hence D(n) and C(n) cannot converge to the same limit.
B-I.
The smallest n is 2. Let D be the midpoint of side AB. Cut AAMB along DM. Then ABMD can be
placed alongside AADM, with side BD atop side AD, so as to form a triangle congruent to
A AMC. Since AAMB need not be congruent to AA MC in a general AA BC, there is no method
with n = 1.
B-2.
Let r -x= + y 2 , R(m, n) - ((x, y): m2 + n2 < x2 + y 2), and
I= fJ A(x,y)e-x -" dxdy.

-x -x
Let £ and E' denote sums over all integers m and over all integers n, respectively. Then
I = £J
' f
R(m, n)
e-x
2
-
2
dx dy
2 0J e-2 rdrdO
= £ £*f2 "[( ie-rI]znd9

2
0f
= 'j 2 "ie e- m2 2
dO
fo
= ££'Xe-2"2=_-(Ee--)(E'e- 2) = 1(2g - 1)2 .
B-3.
Let a(n) [=in [+I and b(n) = [V2n], where [x] is the greatest integer in x. For t in
(1,2,.. ., a(n)), there are t2 - I ordered pairs (c, d) with c and d in X = (1,2,..., n) and
c + d - t2 For t in (I + a(n),2 + a(n),..., b(n)}, there are 2n + I - t2 ordered pairs (c. d)
with c and d in X and c + d = t 2 . Hence the total number F(n) of favorable (c, d) is
a(n) b(n)
2
F(n) = £ (t - 1) + £ (2n + I - t2)
It t-I +a(n)
/a(n~)hn
= 2 £ t2) - bn t2 a(n) + [b(n) - a(n)](2n + 1)
2a(n)[1 + a(n)][l + 2a(n)] - b(n)[l + b(n)][l + 2b(n)]

6 6
-2(n + I)a(n) + (2n + I)b(n).
Since, = F n)/n 2,
lim (prnf) = lim F(n)/n3 /2
- 2 (9 )
ain
-! |---
lil
3 2 l (b(n)
_- lim (())-2
3
~ 2lm a(n)
lm ()+
21
2 lim
b(n)
()
2- I
=1 (V) 3-_
2+ +I4
22V = (C2 1).
SOLUTIONS: THE FORTY-THIRD COMPETITION 125
B-4.
Let P, = (no + k)(n2 + k) . (n, + k). We are given that POIPk for all integers k.
(a) PoIP- and PoIP, together imply PJ21(P_ PI) or (n 2n n2 )I[(nl I - )(n2 - 1)
(n2n- 1)].
No n, can be zero since P, L 0 for k sufficiently large. Thus, for each i, ni > I and n > 2 -
> 0. Hence Po2 > P_ PI > 0. This and P0KP, PI) imply PIPI = 0. Then for some i, In = 1.
(b) Pa is a polynomial in k of degree s. Since Po divides each P., PO also divides the nth
difference
E (- I) (sZ)P, = s!.
t-0
Since P0, > 0, this means that P, < s!. As PO is a product of s distinct positive integers, it
follows that
(n,, n,....n, = (1,2. s).
B-5.
Since the derivatives of xl is negative for x > e,
(I) at > bawhene a < b.
The u's are defined so that uo = e and
(2) x = (Uo) = (u,) ' = (u2 )u' =
Hence
(3) " = (unln U- 1)/In Un
As Y > e', u, = In x > e = u,. Now u, > u, implies In u, > In uo and then (3) with n= I
implies u, < u,. Also, (2) and (1) imply (u,)12 = (uC)"' > (u,)", which gives us u2 > uo. Now
U2 < us and (3) with n = 2 imply U3 > u,. Also (2) and (1) imply (u 2 )"' = (u)u2 < (u 2 )u' and
hence u3 < U1 . Similarly, U2 < U4 < U3. Then an easy induction shows that
e < U2n < U2 ,+2 < Ul,+ I < U,- I forn = 1,2,....
Thus the monotonic bounded sequences uO, U 2, U4 ,... and U1, U3, U5 ,... have limits a and b,
respectively, with e < a < b. Also
a' = lim (U 2 ,)2.' X = lim (u2,, u" = bo.
Then ah = bl, e < a < b, and (I) imply a = b. Hence lim,-uu exists and is the unique real
number g = g(x) with g > e and gg = x. Sincef( v) = y' is continuous and strictly increasing for
v > e, its inverse function g(x) is also continuous.
B-6.
Let s = (a + b + c)/2, I = s - a, u = s - b. v = s - c and similarly for the primed letters.
Using Heron's Formula, the inequality to be proved will follow from
(A) +:/s' 'u'v' < 7(s + s')(t + t')(u + u')(v + v')

for positive s, t, u, v, s', t', u', v'. A simpler analogous inequality that might be helpful is
(B) V + 7 < /(x + x')(y +y') for x, y, x', y'positive.
First we note that (B)follows from the Cauchy Inequality applied to the vectors (Vx, Vxp) and (Vy,
iy ) [and also follows from (Ixy7- V/x3y)2 > 0 or from the Inequality on the Means applied to
xy' and x'y]. Using (B) with x - Xt 7 x' - 17, y - V, y' = V-u7 and reapplying (B) to the
new right side, one has
f4/ + 4Is't'u'v' Ar + )(V + Au)
VV/8(s + s')(t + t') V(u + u')(v + v').
Since here the rightmost part equals the right side of (A), we have proved (A).
Equality holds in (B) if and only if Vx: -=/j : ry and this holds if and only if
x: x' -y: y'. Hence equality occurs in (A) if and only if s: t: u: v = s': t': u': v'. It follows
that equality occurs in the original inequality if and only if a, b, c are proportional to a', b', c'.
THE FORTY-FOURTH WILLIAM LOWELL PUTNAM
December 3, 1983
A-1.
For d and m in Z- (1,2,3,... }, let dim denote that d is an integral divisor of m. For m in
Z+, let T(m) be the number of d in Z+ such that dim. The number of n in Z+ such that nia or
nib is
r(a) + T(b) - T(gcd(a,b)).
Also T(p'q') - (s + 1)(t + 1) for p, q, s, t in Z+ with p and q distinct primes. Thus the desired
count is
T(24. 54°) + r( 2 6. 53°) - T(240 - 530) - 412 + 61 - 31 - 41- 31
- 1681 + 620 - 2301.
A-2.
Let OA be the long hand and OB be the short hand. We can think of OA as fixed and OB as
rotating at constant speed. Let v be the vector giving the velocity of point B under this
assumption. The rate of change of the distance between A and B is the component of v in the
direction of AB. Since v is orthogonal to OB and the magnitude of v is constant, this component
is maximal when 2OBA is a right angle, i.e., when the distance AB is =42 2
Alternatively, let x be the distance AB and 6 = A OB. By the Law of Cosines,
x2 32 + 42 - 2 * 3 - 4cos0 = 25 - 24cos0.
Since dO/dt is constant, we may assume units chosen so that 0 is also time t. Now
dx 24dx _ x sin 1 iO
dO ' dTO 25 -24 coO
Since dx/dO is an odd function of 0, idx/dsl is a maximum when dx/dO is a maximum or a
minimum. Since dx/ds is a periodic differentiable function of 0, d 2x/ds2 = 0 at the extremes for
dx/ds. For such 0,
d2X (d 2dx 2 x 144 sin20
l2cos - x 2 + dX 142 =
dT hd TJ
Then
127
2 12c sin28
n 12 - 12
cosCOS2p = 25 - 24 cos 0,
Cos9 CosO
and it follows that
12cos2 0 - 25cosO + 12 = 0.
The only allowable solution for cos 0 is cos 0 = 3/4 and hence x = /25 -24 cos 0
= 25-18 =W.
A-3.
F(n) =-1 + 2n + 3n2 + - +(p -l)ne2

nF(n) n + 2n2 + ... +(p - 2)nP-2 +(p -1)n'.
Hence (1 - n)F(n) = (1 + n + n2 + *.* +nP-2) - (p - 1)nP-' and similarly
(1 - n)2F(n) - 12- nP -(1- n)(p -1)nP-1 = 1 -p nP-' +(p - 1)nP.
2
Modulo p, nP - n by the Little Fermat Theorem and so (1 - n) F(n) - 1 - n. If neither a nor
b is congruent to 1 (mod p), 1 - a i 1 - b and there are distinct reciprocals (1 - a) I and
(1 - b)- '(mod p); then
f (a) - (1 - a) -', f (b) - (1 - b) -', f (a) gtf (b)(mod p)-
If one of a and b, say a, is congruent to 1, then b l O (mod p) and so f(b) = (I - b) ' O
(mod p) while
f(a) - 1 + 2 + +(p -1) =p(p - 1)/2 - O(mod p).
A4.
Let ( =0
0)for r > m and for r < 0. For i = 0,1,2 let
T,(m) - (i) i + 3) +(i + 6) -(i +

We note that S(m) = T2 (m) + 1. Since (,n) =(n r1) + (n- ),
T 2 (m) = T2 (m -1) + TJ(m -1), T,(m) = T1(m -1) + T0 (m - 1),
TO(m) = TO(m - 1) - T 2 (m - 1).
Let the backwards difference operator V be given by vf (n) = f(n) - f(n - 1). Then
VT 2 (m) = T1 (m -1), VTl(m) = TO(m - 1), VTO(m) = -T 2 (m - 1).
These imply that
V 3 T 2 (m) 2
_ v T(m - 1) = VTo(m - 2) = -T 2 (m - 3) for m > 3.
3
Expanding V T2 (m), this gives us
(R) T,(m) = 3[T 2 (m -1) - T2(m - 2)] for m > 3.
When mi = 6k - 1 with k > 1, we have m > 5. It then follows from (R) that T2 (m) 0-(mod 3)
and hence S(m) - 1 (mod 3). Thus S(m) # 0.
A-5.
Inductively we define a sequence of integers 3 = a,,a 2,a 3 ,... and associated intervals
I= [(an)I'", (1 + aj)l'") such that a, > 3n, a. w n (mod 2), the sequence ((a,)'",} is
11
nondecreasing, and 1,, 2 J,,+ . When this has been done, ((aj) "}, being nondecreasing and
bounded, will have a limit u which is in In for all n. Then (a)"") < u < (1 + a,) 1"" will imply
1
that a,, < uV < 1 + a,, and so [u"] = a. - n(mod 2) for all n.
SOLUTIONS: THE FORTY-FOURTH COMPETITION 129
Let a, = 3. Then 1, = 13,4). Let us assume that we have a, a 2 , .... ak and II, 12,. .'., k with
the desired properties. Let
Jk= [(a)(k +)/ ,(l + a )(A+ \/k)
Then x is in 1Aif and only if xA+' is in J. The length of Jk is
(1 + a)( /A(a*)+ > (1 + ak - ak)(ak) = a/& > ( 3 k)Ilk = 3

Since the length of JA is at least 3, Jk contains an interval Lk = [ak+I,1 + ak +I) for some integer
a + I which is congruent to k + 1 (mod 2). Let
=
=A [(a,,+.)l/(k+I), (1 + ak 1( + )).
Since x E 'A ifand only if Xk I E jk, X E 4k + Iif and only if Xk I E Lk, and Jk 2 Lk, one sees
that 1A2 IA+ . Also
ak + I > (ak )k / -l_[(ak)Il/k] k +I > 3 k+ 1
This completes the inductive step and shows that the desired u exists.
A-6.
Under the change of variables s = u - v and t - u + v, with the Jacobian a(u, v)/a(s, 1) -
1/2, F(x) becomes 1(x)/E(x) where
fo J.P(2 )+(2 ))d

- | Jexp( 1 t3 + 3ts2) dsdt
and E(x) = 2x- 4 exp(x3 ). Since l(x) and E(x) go to + oc as x goes to + oo, one can use
L'Hbpital's Rule and we have limx- F(x) - limx.-X('/E') where
F=| exp( x3 + 3Xs2=ds - exp(x3/4)fx exp(3xs 2 /4)ds
and E' = (6x-2 - 8x- 5)exp(x 3 ). In the integral for I', make the change of variable s = w/ F,
s- dw/ Vx, to obtain
- exp(X 3/4) fXE exp(3W2/4) dw.
Now
|X exp(3w 2 /4) dw
lim F(x) -' Uir -- liM 332 91 2
x- x .mE'
0 x- (6x 2
/ - 8x / )eXp(3x 3 /4)
We can, and do, use L'Hbpital's rule again to obtain
lim F(x) - lim 2(3/2)x 1 12 exp(3x 2 /4) 2

x-x0 x- oo [(27/2) xll2 + **]exp(3x'/4) 9
The diameter of S must be 4 and S must be centered at the center of C. The set of points
inside C nearer to v than to another vertex w is the part of that half-space, bounded by the
perpendicular bisector of the segment vw, containing v which lies within C. The intersection of
these sets is a cube C' bounded by the three facial planes of C through v and the three planes
which are perpendicular bisectors of the edges of C at v. These last 3 planes are planes of
symmetry for C and S. Hence R is one of 8 disjoint congruent regions whose union is the set of
points between S and C, excepting those on the 3 planes of symmetry. Therefore
8 vol(R) - vol(C) - vol(S) - 4 3 - 4v *23,
33
vol( R)-8- -i-
B-2.
A representation for 2n is of the form

2n -eo + 2e, + 4e2 + - +2:e,,
the e, in (0,1,2,3}, and with eo in {0,2}. Then el + 2e 2 + +2 k -lek is a representation for
n if e- 0 and is a representation for n - 1 if eo = 2. Since all representations for n and n - 1
can be obtained this way,
C(2n) - C(n) + C(n -1).
Similarly, one finds that
C(2n + 1) - C(n) + C(n - 1) = C(2n).
Since C(1) = 1 and C(2) = 2, an easy induction now shows that C(n) = [1 + n/2].
B-3.
To satisfy the equation, each y, must have at least 3 derivatives. Here £ will be a sum with i
running over 1,2,3. We have Ty,2 = 1 and E(y,')2 = f Differentiating, one has E2y y,' = 0 and
:2 y,'y;" = f'. Differentiating Ey yj' = 0 leads to Ey y," + E(y,') 2 = o so Ey, y," = -f. Differenti-
ating this gives us Ey,'y" + Eyy,"' = -f'. This and Ey,'yy" = f'/2 leads to Eyy" =- 3f'/2.
Multiplying each term of
y,"' + py," + (y,' + ryi = 0
by y, and summing gives us
-3f'/2 -pf+ q-0 + r -0.
Thus f' + (2/3)pf - (2/3)r and so A = 2/3 - B.
B4.
We can let m - k2 +j, where k and j are integers with 0 < j < 2k, since the next square
after k2 is k 2 + 2k + 1; let this j be the excess for m. We note that [V¼W] = k and f(m) =
k2 + j + k. If the excess j is 0, m is already a square. Let A consist of the m's with excess j
satisfyingO j < k and B consist of the m's with k < j < 2k. If m is in B,
f(m) - k 2 +j + k = (k + 1)2 +(j - k - 1),
with the excess] - k - 1 for f(m) satisfyingO < j - k - 1 < k + 1, and hence f(m) is either a
square or is in A. Thus it suffices to deal with the case in which m is in A. Then [Im + k] = k
and
f2(m) -f(f(m)) =f(m + k) - m + 2k = (k + 1)2 +(j - 1).
2
Hence f (m) is either a square or an integer in A with excess smaller than that of m. Continuing,
one sees that f'(m) is a square for some r with 0 6 r < 2j.
SOLUTIONS: THE FORTY-FOURTH COMPETITION 131
B-5.
By definition of a. and 1lull,
an- /- k) dx + J2 /k+l)(k+1 ) dx
k-I L2n/(2
fn k +1)X n/kl
rIn2k+1 - 1 + 1 - 12k+2 1
k-l 2k 2k + 1 2k + I 2k + 1
- _ (2k + 1)2
" [3 3 5 5 (2n -1) (2n -1)l
-l inn--
k-i 2k(2k + 2) [2 4 4 6 (2n - 2) 2n J
Since
2 2 4 4 6 6
and In x is continuous for x > O,lim,,- a, - ln(4/v).

B-6.
2 2
Since r # 1 and rm- 1- (r - 1)(r'-I + r' + +1) -_O, one has r'-' + r"'
+ - +1-Oandso
2
-1 -r(I + r+ r + . +rm-2
- 1 r(1 + r)(1 + r )(1 + r4 ) 2
.(1 + r(m-1 )/ 2 ),
1 - (r + r )(1 + r )(1 + r ) *(1 + r(-' 1)/ 2 ).
2 2 4
Since r + r2 - rm+l + r2 with m + 1 - 2 (2 k-1 + 1), each of the factors in the last expression
for -1 is a sum of two squares. Their product can be expressed as a sum of two squares by
repeated application of the identity
2 2 2 2 2
(a + b )(c + d ) - (ac - bd)' +(ad+ bc) .
This converts -1 into P2 + Q2 with each of P and Q a polynomial in r with integer coefficients.
THE FORTY-FIFTH WILLIAM LOWELL PUTNAM
December 1, 1984
A-i.
The set B can be partitioned into the following sets:
(i) A itself, of volume abc;
(ii) two a X b x 1 bricks, two a x c x 1 bricks, and two b x c x 1 bricks, of total volume
2ab + 2ac - 2bc;
(iii) four quarter-cylinders of length a and radius 1, four quarter-cylinders of length b and
radius 1, and four quarter-cylinders of length c and radius 1, of total volume (a + b + c)7r;
(iv) eight spherical sectors, each consisting of one-eighth of a sphere of radius 1, of total volume
41r/3.
Hence the volume of B is
abc + 2(ab + ac + bc) + 7r( a + b + c) + 3
A-2.
Let S(n) denote the nth partial sum of the given series. Then
kI[3 k* 3k + 1 2k+I1 3`
and the series converges to hm, -.S(n) = 2.

A-3.
Let N = Mj=- N has rank 2, so that 0 is an eigenvalue of multiplicity 2n 2. Let e denote
the 2n x 1 column vector of 1's. Notice that Ne = n(a + b)e, and therefore n(a + b) is an
eigenvalue. The trace of N is 2na, and therefore the remaining eigenvalue is 2na - n(a + b)
n(a - b). [Note: This corresponds to the eigenvector f, where fj = ( -l)+', i = 1,...,2n.]
The preceding analysis implies that the characteristic equation of N is
det(N - Al) = A2' 2
(A - n(a + b))(A - n(a - b)).
Let A = a - x. Then
2
detM,, = det(N -(a -x)l) = (a x) (a - x - n(a + b))(a -x - n(a - b)).
It follows that
lim 2" = lim (a - x - n(a + b))(a -x - n(a -b)) = n 2 (a2 - b2 ).

- (x - a) 2 2
A-4.
Let = ArcAB, a = ArcDE, and 8 = ArcEA. Then ArcCD =r7- and Arc BC =
7- a -,.
132
SOLUTIONS: THE FORTY-FIFTH COMPETITION 133
The area of P, in terms of the five triangles from the center of the circle is
1sinO + Isina +
) +-sin(ir- sin# + sin( -a - 8).
22
This is maximized when 6 = i7/2 and a = fl = 77/3. Thus, the maximum area is
1 *1 +
1 *1 +
1Vr +
iVT
r +
iVYr = +
3 rT .
2 2 -2 2 22 -22 4
A-5.
For t > 0, let R, be the region consisting of all triples (x, y, z) of nonnegative real numbers
satisfying x + y + z < t. Let
1(t) = fffxly9z8(t _ x y z) 4
dxdydz
R,
and make the change of variables x = tu, y = tu, z = tw. We see that 1(t) = 1(j)t25 .
Let J = fSO1(t)e-'dt. Then
J =f1(1) t25e-'dt = 1(1)F(26) = 1(1)25!.
It is also the case that

N=J
R,
eJ- 'xy 9 zs(t
e X y Z)
4
dxdydzdt.
Let s = t - x - y - z. Then
J =f . . . e-'e-'e ye- x'y 9 zFs4 dXdydzds = F(2)r(1o)r(9)r(5) = 1!9!8!4!.
The integral we desire is 1(1) = J/25! = 1!9!8!4!/25!.
A-6.
(a) All congruences are modulo 10.
LEMMA. f(5n) = 2"f(n).
Proof. We have
( ) =
_) "n'l (5i + 1)(5i + 2)(5i + 3)(5i + 4)
- 2
If i is even, then
1 (5i + 1)(5i + 2)(5i + 3)(5i + 4) I (1 2 3 4) 2,
and if i is odd, then
1(5i + 1)(5i + 2)(5i + 3)(5i + 4) = (6 7 8 9) 2.
Thus the entire product above is congruent to 2". From (*) it is clear that the largest power of 10
dividing (5n)! is the same as the largest power of 10 dividing 1O'n!, and the proof follows.
We now show by induction on 5'± + + 5"' that
f(5' + . + 5') 2-"+''-k
(which depends only on a, + + ak as desired).
This is true for 5"' ± . . . +51k = 1, since f(5 0 ) 20 = 1.
CASE 1. All a, > 0. By the lemma and induction,

f(5-l + * + 5k) = 25~1 +' +5k
5' ((5 + + 5" 1)
-i 2k - 2(a--1)+ +(ak1) (since 25 3 2 for i > 0)
= 2" +' +U
CASE 2. Some a, = 0, say a, = 0. Now

(1 + 5m)! = (1 + Sm)(5m)!,
so f (1 + 5m) = (1 + Sm)f (Sm). But f(5m) is even for m > 1 since (5m)! is divisible by a higher
power of 2 than of 5. But
(1 + 5m) (2j) - 2j,
sof(1 + Sm) -f(5m). Letting m = 5a'2- + *-- +5'k -,I the proof follows by induction.
(b) The least p > 1 for which 2'+P =- 22 for all s > I is p = 4.
B-I.
We have
f(n + 2) -f(n + 1) = (n + 2)! = (n + 2)(n + 1)! = (n + 2)[f(n + 1) -f(n)].
It follows that we can take P(x) = x + 3 and Q(x) = -x - 2.
B-2.
The problem asks for the minimum distance between the quarter of the circle x 2 + Y2 = 2 in
the open first quadrant and the half of the hyperbola xy = 9 in that quadrant. Since the tangents
to the respective curves at (1,1) and (3,3) separate the curves and are both perpendicular to
x = y, the minimum distance is 8.
B-3.
The statement is true. Let 4) be any bisection on F with no fixed points, and set x * y = +(x).
B-4.
Such a function must satisfy
|Xg2 (t) dt 1 x
x =-f g( t) dt,
xg(t)dt x o
or equivalently,
jX 2 (t) dt x t) d]2
[Ag(
Let z(x) - Jofg(t) dt. Then z'(x) = g(x) and we have
t2(z) di=x, x > 0.
Differentiating, we have
I (,)2 x * 2zz' - z2
(1Z = 2 X>
SOLUTIONS: THE FORTY-FIFTH COMPETITION 135
2
x2(z') - 4xzz' + 2z = 0, x > 0,
(xz' - rz)(xz' - r2 z) = 0, x > 0,
where r, = 2 + F2 and r, = 2 - 2.
Now x, z', and z are continuous and z > 0, so the last equation implies that xz'/z - r, where
r = ri or r = r2 . Separating variables, we have z'/z = r/x and it follows that
In z r ln x + CO,
or equivalently, z = Clx', Cl > 0. Differentiating, we have z' = g(x) = Cx'- , C > 0. But g is
continuous on [0, xc) and therefore we cannot have r = r2 (because r2 - 1 = 1 - F < 0). Thus
2
g(x) = Cxl1V , C > O,
and one can check that such g(x) do satisfy all the conditions of the problem.
B-5.
Define
D(x) = (1 - x)(1 - X2 )(1 - X4 ) .. (1 - X2 ' -).
Since binary expansions are unique, each monomial xk (O < k < 2" - 1) appears exactly once in
the expansion of D(x), with coefficient (- l)d(k ). That is,
2' 1
D(x) = ( _ 1 )d(k)Xk
k=O
Applying the operator x d) to D(x) m times, we obtain
2'-1
(x dx ) D(x) = ( 1) d(k) xk
so that
(Xd ) D(x)] = ; - (-1) d(A)2'.
Define F(x) = D(x + 1), so that
(X d ) D(x)] =[(x + 1) d]"F(x)]
But
F(x) n [1 -(x + 1)2 ]= ri [-2- 1 x + O(X2)], (x -0)

a= a-I
=(-1) 2-(- 1)1'xm + O(Xm+I),
and by observing that [(x + 1)d/dx]x" = nx' + nx'', we see that
[(x + 1) d ]J(Ax- + O(x-+')) = m!A + O(x).
So
(x + 1) d I F(x)] =(-1)"2('- 1)/ m! + 0(x)] xo 1) -2"'2 m!.

B-6.
Suppose that i and v are consecutive edges in P,. Then ii/3, (u + f))/3, and e/3 are
consecutive edges in P+,,.l Further,
11 x 6ll = T8
is removed at this corner in making P,+,. But at the next step, the amount from these three
consecutive edges is
11 i u i+ 3l 1 11+ v ell.
2119 9 211 9 911
Thus, the amount removed in the (k + I)st snip is 2/9 times the amount removed in the kth.
Note that one-third of the original area is removed at the first step. Thus, the amount removed
altogether is
-1[1 + (2/9) + (2/9) 2 + .. ]=1 9 3

of the original area. Since the original area is r/i/4, we have
lim Area P. = 4 . = r
The curve in this problem has been studied extensively by Georges de Rham. (See "Un peu de
mathematiques a propos d'une courbe plane," Elem. Math., 2 (1947), 73-76, 89-97; "Sur une
courbe plane," J. Math. Pures Appl., 35 (1956), 25-42; and "Sur les courbes limites de polygones
obtenus par trisection," Enseign. Math., 5 (1959), 29-43.) Among de Rham's results are the
following. The limiting curve is Cl with zero curvature almost everywhere, but every subarc
contains points where the curvature is infinite. Consequently, the curve is nowhere analytic. De
Rham parametrizes pieces of the curve so that the tangent vector is intimately related to the
Minkowski ?-function. If the construction is repeated, but with each edge divided in the ratio
(1/4, 1/2, 1/4) rather than (1/3, 1/3, 1/3), then the resulting limit curve is analytic, consisting
of piecewise parabolic arcs.
WINNING TEAMS 137
APPENDIX
WINNING TEAMS
Twenty-sixth Competition-1%5
Harvard University, Cambridge, Massachusetts
Massachusetts Institute of Technology, Cambridge, Massachusetts
University of Toronto, Toronto, Ontario, Canada
Princeton University, Princeton, New Jersey
California Institute of Technology, Pasadena, California
Twenty-seventh Competition-1966
University of Chicago, Chicago, Illinois
University of Michigan, Ann Arbor, Michigan
Twenty-eighth Competition-1967
Michigan State University, East Lansing, Michigan
University of Michigan, Ann Arbor, Michigan
Twenty-ninth Competition-1968
University of Waterloo, Waterloo, Ontario, Canada
University of California at Los Angeles, Los Angeles, California
Michigan State University, East Lansing, Michigan
University of Kansas, Lawrence, Kansas
Thirtieth Competition-1969
Rice University, Houston, Texas
Yale University, New Haven, Connecticut
Thirty-first Competition-1970
University of Toronto, Toronto, Ontario, Canada
Illinois Institute of Technology, Chicago, Illinois
138 THE WILLIAM LOWELL PUTNAM MATHEMATICAL COMPEl 11 ION
Thirty-second Competition-1971
University of California, Davis, California
Thirty-third Competition-1972
Oberlin College, Oberlin, Ohio
Swarthmore College, Swarthmore, Pennsylvania
Thirty-fourth Competition-1973
University of British Columbia, Vancouver, British Columbia, Canada
Thirty-fifth Competition-1974
University of British Columbia, Vancouver, British Columbia, Canada
Thirty-sixth Competition-1975
Thirty-seventh Competition-1976
Washington University, St. Louis, Missouri
* Case Western Reserve University, Cleveland, Ohio
* Massachusetts Institute of Technology, Cambridge, Massachusetts
* Tied for fourth place.

WINNING TEAMS 139
Thirty-eighth Competition-1977
University of California, Davis, California
Thirty-ninth Competition-1978
Case Western Reserve University, Cleveland, Ohio
Fortieth Competition-1979
Stanford University, Stanford, California
Forty-first Competition-1980
University of Maryland, College Park, Maryland
University of California, Berkeley, California
Forty-second Competition- 1981

Stanford University, Stanford, California
University of Maryland, College Park, Maryland
Forty-third Competition-1982
Forty-fourth Competition-1983

Forty-fifth Competition- 1984

*University of California, Davis, California
*Washington University, St. Louis, Missouri
* Tied for first place.

WINNING INDIVIDUALS 141
WINNING INDIVIDUALS
Twenty-sixth Competition-1965
Andreas R. Blass, University of Detroit
Robert Bowen, University of California, Berkeley
Daniel Fendel, Harvard University
Lon M. Rosen, University of Toronto
Barry Simon, Harvard University
Twenty-seventh Competition-1966
Marshall W. Buck, Harvard University
Theodore C. Chang, Massachusetts Institute of Technology
Robert E. Maas, University of Santa Clara
Richard C. Schroeppel, Massachusetts Institute of Technology
Robert S. Winternitz, Massachusetts Institute of Technology
Twenty-eighth Competition-1967
David R. Haynor, Harvard University
Dennis A. Hejhal, University of Chicago
Peter L. Montgomery, University of California, Berkeley
Richard C. Shroeppel, Massachusetts Institute of Technology
Don B. Zagier, Massachusetts Institute of Technology
Twenty-ninth Competition-1968
Don Coppersmith, Massachusetts Institute of Technology
Gerald A. Edgar, University of California, Santa Barbara
Gerald S. Gras, Massachusetts Institute of Technology
Dean G. Huffman, Yale University
Neal Koblitz, Harvard University
Thirtieth Competition-1969
Alan R. Beale, Rice University
Gerald A. Edgar, University of California, Santa Barbara
Robert A. Oliver, University of Chicago
Steven Winkler, Massachusetts Institute of Technology
Thirty-first Competition-1970
Jockum Aniansson, Yale University
Jeffrey Lagarias, Massachusetts Institute of Technology
Robert A. Oliver, University of Chicago
Arthur Rubin, Purdue University
Steven K. Winkler, Massachusetts Institute of Technology
Thirty-second Competition-1971
Robert Israel, University of Chicago
Dale Peterson, Yale University
Arthur Rubin, Purdue University
David Shucker, Swarthmore College
Michael Yoder, California Institute of Technology
Thirty-third Competition-1972
Ira Gessel, Harvard University
Dean Hickerson, University of California, Davis
Arthur Rothstein, Reed College
Arthur Rubin, California Institute of Technology
David Vogan, University of Chicago
Michael Yoder, California Institute of Technology
Thirty-fourth Competition- 1973

David J. Anick, Massachusetts Institute of Technology
Peter G. de Buda, University of Toronto
Matthew L. Ginsberg, Wesleyan University
Arthur L. Rubin, California Institute of Technology
Angelos J. Tsirimokos, Princeton University
Thirty-fifth Competition- 1974

Thomas G. Goodwillie, Harvard University
Grant M. Roberts, University of Waterloo
Karl C. Rubin, Princeton University
James B. Saxe, Union College
Philip N. Strenski, Armstrong State College
Thirty-sixth Competition-1975
Franklin T. Adams, University of Chicago
David J. Anick, Massachusetts Institute of Technology
Ernest S. Davis, Massachusetts Institute of Technology
Thomas G. Goodwillie, Harvard University
Christopher L. Henley, California Institute of Technology
Thirty-seventh Competition-1976
Philip I. Harrington, Washington University, St. Louis
Christopher L. Henley, California Institute of Technology
Paul M. Herdig, Case Western Reserve University
Nathaniel S. Kuhn, Harvard University
Steven T. Tschantz, University of California, Berkeley
David J. Wright, Cornell University
WINNING INDIVIDUALS 143
Thirty-eighth Competition-1977
Russell D. Lyons, Case Western Reserve University
Stephen W. Modzelewski, Harvard University
Michael Roberts, Massachusetts Institute of Technology
Adam L. Stephanides, University of Chicago
Paul A. Vojta, University of Minnesota, Minneapolis
Thirty-ninth Competition-1978
Randall L. Dougherty, University of California, Berkeley
Mark R. Kleiman, Princeton University
Russell D. Lyons, Case Western Reserve University
Peter W. Shor, California Institute of Technology
Steven T. Tschantz, University of California, Berkeley
Fortieth Competition-1979
Richard Mifflin, Rice University
Mark G. Pleszkoch, University of Virginia
Miller Puckette, Massachusetts Institute of Technology
Charles H. Walter, Princeton University
Forty-first Competition- 1980

Eric D. Carlson, Michigan State University
Daniel J. Goldstein, University of Chicago
Laurence E. Penn, Harvard University
Michael Raship, Harvard University
Forty-second Competition-1981
David W. Ash, University of Waterloo
Scott R. Fluhrer, Case Western Reserve University
Michael J. Larsen, Harvard University
Robin A. Pemantle, University of California, Berkeley
Adam Stephanides, University of Chicago
Forty-third Competition-1982
Noam D. Elkies, Columbia University
Brian R. Hunt, University of Maryland, College Park
Edward A. Shpiz, Washington University, St. Louis
Forty-fourth Competition-1983

Michael J. Larsen, Harvard University
Gregg N. Patruno, Princeton University
Forty-fifth Competition-1984
Benji N. Fisher, Harvard University
Daniel W. Johnson, Rose-Hulman Institute of Technology
Michael Reid, Harvard University
Richard A. Stong, Washington University, St. Louis
INDEX 145
INDEX OF PROBLEMS
Abstract algebra Differential calculus

-binary operations 1971 B-1; 1972 A-2; -differentiation 1967 A-1
1978 A-4 -maxima & minima 1970 A-2: 1973 B-6,
-extension fields 1968 B-3; 1980 A-4 1981 B-2
-finite fields 1979 B-3 -partial derivatives 1967 B-6
-group theory 1968 B-2; 1969 B-2; 1972 -rates 1970 B-4; 1972 B-2
B-3; 1975 B-1; 1976 B-2; 1977 B-6 Differential equations 1975 A-5: 1983 B-3
-mappings 1966 A-5 -higher order 1966 B-6
Area 1978 B-1; 1979 B-5; 1981 B-6; 1984 -homogeneous linear 1979 B-4
A-4; 1984 B-6 -nonhomogeneous linear 1979 B-4
Arithmetic geometric mean inequality 1968 -systems 1969 A-5; 1971 B-5: 1973 A-5:
A-6: 1975 B-6; 1978 A-3 1982 A-4
Arithmetic progressions 1972 A-1; 1978 A-1; Diophantine equations 1971 A-5; 1978 B-4;
1979 A-3 1979 A-1
Balanced triples 1977 B-3 Dissections 1982 B-1
Bernoulli polynomials 1981 B-1; 1982 A-2 Equations
Binary operations 1971 B-1; 1972 A-2; 1978 -functional 1971 B-2; 1979 A-2
A-4; 1984 B-3 -systems 1967 A-6: 1977 A-2; 1980 A-5
Binary representation 1981 B-5; 1973 B-1; Euler's formula for polyhedra 1969 A-3
1984 B-5 Extension fields 1968 B-3; 1980 A-4
Binomial coefficients 1965 B-4; 1967 B-5; Factorials 1984 A-6; 1984 B-1
1971 A-4; 1972 A-1; 1974 A-4; 1974 B-6; Fermat's theorem 1983 A-3
1977 A-5; 1983 A-4 Fields, characteristic 1979 B-3
Calculus (See Differential calculus, Integral Finite fields 1979 B-3
calculus, Real analysis) Functions 1977 A-3
Cauchy-Schwarz inequality 1977 B-5; 1978 -continuous 1966 A-5; 1982 B-5
A-6; 1979 B-6; 1982 B-6 -convex 1980 B-5
Centroids 1982 A-1; 1984 B-4 -linear 1980 B-2
Chessboard 1981 A-2 Games 1971 A-5
Clocks 1983 A-2 Gamma function 1984 A-5
Coloring problems 1979 A-4 Geometry (See Plane geometry, Plane ana-
Combinatorial identities 1965 A-2; 1974 A-4 lytic geometry, Solid geometry)
Combinatorics 1965 A-5; 1965 B5; 1967 Graphs 1965 A-4
A-6; 1973 A-6; 1974 A-1; 1980 B-4 -coloring 1979 A-4
-inequalities 1978 A-6 -Hamiltonian circuits 1968 A-3
-optimization 1974 B-1 -incidence matrices 1965 A-4
Complex numbers 1967 B-1; 1973 B-2; 1979 Greatest integer function 1973 A-3; 1979
B-6 A-S; 1983 A-5; 1983 B-2; 1983 B-4
-algebra 1975 A-4 Group theory 1968 B-2; 1969 B-2; 1972 B-3;
-graphing 1975 A-2 1975 B-1; 1976 B-2; 1977 B-6
Complex variables 1972 B-6 Hamiltonian circuits 1968 A-3
Continuous functions 1966 A-5; 1978 B-5; Heron's formula 1982 B-6
1982 B-5 Higher plane curves
Convex sets 1967 A-5; 1969 B-4; 1979 B-5 -cycloids, etc. 1971 B-S
Decimal representation 1984 A-6 Hyperbolic functions 1979 B-1; 1980 B-I
Determinants 1969 A-2; 1978 A-2; 1984 A-3 Incidence matrices 1965 A-4
Difference equations 1980 B-3 Inequalities
Differences 1971 A-6; 1974 A-6; 1976 B-5; -arithmetic mean-geometric mean 1968
1983 A-4 A-6; 1975 B-6
-Cauchy-Schwarz 1966 B-3; 1977 B-5; -matrices 1968 B-5; 1969 B-6; 1981 B-4
1979 B-6 permutation matrices 1967 A-2
-functional considerations 1967 A-1; 1967 -symmetric matrices 1967 A-2
B-6; 1973 B-4; 1973 B-6; 1974 B-5; Linear functions 1980 B-2
1978 A-5; 1980 A-6 Linear programming 1980 B-2
-geometric (triangle, polygon, etc.) 1966 Locker room problem 1967 B-4
A-2; 1966 B-i; 1971 A-3; 1973 A-1; Mappings 1966 A-5
1982 B-6 -binary operations 1978 A-4
-miscellaneous 1968 A-2; 1978 B-6; 1979 Mathematical induction 1978 B-3; 1978 B-6
A-6; 1980 A-4; 1982 A-5 Matrices 1967 A-2; 1968 B-5; 1969 B-6;
-series 1966 B-3; 1974 B-5; 1980 B-1 1981 B-4; 1984 A-3
Infinite products 1969 B-3; 1970 B-1; 1977 Maxima 1983 A-2
B-1 Maximum & minimum problems 1975 A-3;
Infinite sequences 1965 A-3; 1966 A-1; 1966 1978 B-5; 1979 A-1; 1981 B-2; 1984 A-4;
A-3; 1969 A-6; 1969 B-3; 1969 B-5; 1970 1984 B-2
A-4; 1972 A-3; 1978 B-3; 1979 A-3; 1980 Multivariable calculus 1975 A-3
B-3; 1982 B-5; 1983 A-5; 1983 B-4; 1983 Normal lines 1979 B-1
B-5 Number theory 1966 A-4; 1974 A-3; 1976
Infinite series 1967 A-2; 1967 B-5; 1969 A-4; A-3
1970 A-1; 1972 B-1; 1972 B-6; 1973 A-2; -arithmetic 1975 A-1
1975 B-5; 1976 B-1; 1977 A-4; 1978 B-2; -congruences 1968 B-5; 1969 B-1; 1973
1979 A-6; 1980 B-1; 1981 B-5; 1982 A-6; B-1; 1977 A-5; 1983 A-3
1984 A-2 -Diophantine equations 1971 A-5; 1978
-convergence and divergence 1966 B-3; B-4; 1979 A-1
1969 B-5 -divisibility 1966 B-2; 1966 B-4; 1971
Integral calculus 1980 A-5 B-1; 1971 B-6; 1972 A-5; 1973 B-3;
-definite integrals 1965 B-1; 1967 B-3; 1981 A-1; 1981 B-3; 1982 B-4; 1983
1968 A-1; 1969 A-4; 1970 B-1; 1970 A-1; 1983 A-3
B-2; 1970 B-4; 1972 A-6; 1973 B-4; -Euler phi-function 1968 B-3; 1972 A-5
1976 B-1; 1979 B-2; 1980 A-3; 1980 -Fermat's theorem 1983 A-3
A-6; 1982 A-3; 1983 B-5 -irrational numbers 1974 B-3; 1977 B-3;
-double integrals 1981 A-3; 1981 B-6; 1980 A-4
1982 B-2; 1983 A-6 -least common multiple 1980 A-2
-improper integrals 1968 B-4; 1976 A-5; -multiplicative functions 1967 B-4
1982 A-3 -Pythagorean triples 1965 B-3
-triple integrals 1984 A-5 -relatively prime integers 1974 A-1
Integral equations 1967 A-4; 1980 A-5 -representation of integers 1970 A-3; 1973
Irrational numbers 1974 B-3; 1977 B-3; 1980 B-1; 1981 B-5; 1983 B-2
A-4 sums of divisors 1969 B-1; 1976 B-6
Isoperimetric problems 1972 A-4 Partial derivatives 1967 B-6
Lattice points 1981 A-6 Permutation matrices 1967 A-2
Law of cosines 1972 B-5; 1983 A-2 Permutations 1982 A-6
L'Hopital's rule 1979 B-2; 1983 A-6 Pick's theorem 1971 A-3
Limits 1965 B-4; 1966 A-3; 1966 A-6; 1967 Pigeonhole principle 1971 A-1; 1978 A-1;
B-3; 1969 B-5; 1970 A-4; 1972 A-3; 1974 1980 A-4
B-2; 1976 B-1; 1978 B-3; 1979 B-2; 1981 Plane analytic geometry 1965 A-6; 1972 A-4;
A-1; 1981 A-3; 1981 B-1; 1982 B-3; 1983 1977 A-1
B-5 -ellipses 1976 B-4
Linear algebra -higher plane curves 1971 B-5
-determinants 1969 A-2; 1978 A-2 -hyperbolas 1967 B-2
incidence matrices 1965 A-4 -parabolas 1974 A-5; 1980 A-1
INDEX 147
Plane geometry -measure theory 1972 A-6

-circles 1965 B-6 -minimizing an integral 1978 A-3
-impossible constructions 1968 B-3 -Taylor's remainder theorem 1974 B-5
-lattices 1971 A-3 Recurrence relations 1966 A-3; 1967 A-2;
-minimum problem 1976 A-1 1969 A-6; 1971 B-6; 1973 B-5; 1975 B-5;
-octagons 1978 B-1 1979 A-3; 1980 B-3; 1980 B-6; 1982 B-5;
-pentagons 1984 A-4 1983 B-2; 1983 B-4
-polygons 1966 B-1; 1966 B-5; 1967 B-i; Reflection 1981 A-4
1969 A-3; 1978 B-1; 1981 B-6; 1984 Representation of integers 1983 B-2
B-6 Rolle's theorem 1973 A-4; 1980 A-5; 1981
-quadrilaterals 1970 B-6; 1972 B-5 A-5
-triangles 1965 A-1; 1965 B-3; 1966 A-2; Sequences (See Infinite sequences)
1971 A-3; 1973 A-1; 1981 A-6; 1982 Series (See Infinite series)
B-1; 1982 B-6 Sets 1968 A-3; 1975 B-1; 1980 B-4
Plane geometry and mechanics 1974 A-2 Solid analytic geometry 1970 A-5; 1971 B-4
Points Solid geometry
-collinear 1979 A-4 -covering problems 1975 B-2
Polygons 1984 B-1 -cubes 1983 B-1
Polyhedra 1980 B-2 -lattices 1971 A-1
Polynomials 1967 A-3; 1968 A-5; 1968 A-6; -polyhedra 1980 B-2
1970 B-2; 1971 A-2; 1971 A-4; 1972 B-4; -quadrilaterals 1977 B-2
1973 B-5; 1974 A-6; 1975 A-4; 1976 B-5; -spheres 1968 A-4; 1983 B-1
1978 B-3; 1978 B-5; 1979 A-5; 1980 A-1; -triangles 1975 A-6
1981 A-5; 1983 B-6 Symmetric functions
-irreducible 1979 B-3 -polynomials in several variables 1975 B-3
-two variables 1969 A-1; 1970 B-2; 1976 Tangents 1980 A-1
A-2 Theory of equations
Probability 1968 B-1; 1970 A-6; 1976 B-3; -roots 1968 A-6; 1976 A-4; 1977 A-1
1982 B-3 Topology
Progressions 1972 A-1; 1978 A-1; 1979 A-3 -closed curve in the plane 1977 B-4
Quickest descent problem 1974 A-2 -connectedness 1975 B-4
Rational numbers 1973 B-2; 1978 B-2; 1980 Triangle inequality 1979 A-4
B-6; 1981 B-5 Triangular numbers 1975 A-1
Real analysis 1970 B-3; 1971 A-6; 1972 B-2; Trigonometric functions 1967 A-1
1976 A-6; 1978 A-5; 1983 A-5; 1984 B-4 Trigonometry 1974 B-3
-compact sets 1968 B-6 -law of cosines 1983 A-2
-continuity 1970 B-5; 1972 A-3; 1973 Vectors 1968 A-4; 1975 A-6; 1983 A-2
A-4; 1974 B-4; 1977 A-6; 1979 A-2 Volumes 1983 B-1; 1984 A-I
-limits 1974 B-2 Wallis product 1969 B-3; 1983 B-5

G. L. Alexanderson - The William Lowell Putnam Mathematical Competition. Problems and Solutions - 1965-1984.-Mathematical Association of America (1985) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

G. L. Alexanderson - The William Lowell Putnam Mathematical Competition. Problems and Solutions - 1965-1984.-Mathematical Association of America (1985) PDF

Uploaded by

Copyright:

Available Formats

THE WILLIAM LOWELL PUTNAM

Published and distributed by

The Mathematical Association of America (Incorporated)

Printed in the United States of America

Current printing (last digit):

PREFACE ......................................................... Vii

RECOLLECTIONS OF THE FIRST PUTNAM EXAMINATION BY HERBERT ROBBINS ................ xi

LIST OF PROBLEMS ......................................................................... 3

SOLUTIONS TO THE PROBLEMS IN THE VARIOUS COMPETITIONS

Winning Teams ....................................................... 137

INDEX OF PROBLEMS ................................................ 145

November 20, 1965

(A) lim ) + e(is) + . + e(C(,)

AX ()O) + ("2) x+(n) X+,

(1i) + (3)x (5)x2

November 19, 1966

A-1. Letf(x) =aisin x+a 2 sin 2x+ * * +asin

(b) j S. - = exp(x + x'/2), where exp(x) = eX.

B-2. Let 0•-p-l and 0:5r<1 and consider the identities

(xo, yo)) + (af (0o,yo)) S 16.

Determine how many members J has.

(The integrand is taken to be 1 at x =0.)

T.T,+, = n (n = 1, 2, 3) and lim -= 1.

A-1. Show that the power series for the function

B-5. Let u, denote the "ramp" function

Show that . is associative and commutative.

x*(x*y)=y for all x,y in S,

Show that * is commutative but not necessarily associative.

The first example would generate the series

I + (1/2) - (1/3) - (1/4) - (1/5) - (1/6) + (1/7) + (1/8)

+ (1/9) + (1/10) - (1/11) - (1/12) - -]..1

(b) Is this sufficient condition also necessary?

A-5. A particle moves in 3-space according to the equations:

x = sec t, y = sec t, z = tan t;

whereas if x(0) = y(O) = 1, z(0) = - 1, then

x = 1/(t + 1), y = lI(t + 1), z =-1/(t + 1).

B-5. (a) Let z be a solution of the quadratic equation

B-6. On the domain 0 00! 2n:

I sin20(sjn 3(20) sin 3(40)... sin 3(2- l@0))sin (2n0)

sin 2 8 *sin 2 (20) - sin 2 (40) ... sin2(2n 9 ) : (3/4)5.

I^- E , (. -2k) (n)

\ / f the rolling parabola

\ (0, 1/4), the original position

the fixed parabola

B-3. Prove that if a is a real number such that

cos 7ra = 1/3,

then a is irrational. (The angle nra is in radians.)

f(x, y) = tim g.(x, y) for all (x, y) E R'.

REMARKS You may assume as known Taylor's remainder formula:

as well as the fact that

* X (n) for i = 0,1,2.

-() + (6) + (6) 22.1

A-i. Supposing that an integer n is the sum of two triangular numbers,

z(x) = c Vy,(x)y 2 (x)

satisfies the equation

State clearly the manner in which c depends on y,(x) and y,(x).

sa(a, . a.)[s,(a,,..., a)]

for arbitrary n Ž: k and arbitrary n-tuples a,. a. of positive real numbers.

s3(a,, a2, a3, a.) = a,aa 3 + a,a2 a, + a,aa, + a~a~a,.

him I Ink] -2 [n)

S(., by must J(x,y) be identically zero on S?

For which k is Ik smallest?

(a) Prove that B(a,b)-c implies B(c,c)-c when B is a bypass.

x(xy)=y for all x,y in S,

(ii) x ( v z) (x * y) * z (no case of associativity holds).