Anda di halaman 1dari 235

Contents

Chapter
1.1
1.2
Chapter
2.1

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. Examples for practice . . . . . . . . . . . . . . . . . . . . . . .
Arithmetic problems . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Junior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.2 Senior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.3 Undergraduate problems . . . . . . . . . . . . . . . . . . . .
2.1.4 Olympiad problems . . . . . . . . . . . . . . . . . . . . . . .
2.2 Algebraic problems . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1 Junior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.2 Senior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.3 Undergraduate problems . . . . . . . . . . . . . . . . . . . .
2.2.4 Olympiad problems . . . . . . . . . . . . . . . . . . . . . . .
2.3 Geometric problems . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.1 Junior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2 Senior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.3 Olympiad problems . . . . . . . . . . . . . . . . . . . . . . .
2.4 Analysic problems . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.1 Junior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.2 Senior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.3 Undergraduate problems . . . . . . . . . . . . . . . . . . . .
2.4.4 Olympiad problems . . . . . . . . . . . . . . . . . . . . . . .
2.5 Problems of Other Topics . . . . . . . . . . . . . . . . . . . . . . . .
2.5.1 Junior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.2 Senior problems . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.3 Undergraduate problems . . . . . . . . . . . . . . . . . . . .
2.5.4 Olympiad problems . . . . . . . . . . . . . . . . . . . . . . .
Chapter 2. Exercises for training . . . . . . . . . . . . . . . . . . . . . . .
3.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.1 Exercises from Mathematic Reflextion . . . . . . . . . . . .
3.1.2 Exercises from OP from Around the World . . . . . . . . .
3.2 Problems of Hanoi Open Mathematical Olympiad . . . . . . . . .
3.2.1 Hanoi Open Mathematical Olympiad 2006 . . . . . . . . .
3.2.2 Hanoi Open Mathematical Olympiad 2007 . . . . . . . . .
3.2.3 Hanoi Open Mathematical Olympiad 2008 . . . . . . . . .
3.2.4 Hanoi Open Mathematical Olympiad 2009 . . . . . . . . .
3.2.5 Hanoi Open Mathematical Olympiad 2010. Senior Section
3.3 Singapore Open Mathematical Olympiad 2009 . . . . . . . . . . . .
3.3.1 Junior Section . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.2 Senior Section . . . . . . . . . . . . . . . . . . . . . . . . . .
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

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

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

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

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

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

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

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

1
1
2
14
15
15
36
45
47
60
60
77
86
88
101
101
119
138
153
153
156
159
170
174
174
180
184
185
196
196
196
210
215
215
216
219
221
224
226
226
229
234

Chapter 1
Introduction
1.1

Preface

Although mathematical olympiad competitions are carried out by solving problems, the system of
Mathematical Olympiads and the related training courses cannot involve only the techniques of solving
mathematical problems. Strictly speaking, it is a system of mathematical advancing education.
To guide students who are interested in mathematics and have the potential to enter the world of
Olympiad mathematics, so that their mathematical ability can be promoted efficiently and comprehensively, it is important to improve their mathematical thinking and technical ability in solving mathematical
problems.
Technical ability in solving mathematical problems does not only involve producing accurate and
skilled computations and proofs, the standard methods available, but also the more unconventional,
creative techniques.
It is clear that the usual syllabus in mathematical educations cannot satisfy the above requirements,
hence the mathematical olympiad training books must be self-contained basically.
The book is based on the lecture notes used by the editor in the last 25 years for Olympiad training
courses in BAC GIANG SPECIALIZING UPPER SECONDARY SCHOOL. Its scope and depth
significantly exceeds that of the usual syllabus, and introduces many concepts and methods of modern
mathematics.
The core of each lecture are the concepts, theories and methods of solving mathematical problems.
Examples are then used to explain and enrich the lectures, and indicate their applications. And from
that, a number of questions are included for the reader to try. Detailed solutions are provided in the
book.
The examples given are not very complicated so that the readers can understand them more easily.
However, the practice questions include many from actual competitions which students can use to test
themselves. These are taken from a range of countries, e.g. China, Russia, the USA and Singapore. The
questions are for students to practise, and test students ability to apply their knowledge in solving real
competition questions. Each section can be used for training courses with a few hours per week. The test
questions are not considered part of the lectures, since students can complete them on their own.

Acknowledgments
My great thanks to Doctor of Science, Professor Nguyen Van Mau, and Doctor Associate Professor
Nguyen Vu Luong for their strong support. I would also like to thank my colleagues, MA Bach Dang
Khoa, MA Tran Thi Ha Phuong, MA Nguyen Danh Hao and MA Tran Anh Duc for their careful reading
of my manuscript, and their helpful suggestions. This book would be not written today without their
efficient assistance.
Nguyen Van Tien

Chapter 1. Introduction

1.2

Glossary

Abel summation. For an integer n > 0 and reals a1 , a2 , ..., an and b1 , b2 , ..., bn,
ai bi = bn ai + ((bi bi+1 ) aj )
n

n1

i=1

i=1

i=1

j=1

Angle bisector theorem. If D is the intersection of either angle bisector of angle


ABC with line AC, then
BA DA
=

BC DC

Arithmetic mean-geometric mean (AM-GM) inequality. If a1 , a2 , ..., an are


1 n
n nonnegative numbers, then their arithmetic mean is defined as ak and their
n k=1

geometric mean is defined as (a1 a2 an ) n .


The arithmetic mean - geometric mean inequality states that
1

1
1 n
ak (a1 a2 an ) n .
n k=1

with equality if and only if a1 = a2 = = an . The inequality is a special case of the power
mean inequality.
Arithmetic mean-harmonic mean (AM-HM) inequality. If a1 , a2 , ..., an are n
1 n
positive numbers, then their arithmetic mean is defined as ak and their harmonic
n k=1
1
mean is defined as
The arithmetic mean - harmonic mean inequality states that
1 n 1

n k=1 ak
1 n
1
ak
n
1
1
n k=1

n k=1 ak

( ak ) (
n

or

k=1

1
) n2 .
k=1 ak
n

with equality if and only if a1 = a2 = = an . Like the arithmetic mean-geometric mean


inequality, this inequality is a special case of the power mean inequality.
Bernoullis inequality. For x > 1 and a > 1,
with equality when x = 0.

(1 + x)a 1 + ax,

1.2. Glossary

Binomial coefficient.

Cnk =

n!
,
k!(n k)!

the coefficient of xk in the expansion of (x + 1)n .


Binomial theorem.
(x + y)n = Cnk xnk y k = Cnk xk y nk .
n

k=0

k=0

Brianchons theorem. If hexagon ABCDEF is circumscribed about a conic in the


projective plane such that A D, B E, and C F , then lines AD, BE, and CF concur.
(If they lie on a conic in the afine plane, then these lines either concur or are parallel.)
This theorem the dual to Pascals theorem.
Brocard angle. ( See Brocard points.)
Brocard points. Given a triangle ABC, there exists a unique point P such that
ABP = BCP = CAP and a unique point Q such that BAQ = CBQ = ACQ.
The points P and Q are the Brocard points of triangle ABC. Moreover, ABP and
BAQ are equal; their value is the Brocard angle of triangle ABC.
Cauchy-Schwarz inequality. For any real numbers a1 , a2 , ..., an , and b1 , b2 , ..., bn
n
( a2i ) ( b2i )
i=1
i=1
n

( ai bi ) .
n

i=1

with equality if and only if ai and bi are proportional, i = 1, 2, ..., n.


Centrally symmetric. A geometric figure is centrally symmetric (centrosymmetric)
about a point O if, whenever P is in the figure and O is the midpoint of a segment P Q,
then Q is also in the figure.
Centroid of a triangle. Point of intersection of the medians.
Centroid of a tetrahedron. Point of the intersection of the segments connecting
the midpoints of the opposite edges, which is the same as the point of intersection of the
segments connecting each vertex with the centroid of the opposite face.
Cevas theorem and its trigonometric form. Let AD, BE, CF be three cevians
of triangle ABC. The following are equivalent:
(i) AD, BE, CF are concurrent;

Chapter 1. Introduction
AF BD CE

= 1;
F B DC EA
sin ABE sin BCF sin CAD

= 1.
(iii)
sin EBC sin F CA sin DAB

(ii)

Cevian. A cevian of a triangle is any segment joining a vertex to a point on the


opposite side.
Chinese remainder theorem. Let k be a positive integer. Given integers a1 , a2 , ..., ak
and pairwise relatively prime positive integers n1 , n2 , ..., nk , there exists a unique integer
a such that 0 a < ni and a ai (mod ni ) for i = 1, 2, . . . , k.
k

i=1

Circumcenter. Center of the circumscribed circle or sphere.


Circumcircle. Circumscribed circle.
Complex numbers in planar geometry. If we introduce a Cartesian coordinate
system in the Euclidean plane, we can assign a complex number to each point in the plane
by assigning + i to the point (, ) for all reals and . Suppose that A, B, ..., F are
points and a, b, ..., f are the corresponding complex numbers. Then:

a + (c b) corresponds to the translation of A under the vector BC;


given an angle , b+ei (ab) corresponds to the image of A under a rotation through
about B;
given a real scalar , b + (a b) corresponds to thee image of A under a homothety
of ratio centered at B;
the absolute value of a b equals AB;
(c b)
equals ABC (directed and modulo 2).
the argument of
(a c)
Using these facts, one can translate much of the language of geometry in the Euclidean
plane into language about complex numbers.
Congruence. For integers a, b, and n with n 1, a b (mod n) (or a is congruent to
b modulo n) means that a b is divisible by n.
Concave up (down) function. A function f (x) is concave up (down) on [a, b] R
if f (x) lies under (above) the line connecting (a1 , f (a1 )) and (b1 , f (b1 )) for all
a a1 < x < b1 b.

A function g(x) is concave up (down) on the Euclidean plane if it is concave up (down)


on each line in the plane, where we identify the line naturally with R.
Concave up and down functions are also called convex and concave, respectively.
Convex hull. Given a nonempty set of points S in Euclidean space, there exists
a convex set T such that every convex set containing S also contains T . We call T the

1.2. Glossary

convex hull of S.
Cyclic polygon. Polygon that can be inscribed in a circle.
De Moivres formula. For any angle a and for any integer n,
(cos a + i sin a)n = cos na + i sin na.

Derangement. A derangement of n items a1 , ..., an is a permutation (b1 , b2 , ..., bn ) of


these items such that bi ai for all i. According to a formula of Eulers, there are exactly
n!
derangements of n items.

n!
n! n! n!
+ + + (1)n
1! 2! 3!
n!

Desargues theorem. Two triangles have corresponding vertices joined by lines


which are concurrent or parallel if and only if the intersections of corresponding sides are
collinear.
Directed angles. A directed angle contains information about both the angles
measure and the angles orientation (clockwise or counterclockwise). If two directed angles
sum to zero, then they have the same angle measure but opposite orientations. One often
takes directed angles modulo or 2. Some important features of directed angles modulo
p follow:
If A, B, C, D are points such that ABC and ABD are welldefined,
then ABC = ABD if and only if B, C, D are collinear.
If A, B, C, D are points such that ABC and ADC are welldefined,
then ABC = ADC if and only if A, B, C, D are concyclic.

Because 2() = 2( + ), but + , one cannot divide directed angles by 2. For


2
2
example, if ABC = 2ADC, D lies either on the internal angle bisector of angle ABC,
1
or on the external angle bisector of angle ABC we cannot write ADC = ABC to
2
determine which line D lies on.
These features show that using directed angles modulo allows one to deal with multiple possible configurations of a geometry problem at once, but at the expense of possibly
losing important information about a configuration.
Eulers formula. (for planar graphs)
If F, V, and E are the number of faces, vertices, and edges of a planar graph, then
F + V E = 2. This is a special case of an invariant of topological surfaces called the Euler
characteristic.
Eulers formula. (in planar geometry). Let O and I be the circumcenter and incenter, respectively, of a triangle with circumradius R and inradius r. Then OI 2 = R2 2rR.

Chapter 1. Introduction

Euler line. The orthocenter, centroid and circumcenter of any triangle are collinear.
The centroid divides the distance from the orthocenter to the circumcenter in the ratio
of 2 : 1. The line on which these three points lie is called the Euler line of the triangle.
Eulers theorem. Given relatively prime integers a and m with m 1.
a(m) a(modm), where (m) is the number of positive integers less than or equal to
m and relatively prime to m. Eulers theorem is a generalization of Fermats little theorem.
Excircles or escribed circles. Given a triangle ABC, there are four circles tangent to the lines AB, BC, CA. One is the inscribed circle, which lies in the interior of
the triangle. One lies on the opposite side of line BC from A, and is called the excircle
(escribed circle) opposite A, and similarly for the other two sides. The excenter opposite
A is the center of the excircle opposite A; it lies on the internal angle bisector of A and
the external angle bisectors of B and C.
Excenters. See excircles.
Exradius. The radius of the three excircles of a triangle.
n

Fermat number. A number of the form 22 for some positive integer n.


Fermats little theorem. If p is prime, then ap a(modp) for all integers a.
Feuerbach circle. The feet of the three altitudes of any triangle, the midpoints of
the three sides, and the midpoints of segments from the three vertices to the orthocenter,
all lie on the same circle, the Feuerbach circle or the nine-point circle of the triangle.
Let R be the circumradius of the triangle. The nine-point circle of the triangle has
radius R/2 and is centered at the midpoint of the segment joining the orthocenter and
the circumcenter of the triangle.
Feuerbachs theorem. The nine-point circle of a triangle is tangent to the incircle
and to the three excircles of the triangle.
Fibonacci sequence. The sequence F0 , F1 , ... defined recursively by F0 = 0, F1 = 1,
and
Fn+2 = Fn+1 + Fn for all n 0.
Generating function. If a0 , a1 , a2 , ... is a sequence of numbers, then the generating
function for the sequence is the infinite series
a0 + a1 x + a2 x2 + .
If f is a function such that
f (x) = a0 + a1 x + a2 x2 + ,

1.2. Glossary

then we also refer to f as the generating function for the sequence.


Graph. A graph is a collection of vertices and edges, where the edges are distinct
unordered pairs of distinct vertices. We say that the two vertices in one of these unordered
pairs are adjacent and connected by that edge.
The degree of a vertex is the number of edges which ontain it.
A path is a sequence of vertices v1 , v2 , ..., vn such that vi is adjacent to vi+1 for each i.
A graph is called connected if for any two vertices v and w, there exists a path from
v to w.
A cycle of the graph is an ordered collection of vertices v1 , v2 , ..., vn such that v1 vn
and such that the (vi , vi+1 ) are distinct edges.
A connected graph which contains no cycles is called a tree, and every tree contains
at least two leaves, vertices with degree 1.
Harmonic conjugates. Let A, C, B, D be four points on a line in that order.
If the points C and D divide AB internally and externally in the same ratio, (i.e.,
AC CB = AD DB), then the points C and D are said to be harmonic conjugates
of each other with respect to the points A and B, and AB is said to be harmonically
divided by the points C and D. If C and D are harmonic with respect to A and B, then
A and B are harmonic with respect to C and D.
Harmonic range. The four points A, B, C, D are referred to as a harmonic range,
denoted by (ABCD), if C and D are harmonic conjugates with respect to A and B.
Hellys theorem. If n > d and C1 , ..., Cn are convex subsets of Rd , each d+1 of which
have nonempty intersection, then there is a point in common to all the sets.
Herons formula. The area of a triangle with sides a, b, c is equal to

a+b+c
s(s a)(s b)(s c), where s =

H
olders inequality. Let w1 , ..., wn be positive real numbers whose sum is 1. For
any positive real numbers aij ,
( aij )
n

i=1

j=1

wi

i
aw
ij

j=1 i=1

Homothety. A homothety (central similarity) is a transformation that fixes one


point O (its center) and maps each point P to a point P for which O, P, P are collinear
and the ratio OP OP = k is constant (k can be either positive or negative), where k is
called the magnitude of the homothety.
Homothetic triangles. Two triangles ABC and DEF are homothetic if they have
parallel sides. Suppose that AB DE, BC EF, and CA F D. Then lines AD, BE, and
CF concur at a point X, as given by a special case of Desargues theorem. Furthermore,

Chapter 1. Introduction

some homothety centered at X maps triangle ABC onto triangle DEF.


Incenter. Center of inscribed circle.
Incircle. Inscribed circle.
Inversion of center O and ratio r. Given a point O in the plane and a real number
r > 0, the inversion through O with radius r maps every point P O to the point P on

the ray OP such that OP OP = r 2 . We also refer to this map as inversion through ,
the circle with center O and radius r. Key properties of inversion are:
1. Lines through O invert to themselves (though the individual points on the line are
not all fixed).
2. Lines not through O invert to circles through O and vice versa.
3. Circles not through O invert to other circles not through O.
4. A circle other than inverts to itself (as a whole, not point-by-point) if and only
if it is orthogonal to , that is, it intersects and the tangents to the circle and to at
either intersection point are perpendicular.
Isogonal conjugate . Let ABCbe a triangle and let P be a point in the plane which
does not lie on any of the lines AB, BC, and CA. There exists a unique point Q in the
plane such that ABP = QBC, BCP = QCA, and CAP = QAB, where the
angles in these equations are directed modulo . We call Q the isogonal conjugate of P .
With this definition, we see that P is also the isogonal conjugate of Q.
Jensens inequality. If f is concave up on an interval [a, b] and 1 , 2 , ..., n are
nonnegative numbers with sum equal to 1, then
1 f (x1 ) + 2 f (x2 ) + + n f (xn ) f (1 x1 + 2 x2 + + n xn )

for any x1 , x2 , ..., xn in the interval [a, b]. If the function is concave down, the inequality
is reversed.

a
Kummers Theorem. Given nonnegative integers a and b and a prime p, pt Ca+b
if
and only if t is less than or equal to the number of carries in the addition a + b in base p.

Lattice point. In the Cartesian plane, the lattice points are the points (x, y) for
which x and y are both integers.
Law of cosines. In a triangle ABC, CA2 = AB 2 + BC 2 2AB BC cos ABC, and
analogous equations hold for AB 2 and BC 2 .
Law of quadratic reciprocity. If p, q are distinct odd primes, then
(p1)(q1)
p q
( ) ( ) = (1) 4 .
q p

1.2. Glossary

q
p
where ( ) and ( ) are Legendre symbols.
q
p

Law of sines. In a triangle ABC with circumradius equal to R one has

sin A sin B sin C


=
=
= 2R.
BC
AC
AB
Legendre symbol. If m is an integer and n is a positive prime, then Legendre symm
bol ( ) is defined to equal 0 if n m, 1 if m is a quadratic residue modulo n, and 1 if
n
m is a quadratic nonresidue modulo n.
Lucass theorem. Let p be a prime; let a and b be two positive integers such that
a = ak pk + ak1 pk1 + + a1 p + a0 ,
b = bk pk + bk1 pk1 + + b1 p + b0 ,

where 0 ai , bi < p are integers for i = 0, 1, ..., k. Then

Cab11 Cab00 (mod p)


Cabk2
Cab Cabkk Cabk1
k2
k1

Matrix. A matrix is a rectangular array of objects. A matrix A with m rows and


n columns is an m n matrix. The object in the ith row and j th column of matrix A is
denoted ai,j . If a matrix has the same number of rows as it has columns, then the matrix
is called a square matrix. In a square n n matrix A, the main diagonal consists of the
elements a1,1 , a2,2 , ..., an,n .
Menelaus theorem. Given a triangle ABC, let F, G, H be points on lines BC, CA, AB,
respectively. Then F, G, H are collinear if and only if, using directed lengths,
AH BF CG

= 1.
HB F C GA

Minkowskis inequality. Given a positive integer n, a real number r 1, and


positive reals a1 , a2 , ..., an and b1 , b2 , ..., bn , we have
1
r

1
r

1
r

((ai + bi )r ) ( ari ) + ( bri ) .


n

i=1

i=1

i=1

Multiset. Informally, a multiset is a set in which an element may appear more than
once. For instance, {1, 2, 3, 2} and {2, 2, 2, 3, 1} are distinct multisets.
Nine point circle. (See Feuerbach circle.)

Orbit. Suppose that S is a collection of functions on a set T , such that S is closed


under composition and each f S has an inverse. T can be partitioned into its orbits
under S, sets of elements such that a and b are in the same set if and only if f (a) = b for
some f S.

10

Chapter 1. Introduction

Order. Given a nonzero element g of a finite field, there exists a smallest positive
integer d, named the order of g, such that g d = 1.
Orthocenter of a triangle. Point of intersection of the altitudes.
Pascals theorem. If ABCDEF is a hexagon inscribed in a conic in the projective
plane, such that each pair of opposite sides intersects at most one point, then the three
intersection points formed in this manner are collinear. (If the hexagon is inscribed in a
conic in the afine plane, then either the above result holds, or else each pair of opposite
sides is parallel.)
This theorem the dual to Brianchons theorem.
Pells equations. If D is a prime congruent to 3 modulo 4, then the Diophantine
equation
x2 Dy 2 = 1
in x and y is known as a Pells equation. This equation has infinitely many integer solutions in x and y.
Phi function. For all positive integers n, (n) is defined to be the number of integers in {1, 2, ..., n} which are relatively prime to n. This function is multiplicative, that
is, (ab) = (a)(b) for all a, b relatively prime.
Periodic. Function f (x) is periodic with period T > 0 if
f (x + T ) = f (x) for all x.

Permutation. Let S be a set. A permutation of S is a one-to-one function S S


that maps S onto S. If S = {x1 , x2 , ..., xn } is a finite set, then we may denote a permutation of S by {y1, y2 , ..., yn }, where yk = (xk ).

Picks theorem. Given a non self-intersecting polygon P in the coordinate plane


whose vertices are at lattice points, let B denote the number of lattice points on its
boundary and let I denote the number of lattice points in its interior. The area of P is
given by the formula
1
I + B 1.
2

Pigeonhole principle. If n objects are distributed among k < n boxes, some box
contains at least two objects.
Pole-polar transformation. Let C be a circle with center O and radius R. The
pole-polar transformation with respect to C maps points different from O to lines, and
lines that do not pass through O to points. If P O is a point then the polar of P is the

11

1.2. Glossary

line p that is perpendicular to ray OP and satisfies


d(O, P )d(O, p ) = R2 ,

where d(A, B) denote the distance between the objects A and B.


If q is a line that does not pass through O, then the pole of q is the point Q that has
polar q. The pole-polar transformation with respect to the circle C is also called reciprocation in the circle C.
Polynomial in x of degree n. Function of the form
f (x) = a0 + a1 x + a2 x2 + + an xn = ak xk

with an 0.

k=1

Power of a point theorem. Suppose that we are given a fixed point P which
lies either outside, on, or inside a fixed circle C with center O and radius r. Draw a line
through P which intersects the circle at X and Y . The power of the point P with respect
to C is defined to be the product of the signed distances P X and P Y.
The power of a point theorem states that this quantity is a constant; i.e., it does not
depend on which line was drawn. More precisely,
P X P Y = P O2 r2
no matter which line is drawn.
Power mean inequality. Let a1 , a2 , ..., an be any positive numbers for which
a1 + a2 + + an = 1. For positive numbers x1 , x2 , ..., xn we define
M = min{x1 , x2 , ..., xk },
M+ = max{x1 , x2 , ..., xk },
M0 = xa11 xa22 xann ,

Mt = (a1 xt1 + a2 xt2 + + ak xtk ) t ,


1

where t is a non-zero real number. Then

M Ms Mt M+ for all s t.
Primitive element. For each prime p, a field F with p elements contains an element
g, called a primitive element of F , with the following property: for any nonzero element
h of F , there exists an integer k such that g k = h.
Projective plane. Let K be a field. The projective plane over K is the set of equivalence classes of K3 {(0, 0)}, under equivalence by scalar multiplication (that is, where
(a, b, c) and (d, e, f ) are equivalent if and only if (a, b, c) = (d, e, f ) for some K).
The elements of K are called points, and the equivalence class containing (a, b, c) is
often denoted [a, b, c] or [a b c]. Also, given (, , ) K3 {(0, 0)}, the set of solutions
[x, y, z] to
x + y + z = 0

12

Chapter 1. Introduction

is called a line in the projective plane over K. Any two distinct points (resp. lines) are
said to intersect in or lie on a unique line (resp. point).
Ptolemys theorem. In a convex cyclic quadrilateral ABCD,
AC BD = AB CD + AD BC.
Radical axis. Let 1 and 2 be two non-concentric circles. The locus of all points of
equal power with respect to these circles is called the radical axis of 1 and 2 .
Radical axis theorem. Let 1 , 2 , 3 be three circles whose centers are not collinear.
There is exactly one point whose powers with respect to the three circles are all equal.
This point is called the radical center of 1 , 2 , 3 .
Root of an equation. Solution to the equation.
Root of unity. Solution to the equation z n 1 = 0.
Root Mean Square-Arithmetic Mean Inequality. For positive numbers x1 , x2 , ..., xn ,

x21 + x22 + + x2n x1 + x2 + + xn

n
n
Sigma function. For all positive integers n, (n) is defined to be the sum of all
positive integer divisors of n. This function is multiplicative, that is, (ab) = (a)(b) for
all a, b relatively prime.
Simson line. For any point P on the circumcircle of ABC, the feet of the perpendiculars from P to the sides of line called the Simson line of P with respect to ABC all
lie on a line called The Simson line of P with respect to ABC.
Solid triangle inequality. Given four points A, B, C, P in three-dimensional space
which are not coplanar, we have
AP B + BP C > AP C.
Stewarts theorem. In a triangle ABC with cevian AD, write
a = BC, b = CA, c = AB, m = BD, n = DC, and d = AD.
Then

d2 a + man = c2 n + b2 m.

This formula can be used to express the lengths of the altitudes and angle bisectors
of a triangle in terms of its side lengths.

13

1.2. Glossary

Thue-Morse sequence. The sequence t0 , t1 , ..., defined by t0 = 0 and the recursive


relations
t2k = tk , t2k+1 = 1 t2k for all k 1.
The binary representation of n contains an odd number of 1 s if and only if tn is odd.
Triangular number. A number of the form
teger.

n(n + 1)
, where n is some positive in2

Trigonometric identities.
sin2 x + cos2 x = 1,
1 + cot 2 x =

1
= csc2 x,
2
sin x
1
2
= sec 2 x;
tan x + 1 =
2
cos x

addition and subtraction formulas:


sin(a b) = sin a cos b cos a sin b,
cos(a b) = cos a cos b sin a sin b,
tan a tan b
tan (a b) =
;
1 tan atan b
double-angle formulas:
sin 2a = 2 sin a cos a =

2tan a
,
1 + tan 2 a

cos 2a = cos2 a sin2 a = 2 cos2 a 1 = 1 2 sin2 a =


tan 2a =

2tan a
;
1 tan 2 a

triple-angle formulas:
sin 3a = 3 sin a 4 sin3 a,
cos 3a = 4 cos3 a 3 cos a,
3tan a tan 3 a
tan 3a =
;
1 3tan 2 a
half-angle formulas:
sin2

a 1 cos a
=
2
2

1 tan 2 a
1 + tan 2 a

14

Chapter 1. Introduction
cos2

a 1 + cos a
=
;
2
2

sum-to-product formulas:
sin a + sin b = 2 sin

ab
a+b
cos
,
2
2
ab
a+b
cos
,
cos a + cos b = 2 cos
2
2
tan a + tan b =

sin(a + b)
;
cos a cos b

difference-to-product formulas:
sin a sin b = 2 cos

ab
a+b
sin
,
2
2
a+b
ab
cos a cos b = 2 sin
sin
,
2
2
tan a tan b =

sin(a b)
;
cos a cos b

product-to-sum formulas:
2 sin a cos b = sin(a + b) + sin(a b),
2 cos a cos b = cos(a + b) + cos(a b),
2 sin a sin b = cos(a + b) + cos(a b).
Wilsons theorem. If n > 1 be a positive integer, then

if and only if n is prime.

(n 1)! = 1 (mod n)

Zeckendorf representation. Let F0 , F1 , ... be the Fibonacci numbers 1, 2, .... Each


nonnegative integer n can be written uniquely as a sum of nonconsecutive positive Fibonacci numbers; that is, each nonnegative integer n can be written uniquely in the form

n = k Fk
k=0

where ak {0, 1} and (ak , ak+1 ) (1, 1) for each k.


This expression for n is called its Zeckendorf representation.

Chapter 2
Examples for practice
2.1
2.1.1

Arithmetic problems
Junior problems

Example 2.1. Prove that among seven arbitrary perfect squares there are two whose
diference is divisible by 20.
First solution
It is easy to check that perfect squares can give one of the following residues:
1, 2, 4, 8, 16 (mod 20).

By the Pigeonhole principle we conclude that among seven perfect squares we must
have at least two that have the same residue modulo 20. Hence their diference is divisible
by 20 and our proof is complete.
Second solution
Note that for all integer x we have x2 1, 2, 4, 8, 16 (mod 20) and we have six distinct
possible residues. If we have seven arbitrary perfect squares
x21 , x22 , x23 , x24 , x25 , x26 , x27 ,
by the pigeonhole principle, there are two squares x2i and x2j with the same residue and
they satisfy the requirement.
Third solution
Observe that by the Pigeonhole Principle, there are at least four perfect squares which
all have the same parity. Now, note that for any integer n, we have n2 = 1, 0, 1 (mod 5).
Again by the Pigeonhole Principle, out of these four perfect squares, we have at least two
perfect squares, say a2 and b2 , such that a2 = b2 (mod 5). This implies that 5 (a2 b2 ).
Also, 2(a b) and 2(a + b) since both a and b have the same parity. Hence, 4 (a2 b2 ),
but gcd(5, 4) = 1, thus we have 20 (a2 b2 ), and we are done.
Example 2.2. Find all nine-digit numbers aaaabbbb that can be written as a sum of fifth
powers of two positive integers.
Solution
15

16

Chapter 2. Examples for practice

By Newtons binomial formula, (5u+v)5 = v 5 (mod 25) for any integers u, v, or any fifth
power n5 leaves remainders 7, 1, 0, 1, 7 modulus 25, when n is respectively congruent
to 3, 4, 0, 1, 2 modulus 5. Since aaaabbbbb = m5 + n5 for integers m, n, and m5 + n5 leaves
remainders 0, 1, 2, 6, 7, 8, 14, then b {0, 1, 3, 4, 7, 9} for the last two digits of m5 +n5
to be equal.
Since 15 = 1 (mod 11), 25 = 32 = 1 (mod 11), 35 = 243 = 1 (mod 11), 45 = 1024 =
1 (mod 11) and 55 = 3125 = 1 (mod 11), then m5 + n5 leaves re- mainders 2, 1, 0, 1, 2
modulus 11. Moreover, aaaabbbb0 is clearly a multiple of 11, and since no digit may be
congruent to 1 modulus 11, then b {9, 0, 1, 2}. Together with the previous result, we
may conclude that the possible values for b are b = 9 with wlog m = 0 (mod 5) and
n = 4 (mod 5) and exactly one of m, n is a multiple of 11, or b = 0 with m + n a multiple
of 5, or b = 1 with m, n = 3 (mod 5) and exactly one of m, n a multiple of 11.
If b = 1, clearlym, n cannot finish in the same digit or otherwise m5 + n5 would finish
in 8, hence wlog m ends in 3 and n ends in 8. Since 885 > 999999999, m = 33 must be a
multiple of 11, and n = 10u + 8 for some digit u. Now, 85 ends in 68 and 333 ends in 93,
but (10u + 8)5 85 is a multiple of 100, so m5 + n5 formed in this way ends in 61, not in
11. No solution is possible in this case.
If b = 9, either m is a multiple of 10 and n5 finishes in 99999, or m is odd and finishes
in 5 and n5 must finish in 4. In the first case, a digit u must exist such that (10u+9)5 must
end in 99999, or since 95 = 59049, 50 94 u = 328050u, hence 50u must end in 950, yielding
u = 1 (mod 20), impossible for a digit. In the second case, m = 10u + 5 and n = 10v + 4,
and m5 + n5 ends in the same two digits as 55 + 45 + 55 10u + 44 50v, which are also the same
three digits as
125 + 24 + 250u + 800v = 800v + 250u + 149.
Clearly u must be odd, and 250u ends in 750 or 250, and 800v must finish respectively
in 100 (impossible) or in 600, hence v = 2 or v = 7. But 745 > 999999999, or n = 24 not a
multiple of 11, yielding m = 55. Now, 555 + 245 = 511246999, and no solution exists in this
case.
Finally, if b = 0, then m + n is a multiple of 5 with multiplicity 2 = a = 1, because
if m + n is a multiple of 53 , either m + n = 250 and either m5 or n5 exceeds 1010 , or
m + n = 125, and wlog m5 ends in 0 and n5 ends in 5, absurd. Note that 55 divides
m5 + n5 = (m + n)5 5mn(m + n)3 + 5m2 n2 (m + n), yielding either a = 1 and m2 n2 is a
multiple of 54 , or a = 2 and m2 n2 is a multiple of 52 . In either case, either m or n is a
multiple of 5, hence so is the other, and clearly one ends in 0 iff the other one ends in 0,
hence either m = 10u + 5 and n = 10v + 5 or m = 10u and n = 10v. In the first case, the last
four digits of m5 + n5 are also the last four digits of 53 1000(u2 + v 2 ) + 55 10(u + v) + 2 55 ,
hence the last four digits of 5000(u2 + v 2 1) + 1250(u + v + 1). Now, u + v + 1 must be a
multiple of 4 for the last three digits to be 0, hence u2 + v 2 1 is even, and 1250(u + v + 1)
must end in 0000, yielding u + v + 1 = 8, since u + v + 1 = 16 produces either u or v = 8, and
855 > 999999999. Trying all possible alternatives with u + v = 7, we find that u = 6 results
in m5 > 999999999, u = 5 and v = 2 results in m5 + n5 = 513050000, and u = 4 and v = 3
results in 237050000, so no solution exists in this case. Finally, when m = 10u and n = 10v,
the problem is equivalent to finding all digits u, v such that u5 + v 5 = aaaa. Combining the
numbers whose fifth powers do not exceed 104, we find that this result is only true when
65 + 15 = 7777.
We conclude that the only number of the form aaaabbbbb that is the sum of two fifth

17

2.1. Arithmetic problems

powers is 777700000 = 605 + 105 .

()

Example 2.3. Let n be an integer greater than 2. Find all real numbers x such that
{x} = {nx}, where {a} denotes the fractional part of a.
Solution
Lets suppose k = x < k + 1 and then nk = nx < nk + n. We write
n1
n1
r+1
r
) = Ir .
[k, k + 1) = [k + , k +
n
n
r=1
r=1

r
For x Ir the inequality {x} {nx} reads as x k nx nk r that is x k +
n1
The compatibility conditions are
x Ir k +

r
r+1
r
k+
k+
n
n1
n

and we conclude that:


1) if x I0 , it satisfies {x} = {nx},

2) if x Ir , r 1, only those x in the subinterval Jr Ir , Jr = [k +

satisfy the inequality.

r+1
r
,k +
)
n1
n

()

Example 2.4. Find all quadruples (x, y, z, w) of integers satisfying the system of equations
x + y + z + w = xy + yz + zx + w 2 w = xyz w 3 = 1.
First solution
Note that
(x + y)(y + z)(z + x) = (x + y + z)(xy + yz + zx) xyz = 2.

This implies that (x + y, y + z, z + x) equals some permutation of (1, 1, 2), (1, 1, 2)


or (1, 1, 2). It follows that (x, y, z) equals some permutation of (0, 1, 1), (1, 1, 2) or
(0, 1, 2).
The first case implies that w = 1 (x + y + z) = 3 and w 3 = 1 + xyz = 1, which is a
contradiction. The second case implies that
w = 1 (x + y + z) = 1, w 2 w = 1 (xy + yz + zx) = 2

and w 3 = xyz + 1 = 1, from where it follows that w = 1. The third case implies that
w = 1(x+y+z) = 2 and w 3 = xyz+1 = 1, which is a contradiction. It follows that all posible quadruples (x, y, z, w) are (1, 1, 2, 1) with all posible permutations among x, y, z ()
Second solution
Note first that

(x+w +1)(y +w +1)(z +w +1) = w 3 1+(w +1)(w 2 +w 1)+(w +1)2 (w 1)+(w +1)3 = 2,
or x + w + 1 divides 2, hence x may take one of the following values:

18

Chapter 2. Examples for practice


x = w 3; then y + z = 1 w x = 2, and

yz = 1 + w w 2 + 2(w + 3) = w 2 + 3w + 5,

leading to w 3 1 = xyz = (w + 3)(w 2 3w 5) = w 3 14w 15, or w = 1, for x = 2 and


yz = 1. Note that (y z)2 = (y + z)2 4yz = 0, or y = z = 1 because y + z = 2.
x = w 2; then y + z = 1 w x = 1, and yz = 1 + w w 2 + (w + 2) = 1 + 2w w 2 ,
leading to
w 3 1 = xyz = (w + 2)(w 2 2w 1) = w 3 5w 2;

no solution in integers exists in this case because no integer w satisfies 5w = 1.


x = w; then y + z = 1 w x = 1, and yz = 1 + w w 2 w = w 2 1, leading to
w 3 1 = xyz = w 3 + w, or w = 1, for x = 1 and yz = 2.
Note that (y z)2 = (y + z)2 4yz = 1 + 8 = 32 for y z = 3, with solutions y = 1 and
z = 2, or y = 2 and z = 1.
x = 1 w; then y + z = 1 w x = 2, and
yz = 1 + w w 2 + 2(1 w) = 1 w w 2 ,

leading to w 3 1 = xyz = (w 1)(w 2 + w 1) = w 3 2w + 1, or w = 1, for x = 0 and yz = 1.


Note that (y z)2 = (y + z)2 4yz = 8 is not a perfect square, hence no solution in integers
exist in this case.
Restoring generality, all solutions are w = 1, and (x, y, z) a permutation of (1, 1, 2).
()
Example 2.5. Find the sequences of integers (an )n0 and (bn )n0 such that

n
1+ 5
(2 + 5) = an + bn
2
for each n 0.

First solution
Let
pn =

(2 +

n
n
5) + (2 5)
2

n
n
5) (2 5)
qn =

2 5

n
for n = 1, 2, . . .. Then (2 + 5) = pn + qn 5, n = 0, 1, 2, . . . and both obtained sequences
satisfy the same recurrence
xn+1 = 4xn + xn1 , n N
(2.1)
and

(2 +

with initial conditions p0 = 1, p1 = 2, q0 = 0, q1 = 1. It is clear that (pn )n0 and (qn )n0 are
sequences of nonnegative integers and since

bn
1+ 5
bn
5 = pn + qn 5
an + bn
= pn + qn 5 (an + ) +
2
2
2

19

2.1. Arithmetic problems

bn

an + 2
b

a
{ n
bn

= pn
= qn

= pn qn
n N.
= 2qn

we have that (an )n0 and (bn )n0 are sequences of integers and can be defined independently by recurrence (2.1) with initial conditions a0 = 1, a1 = 1, b0 = 0, b1 = 2. In explicit
form

n
n
n
n
( 5 1) (2 + 5) + ( 5 + 1) (2 5)
(2 + 5) (2 5)
; an =
.
()

bn =
5
5
Second solution

Using Newtons
binomial
formula,
exchanging
5
by

5 in the first term results in

exchanging 5 by 5 in the second term, i.e.,


yielding

(2 +

bn =

n
n
n

n
5) + (2 5) = 2an + bn and (2 + 5) (2 5) = bn 5,

(2 +

n
n
n
( 5 1) (2 + 5) + ( 5 + 1) (2 5)
5) (2 5)
; an =
.

5
5

()

The fact that an , bn are integers for all n may be easily proved considering that they
are solutions of the recursive equations an = 4an1 + an2 and bn = 4bn1 + bn2 , with initial
conditions a0 = a1 = 1 and b0 = 0, b1 = 2.
()
Example 2.6. Prove that for each positive integer k the equation
x31 + x32 + + x3k + x2k+1 = x4k+2

has infinitely many solutions in positive integers with x1 < x2 < < xk+1 .
First solution
Since

13 + 23 + + k 3 =

then by substitution

k 2 (k + 1)2
= (1 + 2 + + k)2
4

x1 = x, x2 = 2x, ..., xk = kx, xk+1 = (1 + 2 + + k)y 2 , xk+2 = (1 + 2 + + k)y

in original equation, we obtain

(13 + 23 + + k 3 ) x3 + (1 + 2 + + k)2 y 4 = (1 + 2 + + k)4 y 4 x3 = ay 4

20

Chapter 2. Examples for practice

(k 1)(k + 2)(k 2 + k + 2)
. Since x = a7 n4 , y = a5 n3 for
4
any positive integer n , then x3 = ay 4 . Hence
where a = (1 + 2 + + k)2 1 =

x1 = a7 n4 , x2 = 2a7 n4 , ..., xk = ka7 n4 , xk+1 = (1 + 2 + + k)a1 0n6 , xk+2 = (1 + 2 + + k)a5 n3

for any positive integer n is a solution to the original equation and obviously, x1 < x2 <
< xk+1 .
()
Second solution
For any positive integer n we have the well-known identity:

that is

13 + 23 + + n3 + (n + 1)3 + + (n + k)3 = (

(n + k)(n + k + 1)
) ,
2
2

(n + k)(n + k + 1)
n(n + 1)
) + (n + 1)3 + + (n + k)3 = (
)
(
2
2
2

Consider the positive integers n such that the triangular number tn+k = (n+k)(n+k+1)
is
2
2
a perfect squares. There are infinitely many such integers since the relation tn+k = u is
equivalent to the Pells equation (2n + 2k + 1)2 2u2 = 1. The fundamental solution to this
Pell equation is (3, 2), hence all these integers are given by the sequence (ns ), where

2ns + 2k + 1 + us 2 = (3 + 2 3)s

for s big enough such that ns 1.


We can take

x1 = ns + 1, , xk = ns + k, xk+1 = ns (ns + 1), xk+2 = us .

It is clear that for s big enough we have ns 1 and

infinite family of solutions.

n(n + 1)
> n + k, hence we get an
2
()

Example 2.7. Find all triples (x, y, z) of integers satisfying the system of equations

z2

(x2 + 1) (y 2 + 1) +

10

(x + y)(xy 1) + 14z

= 2010

= 1985.

Solution
Note that z = 10k for some integer k because

z2
= 2010 (x2 + 1) (y 2 + 1)
10

is an integer. Let p = x + y and q = xy 1. Then

(x2 + 1) (y 2 + 1) = x2 y 2 + x2 + y 2 + 1 = (xy 1)2 + (x + y)2 = p2 + q 2

21

2.1. Arithmetic problems

and the system becomes


p2 + q 2 + 10k 2
{
pq + 140k

= 2010
p2 + q 2
{
= 1985
pq

= 2010 10k 2
= 1985 140k

(iii)

Since (p q)2 = 2010 10k 2 2(1985 140k) = 10(k 14)2 then only k = 14 can provide
solvability to (iii). And for k = 14, (iii) becomes
{

Hence,

p2 + q 2
pq

x+y
{
xy

= 50
p = q = 5.
= 25

=5
x =4
x =1
{
or {
=4
y =1
y =4

and triples (4, 1, 140), (1, 4, 140) are all integer solutions of the original system in integers.
()

Example 2.8. Let n be a positive integer. Find the least positive integer a such that the
system
x + x2 + + xn = a
{ 12
x1 + x22 + + x2n = a.
has no integer solutions.

Solution
First, we notice that if xi 0, 1 for an integer component xi then x2i > xi and we have
a contradiction
a = x21 + x22 + + x2n > x1 + x2 + + xn = a.

Hence any component xi is 0 or 1 and the system has integer solutions for a = 1, ..., n:
take x1 = = xa = 1 and xa+1 = = xn = 0. Therefore the least positive integer a such
that the system has no integer solutions is n + 1:
x1 + x2 + + xn x21 + x22 + + x2n n < a = n + 1.

()

Example 2.9. Let n be a positive integer relatively prime with 10. Prove that the hundreds
digit of n20 is even.
Solution
If n is prime with 10 it is prime with 5 and 2.
Since (52 ) = 54 = 20 and (23 ) = 22 1 = 4, where (n) is Eulers totient function, then
n20 = 1 (mod 25) and n20 = (n4 )5 = 15 = 1 (mod 8), or n20 = 1 (mod 200), and uniqueness
of this residue modulus 200 is guaranteed by the Chinese Remainder Theorem, hence the
last three digits of n20 are a01, where a is even. The conclusion follows.
()
Example 2.10. Find all integers n for which 9n + 16 and 16n + 9 are both perfect squares.
Solution
If 9n + 16 and 16n + 9 are both perfect squares then n 0 and the number
pn = (9n + 16)(16n + 9) = (12n)2 + (92 + 162 )n + 122

22

Chapter 2. Examples for practice

is also a perfect square. Since

(12n + 12)2 (12n)2 + (92 + 162 )n + 122 < (12n + 15)2

it follows that if n > 0 then we must have pn = (12n + 13)2 or pn = (12n + 14)2 . The former
condition gives n = 1 and the latter, n = 52. Therefore, n = 0, n = 1, and n = 52 are the
only integers n for which the expressions 9n+16 and 16n+9 simultaneously return perfect
squares.
()
Example 2.11. Find all n for which there are n consecutive integers whose sum of squares
is a prime.
Solution
If n = 1 any integer square cannot be a prime, so n is at least 2. In cases n = 2 and
n = 3 we can find consecutive numbers 1, 2 and 2, 3, 4 for which sum of squares is 5 and 29,
so prime numbers. Suppose now n > 3. If a Z such that A = a2 + (a + 1)2 + + (a + n 1)2
is prime, then
A = na2 + 2(1 + + (n 1))a + 12 + + (n 1)2 = n (a2 + (n 1)a +

(n 1)(2n 1)
) = p B.
6

Case 1: n is prime.
Because n > 3, we have 2 n, 3 n, so 6 n, and because A is prime it follows
6 (n 1)(2n 1), B = 1 and A = n. But A is a sum of n > 3 integer consecutive squares,
so A > n, contradiction.
Case 2: n is not prime.
If p prime, p n, p > 3, we get

m(pm 1)(2pm 1)
) = pB = p,
6
because A is prime. But A is a sum of n > 3 integer consecutive squares, so A > n > p,
contradiction. We get n = 2a3 b, a = 1, b = 1.
If a = 2, n = 4m, A = 2(2ma2 + 2m(4m 1)a + 3) = 2B = 2, because A is prime. But for
the same above reason, A > n > 2, contradiction.
If b = 2 we have the same approach. If n = 2 3 = 6, we can find integers 1, 0, 1, 2, 3, 4
for which the sum of squares is 31, prime number. Finally the answer is n {2, 3, 6}. ()
n = pm, A = p (ma2 + m(pm 1)a +

Example 2.12. Find all positive integers n for which (n2)!+(n+2)! is a perfect square.
Solution
Note the identity

(n 2)! + (n + 2)! = (n 2)! (n2 + n 1)2

Thus, if (n 2)! + (n + 2)! is a perfect square, then (n 2)! must also be a perfect
square. For n = 2, 3 we have that (n 2)! = 1, so n = 2 and n = 3 are solutions. In the
case of n 4, we note that the largest prime number smaller than or equal to n 2, which
we will denote by p, can only appear once in the prime factorization of (n 2)!, implying
that (n 2)! is never a perfect square for n 4. If it were to appear at least twice in the
factorization, then it is necessary to have n 2 = 2p. But, by Bertrands postulate, we
know that there is a prime number in between p and 2p, contradicting ps maximality.
Hence, n = 2 and n = 3 are the only solutions.
()

2.1. Arithmetic problems

23

Example 2.13. Find all primes p and q such that both pq 555p and pq +555q are perfect
squares.
First solution
Since pq 555p = p(q 555) is a perfect square, p divides q 555 and q > 555. Therefore
there exists an integer a 1 such that q 555 = ap
()
Likewise q divides p + 555, so there exists an integer b 1 such that p + 555 = bq ()
From () and () it follows that
p + 555 = b(555 + ap) (1 ab)p = 555(b 1) = 0 1 ab = 0 a = 1, b = 1.

Therefore q p = 555, so p = 2 (otherwise q p would be an even number) and q = 557. ()


Second solution
We have pq 555p = p(q 555) and pq + 555q = q(p + 555), hence p(q 555) and
q(p + 555), so
pq(q 555)(p + 555) = pq + 555q 555p 5552 .

Thus pq (3 5 37(p + 555 q)) . But note thatp + 555 = qk hence p(3 5 37 (k 1)).
If p = 3 3(q 555) 3q q = 3 contradiction since q > 555. Analogously for p = 5
and p = 37. Thus we arrive to p(k 1), hence k = ph + 1.
But q 555 = pr and hence
p + 555 = q(ph + 1) p = pqh + pr 1 = qh + r h = 0, r = 1 k = 1

Hence q = p + 555, if p 3 we obtain that p + 555 is even, contradiction, so p = 2 and


q = 557 is the solution to our problem.
()

Example 2.14. Find all pairs (x, y) for which x! + y! + 3 is a perfect cube.

Solution
Let x!+y!+3 = z 3 with x, y 7. We have that z 3 = 3 mod 7, so this gives a contradiction
since the cubic residues modulo 7 are only 0, 1, 6. Thus, we can suppose without loss of
generality that x 6.
Now let us do the case work:
A) x = 1
y! + 4 = z 3 and y 7 z 3 = 4 mod 7; impossible, so y 6, but for any value of y we
obtain a cube.
B) x = 2
y! + 5 = z 3 and y 7 z 3 = 5 mod 7; impossible, so y 6, and we obtain the solution
y = 5.
C) x = 3
y! + 9 = z 3 and y 7 z 3 = 2 mod 7; impossible, so y 6, and we obtain the solution
y = 6.
D) x = 4
y! + 27 = z 3 ; we postpone this for now.
E) x = 5
y! + 123 = z 3 and y 7 z 3 = 4 mod 7; impossible, so y 6, and we obtain the solution
y = 2.

24

Chapter 2. Examples for practice

F) x = 6
y! + 723 = z 3 and y 7 z 3 = 2 mod 7; impossible, so y 6, and we obtain the solution
y = 3.
Now, lets turn back to D). First, note that the case y = 1, 2, 3, ..., 8 can be easily
checked manually. For y > 8 we have 81y!; so z = 3z1 and y!/27 = z13 1. Now, 3z13 1,
soz = 1 mod 3.
But if z = 3k + 1; then
y!
= (3k + 1)3 1 = 27k 3 + 27k 2 + 9k = 9k(3k 2 + 3k + 1).
27

Thus y! = 243k(3k 2 + 3k + 1). Furthermore, let f (n) to be equal to the power of 3


dividing n and note that f (n) = log3 n, so
f (y!) = [

243k(3k 2 + 3k + 1)
243k(3k 2 + 3k + 1)
243k(3k 2 + 3k + 1)
]+[
]+[
] + ,
3
9
27

where [x] denotes as usual the integer part of x.


But f (y!) = f (243k(3k 2 + 3k + 1)) = 5 + f (k) = 5 + log3 k, so we can see that for k 1,
[

243k(3k 2 + 3k + 1)
] = 81k(3k 2 + k + 1) > 5 + log3 k
3

and hence the equation has no solution. Finally, we get that the solutions to original
equation are (x, y) = (2, 5); (x, y) = (5, 2); (x, y) = (3, 6); (x, y) = (6, 3).
()

Example 2.15. Prove that there are infinitely many pairs (p, q) of primes such that p6 +q 4
has two positive divisors whose difference is 4pq.
Solution
Let p = 2 and let q be odd prime number. Then

and we have

p6 + q 4 = 64 + q 4 = (8 + 4q + q 2 ) (8 4q + q 2 )
8 + 4q + q 2 8 4q + q 2 = 8q = 4pq.

()

Example 2.16. Prove that there is no n for which (k 4 + k 2 + 1) is a perfect square.


n

k=1

First solution
Let g(k) = k 4 + k 2 + 1. If we evaluate g(1), g(2), g(3), etc., we notice an interesting
pattern:
g(1) = 1 3, g(2) = 3 7, g(3) = 7 13, g(4) = 13 21,
and so on. This is because g(k)factors as g(k) = (k 2 k + 1)(k 2 + k + 1), and

g(k + 1) = ((k + 1)2 (k + 1) + 1)((k + 1)2 + (k + 1) + 1) = (k 2 + k + 1)(k 2 + 3k + 3).

So multiplying g(1), g(2), ..., g(n), we see that there are a sequence of squares in the
product (32 , 72 , 132 , ...), but with a final factor of n2 + n + 1 left over at the end. It follows
that since n2 + n + 1 cannot be a square (since it lies between n2 and (n + 1)2 ), the product

25

2.1. Arithmetic problems

cannot be a square.

()

Second solution

First, note that (k 4 + k 2 + 1) = 3, which is not a perfect square. Now, suppose n > 1.
1

Because

k=1

k 4 + k 2 + 1 = (k 2 + k + 1)(k 2 k + 1)

and (k + 1)2 (k + 1) + 1 = k 2 + k + 1, it follows that

(k 4 + k 2 + 1) = (n2 + n + 1) (k 2 + k + 1)2 .
n

n1

k=1

k=1

Moreover, because the difference between (n + 1)2 and n2 is 2n + 1, n2 + n + 1 cannot

be a perfect square for any n > 1. Thus, there is no n for which (k 4 + k 2 + 1) is a perfect
n

k=1

square.

()

Example 2.17. Call penta -sequence a sequence of consecutive positive integers such that
each of them can be written as a sum of five nonzero perfect squares. Prove that there are
infinitely many penta-sequences of length 7.
Solution
Consider, for each non-negative integer n, the numbers
(2n + 6)2 + (n + 5)2 + (n + 4)2 + (n + 2)2 + (n + 2)2 = 8n2 + 50n + 85,
(2n + 7)2 + (n + 4)2 + (n + 4)2 + (n + 2)2 + (n + 1)2 = 8n2 + 50n + 86,
(2n + 6)2 + (n + 5)2 + (n + 4)2 + (n + 3)2 + (n + 1)2 = 8n2 + 50n + 87,
(2n + 7)2 + (n + 5)2 + (n + 3)2 + (n + 2)2 + (n + 1)2 = 8n2 + 50n + 88,
(2n + 8)2 + (n + 4)2 + (n + 2)2 + (n + 2)2 + (n + 1)2 = 8n2 + 50n + 89,
(2n + 5)2 + (n + 6)2 + (n + 4)2 + (n + 3)2 + (n + 2)2 = 8n2 + 50n + 90,
(2n + 8)2 + (n + 4)2 + (n + 3)2 + (n + 1)2 + (n + 1)2 = 8n2 + 50n + 91.

Clearly, for each non-negative integer n, these numbers form different penta- sequences
of length 7. The conclusion follows.
()

n + n + 2009 is an integer.
Example 2.18. Find all positive integers n for which
First solution

n + n + 2009 = m for some integer m. Then


Suppose

m4 2009
n=
2m2

For n to be an integer, m must be odd with m2 2009.


Because 2009 = 72 41, it follows that m = 7
and n = 16.

n + n + 2009 is an integer.
Thus n = 16 is the only integer n for which

()

26

Chapter 2. Examples for practice

Second solution

Suppose there is a positive integer k such that n + n + 2009 = k 2 .

2009
Taking the conjugate we obtain n + 2009 n = 2
k

2009
2
Subtracting the above equations we get 2 n = k 2
k

From this equation we infer that n must be rational and hence integer, since n is
integer. Thus, k 2 is a divisor of 2009. However, the only square numbers that divide 2009
are 12 and 72 . It is clear that k = 1. If k = 7 we get n = 16 and it is easily verifed that

16 + 16 + 2009 = 7 is an integer.
()
Third
solution



For
n + n + 2009 to be an integer n, n + 2009, n+ n + 2009 must be perfect
squares. So let n = a2 , n + 2009 = b2 with a, b positive and a + b a perfect square. We have
b2 a2 = 2009 = 72 41 giving the systems
b a = 41
{
b + a = 49

or

b a = 2009
{
b + a = 1.

First system gives the solution n = 16. Second system implies a = 1004, contradiction.
Therefore, n = 16 is the only solution.
()

Example 2.19. How many positive integers n less than 2012 are divisible by [ 3 n].
Solution

Let a = [ 3 n] . Then a 3 n < a + 1, i.e. a3 n < (a + 1)3 . Because [ 3 n] divides n, it


follows that n = ab for some positive integer b. We get a3 = ab < (a + 1)3 , hence
1
(a + 1)3
= a2 + 3a + 3 + a2 + 3a + 4.
a
a

All positive integers n with the property that [ 3 n] divides n are given by
a2 b <

a3 , a3 + a, a3 + 2a, ..., a3 + (3a + 3)a, a Z .

Note that the largest a with a3 < 2012 is a = 12 and we have 123 + 12k 2012 is equivalent
2
to 12k 284, i.e. k 23 + , hence k = 23. The desired number is given by
3
(3a + 4) + 24 = 3
11

a=1

11 12
+ 44 + 24 = 198 + 44 + 24 = 266.
2

()

Example 2.20. Find all n for which the number of diagonals of a convex n-gon is a
perfect square.
Solution
Since the number of diagonals of a convex n-gon is
integer solutions to the Diophantine equation
n(n 3)
= m2
2

n(n 3)
, we seek to find positive
2

27

2.1. Arithmetic problems

that is

n2 3n 2m2 = 0

for n 3 and m 0. Note that n2 + m2 = 0 (mod 3) which implies that both n and m are
divisible by 3. So, by letting n = 3N and m = 3M , the equation becomes
N 2 N 2M 2 = 0

that is

(2N 1)2 2(2M)2 = 1.

Since the non negative solutions of the Pells equation X 2 2Y 2 = 1 are


k

Xk + Yk 2 = (3 + 2 2)

for k 0, we have that (note that Xk is odd)


nk = 3Nk =

3(Xk + 1)
for k 0,
2

that is 3, 6, 27, 150, 867, 5046....

()

Example 2.21. For an even integer n consider a positive integer N having exactly n2
divisors greater than 1. Prove that N is the fourth power of an integer.
First solution
The following results will be of use.
Lemma 1. If an odd prime p divides x2 + 1 for some integer x, then p 1 mod 4.
Proof: Since p is odd, p is of the form 4y + 1 or 4y + 3; showing that p = 3 mod 4 is
impossible will establish the claim. Assuming to the contrary that p = 4y + 3, we have
x2 = 1 mod p; (x2 )2y+1 = x4y+2 = xp1 = (1)2y+1 = 1 mod p

which is impossible since (noting that p cannot divide x) xp1 = 1 (by Fermats little
theorem).
()
Lemma 2. If 4x2 + 1 = f1 f2 ...fk , then for each factor fi , fi = 1 mod 4.
Proof. Since 4x2 + 1 is odd, it can only have odd divisors. Each factor is either 1
or some product of (odd) prime factors of 4x2 + 1, each of which must be congruent to
1 mod 4 by lemma 1. It follows that every factor is congruent to 1 mod 4.
()
Considering now the original problem, let n = 2m, then the condition that N has n2
divisors greater than 1 is equivalent to
(N) = (e1 + 1)(e2 + 1)...(ek + 1) = 4m2 + 1

where is the number of divisors function and the prime factorization of N is


N = pe11 pe22 ...pekk .

28

Chapter 2. Examples for practice

By lemma 2 each of the factors ei + 1 is of the form 4yi + 1; therefore each ei is a


multiple of 4. It follows that N is a fourth power.
()
Second solution
Let N be factored (into primes) as N = pe11 pekk .
It is well known that N has (e1 + 1) (ek + 1) = n2 + 1 divisors (including 1). Since n
is even, n2 + 1 1 ( mod 4), so ei + 1 must be either 1 mod 4 or 3 mod 4 for 1 i k. We
will show that n2 + 1 has no factors of the form 3 mod 4; thus ei + 1 1 ( mod 4) so ei is
divisible by 4 for all i, implying that N is a fourth power.
Proof: Assume for a contradiction that there exists some number p such that p = 4k+3
and n2 + 1 is divisible by p. WLOG let p be prime; otherwise it must have some prime
factor also of the form 3 mod 4 (since p is 3 mod 4, its prime factors cannot all be 1 mod 4).
Thus n2 1 ( mod p). Note that n must then be relatively prime to p; thus by Fermats
Little Theorem, np1 1 ( mod p). However,
np1 n4k+2 (n2 )

2k+1

(1)2k + 1 1 ( mod p).

Thus 1 1 ( mod p), but since p 3, we have the desired contradiction.

()

Example 2.22. Let ABCD be a quadrilateral inscribed in a circle and circumscribed


about a circle such that the points of tangency form a quadrilateral A1 B1 C1 D1 . Prove that
A1 C1 B1 D1 .
Solution
Let the tangency points of AB, BC, CD, and DA intersect the inner circle at point
A1 , B1 , C1 , and D1 , respectively, and let P be the intersection point of A1 C1 and B1 D1 .
Then, since ABCD is cyclic,
180o = DAB + BCD

C
A
1 D1 + D1 A1 + A1 B1 B1 C1
1 B1 + B1 C1 + C1 D1 D1 A1
+
=
2
2

=A
1 B1 + C1 D1 = 2A1 P B1 .

denotes the (minor) arc between points X and Y. Thus A1 P B1 = 90o, and
where XY
A1 C1 B1 D1 as desired.
()
Example 2.23. Solve in prime numbers the equation: xy + y x = z.

First solution
We first note that neither x and y equal nor are they both odd, for otherwise z wouldnt
be prime. Without loss of generality, suppose that x = 2. The equation then reduces to
2y + y 2 = z, or,
z = 2y 2 + y 2 + 2 = 2(2y1 1) + (y + 1)(y + 2) 3y.
First, (y + 1)(y + 2) is divisible by three since all primes greater than 3 are either one
or two less than a multiple of three (if some prime was three less, then it would contradict
the prime property). Second, since y is an odd prime of the form 2k + 1 (for some speciff
c integer k), we rewrite the first term as
2(2k+1)1 1 = 4k 1.

29

2.1. Arithmetic problems

Since 4 1 mod 3, it is easy to show that 4k 1 mod 3. Thus, 4k 1 is divisible by three.


Weve shown that z is divisible by three for y > 3. It follows that the only solution in
prime numbers is x = 2, y = 3, z = 17 or x = 3, y = 2, z = 17.
()
Second solution
We must have one of x, y even and the other odd (since otherwise z would be even
and greater than 2). Therefore a solution must be of the form
2 p + p2 = z

where p is an odd prime and z is prime. We see that 23 + 32 = 17 is a solution; are there
any others? No: all primes greater than 3 are of the form 6n 1, and since

for p odd, and

2p = (1)p = 1 mod 3

(6n 1)2 = 36n2 12n + 1 = 1 mod 3,

it follows that 26n1 + (6n 1)2 is divisible by 3.

()

Third solution
First of all, we can observe that x, y cant be both odd primes, or z would be even;
so, one of them must be equal to 2, say x. We now have the equation 2y + y 2 = z. The
following lemmas hold:
Lemma 1 . If a is an odd number then 2a 2 ( mod 3).
Proof. 2 0 ( mod 3) 2k 0 ( mod 3) (2k 1) (2k + 1) 0 ( mod 3)
Thus
22k 1 ( mod 3) 22k + 1 = 2 ( mod 3).
Lemma 2. If p = 3 is a prime then p2 = 1 ( mod 3).

Proof. If p is such a prime then (p, 3) = 1 we can apply Fermats Theorem, so


p31 p2 1 ( mod 3).

Lemma 3. If p > 3 is a prime then 2p + p2 = 0 ( mod 3).

Proof. It follows from Lemma 1 and Lemma 2.

()

()
()

It follows from Lemma 3 that y must be equal to 3: the only one solution is (2, 3, 17),
or (3, 2, 17) by symmetry between x and y.
()
Example 2.24. Let a and b be integers such that b a is an odd prime. Prove that
P (x) = (x a)(x b) p is irreducible in Z[x] for any prime p.

30

Chapter 2. Examples for practice


First solution
We want to show that
(x a)(x b) p = x2 (a + b)x + ab p = 0

has no solutions in integers when b a is an odd prime and p is prime. The discriminant
of this quadratic is
(a + b)2 4ab + 4p = (a b)2 + 4p = q 2 + 4p

where q = b a, so the problem is equivalent to showing that q 2 + 4p cannot be a square.


Suppose to the contrary that for some integer y
q 2 + 4p = y 2 4p = y 2 q 2 = (y q)(y + q)

In order for (y q)(y + q) to be divisible by 4, y must have the same parity as q, so


letting y = q + 2t for some positive integer t we now have
4p = 2t(2q + 2t) p = t(q + t).

Since q is an odd prime we cannot have t = 1 (since then p = q + 1 which is imposible


for p, q prime with q odd), therefore t > 1, which is also impossible since then the prime p
would factor. It follows that the discriminant cannot be a square, therefore the polynomial
is irreducible over the integers.
()
Second solution
Assume for a contradiction that P (x) = (x a)(x b) p is indeed reducible over Z[x]
for some prime p. Then, since P (x)is a quadratic, there must be some integers r and s
such that P (x) = (x r)(x s). Thus, P (r) = 0, or (r a)(r b) = p. However, since
(r a) (r b) = a b is odd, their product, p, must be even, so p = 2.
Let x = r a and y = r b, and without loss of generality let x < y. Since xy = p > 0,
either x, y are both positive or x, y are both negative. Without loss of generality let them
both be positive. Then since x y = a b, which is an odd prime, xy 1 (1 + 3) > 2 = p,
the desired contradiction.
()
Example 2.25. Let p and q be odd primes. Prove that for any odd integer d > 0 there is
an integer r such that the numerator of the rational number
[n r (mod q)]
nd
n=1
p

is divisible by p, where [Q] is equal to 1 or 0 as the proposition Q is true or false.

Solution
If p < q, chose r = p. Then, n r (mod q) is always false for n = 1, 2, ..., p 1, and the
rational number has numerator 0 divisible by p.
If p > q, positive integers a, b exist such that p = aq + b with b < q. If b is even, take
q+b
b
. In either case, we have that b 2r 0 (mod q), or
r = , and if b is odd, take r =
2
2
if n r (mod q), then p n b r r (mod q).

31

2.1. Arithmetic problems

Since n and p n cannot be equal because p is odd, all numbers n {1, 2, ..., p 1}
such that n = r (mod q) may be grouped up in distinct pairs of the form (n, p n). For
each one of these pairs, their contribution to the total sum is

But for odd d,

1
nd + (pn)d
1
+
=

nd (p n)d nd (p n)d

nd + (p n)d = p (nd1 nd2 (p n) + nd3 (p n)2 ... + (p n)d1 ) ,

and the numerator of this fraction is divisible by p, but not its denominator, since n and
p n are both smaller than prime p. Adding any number of such fractions, a common
factor p will always appear in the numerator, but never in the denominator.
()
Example 2.26. Find all integers that can be represented as a3 + b3 + c3 3abc for some
positive integers a, b, and c.
Solution
Let us say an integer is nice if it can be represented as a3 + b3 + c3 3abc for some
positive integers a, b, c. Assume without loss of generality that b = a + x and c = a + x + y,
for some nonnegative integers x, y. Therefore,
a3 + b3 + c3 3abc = (3a + 2x + y)(x2 + xy + y 2 ).

For x = y = 0 it follows that 0 is nice. Suppose that x, y are not both zero. Since
(3a + 2x + y)(x2 + xy + y 2) > 0 we have that any nonzero nice integer is nonnegative. Let
us prove first that 1 and 2 are not nice. We have that
(3a + 2x + y)(x2 + xy + y 2) > 3a(x2 + xy + y 2) 3,

from where it follows the claim. Let us prove also that any nice integer divisible by 3 must
be divisible by 9. We have that
0 (3a + 2x + y)(x2 + xy + y 2 ) (y x) (x y)2 (y x)3 (mod 3),

from where it follows that x y (mod 3). Therefore,

3a + 2x + y x2 + xy + y 2 0 (mod 3),

which implies the claim. Let us prove that 9 is not nice. From the previous result we have
that x y (mod 3), from where it follows that
(3a + 2x + y)(x2 + xy + y 2 ) = (3a + 3) 3 > 9.

Let us proceed to find which integers are nice. Taking x = 0, y = 1 it follows that any
positive integer of the form 3a + 1 is nice. Taking x = 1, y = 0 it follows that any positive
integer of the form 3a + 2 is nice. Taking x = y = 1 it follows that any positive integer
of the form 9(a + 1) is nice. From these we conclude that all the nice integers are 0, any
positive integer greater than 3 of the form 3a + 1 or 3a + 2, and the integers greater than
9 of the form 9a, and we are done.
()

32

Chapter 2. Examples for practice

Example 2.27. Find all


positive integers n such that a divides n for all odd positive
integers a not exceeding n.

First solution

If 1 is the largest odd integer not exceeding n, the result is trivially true, and n < 3,
or n 8. Assume
2m + 1 is the largest odd integer
now that m 1 isan integer such that
2
not exceeding n. Then, 2m + 3 > n 2m + 1, or 4m + 12m + 9 > n 4m2 + 4m + 1. Since
2m + 1 and 2m 1 are positive odd integers with difference 2, they are coprime, and if
both divide n, then their product 4m2 1 must also divide n, which is larger than 4m2 1.
Therefore, n 2(4m2 1), and 4m2 + 12m + 9 > 8m2 2, or 4m2 12m 11 < 0. Now, if
m 4, then 4m2 12m 11 =
(m2 11) + 3(m 4)m > 0, and necessarily m 3. Assume
that m = 3. Then 2m+3 = 9 > n, and n < 81, but n must be divisible by 3, 5 and 7, which
are coprime. Therefore, n mustbe divisible by 105, which is absurd, and m 2. Assume
now m = 2. Then, 2m + 3 = 7 > n 5 = 2m + 1, and 25 n < 49, but n must be divisible
by 3 and 5, which are coprime. Therefore, n must be divisible by 15, or n = 30, 45. Assume
next that m = 1. Then, 9 n < 25 and n must be divisible by 3, or n = 9, 12, 15, 18, 21, 24.
The integers that we are looking for are then 1 through 9, 12, 15, 18, 21, 24, 30 and 45. ()
Second solution
For n 25 such integers can be computed as 1, 2, 3, 4, 5, 6, 7, 8, 9, 12, 15, 18, 21, and
24. In order to get some intuition, consider what happens for n 25: up to n = 72 = 49,
these are the integers divisible by 3 and 5, that is multiples of 15, which are 30 and 45
in this range. After n = 49, up to n = 92 = 81, these are the integers divisible by 3, 5, and
7, that is multiples of 105, a contradiction. Recall Bertrands postulate, that is there is
always a prime between m and 2m, where m is any integer with m > 1. Using induction
we can
see that
every time we jump from n to 4n we get at least one more prime in the
range [ n, 2 n] .
This prime is greater than 4 for n 9, so the product of primes that must divide n
grows faster than n. In summary, the only
positive integers n such that a divides n for all
odd positive integers a not exceeding n are
1, 2, 3, 4, 5, 6, 7, 8, 9, 12, 15, 18, 21, 24, 30, 45.

Example 2.28. Find all primes q1 , q2 , ..., q6 such that q12 = q22 + + q62 .

()

Solution
Every square is 0 or 1 modulo 3 and clearly q1 3. Suppose there are 0 a 5 primes
between q2 , ..., q6 not equal to 3. Then 1 = 1 a + 0(5 a) (mod 3), from which a = 1 or
a = 4.
Suppose a = 1.
Then q12 = q22 + 4 32or(q1 q2 )(q1 + q2 ) = 36. Since q1 + q2 > q1 q2 , then q1 + q2 can be
only 9, 12, 18, 36 and its easy to see that there are no solutions.
Suppose a = 4.
Then q12 = q22 + q32 + q42 + q52 + 9. Since qi are primes, their quadratic residues modulo 8
are 1 if qi is odd or 4 if qi = 2. Clearly q1 = 2 and suppose that there are 0 b 4 primes
between q2 , ..., q5 not equal to 2. Then 1 = 1 + 1 b + 4(4 b) (mod 8) or 3b = 0 (mod 8),
which has the only solution b = 0.

33

2.1. Arithmetic problems

Hence the solutions to the problem are (5, 2, 2, 2, 2, 3) and its permutations with 5
fixed.
()

Example 2.29. Prove that the equation x3 + y 3 + z 3 + w 3 = 2008 has infinitely many
solutions in integers.
First solution A Since 2008 = 8 251 = 23 251, it suffices to show that
x3 + y 3 + z 3 + w 3 = 251

has infinitely many solutions over the integers. Let us note that

and thus we are done.

(30n3 + 5)3 (30n3 5)3 (30n2 )3 + 1 = 251

()

Second solution
For every n Z, the 4-tuple

(x = 10 + 60n3 , y = 10 60n3 , z = 2, w = 60n2 )

provides us with a solution to the given equation. Since n3 = m3 implies n = m, we have


that no two of these solutions can be identical: this terminates our proof.
()
Example 2.30. Let n be an integer. Find all integers m such that am + bm an + bn for
all positive real numbers a and b with a + b = 2.
First solution
First we will prove that for any positive real numbers x, y and any natural m n
following inequality holds

We have

(xn + y n )(xmn + y mn )

xm + y m

(2.2)

2(xm +y m )(xn +y n )(xmn +y mn ) = xm +y m xn y mn xmn y n = (xn y n )(xmn y mn ) 0.


Combination (2.2) with PM-AM inequality

x + y mn
xmn + y mn
(
)
2
2

gives inequality

xm + y m (xn + y n ) (

x + y mn
)
2

Applying (2.3) to positive real numbers a and b with a + b = 2 we obtain


a + b mn
)
= an + bn .
a + b (a + b ) (
2
m

(2.3)

34

Chapter 2. Examples for practice

Suppose that there is natural m < n such that for any positive real numbers a and b
with a + b = 2 holds inequality
am + bm an + bn am + (2 a)m = an + (2 a)n .

Then we obtain inequality

2m = lim(am + (2 a)m ) = lim(an + (2 a)n ) = 2n


a2

a2

which contradict to inequality 2m < 2n m < n.


Thus, answer is {mm N and m n}.

()

Second solution
First, let us prove that for any nonnegative positive integer k and for any positive
integer n, an+k + bn+k an + bn , if a + b = 2 and a 0, b 0. To prove this, we use
mathematical induction on k.
The case k = 0 is readily checked.
Assume that the statemnt is true for k = l and let us prove, that it is true too for
k = l + 1.
We have to prove that
an+l+1 + bn+l+1 an+l + bn+l .
If we have that a + b = 2 the inequalities is equal to

2an+l+1 + 2bn+l+1 (an+l + bn+l )(a + b), or an+l+1 + bn+l+1 an+1 b + bn+1 a.

The last inequality is equivalent to

(a b)2 (an+l1 + an+l2 b + + bn+l2 a + bn+l1 ) 0

and since (a b)2 0 and an+l1 + an+l2 b + + bn+l2 a + bn+l1 > 0 we have that inequality
holds if and only if a = b or k = 0.
Thus an+l+1 + bn+l+1 an+l + bn+l an + bn or an+l+1 + bn+l+1 an + bn . This completes
the induction and proves that the statemnt is true for all nonnegative integers. The last
result implies that the given relation is only satisfied for all integers m n.
()
Third solution
Let n = 0 or n = 1, we need to find all integers m such that am + bm 2. Let f (x) =
m
x + (2 x)m with 0 < x < 2. Since g(x) = xm is convex in this interval we have that
f (x) = g(x) + g(2 x) 2g(1) = 2 by Jensens inequality, so all integers m satisfy. To note
that g (x) = m(m 1)xm2 and m(m 1) 0. If n 2 then m n. Let f (x) = xn + (2 x)n .
Since f (x) = f (2 x) it suffices consider 1 x < 2 fixed. Now let g(n) = xn + (2 x)n ,
we will prove that g(n) is non decreasing. We have g (n) = xn ln (x) + (2 x)n ln (2 x).
Now suppose n 2 fixed, that is to say h(x) = xn ln (x) + (2 x)n ln (2 x). We need
to prove that h(x) 0. Since t(x) = xn ln (x) is convex if 1 x < 2 we have that h(x) =
t(x) + t(2 x) 2t(1) = 0 by Jensens inequality and we are done. To note that
t (x) = xn2 [n(n 1) ln (x) + 2n 1] > 0.

If n 1 then m n.
The idea is the same, but in this case we consider 0 < x 1.

()

35

2.1. Arithmetic problems

Example 2.31. Let a, b, c, d be integers such that a + b + c + d = 0.


Prove that a5 + b5 + c5 + d5 is divisible by 30.
First Solution
Because x5 x (mod 5), x3 x (mod 3), x2 x (mod 2) then x5 x3 x (mod 3) and
x5 x (mod 2) for any integer x. Therefore,
a5 + b5 + c5 + d5 a + b + c + d 0 (mod k)

where k = 2, 3, 5. Thus, a5 + b5 + c5 + d5 30.


For any natural m > 1 product of m consequtive integers allways divisible by m, then
2
x x = (x 1)x divisible by 2, x3 x = (x 1)x(x + 1) divisible by 3 and
x5 x = x (x2 1) (x2 + 1)

= x (x2 1) (x2 4 + 5)

= (x 2)(x 1)x(x + 1)(x + 2) + 5x (x2 1) divisible by 5.

()

Second solution
We will prove that x5 x (mod 30). First note that x5 x = (x 1) x (x + 1) (x2 + 1).
Therefore we want to show that (x1) x (x+1) (x2 +1) is always divisible by 30 = 2 3 5.
In the following assume x 1 > 0. Then there will always be factors of 2 and 3 among
x 1, x, and x + 1, and in some cases a factor of 5 as well. The only triples x 1, x, x + 1
for which there is not a factor of 5 are those of the form
5 n + 1, 5 n + 2, 5 n + 3, or 5 n + 2, 5 n + 3, 5 n + 4.
In the first case, x = 5 n + 2, so x2 + 1 = 25 n2 + 20 n + 5, which gives a factor of 5. In
the second case, x = 5 n + 3, so x2 + 1 = 25 n2 + 30 n + 10, which also gives a factor of 5.
It follows that x5 x is always divisible by 30 and hence the conclusion.
()
Third solution
We will prove a lemma first:
Lemma. Let x be an integer, then 30 (x5 x).
Proof. We have x5x = (x1)x(x+1) (x2 + 1) . Because 6(x1)x(x+1) so 6 (x5 x) .
If x 0 (mod 5) then 5x.
If x 1 (mod 5) then 5(x 1).
If x 2 (mod 5) then 5 (x2 + 1) .
If x 3 (mod 5) then 5 (x2 + 1) .
If x 4 (mod 5) then 5(x + 1).
So for all integer x, 5 (x5 x) . But gcd(5, 6) = 1, it follows that 30 (x5 x) . Coming
back to the problem, from the Lemma, we have
a5 + b5 + c5 + d5 = a5 + b5 + c5 + d5 (a + b + c + d) = a5 a + b5 b + c5 c + d5 d 30.
So a5 + b5 + c5 + d5 is divisible by 30.

()

Example 2.32. Find all primes p and q such that 24 does not divide q + 1 and p2 q + 1 is
a perfect square.

36

Chapter 2. Examples for practice


Solution
If one of p or q is even. Since x2 = p2 q + 1, x is odd, ie.

x2 = 1 (mod 8) p2 q + 1 = 1 (mod 8) p2 q = 0 (mod 8)

Because both p and q are primes, and 2 is the only even prime, p = q = 2
If both p, q > 2 in that case x > 2. We have
p2 q + 1 = x2 p2 q = (x 1)(x + 1).

Since x is even, x 1 and x + 1 will be both odd, and they are relatively prime because
gcd(x 1, x + 1) = gcd(x 1, 2) = 1 So,
(1)

x 1 = q, x + 1 = p2 or (2)

x 1 = p2 , x + 1 = q.

Since x > 2 and the gcd(x 1, x + 1) = 1 the other cases are not possible.
(1) q + 1 = p2 1

If p = 3, we have solution q = 7.
If p 3, then p2 1 (mod 3) and p2 1 (mod 8) which contradicts with the condition
that 24 doesnt divide q + 1.
(2) p2 + 2 = q
For p = 3, we have solution q = 11.
If p 3, then p2 + 2 = 0 (mod 3) which cannot be prime.
So, all the solutions are: (p, q) {(2, 2), (3, 7), (3, 11)}.

2.1.2

()

Senior problems

Example 2.33. Find the least odd positive integer n such that for each prime p, the
n2 1
+ np4 + p8 is divisible by at least four primes.
number M =
4
First solution
Let n = 2k + 1 with k nonnegative integer. For k = 0, 1, 2, 3 it is easy to see that when
p = 2 there are less than four prime divisors.
M = p8 + np4 +
= (p4 +

n2 1
n 2 1
= (p4 + )
4
2
4

n1
n+1
) (p4 +
) = (p4 + k)(p4 + k + 1).
2
2

Let k = 4, then M = (p4 + 4)(p4 + 5) = (p2 + 2p + 2)(p2 2p + 2)(p4 + 5).


If p = 2, then M is divisible by 2, 3, 5, 7. If p is odd we have
(p2 + 2p + 2, p2 2p + 2) = (p2 + 2p + 2, 4p) = 1,

(p2 + 2p + 2, p4 + 5) = (p2 + 2p + 2, p4 + 5 p4 8p2 4 4p3 4p)

37

2.1. Arithmetic problems

= (p2 + 2p + 2, 4p3 + 8p2 + 4p + 1)


= (p2 + 2p + 2, 4p3 + 8p2 + 4p + 1 4p3 8p2 4p)
= (p2 + 2p + 2, 1) = 1,

and

(p2 2p + 2, p4 + 5) = (p2 2p + 2, 4p3 8p2 + 4p + 1) = (p2 2p + 2, 1) = 1.

Thus p2 + 2p + 2, p2 2p + 2 and p4 + 5 are pairwise coprime. As p4 + 5 = 2 (mod 4) for


all odd p, then 21 is the greatest power of 2 dividing p4 + 5. Since both p2 + 2p + 2 and
p2 2p + 2 are odd, there is another prime different from 2 and from all the divisors of
p2 + 2p + 2 and p2 2p + 2 which divides p4 + 5, and so n = 9 is the least desired number.
()
Second solution
Let n = 2k + 1, then
n2 1
+ np4 + p8 = k(k + 1) + (2k + 1)p4 + p8 = (p4 + k)(p4 + k + 1).
4

Note that for k = 0, 1, 2, 3 the result does not hold for p = 2. We prove that k = 4 is the
least integer that satisfies the condition. For k = 4 we have
(p4 + 4)(p4 + 5) = (p2 + 2p + 2)(p2 2p + 2)(p4 + 5).

Since (p2 + 2p + 2)(p2 2p + 2) = (p4 + 5) 1 we have that

(p2 + 2p + 2, p4 + 5) = (p2 2p + 2, p4 + 5) = 1.

This implies that any prime that divides (p2 + 2p + 2)(p2 2p + 2) does not divide p4 + 5
and viceversa. Then, it is enough to prove that two primes divide (p2 + 2p + 2)(p2 2p + 2)
and another two divide p4 + 5.
For p = 2 the result holds. Assume that p is an odd prime. Note that 2 (p4 + 5) . To
prove that another prime divides p4 + 5 it is enough to prove that 4 p4 + 5. This results
follows from the fact that 4p4 +3. In order to prove that two primes divide (p2 +2p+2)(p2
2p + 2) it is enough to prove that (p2 + 2p + 2, p2 2p + 2) = 1. Let (p2 + 2p + 2, p2 2p + 2) = d.
Note that d is odd and that d4p. This implies that dp. If d = p then p (p2 + 2p + 2) , which
is a contradiction. Therefore, d = 1, as we wanted to prove. This implies that k = 4 is the
least integer value, from where we conclude that n = 9 is the least odd positive integer
that satisfies the condition.
Example 2.34. (a)
integer an such that

Prove that for each positive integer n there is a unique positive


(1 +

5) = an + an + 4n .

(b) When n is even, prove that an is divisible by 5 4n1 and find the quotient.
First solution
(a) Let 2 = 1 + 5. With this we have
2n n =

an + an + 4n .

(2.4)

38

Chapter 2. Examples for practice

(2.4) Squaring both sides leads to

4n (2n 1) = 2an + 2 an (an + 4n ).

(2.5)

(2.5)
Subtracting 2an from both sides and squaring the resulting value leads to

This is reduced to
an = 4

n1

[4n (2n 1) 2an ]2 = 4an (an + 4n ).

(2.6)

2n 1 2
2
) = 4n1 (n (1)n n ) = 4n1 (2n + 2n 2) = 4n1 (L2n 2)
(
n

where Lm is the mth Lucas number. Hence it has been shown that an is a positive integer
and is given by
an = 4n1 (L2n 2) .
(b) If n is an even value, say n = 2m, then

a2m = 42m1 (L4m 2) = 42m1 5F2 = 5 4m1 (2m F2m )2 .

From this relation it is shown that a2m is divisible by 5 4m1 and has the quotient
value (2m F2m )2 .
()
Second solution

n
(a) Let (1 + 5) = xn + yn 5, where xn , yn are positive integers, n = 1, 2, ... Then

hence

(1

5) = xn yn 5, n = 1, 2, ...

x2n 5yn2 = (4)n , n = 1, 2, ...

If n is even, consider an = x2n 4n and we have

an + an + 4n = x2n 4n + x2n = 5yn2 + x2n = yn 5 + xn = (1 + 5)n .

(i)

If n is odd, consider an = 5yn2 4n and we have

an + an + 4n = 5yn2 4n + 5yn2 = x2n + 5yn2 = xn + yn 5 = (1 + 5)n .


(b) If n is even, then we have an = x2n 4n = 5yn2 , where

n
n
n
n
2n
1+ 5
1 5
1
) (
) ) = 2n1 Fn ,
yn = ((1 + 5) (1 5) ) = ((
2
2
2 5
2 5

where Fn is the nth Fibonacci number. In this case we get an = 5 4n1 Fn2 , hence 5 4n1 an
and the quotient is Fn2 .
()

39

2.1. Arithmetic problems

Example 2.35. If a1 , a2 , . . . , ak (0, 1), and k, n are integers such that k > n 1, prove
that the following inequality holds
min{a1 (1 a2 )n , a2 (1 a3 )n , ..., ak (1 a1 )n }

Solution ( Method Reductio ad absurdam)


Lets suppose that the inequality doesnt hold. Therefore
a1 (1 a2 )n >

a2 (1 a3 )n >

nn

(n + 1)n+1

nn
(n + 1)n+1
nn
(n + 1)n+1

ak (1 a1 )n >
Multiplying these relations up, we get

nn

(n + 1)n+1

nn
]
a1 a2 . . . ak (1 a1 ) (1 a2 ) . . . (1 ak ) > [
(n + 1)n+1
n

But, for a (0, 1), we have a(1 a)n


Lets prove this inequality.

nn
(n + 1)n+1

n times

1
1 na + (1 a) + + (1 a)
n
n
a (1 a) = (na) (1 a)

n
n
n+1

1
n n+1
nn
(
) =

n n+1
(n + 1)n+1

1
The equality holds for: na = 1 a a =
(0, 1).
n+1
Using the proved inequality for a1 , a2 , ..., ak , we get:
a1 (1 a1 )n

a2 (1 a2 )n

nn
(n + 1)n+1
nn
(n + 1)n+1

ak (1 ak )n

nn

(n + 1)n+1

()

(by AM - GM)

40

Chapter 2. Examples for practice


Multiplying these relations up, we get
a1 a2 . . . ak (1 a1 )n (1 a2 )n . . . (1 ak )n [

nn
] .
(n + 1)n+1

This inequality contradicts (*), which follows from the initial assumption.
Therefore, that assumption is false.

()

Example 2.36. Is there an integer n such that exactly two of the numbers:
n + 8, 8n 27, 27n 1 are perfect cubes?
Solution
The first positive perfect cubes are
1, 8, 27, 64, 125, 216, 343, 512, 729, 1000, 1331, 1728, 2197, 2744, 3375, 4096,
the difference of any two perfect cubes one of which is in absolute value larger than 4096
being clearly larger than 4096 3375 = 721.
Assume that 8n 27 = u3 and 27n 1 = v 3 are both perfect cubes for some integers u, v. In this case, (2v)3 (3u)3 = 272 8 = 721, and 721 must be the difference
between two perfect cubes. By inspection, the pairs of perfect cubes for which this
happens are (4096, 3375), (729, 8), (8, 729) and (3375, 4096), yielding respectively
2v = 16, 9, 2, 15 and 3u = 15, 2, 9, 16, with solutions (u, v) = (5, 8) and (u, v) = (3, 1)
(the other two cases yield non-integral values for u, v). It follows respectively that n = 19
and n = 0, for n + 8 = 33 and n + 8 = 23 , perfect cubes in both cases, or whenever 8n 27
and 27n 1 are simultaneously perfect cubes, so is n + 8.
We proceed similarly in the cases where n + 8 = w 3 and 8n 27 = u3 are both perfect
cubes (hence (2w)3 u3 = 91 is the difference between two perfect cubes) and n + 8 = w 3
and 27n 1 = v 3 are both perfect cubes (hence (3w)3 v 3 = 217 is the difference between
two perfect cubes). In both cases, we find by inspection that the only integral values of
w, u and w, v that make this possible, also yield, respectively, 27n 1 = v 3 a perfect cube,
and 8n 27 = u3 a perfect cube. In fact, the values obtained for u, v, w, n in these two
cases are exactly the values found in the case where we assume that 8n 27 = u3 and
27n 1 = v 3 are both perfect cubes.
It follows that either at most one of the given numbers is a perfect cube, or all three
are perfect cubes, hence no n exists such that exactly two of the given numbers are perfect
cubes.
()
Example 2.37. Find all pairs (x, y) of positive integers such that x2 + y 2 = p6 + q 6 + 1, for
some primes p and q.
Solution
Suppose that 3 q and 3 p, then 3 is coprime with both p and q (because they are
prime). So 9 is coprime with both p and q.
Then, being (9) = 6 we have that p6 + q 6 + 1 3 (mod 9) by Euler theorem. But then
we should have x2 + y 2 0 (mod 3) which implies that x 0 (mod 3) and y 0 (mod 3).
Hence x2 + y 2 0 (mod 9) and 3 0 (mod 9), contradiction.
Then we should have that without loss of generality p = 3.

41

2.1. Arithmetic problems

Now 36 + q 6 + 1 2 + q 6 (mod 4). So if q is odd, then x2 + y 2 3 (mod 4), contradiction.


Then q = 2. So we have to solve x2 + y 2 = 794 2 (mod 397).
But we have that 397 1 (mod 4) and it is prime. So we get this factorization in primes
of Z[i]
794 = (1 + i)(1 i)(6 + 19i)(6 19i).

Therefore the unique solutions are (25, 13) and(13, 25).

()

a3 + b3
for some positive
Example 2.38. Find the least prime that can be written as
2011
integers a and b.
Solution
Since 2011 is prime (it is not divisible by any prime up to 43, and it is less than
2
47 = 2209), then a3 + b3 = (a + b)(a2 ab + b2 ) is the product of two primes. Since a, b 1,
clearly a + b > 1. If a2 ab + b2 = 1, then ab (a b)2 + ab = 1, or a = b = 1, and a3 + b3 = 2 is
clearly not a multiple of 2011, hence either a + b is the lowest prime and a2 ab + b2 = 2011,
or a + b = 2011 and a2 ab + b2 is the lowest prime. Note however that
a2 ab + b2 =

3(a b)2 + (a + b)2 (a + b)2

4
4

or if a+b = 2011, then a2 ab+b2 > 2011500, while if a2 ab+b2 = 2011, then a+b 4 2011
and a + b < 90 because 2011 < 452 = 2025.
Let us therefore look for values a, b such that a + b = p is prime and a2 ab + b2 = 2011.
p2 32011
Clearly, 3ab = (a + b)2 (a2 ab + b2 ) = p2 2011, or p 47. Note
is always
3
an integer since p2 2011 1 (mod 3) for primes p 3. Now, if p 3 (mod 50), then
p2 32011
p2 9 (mod 100), and p2 2011 2(mod100), or ab =
66(mod100).
3
Then (ab)2 = p2 4ab 964 45 (mod 100) cannot be a perfect square, or p 47, 53,
and p 59.
Note however that for (a, b) a permutation of (10, 49), we have
493 + 103 118649
=
= 59,
2011
2011

and this is therefore the lowest prime of this form that can be found.

()

Example 2.39. Let x, y, z be integers such that 3xyz is a perfect cube. Prove that (x +
y + z)3 is a sum of four cubes of nonzero integers.
First solution
If xyz, x + y z, y + z x, z + x y 0, note that

(x + y + z)3 = (x + y z)3 + (y + z x)3 + (z + x y)3 + 24xyz,

where the first three terms in the RHS are clearly nonzero cubes, and 24xyz = 23(3xyz)
is also a nonzero cube. This is however not a solution when at least one of
xyz, x + y z, y + z x, z + x y

42

Chapter 2. Examples for practice

is zero. In that case, we can use that


73 = 343 = 216 + 125 + 1 + 1 = 63 + 53 + 13 + 13 ,

or if x + y + z = s 0, then

(x + y + z)3 = s3 = (7s)3 + (6s)3 + (5s)3 + (s)3

is the sum of four nonzero cubes. Finally, if x + y + z = 0. Then for any a, b 0,


(x + y + z)3 = 0 = a3 + b3 + (a)3 + (b)3

is the sum of four nonzero cubes. Note therefore that for any three integers x, y, z, (x+y+z)3
may be written as the sum of four nonzero cubes, regardless of whether 3xyz is a perfect
cube or not.
()
Second solution
Let 3xyz = w 3 . We have the identity

(x + y + z)3 = (x + y z)3 + (y + z x)3 + (z + x y)3 + (2w)3

which can be proved expanding the cubes in both sides.


Now if x + y z = 0 then (x + y + z)3 = 8z 3 but 1 = (1)3 + (7)3 + (5)3 + (6)3 and hence
8z 3 = (2z)3 + (14z)3 + (10z)3 + (12z)3 ,

analogously if y + z x = 0 or z + x y = 0.

()

2
2
Example 2.40. Let a and b be integers such
that a m b n = a b, where m and n are
consecutive integers. Prove that gcd(a, b) = a b.

Solution
Let d = gcd(a, b) and write a = da and b = db with gcd(a , b ) = 1.
From a2 m b2 n = a b we get (ma 1)a = (nb 1)b , so a (nb 1)b and gcd(a , b ) = 1.
This implies that a (nb 1), and consequently there exists an integer such that
nb 1 = a and ma 1 = b or,
ndb a = 1, and mda b = 1.

(1)

In particular, gcd(d, ) = 1. Now subtracting the two equalities in (1) we find that
(nd + )b = (md + )a ,

and again, since a (nd + )b and gcd(a , b ) = 1, we conclude that a (nd + ) and that
there exists an integer such that
nd + = a, and md + = b.

(2)

On the other hand, recalling that m and n are consecutive integers we get from (2)
that
d = (a b ), with = n m {1, +1},
(3)

Whereas from (2) we have = (a n(a b )). So, is a common divisor of and
d which are coprime, hence {1, +1}.
Finally, multiplying both sides of (3) by d and using the fact that = 1, we conclude
that d2 = a b, which is the desired conclusion.
()

2.1. Arithmetic problems

43

Example 2.41. Prove that there are infinitely many positive integers n that can be expressed as a4 +b4 +c4 +d4 4abcd, where a, b, c, d are positive integers, such that n is divisible
by the sum of its digits.
First solution
For j 1, let n = 196 104j2 + 2 104j4 , then

n = (10j )4 + (10j )4 + (10j1)4 + (10j1 )4 4(10j )(10j )(10j1 )(10j1 ),

and n is divisible by the sum of its digits (18) since n is even and also divisible by 9 by
the decimal integer divisibility test for 9 (i.e. the sum of the digits is divisible by 9). ()
Second solution
Note that it suffices to find one, since if n = a4 + b4 + c4 + d4 4abcd, then

n = (10m a)4 + (10m b)4 + (10m c)4 + (10b d)4 4(10m a)(10m b)(10m b)(10m c) = 104m n

has the same sum of digits as n, and if n is divisible by the sum of its digits, so is n .
Note now that
(105 )4 + (104 )4 + (2 103 )4 + 14 4 2 105+4+3 = 1020 + 1016 + 8 1012 + 1

has sum of digits 11, and is clearly divisible by 11 since the sum of its digits in odd
positions is 11, and the sum of its digits in even positions is 0. The conclusion follows. ()
Example 2.42. Find all triples (n, k, p), where n and k are positive integers and p is a
prime, satisfying the equation
n5 + n4 + 1 = pk .
First solution
Its easy to see that (1, 1, 3) and (2, 2, 7) are solutions. Well prove that there arent
any other.
n5 + n4 + 1 = (n2 + 1)2 n2 + n5 n2
= (n2 + 1 n)(n2 + 1 + n) + n2 (n 1)(n2 + 1 + n)
= (n2 + n + 1)(n3 n + 1).

Suppose n > 2.
Then n3 n + 1 (n2 + n + 1) = n(n + 1)(n 2) > 0. Hence we have n2 + n + 1 = pr and
3
n n + 1 = ps where r + s = k and s > r. Subtracting the first relation from the second we
get
n(n + 1)(n 2) = pr (psr 1).

Clearly r > 0 and p doesnt divide n. If n + 1 = pr it contradicts the first relation, and
so p divides both n + 1 and n 2. But (n + 1, n 2) = (n + 1, 3) can be 3 or 1, from which
p = 3.
Then n5 + n4 + 1 = 0 (mod 9), but checking this congruence for every residue modulo 9
its easy to see that there are no solutions.
()

44

Chapter 2. Examples for practice

Second solution
First note that
n5 + n4 + 1 = (n2 + n + 1) (n3 n + 1) = r(n) s(n),

and that s(n) > r(n) for n 3. Next note that for n 3, we cannot have r(n) = pa and
s(n) = pb for any prime p and positive integers a, b, b > a, since otherwise we would have
s(n)
s(n)
(n 2)
(n 2)
= p(ba) , but
= n1
, and 0 <
< 1 for n 3. Therefore any
r(n)
r(n)
r(n)
r(n)
solution must satisfy n < 3, that is there are at most two solutions. Finally, it is trivially
verified that n = 1 and n = 2 both give solutions:
(n, k, p) = (1, 1, 3), and (n, k, p) = (2, 2, 7).
()
Third solution
For n = 1 we have p = 3 and k = 1. Now suppose that n 2, since
pk = n5 + n4 + 1 = (n2 + n + 1)(n3 n + 1)

there exist two positive integers r and s such that n2 + n + 1 = pr and n3 n + 1 = ps with
r s.
We have that gcd(n2 + n + 1, n3 n + 1) = gcd(ps , pr ) = ps . Because
n3 n + 1 = (n 1)(n2 + n + 1) (n 2) and n2 + n + 1 = (n 2)(n 3) + 7,

we have that gcd(n2 + n + 1, n3 n + 1) = gcd(n2 + n + 1, n 2) = gcd(n 2, 7) therefore


ps = 1 or ps = 7. Examining these two cases we obtain two solutions
(n, p, k) {(1, 3, 1); (2, 7, 2)}.

()

k
Example 2.43. Let n be an integer greater than 2. Prove that Cn1
= (1)k (mod n) for
each k = 1, 2, ..., n 1, if and only if n is a prime.

Solution
Clearly,

(n 1) (n 2) ... (n k)

k!
k
If n is prime, for any k < n, then k! is prime with n, and Cn1
(1)k (mod n) is
equivalent to
(n 1) (n 2) ... (n k) (1)k k! (mod n).
k
=
Cn1

This is clearly true since n j (j) (mod n) for each j {1, 2, ..., n 1}.
If n is not prime, let k be the least divider of n which is larger than 1. Then, since
(k 1)! is prime with n, and n j (j) (mod n) for all j {1, 2, ..., k 1}, the result
k
Cn1
= (1)k (mod n) is equivalent to
or equivalently,

nk
1 (mod n),
k

n
0 (mod n), absurd. This concludes the proof.
k

()

45

2.1. Arithmetic problems

2.1.3

Undergraduate problems

Example 2.44. For all integers k, n 2 prove that

n
n 1
n
)+1
1 + log (1 +
k n
k1

(2.7)

First solution
n
n
Let = x, 0 < x = . Consider the function
k
2
f (x) =

nx
1
1
ln (1 +
) + 1 (1 + x) n
n
nx

and we study it for all n 2. We have

n
1
n2
1+x
1) lim f (x) = 0 ; 2) f (x) = (

)
x0
n (nx + n x)(n x)
1+x

Recall Bernoullis inequality: (1 + x)a 1 + ax for 0 a 1. Hence

x
n
1+x1+
n
To prove that the derivative of f is positive for any 0 < x < n and for any fixed n 2,
it is enough to prove that
1 + nx
n2

(2.8)
(nx + n x)(n x) 1 + x
We have
(2.8) n3 (x + 1) (nx + n x)(n2 x2 )
n3 x + n3 n3 x + n3 n2 x nx3 nx2 + x3
n2 x + nx2 + nx3 x3 ,

which is clearly true. Thus the derivative is positive and therefore f (x) 0. The proof is
completed.
()
Second solution
We will use the following simple result.
Lemma 1. For all positive a
1
1
< ln (a + 1) ln a <
a+1
a

Proof. Let f (x) = ln x. Then f (x) =

exists c (a, a + 1) such that

1
and due to the Mean-Value theorem there
x

ln (a + 1) ln a
1
= f (c) =
= ln (a + 1) ln a.
c
(a + 1) a

Since c (a, a + 1), it must be that

1
1 1
< <
a+1 c a

46

Chapter 2. Examples for practice


Combining these two results yields the desired inequalities.
Returning to the problem, by the AM-GM inequality we have

k+n
k+1k+2
k+n
k+n
k+1 k+2
n
n
+
++
n

=n
.
k
k+1
k+n1
k k+1 k+n1
k
We also have
k+n
1
1
1
k+1 k+2
+
++
=n+( +
++
) = n + A.
k
k+1
k+n1
k k+1
k+n1

Due to the result of the lemma, we have

A ln k ln (k 1) + ln (k + 1) ln k + . . . + ln (k + n 1) ln (k + n 2)
n
)
= ln (k + n 1) ln (k 1) = ln (1 +
k1
Combining the obtained results, we finally obtain

n 1 k+1 k+2
k+n
n
1+ (
+
++
)
k n
k
k+1
k+n1
n
1
n
1
)) = 1 + ln (1 +
).
(n + ln (1 +
n
k1
n
k1

and the inequality is proved.

()

Example 2.45. Find the minimum of f (x, y, z) = x2 + y 2 + z 2 xy yz zx over all triples


(x, y, z) of positive integers for which 2010 divides f (x, y, z).
Solution
If wlog z is odd and x, y are even, then z 2 is the only odd term, f (x, y, z) is odd, hence
not a multiple of 2010, while if wlog x, y are odd and z is even, then x2 , y 2 , xy are the
only odd terms, f (x, y, z) is again odd. Therefore, x, y, z have the same parity, and we
yz
xy
,v =
may define u =
2
2
3s2 + d2 = 4(u2 + v 2 + uv) = (x y)2 + (y z)2 + (x y)(y z) = f (x, y, z),

where s = u + v and d = u v, and if 2010 divides f (x, y, z), then 4020 = 22 3 5 67 divides
f (x, y, z). Now, any perfect square leaves a remainder equal to 1, 0, 1 modulus 5, hence
if d, s are not both multiples of 5, then 3s2 + d2 cannot be a multiple of 5, hence 52 divides
d
s
3s2 + d2 = f (x, y, z), and 20100 divides f (x, y, z). Defining s1 = and d1 = , we find that
5
5
3s21 + d21 =

f (x, y, z) 20100k
=
= 804k.
25
25

But taking s1 = 16, d1 = 6, we find 3s21 + d21 = 768 + 36 = 804, or


f (x, y, z) = 25 804 = 20100,

47

2.1. Arithmetic problems

with equality for example for s = 80 and d = 30, ie u = 55 and v = 25, or


f (z + 160, z + 50, z) = 20100

for all positive integer z as it is easily checked by direct calculation.


()
Note
We have restricted ourselves to positive values of f (x, y, z), since clearly f (x, x, x) = 0
is a multiple of 2010 for all positive integer x, making the problem trivial.

2.1.4

Olympiad problems

Example 2.46. Find all positive integers n for which


1
1
1
P = (14 + ) (24 + ) (n4 + )
4
4
4

is the square of a rational number.

First solution
1
1
1
Let P = nk=1 (k 4 + ) and let ak = k 2 k + , k = 1, 2, ..., n. Since ak+1 = k 2 + k + and
4
2
2
k4 +
then

1
1 2
= (k 2 + ) k 2 = ak ak+1 , k = 1, 2, ..., n
4
2

1
(2n2 + 2n + 1) Q2
4
where Q = nk=2 ak . Therefore, P is the square of a rational number if and only if 2n2 +2n+1
is the square of a positive integer, i.e. if and only if 2n2 + 2n + 1 = m2 for some positive
integer m. Therefore 2m2 (2n + 1)2 = 1 and then our problem is finding the solutions to
the equation
x2 2y 2 = 1
P = a1 an+1 Q2 =

in positive integers. Let

Z ( 2) = {x + y 2 x, y Z} ; N ( 2) = {x + y 2 x, y N}

and let s = 3 + 2 2 and for any z = x + y 2 Z ( 2) denote z = x y 2 . In this notation

equation x2 2y 2 = 1, x, y N becomes zz = 1, z N ( 2) . Denote the set of all such


solutions by Sol, i.e.

Sol = {z z N ( 2) and zz = 1}

Note that for z0 = 1 + 2 N ( 2) we have z0 z 0 = 1 and since ss = 1 then for


zk = sk z0 we also have zk z k = 1, k N. Also it is clear that z0 is smallest element in
Sol. Note that if z Sol (that is zz = 1, z N ( 2)) and z z0 then sz Sol. Indeed,

48

Chapter 2. Examples for practice

sz sz = ss zz = 1 (1) = 1. It remains to prove sz N ( 2) . Let z = x + y 2 then


x2 = 2y 2 1 and

sz = (3 2 2) (x + y 2) = 3x 4y + (2x + 3y) 2.

Since z z0 z > z0 x, y 2, and moreover, x, y 3 because x is odd and 2y 2 1


isnt square of integer for y = 2, we have
and

3x 4y 9x2 16y 2 9 (2y 2 1) 16y 2 2y 2 9 y 3


3y 2x 9y 2 4x2 9y 2 4 (2y 2 1) y 2 + 4 0.

We will prove that Sol = {zk k N}. Suppose that exist z Sol which
not belong to the sequence z0 < z1 < z2 < < zk <, . . . Since (zk )k0 is unbounded
from
above then there is k such that zk < z < zk+1 . Since z0 < sk z < z1 = sz0 = 7 + 5 2 then
sk+1 z < z0 and sk+1 z Sol. That is the contradiction because z0 is smallest element in Sol.
Thus,
Sol = {zk k N}.

Let zk = xk + yk 2, k N . Then
x
zk+1 = szk { k+1
yk+1

= 3xk + 4yk
= 2xk + 3yk

xk+2 3xk+1 = 2xk + 3(xk+1 3xk )


xk+2 6xk+1 + 7xk = 0, x0 = 1, x1 = 7.

Since x = 2n + 1 then all natural n for which 2n2 + 2n + 1 is a square of integer should
be elements of set
xk 1
, k = 1, 2, . . .}
2
By substitution xk = 2nk + 1 in xk+2 6xk+1 + 7xk = 0 we obtain
{nk nk =

2nk+2 + 1 12nk+1 6 + 14nk + 7 = 0 nk+2 = 6nk+1 7nk 1

by

Thus, all solutions of problem are the terms of the sequence (nk )k1 defined recursively
nk+2 = 6nk+1 7nk 1, n0 = 0, n1 = 3.

In particular, n1 = 3, n2 = 17, n3 = 80, n4 = 360, . . .

Second solution
Multiplying by 4 does not affect the property of our number M , we can consider
M = (4k 4 + 1)

Observe,that our number can be written in this form:


(4k 4 + 1) = ((2k 2 + 2k + 1)(2k 2 2k + 1))

()

49

2.1. Arithmetic problems

As the factors will be repeated, we will obtain squares, except the last factor.
(4k 4 + 1) = ((2k 2 + 2k + 1)(2k 2 2k + 1)) = (2n2 + 2n + 1) (2k 2 2k + 1)

Let 2n2 + 2n + 1 be equal to c2 . Then, (2n + 1)2 + 1 = 2(2n2 + 2n + 1) = 2c2


Consequently,we have the equation (2n + 1)2 2c2 = 1
Using Pells equation,we can consider two general cases:
x2 2y 2 = 1

and

(2.9)

x2 2y 2 = 1

Obviously,that minimal roots of first equation are x0 = 1, y0 = 1 and the minimal roots
of the second ecuation are x0 = 3, y0 = 2

(2.10)
xk 1 + yk 2 = (3 1 + 2 2)n (1 1 + 2) = (1 + 2)2k+1 = A
Consequently,

xk 1 yk 2 = (1 2)2k+1 = B

is the result of division of (2.10) by (2.9) Then,


2n + 1 = xk =

(A + B 2)
(A + B)
n=

2
4

Substitute the values of A and B and derive

(1 + 2)2k+1 (1 2)2k+1 2
n=

()

Example 2.47. Find all pairs (m, n) of positive integers such that f (f (nm )) = n, where
f Eulers totient function.
Solution
We will show that the pairs are (m, 1) for all positive integers m, (3, 2) and (2, 4).
We first note that:
1) since f (x) is odd iff x = 1 or x = 2 then f (f (nm )) = n for n odd iff n = 1 and for
all positive integers m;
2) since f (n) < n when n > 1, then if m = 1 then n = 1.
So we can assume that n = 2k j, with k 1, j 1 odd and m 2.
We first consider the case when j 1. Then, by 1), f (jm) = 2s h with s 1, h 1 odd
and
f (f (nm )) = f (f (2km j m )) = f (2km1 2s h) = 2km+s2 f (h) = n = 2k j.

Hence km + s 2 k, that is k(m 1) 2 s 1 which implies that k = 1, m = 2, and


s = 1. Then f (h) = j and by 1) we have h = j = 1 which is a contradiction.
Finally assume that j = 1. Then n = 2k and f (f (2km )) = 2km2 = 2k which implies that
k(m 1) = 2, that is (n, m) = (2, 3) or (n, m) = (4, 2).
()

50

Chapter 2. Examples for practice

Example 2.48. Find the least integer n 2011 for which the equation
x4 + y 4 + z 4 + w 4 4xyzw = n

is solvable in positive integers.


Solution
Define x2 y 2 = a, z 2 w 2 = b and xy zw = c, or

a2 + b2 + 2c2 = x4 + y 4 + z 4 + w 4 4xyzw = n.

Every square of an odd number 2m+1 is 4m(m+1)+1, or since exactly one of m, m+1
is even, every odd perfect square leaves a remainder of 1 when divided by 8, hence 2c2
leaves a remainder of 0 or 2 when divided by 8 because every even perfect square is a
multiple of 4. By the same reason, a = x2 y 2 cannot leave a remainder of 2 when divided
by 4, or a2 and b2 leave remainders of 0 or 1 when divided by 8. It follows that the possible
remainders of n when divided by 8 are 0 (when a, b, c are all even), 1 (when one of a,
b is odd, and the other two and c are even), 2 (when either a, b are odd and c is even,
or a, b are even and c is odd), 3 (when c and one of a, b are odd, and the other one
is even), or 4 (when a, b, c are all even). Therefore, the lowest integers n = 2011 such
that the proposed equation has a solution could be 2011, 2012, 2016, 2017. Note that
194 + 3 184 4 19 183 = 2017, or there is a solution for n = 2017. We will next prove
that no solution exists for n = 2011, 2012, 2016.
By the previous arguments, since n = 2016 is a multiple of 8, there may be solutions
only when a, b, c are all even. This means that x, y have the same parity, and so do z, w.
Since xy, zw must also have the same parity, either x, y, z, w are all even, or are all odd.
y
z
w
x
In the first case, define integers x = , y = , z = and w = , hence
2
2
2
2
x4 + y 4 + z 4 + w 4 4x y z w =

2016
= 126 = 6 (mod 8),
16

and no solution exists in this case. When x, y, z, w are all odd, a = x2 y 2 and b = z 2 w 2
are multiples of 8. By sheer trial (we need only to try a b {0, 8, 16, 24, 32, 40}), we
find that no such multiples of 8 exist such that 2016 a2 b2 is twice a perfect square, or
no solutions exist in this case either.
Again by the previous arguments, since n = 2012 = 4 (mod 8), we would need a, b odd,
or x, y have opposite parity, and xy is even, and similarly zw is also even, hence c is even,
contradiction. No solution exists either with n = 2012.
Finally, n = 2011 = 3 (mod 8), or wlog b, c are odd and a is even. Since b = z 2 w 2 is
odd, z, w have opposite parity, hence zw is even, and xy is odd, hence x, y are both odd,
and a = x2 y 2 is a multiple of 8. Again by sheer trial, we take a = 0, 8, 16, 24, 32, 40 and
try to express 2016 a2 as the sum of an odd perfect square b2 and twice another odd
perfect square 2c2 , finding only the following solutions:
a = 0, b = 43, c = 9. Since b = 43 = (z +w)(z w) is prime, it follows that zw = 2122 =
462, and a = 0 results in x = y, or x2 = 462 9, false since 453, 471 are not perfect squares.
a = 8, b = 37, c = 17. Similarly zw = 19 18 = 342, xy = 3 1 = 3, and c xy zw.
a = 8, b = 43, c = 7. Then, zw = 462, xy = 3, and again c xy zw.
a = 40, b = 13, c = 11. Then zw = 42, xy = 99 or xy = 21, and again c xy zw in
either case.

51

2.1. Arithmetic problems

a = 40, b = 19, c = 5. Then zw = 90, xy = 99 or xy = 21, and c xy zw. No solutions


exist either when n = 2011.
We conclude that the minimum such n is n = 2017.
()

Example 2.49. Let n be a positive integer. Prove that each odd common divisor of
n
n
n
n
; . . . ; Cn+1
; C2n1
; C2n1
C2n

is a divisor of 2n 1.
k
k = C k1 , we get that any odd common
First solution Using the identity Cm+1
Cm
m
divisor of the numbers is also a common divisor of
n
n1
n1
n1
.
; Cn+1
; . . . ; Cn+1
; C2n2
C2n1
k
Repeating this argument till we get Cn+1
for some k, we get that any odd common
divisor of the original set of numbers is also a common divisor of
n
2
1
; . . . ; Cn+1
; Cn+1
Cn+1

and of their sum, which equals 2n+1 2 = 2(2n 1). Therefore, we conclude that any
odd common divisor of the original set also divides 2n 1, and were done.
()
Second solution
By Vandermondes identity,
nj j
n
= Cn+1
Cn+i+1
Ci .
i

j=0

By using this formula for i = 0, . . . , n, we find that

1
n1
n
n
n
n
)
, . . . , Cn+1
, Cn+1
) = gcd (Cn+1
, . . . , C2n
, Cn+2
a = gcd (Cn+1

Therefore, if d is divisor of a then d divides the sum

k
= 2n+1 2 = 2(2n 1)
Cn+1
n

k=1

It follows that if d is odd then d divides 2n 1.

()

Example 2.50. Find all pairs (x, y) of positive integers such that x3 y 3 = 2010(x2 + y 2 ).

First solution
Write x = du, y = dv with d 1 and u, v relatively prime positive integers. The equation
becomes d(u3 v 3 ) = 2010(u2 + v 2 ). Thus u2 + uv + v 2 divides 2010(u2 + v 2 ) and since it is
relatively prime to u2 + v 2 , we deduce that u2 + uv + v 2 divides 2010 = 2 3 5 67. We
claim that A = u2 + uv + v 2 actually divides 67. It is immediate that A is odd (if not, u, v
must be both even). Next, it is easy to see that if 5 divides A, then 5 divides both u, v, a
contradiction. Finally, if 3 divides A, we must have u = v (mod 3) and so u3 v 3 = (u v)A
is a multiple of 9. Thus 2010(u2 + v 2 ) is a multiple of 9, which is not the case. Thus A
is a divisor of 67 and since u, v 1, we deduce that u2 + uv + v 2 = 67. Clearly u v, then
67 3v 2, thus v = 4. Considering each case, we deduce that v = 2, u = 7 and so d = 318.
Hence there is one solution, x = 7d, y = 2d with d = 318.
()

52

Chapter 2. Examples for practice

Example 2.51. Let m and n be positive integers. Prove that for each odd positive integer
b there are infinitely many primes p such that pn = 1 (mod bm ) implies bm1 n.
Solution
Let b = pa11 pa22 pakk be the canonical factorization of b. Because b is odd pi > 2 for
i = 1, 2, . . . , k.
Let P = p1 p2 pk . Consider the sysmte of conguences
x pi + 1 (mod p2i ), i = 1, 2, . . . , k.

()

By the Chinese Remainder Theorem the system (*) has solution. Let that solution be x0 .
We have x0 = pi + 1 (mod p2i ) for all i. If x x0 (mod P 2 ), then x is solution of system (*).
Claim. If x = x0 (mod P 2 ), then from condition xn = 1 (mod bm ) it follows that
bm1 n.
Proof. Suppose p P is a prime and a is a positive integer. Let vp (a) be the degree
of p in the canonical factorization of a. For each i, 1 = i = k, pi x0 1. We know that pi is
odd, therefore by a well knwon lemma vpi (xn 1) = vpi (x 1) + vpi (n).
But x x0 pi + 1 (mod p2i ), so vpi (x 1) = 1.
Hence vpi (n) = vpi (xn 1) 1 mai 1 (m 1)ai (because xn 1 (mod bm )).
(m1)ai
n. It is true for all i, and gcd(pi, pj ) = 1 whenever i j, therefore
So pi
bm1 = pi
k

i=1

(m1)ai

n.

Now from Dirichlets theorem there are infinitely many primes p such that p x0 (mod P 2 ). ()

Example 2.52. Let a and b be positive integers such that a does not divide b and b does
not divide a. Prove that there is an integer x such that 1 < x a and both a and b divide
x(b)+1 x, where is Eulers totient function.
Solution
We have, x (x(b) 1) is divisible by x and when gcd(b, x) = 1; applying Eulers theorem
this is also divisible by b.
a
a
we get, using gcd (
, b) = 1, that
So, letting x =
gcd(a, b)
gcd(a, b)

(b)

a
a
a
(
)
1 0 mod (
, b) 0 mod lcm(a, b).

gcd(a, b) gcd(a, b)
gcd(a, b)

Therefore the required x is

a
since this also satisfies 1 < x a.
gcd(a, b)

Example 2.53. Find all triples (x, y, z) of integers such that

x3 + y 3 + z 3
xyz = 2010 max{ 3 x y, 3 y z, 3 z x}
3

()

53

2.1. Arithmetic problems

Solution

Denote u = 3 x y, v = 3 y z, and w = 3 z x, hence clearly u3 + v 3 + w 3 = 0, which is


known to have integral solutions only if at least one of u, v, w is zero, while
x3 + y 3 + z 3
(x + y + z)(x2 + y 2 + z 2 xy yz zx)
xyz =
3
3
=

(x + y + z)(u6 + v 6 + w 6 )
, and
6

(x + y + z)(u6 + v 6 + w 6 ) = 12060u,

where we have assume wlog by cyclic symmetry in the variables that u = max{u, v, w}.
Note also that u 0, with equality iff u = v = w = 0, since if u 0, then
0 = u3 + v 3 + w 3 0,

equality must hold, hence u = v = w = 0. We then have two possible cases:


If u = 0, then u = v = w = 0, or x = y = z. Note that any x = y = z results in both
sides of the proposed equation being zero, hence (x, y, z) = (r, r, r) is a solution for any
integer r.
If u > 0, and since one of v, w is necessarily zero, while v 3 + w 3 = u3 , then the other
one necessarily equals u, ie (u, v, w) is some permutation of (u, 0, u), yielding
(x + y + z)u5 = 6030 = 2 32 5 67,

or since the RHS is not divisible by any fifth power, u = 1, and x + y + z = 6030, where
x = y + 1. Now, since one of v, w is zero, either z = y or z = x, yielding respectively
3y + 1 = 6030 or 3y + 2 = 6030, impossible since 6030 is a multiple of 3, but 3y + 1, 3y + 1
are not.
It follows that (x, y, z) is an integral solution of the proposed equation if x = y = z = r
for some integer r.
()
Example 2.54. Prove that the equation

x3 + y 3
= 2010
xy
is not solvable in positive integers.
First solution
Assume that x, y have opposite parity, then x3 + y 3 is odd, and 2010 is odd, contradiction, hence x, y have the same parity and x y is even. If x, y are both odd,
x3 + y 3
x2 xy + y 2 =
is odd, hence 2 divides x + y with multiplicity a + 1, where a is
x+y
the multiplicity with which 2 divides x y.
Now, (x + y) + (x y) = 2x is divided by 2 with multiplicity 1, hence a = 1, or if x, y are
both odd, xy is divisible by 2 but not by 4, and x+y is divisible by 4 but not by 8. If x, y
are both even, and both are divided by 2 with different multiplicity, then x + y and x y
xy(x + y)
= 2010 x2 + y 2 is a
are both divided by the lowest of both multiplicities, hence
xy

54

Chapter 2. Examples for practice

multiple x y of 4, or since x2 , y 2 are also multiples of 4, 2010 is a multiple of 4, absurd.


Therefore, x, y are divisible by 2 with the same multiplicity a, and x + y, x y are divisible
by 2 with multiplicity at least a + 1, hence x y is divisible by 2 with multiplicity at least
3a 3.
It is well known that any perfect cube leaves remainder 1, 0, 1 modulus 9, or since
x3 + y 3 is a multiple of 3 because 3 divides 2010, then x3 + y 3 is a multiple of 9, and x y
is a multiple of 3 because 2010 is not a multiple of 9. If x, y are not multiples of 3, since
x y divides xy(x + y), then x + y must be a multiple of 3, or 2x = (x + y) + (x y) is a
multiple of 3, contradiction, hence x, y are multiples of 3, hence x3 + y 3 is a multiple of
27, and x y must be a multiple of 9.
x3 + y 3
> x2 , we have x 44, or since y 1, then x y 43
Since 452 = 2025 > 2010 =
xy
must be an even multiple of 9, divisible by 2 if x, y are both odd, or divisible by at least
23 = 8 if x, y are both even. We conclude that x y = 18, and x, y are both odd. Writing
x as a function of y, the proposed equation becomes
y3
+ y 2 + 18y = 562,
27

z 3 + 9z 2 + 54z = 562,

where we have defined z =

y
because y is clearly divisible by 3. Note now that the LHS
3
increases strictly with z, and if z = 5 then z 3 + 9z 2 + 54z = 125 + 225 + 270 = 620 > 562,
while if z = 4 then z 3 + 9z 2 + 54z = 64 + 144 + 216 = 424 < 562. It follows that no positive
integral solutions exist for the proposed equation.
()
Second solution
Assume that the equation is solvable in positive integers. It is clear that x > y. We can
write
x3 + y 3 x3 y 3
>
= x2 + xy + y 2 = (x y)2 + 3xy > (x y)2 ,
2010 =
xy
xy

and get x y < 2010. It follows x y 44.


On the other hand, we have
2010 =

x3 + y 3 x3 + y 3 3 x3 + y 3
3 x+y 3

=
(
)
(
)
xy
44
44
3
44
3

hence we obtain x + y 96. The equation is equivalent to

(x + y)(x2 xy + y 2) = 2 3 5 67 (x y).

If x + y is divisible by 67, then necessarily x + y = 67, since x + y 96 and 67 is


a prime number. In this case we get x2 xy + y 2 = 30(x y) and x + y = 67, hence
(x + y)2 3xy = 30(x y). That is 672 = 30(x y) + 3x(67 x), equation with no integer
solutions, since 67 is not divisible by 3.
If x2 xy + y 2 is divisible by 67, then x2 xy + y 2 = 67k for some positive integer k.
The equation is equivalent to
k(x + y) = 30(x y),

55

2.1. Arithmetic problems

that is (30 k)x = (30 + k)y. It follows y =


and we get

30 k
x, hence
30 + k

x2 [(30 + k)2 (30 k)(30 + k) + (30 k)2 ] = 67k(30 + k)2 .


x2 (3k 2 + 302) = 67k(30 + k)2 .

It is clear that k is divisible by 3, hence we have k = 3a for some positive integer


1 a 9. Then
x2 (3a2 + 100) = 67 a (a + 10)2 .

Because x2 cant be divisible by 67 it follows that 3a2 +100 is divisible by 67. Replacing
a = 1, 2, ..., 9 is easy to see that 3a2 + 100 has no this property, hence the equation is not
solvable.
()
Example 2.55. For each positive integer n define
an =

(n + 1)(n + 2)(n + 2010)


.
2010!

Prove that there are infinitely many n such that an is an integer with no prime factors
less than 2010.
First solution
Let N be the least common multiple of {1, 2, 3, ..., 2012}, P the product of all primes
less than 2012, and take n = kNP , where k is any positive integer. Note that, for any
N
n + m kNP + m
=
= kP +1 is an integer relatively
m {1, 2, ..., 2012}, m divides N , and
m
m
m
prime with P , hence relatively prime with all primes less than 2012. We may therefore
express, for all positive integers k, akN P as a product of integers, all relatively prime with
all primes less than 2012. The conclusion follows.
()
Second solution
Let P be the set of primes less than 2012. If p P then the base p representation of
2012 has at most 11 digits:
(2012)p = c10 c9 ...c1 c0 .
For any a > 10, let n = pa .
pP

Hence the first 11 digits of n+2012 and 2012 in base p coincide, and by Lucas Theorem
10 1
0
2012
C1 C00 = 1 (mod p)
C10
Cs0 C11
Cn+2012

which means that p does not divide an for any p P .

Example 2.56. Determine all primes that do not have a multiple in the sequence
an = 2n n2 + 1, n 1.

56

Chapter 2. Examples for practice

Solution
We will prove that 2 and the primes congruent to 1 modulo 8 are all the primes that
do not have a multiple in the sequence.
First, note that it is clear that 2 satisfies the statement. Next, consider the primes of
the form 4k +1. Let p be such a prime and recall that in this case there is a positive integer
p1
!) . Take n = (p 1)(p q); by
q, q < p, such that p (q 2 + 1) (for example, check q =
2
Fermats little theorem,
2n n2 + 1 n2 + 1 q 2 + 1 0 (mod p),

and this proves the claim in this case.


Furthermore, let p be a prime of the form 8k + 3, for some positive integer k. In this
case, 1 and 2 are not quadratic residues modulo p, which implies that 2 is a quadratic
residue modulo p, i.e. there exists a positive integer q, q < p such that p (q 2 + 2.) Now,
take n = (p 1)(p q 1) 1, and see that Fermats little theorem gives
2n n2 + 1 2p2 q 2 + 1 0 (mod p),

which proves the claim in this second case.


Finally, consider a prime p = 8k 1 for some positive integer k, and note that here 2 is
a quadratic residue, whereas 1 is not a quadratic residue. This clearly implies that the
congruence 2n n2 + 1 = 0 does not have solution, thus yielding our conclusion.
()

6
Example 2.57. Let a1 , a2 , ..., a2009
be distinct
positive integers not exceeding 10 . Prove
that there are indices i, j such that iai jaj 1.

Solution
Let

( x + 1)2 ( x 1)2 3(x + 1) + 8030 x

f (x) =
2006
2009
2006 2009
6
Then f is an
increasing
function2 of x > 0. Hence f (a1 ) f (10 ) < 3, and therefore the
2
( a1 1) ( a1 + 1)
interval I = [
,
] contains at most three integers. By the pigeonhole
2009
2006

principle, there is an i {2006, ..., 2009} such that ai I , that is, iai a1 > 1, completing the proof.
()

Stronger result. Assume first that a1 1742 . Let

( x + 174)2 ( x 174)2

f (x) =
1201
2009
Then f is an increasing
function
of x > 0. Hence f (a1 ) f (106 ) < 808, and therefore

( a1 174)2 ( a1 + 174)2
the interval I = [
,
] contains at most 808 integers. By the pi2009
1201
geonhole
there is an i {1201, 1202, ..., 2009} such that ai I , which implies
principle,

that iai a1 > 174.


Consider next that a1 < 1742, then there is aj {1836, 1837, ..., 2009} such that aj 174.
Thus

jaj a1 > 1836 174 1742 > 174.

57

2.1. Arithmetic problems

Example 2.58. Let S(n) be the number of pairs of positive integers (x, y) such that
xy = n and gcd(x, y) = 1. Prove that
S(d) = (n2 ),
dn

where (s) is the number of divisors of s.

Solution
Let n = pa11 ...pakk be the prime decomposition of n. It is well known that
(n) = (a1 + 1)(ak + 1) (n2 ) = (2a1 + 1)(2ak + 1).

On the other hand, if for each subset M of {pa11 , ..., pakk } we put x = mM m and y =

n
,
x
then xy = n and gcd(x, y) = 1. Clearly all such pairs (x, y) may be obtained in this way,
hence S(n) = 2k .
The divisors of n are the summands in the expansion of the product
(1 + p1 + p21 + + pa11 )(1 + p2 + + pa22 )(1 + pk + + pakk ).

Now, if we substitute each pai i (with ai > 0) in the above expresion by 2, the expansion
will contain a summand 2r corresponding to each divisor of n with exactly r distinct prime
factors, i.e., the result will be dn S(d). Therefore,
S(d) = (1 + 2 + 2 + + 2)(1 + 2 + 2 + + 2)


dn
ak 2 s

a1 2 s

= (1 + 2a1 ) (1 + 2a2 ) (1 + 2ak ) = (n2 ).

Example 2.59. Let a1 , a2 , ..., an be integer numbers, not all zero, such that
a1 + a2 + + an = 0.
Prove that
for some k {1, 2, ...n}.

a1 + 2a2 + ... + 2k1 ak >

2k
,
3

First solution. Assume that


a1 + 2a2 + ... + 2k1 ak

2k
for all k {1, 2, ..., n},
3

the ai being integers. We shall prove by induction that


a1 = a2 = = an = 0.

For k = 1, the result is trivial, since a1 =

2
< 1 directly results in a1 = 0.
3

()

58

Chapter 2. Examples for practice


If the result is true for i = 1, 2, ..., k 1, then
2k
a1 + 2a2 + + 2k1 ak = 2k1 ak ,
3

2
< 1, and again ak = 0. All the ai are then zero, which is not true.
3
The result follows.
Second solution
Let k be the smallest integer such that ak is non-zero, that is,

yielding ak

()

a1 = a2 = ... = ak1 = 0 (if k > 1)

and ak > 0. Since ak is an integer, ak 1. Therefore

a1 + 2a2 + ... + 2k1 ak = 2k1ak , and 2k1 ak = 2k1 >

2 k1 2k
2 =

3
3

()

Example 2.60. Let n be an integer greater than 1. Find the least number of rooks such
that no matter how they are placed on an n n chessboard there are two rooks that do not
attack each other, but at the same time they are under attack by third rook.
First solution
We show that the least number of rooks is 2n 1. The standard algebraic notation
of the n n chessboard is used. By placing the rooks on a12 , a13 , ..., a1n , a21 , a31 , ..., an1 ,
we see that 2n 2 rooks are not sufficient. We will prove by induction that 2n 1 rooks
are sufficient. For n = 2, the result is clear. We now suppose that the result is true for
n = k 1. By placing the 2k + 1 rooks on the (k + 1) (k + 1) chessboard, there is at least
one row containing one rook or no rooks. Otherwise the total number of rooks is greater
than or equal to 2k +2, which is not true. Similarly there is at least one column containing
one rook or no rooks. Select any such row and any such column and delete them from
the (k + 1) (k + 1) chessboard. We combine the undeleted parts of the (k + 1) (k + 1)
chessboard to obtain a k k chessboard which contains at least 2k 1 rooks. Select any
2k 1 rooks. By the induction assumption, they are sufficient. It follows that 2k 1 rooks
are sufficient for the (k + 1) (k + 1) chessboard. This completes the solution.
()
Second solution
We will show that the least number of rooks such that the property holds in a m n
chessboard is m + n 1.
If the rooks are less than m+n1, we can place them along the first column and along
the first row but not at the top left corner (there are m + n 2 places). The property does
not hold for this displacement.
The thesis holds trivially when m + n 6, n > 1 and m > 1. Now we consider a m n
chessboard with m + n > 6, n > 1 and m > 1, and we assume that our thesis holds for any
m0 n0 chessboard such that m0 + n0 < m + n and m0 > 1, n0 > 1. We can assume without
loss of generality that n m and therefore n > 3. Since we have at least m + n 1 = m + 1
rooks, then there is a row with at least two rooks. If there is at least another rook in
the corresponding colums then the property holds. Otherwise we can cancel these two

59

2.1. Arithmetic problems

columns obtaining a m (n 2) chessboard with at least m + (n 2) 1 rooks. Since


m + n > m + (n 2), m > 1 and n 2 > 1, by the inductive hypothesis, the property holds
in this smaller checkboard and therefore it holds also in the initial one.
()
Example 2.61. Let n be a positive integer. Prove that
a) there are infinitely many triples (a, b, c) of distinct integers such that min(a, b, c) = n
and abc + 1 divides one of the numbers (a b)2 , (b c)2 , (c a)2 .
b) there is no triple (a, b, c) of distinct positive integers such that abc + 1 divides more
than one of the numbers (a b)2 , (b c)2 , (c a)2 .
Solution
For any positive integer n, the following relation holds:

(n3 + 6n2 + 10n + 4)(n2 + 4n + 3)n + 1 = n6 + 10n5 + 37n4 + 62n3 + 46n2 + 12n + 1
= (n3 + 5n2 + 6n + 1)2
= (n3 + 6n2 + 10n + 4) (n2 + 4n + 3)2 .

Taking a = n3 + 6n2 + 10n + 4, b = n2 + 4n + 3 and c = n, clearly min(a, b, c) = c = n, and


abc + 1 divides (a b)2 = abc + 1. The conclusion of part a) follows.
()

Assume now wlog that a > b > c. If abc + 1 divides (b c)2 , then a > b > b c, and
abc + 1 (b c)2 < ab, clearly impossible.
If abc + 1 divides (a b)2 and (a c)2 , then it divides
(a c)2 (a b)2 = (2a b c)(b c) < 2ab,

and since 2a > 2a b c > 0 and b > b c > 0, then


resulting in c = 1 and

2ab > (2a b c)(b c) abc + 1,

(2a b 1)(b 1) = ab + 1, or a =

b2
4
= b+2+

b2
b2

Since b 2 divides 4, then b = 3, 4, 6, yielding respectively a = 9, 8, 9; however, in none


of these cases does abc + 1 divide (a b)2 or (a c)2 .
The conclusion of part b) follows.
()

60

Chapter 2. Examples for practice

2.2
2.2.1

Algebraic problems
Junior problems

Example 2.62. Let a, b, c be real numbers such that a3 bc.


1
1

Prove that b2 + c2 whenever a6 + b6 + c6


3
27
First solution
1
Assume the contrary that b2 + c2 < Then note that a6 (bc)2 , thus
3
1
a6 + b6 + c6 (bc)2 + b6 + c6
27
= (b2 + c2 )3 + (bc)2 (1 3(b2 + c2 ))
(b2 + c2 )3 + (

b2 + c2 2
) (1 3(b2 + c2 ))
2

(b2 + c2 )2
(4(b2 + c2 ) + 1 3(b2 + c2 ))
4
2
( 13 )
(b2 + c2 )2
1
1
2
2
(1 + (b + c )) <
=
(1 + ) = ,
4
4
3
2t

1
a contradiction. Thus b2 + c2 , and we are done.
3
Example 2.63. Let a, b, c, d be positive real numbers. Prove that

Solution

()

a+b 3
c+d 3
a2 + d2 3
b2 + c2 3
) +(
) (
) +(
)
2
2
a+d
b+c

Using the well known inequality (

x + y 3 x3 + y 3
)
, we have that
2
2
c + d 3 a3 + b3 c3 + d3
a+b 3
) +(
)
+
(
2
2
2
2

with equality if and only if a = b and c = d. Now notice that if we let

But,

a2 + d2 3
x2 + 1 3
a3 + d3
(
) x3 + 1 2 (
)
2
a+d
x+1

a
= x, we have
d
(2.11)

(2.13) x6 3x5 + 3x4 2x3 + 3x2 3x + 1 0 (x 1)4 (x2 + x + 1) 0 is true,

so we obtain that

a3 + d3
a2 + d2 3
(
)
2
a+d
with equality if and only if a = d. Analogously
b2 + c2 3
b3 + c3
(
)
2
b+c

61

2.2. Algebraic problems

with equality if and only if b = c.


Hence the original inequality is true with equality if and only if a = b = c = d.

()

Example 2.64. Let x, y, z be real numbers. Prove that


2
(x2 + y 2 + z 2 )2 + xyz(x + y + z) (xy + yz + zx)2 + (x2 y 2 + y 2 z 2 + z 2 x2 ).
3

First solution
The original inequality will follow from the following sharper inequality

2
2
(x2 + y 2 + z 2 ) + xyz(x + y + z) x2 y 2 + y 2z 2 + z 2 x2 (xy + yz + zx)2 .
3

(2.12)

Indeed, for any real u, v, w we have

Then

u2 + v 2 + w 2 uv + vw + wu (u + v + w)2 3(uv + vw + wu)


(u v)2 + (v w)2 + (w u)2 0.
(x2 + y 2 + z 2 ) 3 (x2 y 2 + y 2 z 2 + z 2 x2 )
2

and
Therefore,

x2 y 2 + y 2z 2 + z 2 x2 xyz(x + y + z).

LHS(2.14) 2 (x2 y 2 + y 2 z 2 + z 2 x2 ) + xyz(x + y + z)

x2 y 2 + y 2z 2 + z 2 x2 + 2xyz(x + y + z)
2
= (xy + yz + zx)2 (xy + yz + zx)2 = RHS(2.14).
3

()

Remark. Equality in (2.14) occurs if and only if x = y = z and in original inequality


equality occurs if and only if x = y = z = 0.
Second solution
By expanding the terms (x2 + y 2 + z 2 )2 and (xy + yz + zx)2 and equating the terms on
both sides leads to
1
1
x4 + y 4 + z 4 + (x2 y 2 + y 2z 2 + z 2 x2 ) xyz(x + y + z) 0.
3
3

By the AM-GM inequality we have

x2 y 2 + y 2z 2 + z 2 x2 xyz(x + y + z).
Hence it is enough to prove that
x4 + y 4 + z 4 = 0

which is obvious. Equality occurs if and onyl if x = y = z = 0.

()

62

Chapter 2. Examples for practice

Example 2.65. Let H(x, y) =

2xy
be the harmonic mean of the positive real numbers
x+y

x and y. For n 2, find the


greatest constant C such that for any positive real numbers a1 , . . . , an , b1 , . . . , bn the
following inequality holds

Solution
Let
S=
Note that

1
1
C

++

H(a1 + + an , b1 + + bn ) H(a1 , b1 )
H(an , bn )
H(a1 + + an , b1 + + bn )
H(a1 + + an , b1 + + bn )
++
H(a1 , b1 )
H(an , bn )
S=
=

an + bn
(a1 + + an )(b1 + + bn ) a1 + b1
++
]
[
(a1 + + an ) + (b1 + + bn ) a1 b1
an bn

(a1 + + an )(b1 + + bn ) 1
1
1
1
[ + + + ]
(a1 + + an ) + (b1 + + bn ) a1
an b1
bn

1
1
1
1
+ + +
a1
an b1
bn
n2
=
1
1
+
a1 + + an b1 + + bn
The last inequality is true because

1
1
1
+ n2
;
a1
an
a1 + + an
1
1
1
+ n2
b1
bn
b1 + + bn
by the AM-HM inequality. Equality occurs if and only if a1 = = an and b1 = = bn so
C = n2 .

sec a
sec b

and sin2 b + cos 2a


.
Example 2.66. Let a, b (0, ) such that sin2 a + cos 2b
2
2
2
Prove that
1
cos6 a + cos6 b
2
Solution
We will use the following well-known trigonometric identities
(a) sin2 x = 1 cos2 x, (b) cos 2x = 2 cos2 x 1, (c)

1
= cos x.
sec x

The inequalities can be written as

cos2 b cos a cos3 a

1
2

(2.13)

63

2.2. Algebraic problems

and

cos2 a cos b cos3 b

1
2

(2.14)

The signs of the inequalities are preserved because cos x is positive when x (0, ).
2
Now by squaring both sides of (2.13) and (2.14) and adding them we get
cos6 a + cos6 b

Example 2.67. Solve in positive real numbers the system of equations

Solution
We have

x1 + x2 + + xn

1
1
1
1

+
+

+
+

xn x1 x2 xn
x1 x2

()

=1

= n3 + 1.

1
1
1
n2
+ +

= n2
x1 x2
xn x1 + x2 + + xn
and

Thus,

1
1
= nn

x1 + x2 + + xn n
x1 x2 xn
)
(
n
n3 + 1 nn + n2

which implies that n 2. If n = 1 we get a contradiction. For n = 2 we get

x1 + x2

1
1
1

+
+

x1 x2 x1 x2

=1

=9

1 2
2 1
which is (n, x1 , x2 ) {(2, , ) , (2, , )} .
2 3
3 3

()

Example 2.68. Let x, y, z be nonnegative real numbers such that ax + by + cz = 3abc for
some positive real numbers a, b, c. Prove that

x+y
y+z
z+x
1
+
+
+ 4 xyz (abc + 5a + 5b + 5c).
2
2
2
4
Solution
From the given condition,

3a

ax y z
+ + ;
bc c b

64

Chapter 2. Examples for practice


3b

x by z
+
+ ;
c ca a
x y cz
3c + +
b a ab

Then
3(a + b + c)
hence

x+y y+z z+x


ax by cz
+
+
+( +
+ )
c
a
b
bc ca ab

y+z
z+x
x+y
+ 2c) + (
+ 2a) + (
+ 2b)
c
a
b
ax by cz
+ abc +
+
+
bc
ca
ab

2 2(x + y) + 2 2(y + z) + 2 2(z + x) + 4 4 xyz

abc + 5(a + b + c) (

and the conclusion follows. The equality holds if and only if


x + y = 2c2 , y + z = 2a2 , z + x = 2b2 and abc =

ax by cz
=
=
bc ca ab

This implies
b2 c2 = c2 a2 = 2c2 , c2 a2 + a2 b2 = 2a2 , a2 b2 + b2 c2 = 2b2 ,

that is b2 + a2 = c2 + b2 = a2 + c2 = 2, implying a = b = c = 1 and x = y = z = 1. If


a = b = c = x = y = z = 1, the equality, as well as the condition of the problem, hold. ()
Example 2.69. Solve in real numbers the system of equations

(x + y)(y 3 z 3 ) = 3(z x)(z 3 + x3 )

(y + z)(z 3 x3 ) = 3(x y)(x3 + y 3 )

3
3
3
3

(z + x)(x y ) = 3(y z)(y + z ).

Solution
Without loss of generality assume that x = 0, then it is clear that y = z = 0. Because
the given system is symmetric we can assume that x, y, z 0. Assume that x = y, then
x = z or y = z. If we assume that y = z then the first equation becomes 4x4 = 6x4 ,
contradiction. Now x y z 0 and after multiplying all three equations we get
3(x2 xy + y 2 )3(y 2 yz + z 2 )3(z 2 zx + x2 )
=1
(x2 + xy + y 2 )(y 2 + yz + z 2 )(z 2 + zx + x2 )

which can be written as


(1

2xy
2yz
2zx
1
) (1
) (1
)=

2
2
2
(x y) + 3xy
(y z) + 3yz
(z x) + 3zx
27

It is not difficult to see that each factor of the LHS is greater than

a contra- diction. So the only possible solution is x = y = z.

1
which leads to
3
()

65

2.2. Algebraic problems

Example 2.70. Let a, b, c, d be distinct real numbers such that

Prove that

1
1
1
1

+
+
+
0
3
3
3
3
ab
bc
cd
da

3
3
3
3
a b + b c + c d + d a 0.

Solution

3
3
3
3
Let a b, b c, c d, d a by w, x, y, z, respectively and let S = w + x + y + z.
So that,
1 1 1 1
w 3 + x3 + y 3 + z 3 = 0 () and + + + 0 ()
x y z u
Since a, b, c and d are distinct, we can conclude that w, x, y, z 0, and thus,
1 1 1 1
wxy + wxz + wyz + xyz = wxyz ( + + + ) 0.
x y z u

Furthermore, using (*), we have that,

( )

S 3 = 6(wxy + wxz + wyz + xyz) + 3(w 2 + x2 + y 2 + z 2 )S


S(S 2 3(w 2 + x2 + y 2 + z 2 )) = 6(wxy + wxz + wyz + xyz)

Which implies that S 0, otherwise the right hand side would be 0 which is a contradiction of (* * *), and were done.
()
Example 2.71. If x, y, z > 0 and x + y + z = 1, find the maximum of
E(x, y, z) =

xy
yz
zx
+
+

x+y y+z z+x

Solution
By the AM GM inequality,
(

x+y 2
y+z 2
z+x 2
)
)
)
(
(
2
2
2
E(x, y, z)
+
+
x+y
y+z
z+x
=

x+y y+z z+x x+y+2 1


+
+
=
=
4
4
4
2
2

Therefore the maximum of E(x, y, z) is

1
1
, occurring for example if x = y = z =
2
3

Example 2.72. Let a and b be rational numbers such that

Prove that a2 + b2 17.

a2

47
52
and b 2

2
3ab
b 3a2

()

66

Chapter 2. Examples for practice

First solution
Let X = a3 3ab2 and Y = 3a2 bb3 . From the hypothesis, we have X 47 and Y 52.
Observing that X + iY = (a + ib)3 (easily checked), we deduce that
a + b = a + ib = ((a + ib) ) = ((X + Y ) ) = (X 2 + Y 2 ) 3 = (472 + 522 ) 3 .
2

2
3

1
2

2
3

Since 472 + 522 = 4913 = 173 , it follows that a2 + b2 17.

()

Second solution
Note that a = b = 0 clearly satisfies a2 + b2 < 17, therefore a2 3b2 and b2 3a2 are
nonzero rationals, since otherwise 3 would be the square of a rational, absurd. Define
x = a2 , y = b2 , hence x(x 3y)2 472 and y(3x y)2 522 . Adding these two inequalities
results in (x + y)3 472 + 522 = 173 , or x + y = a2 + b2 17. Note that equality is reached
whenever equality holds simultaneously on both inequalities given as condition in the
problem statement, for example when a = 1 and b = 4.
()
Example 2.73. Let a, b, c be positive real numbers. Prove that
a2 b2 (b c)
0.
a+b
cyc

Solution
Clearing the denominators, the inequality is equivalent to
(a4 c2 b + c4 b3 ) 2a3 b2 c
cyc

but this follows from the AM-GM:


Example 2.74. Let
fn (x, y, z) =

cyc

(a4 c2 b + c4 b3 )
a3 b2 c.
2

(x y)z n+2 + (y z)xn+2 + (z x)y n+2

(x y)(y z)(x z)

()

Prove that fn (x, y, z) can be written as a sum of monomials of degree n and find
fn (1, 1, 1) for all positive integers n.
Solution
It is easily verified that

1
1
1
+
+
= 0.
(x y)(x z) (y x)(y z) (z y)(z x)

This allows one to transform fn (x, y, z) as follows:


f n(x, y, z) =

xn+2
y n+2
z n+2
+
+
(x y)(x z) (y x)(y z) (z y)(z x)

67

2.2. Algebraic problems

=
=
=
=

xn+2
y n+2
z n+2
xn+2
+
+
+
(y x)(y z) (z y)(z x) (y x)(y z) (z y)(z x)
z n+2 xn+2
y n+2 xn+2
+
(y x)(y z) (z y)(z x)

y n+1 + y n x + + xn y + xn+1 z n+1 + z n x + + xn z + xn+1


+
(y z)
(z y)
(y n+1 z n+1 ) + x(y n z n ) + + xn (y z)
yz

and finally
fn (x, y, z) = (y n +y n1 z++yz n1 +z n )+x(y n1 +y n 2z++yz n2 +z n1 )++xn1 (y+z)+xn .

Thus, fn (x, y, z) is the sum of all monomials xa y bz c over all triples (a, b, c) of nonnegative integers satisfying a + b + c = n. This answers the first part of the question. As for
the second part, fn (1, 1, 1) is just the total number of monomials in the above sum, that
is,
(n + 1)(n + 2)
(n + 1) + n + + 2 + 1 =

().
2
Example 2.75. Let a, b, c be nonzero real numbers such that ab + bc + ca 0. Prove that
a2

ab
bc
ca
1
+ 2 2+ 2

2
2
+b
b +c
c +a
2

Solution
We have
ab
(a + b)2
1
3
3
ab
(
(
=
+
)

=
)

2
2
2
2
2
2
2
2 cyc 2 (a + b )
2
cyc a + b
cyc a + b

(
cyc

=1+

(a + b)2
3 2 (a2 + b2 + c2 ) + 2(ab + bc + ca) 3
)

2 (a2 + b2 + c2 )
2
2(a2 + b2 + c2 )
2

(ab + bc + ca) 3
1

2
2
2
(a + b + c ) 2
2

where in the last step we have used the fact that ab + bc + ca 0.

()

Example 2.76. If x and y are positive real numbers such that

(x + x2 + 1) (y + y 2 + 1) = 2012,
find the minimum possible value of x + y.

Solution

z2 1

Let z = x + x2 + 1. We have z > 0 and x =


2z

(i)

68

Chapter 2. Examples for practice


From hypothesis y +
From (i) and (ii),

2012
20122 z 2
, we get y =

y2 + 1 =
z
2 2012 z

(ii)

2011
2012
2011 2012
z 2 1 20122 z 2
+
=
(z +
)

x+y =
2z
2 2012 z 2 2012
z
2012

The equality occurs for z =


obtain

2012
or equivalently z 2 = 2012. Then from (i) and (ii) we
z
2011
x=y=
2 2012

2011 2012

So min(x + y) =
2012

()

Example 2.77. Let a, b, c be real numbers greater than 1 such that


c+a
a+b
b+c
+ 2
+ 2
1.
2
a 1 b 1 c 1
Prove that

bc + 1 2
ca + 1 2
ab + 1 2 10
( 2
) +( 2
) +( 2
)

a 1
b 1
c 1
3

First solution Observe that


(

Therefore

Now observe that

b + c 2 (b2 1)(c2 1)
bc + 1 2
)

(
) =
;
a2 1
a2 1
(a2 1)2

ca + 1 2
c + a 2 (c2 1)(a2 1)
)

(
) =
;
b2 1
b2 1
(b2 1)2
ab + 1 2
a + b 2 (a2 1)(b2 1)
)

(
) =

c2 1
c2 1
(c2 1)2

(b2 1)(c2 1)
b+c 2
bc + 1 2
)
=
(
)
+

a2 1
a2 1
(a2 1)2
(

bc + 1 2
)
a2 1

bc + 1 2
)
1
a2 1

3
3

(2.15)

(2.16)

and by A.M-G.M inequality we get

(b2

(a2 1)2 (b2 1)2 (c2 1)2


1)
3

3
= 3.
(a2 1)2
(a2 1)2 (b2 1)2 (c2 1)2
1)(c2

1 10
bc + 1 2
) 3+ =

By virtue of (2.15),(2.16) and (2.17) we obtain ( 2


a 1
3 3

(2.17)
()

69

2.2. Algebraic problems

Example 2.78. Let a, b, c be positive real numbers such that a + b + c + 2 = abc. Find the
minimum of
1 1 1
+ +
a b c
First solution
Note that a + b + c + 2 = abc implies
Then

3 + 2(a + b + c) + (ab + bc + ca) = abc + ab + bc + ca + a + b + c + 1.


(a + 1)(b + 1)(c + 1) = ((a + 1)(b + 1))
cyc

and dividing by the left-hand side, we obtain

1
=1
cyc a + 1
cyc

1
1+

1
a

= 2.

By the Cauchy-Schwarz inequality


1
(1 + )
a cyc
cyc
hence the minimum is

1 3
9
1
9 (1 + )
1
a
2
2
cyc a
cyc
1+
a
1

()

Second solution
1
1
1
Let x =
,y =
,z =
By trivial algebra we observe that a + b + c + 2 = abc
1+a
1+b
1+c
is equivalent to x + y + z = 1 and then
a=

1x y+z
1y x+z
1z y+x
=
,b =
=
,c =
=
.
x
x
y
y
z
z

In terms of the variable x, y, z the inequality is x + y + z = 1 and


y
z
x
+
+
y+z x+z x+y

3
That the minimum of the above expression is is the content of Nesbitts inequality
2
which is well known. One of the many proof available is
x
y
z
(x + y + z)2
1
3
+
+

y + z x + z x + y 2(xy + yz + zx) 2(xy + yz + zx) 2

having employed Cauchy-Schwarz. Thus we have xy + yz + zx

1
which is obvious. ()
3
Example 2.79. Let a1 , a2 , ..., an be positive real numbers. Prove that
a2
an
a1 + + an
a1

+
++

2
2
2
(1 + a1 )
(1 + a1 + a2 )
(1 + a1 + + an )
1 + a1 + + an

70

Chapter 2. Examples for practice


Solution
We will prove the result by induction. For n = 1, the result is equivalent to
a1
a1
,

2
(1 + a1 )
1 + a1

clearly true with a strict inequality since 1 + a1 > 1. If the result is true with a strict
inequality for n 1, then for the result to be true with a strict inequality for n, it suffices
to show that
an
sn
sn an
+
(sn an )(1 + sn )2 (sn + s2n an )(1 + sn an )

1 + sn an (1 + sn )2 1 + sn

where we have denoted sn = a1 +a2 ++an , and which after performing the products in both
sides and simplifying, transforms into a2n 0, trivially true. The conclusion follows, the
inequality being strict for all n. One may get arbitrarily close to equality when a1 , a2 , ..., an
all tend to 0, both sides being arbitrarily close to a1 +a2 ++an , which is in turn arbitrarily
close to zero.
()
Example 2.80. Let a b c > 0. Prove that

1 1 1
(a b + c) ( + ) 1.
a b c

Solution
We write the inequality as follows

1
1 1 1
+

a b c ab+c
It is equivalent to

()

a+c
a+c

ac
b(a b + c)

Therefore it is enough to check that ac b(a b + c) or (b a)(b c) 0.


The last inequality is true due to the given condition. Hence, the proof is completed.

Example 2.81. Let a, b, c be positive real numbers. Prove that


1
1
1
1
1
1
+
+

+
+

10a + 11b + 11c 11a + 10b + 11c 11a + 11b + 10c 32a 32b 32c
Solution
Recall the well-known inequality,
1
n2
,
n
k=1 ai
k=1 ai
n

we have

322
1 1
1 1
1
1
++ + ++ + ++
a
a b
b c
c 10a + 11b + 11c

10

times

11

times

11

times

71

2.2. Algebraic problems

which is equivalent to

11
11
32
10
+
+

32a 32b 32c 10a + 11b + 11c

Similarly

10
11
32
11
+
+

32a 32b 32c 10a + 11b + 11c

and

11
11
10
32
+
+

32a 32b 32c 10a + 11b + 11c


Adding the three inequalities we get the desired result.

()

Example 2.82. Let x, y, z be positive real numbers. Prove that

1
1
1
1
1
1
1 1
1
+ +y
+ +z
+
2(x2 y 2 + y 2z 2 + z 2 x2 ) ( 3 + 3 + 3 ) x
x
y
z
y z
z x
x y
Solution
Note that by the Cauchy-Schwarz Inequality we have that

x
cyc

1 1
1
1
x2 ( + ) =
2 x2 1 ,
+
y z
z
cyc
cyc y
cyc
cyc x

so it would suffice to prove that

(x2 y 2 + y 2 z 2 + z 2 x2 ) (

1
1
1 1 1
1
+ 3 + 3 ) (x2 + y 2 + z 2 ) ( + + )
3
x
y
z
x y z

1
1
1
Now, let a = , b = , c = ; then the inequality to be proven can be rewritten as
x
y
z
(

1
a2 b2

1
b2 c2

1
c2 a2

which is equivalent with

i.e.

) (a3 + b3 + c3 ) (

1
1 1
+ 2 + 2 ) (a + b + c)
2
a
b
c

(a2 + b2 + c2 ) (a3 + b3 + c3 ) (a2 b2 + b2 c2 + c2 a2 ) (a + b + c),


a5 + a3 (b2 + c2 ) a3 (b2 + c2 ) + ab2 c2 ,
cyc

cyc

cyc

cyc

which turns out to be just the immediate

a5 ab2 c2 = abc(ab + bc + ca),


cyc

cyc

which can be seen for example as a consequence of the AM-GM Inequality.

()

Example 2.83. Let x, y, z be positive real numbers with x 2, y 3 and x + y + z = 11.


Prove that xyz 36.

72

Chapter 2. Examples for practice


Solution
By AM-GM inequality, we have
xyz =
=

1 3x + 2y + z 3
1
(3x) (2y) z (
)
6
6
3

1 2x + y + 11 3 1 2 2 + 3 + 11 3
(
) (
) = 36.
6
3
6
3

We have equality if and only if x = 2, y = 3, z = 6.

()

Example 2.84. Let a, b, c be positive real numbers. Prove that


(a + b)2 c2
+
4b.
c
a

First solution
We will use the following lemma.
Lemma 2.

Let a, b, , be positive real numbers then


a2 b2 (a + b)2
+


+
Proof. The lemma can be proved by straight forward calculations.
Based on the lemma we have that

so it suffices to prove that

()

()

(a + b)2 c2 (a + b + c)2
+
,
c
a
a+c

(a + b + c)2
4b.
a+c
The preceding inequality is equivalent to (a b + c)2 0, and the problem is solved.
Second solution
By the Cauchy-Schwarz inequality, we have

c
LHS (c + a)

(a + b)2
c2
+ a
c
a

= (a + b + c)2 = (b + (a + c))2

2
= (2 b (a + c)) = 4b (a + c)
= RHS (a + c).

(and by AM-GM inequality)

73

2.2. Algebraic problems

Dividing both sides by (a + c) we obtain the desired result.

()

Third solution
After clearing denominators we need to prove that

f (b) = ab2 + (2a2 4ac)b + a3 + c3 0.

Considering f as a quadratic function in b we can easily check that the discriminant


= 4ac(2a c)2 0. Since the leading coefficient of this quadratic
function is positive, we conclude that f (b) 0 for all b 0.
()

(2a2 4ac)2 4a(a3 + c3 )

Fourth solution
The proposed inequality follows from:

a3 + 2a2 b + ab2 + c3 4abc ab2 + 2(a 2c)b + a2 + c3


(a + b)2 c2
+ 4b =
=
c
a
ac
ac
=

=
=
=
=

a [b2 + 2(a 2c)b + (a 2c)2 (a 2c)2 + a2 ] + c3


ac

a [(b + a 2c)2 + 4ac 4c2 ] + c3


ac

a(a + b 2c)2 + 4a2 c 4ac2 + c3


ac

a(a + b 2c)2 + c(4a2 4ac + c2 )


ac

(a + b 2c)2 (2a c)2


+
0.
c
a

()

Example 2.85. Let a, b, c be positive real numbers. Prove that


2a2

b
c
1
a
+ 2 2
+ 2

2
2
+ b + 3 2b + c + 3 2c + a + 3 2

Solution
We have
2a2 + b2 + 3 = 2(a2 + 1) + (b2 + 1) 4a + 2b

Similarly, we have

2b2

a
a

.
2a2 + b2 + 3 4a + 2b

b
b
c
c
;

,
2
2
2
+c +3
4b + c 2c + a + 3 4c + a

and we obtain
2a2

a
b
c
a
+ 2 2
+ 2

.
2
2
+ b + 3 2b + c + 3 2c + a + 3 cyc 4a + 2b

(i)

74

Chapter 2. Examples for practice

Now we prove

a
1
a

1.
2
cyc 2a + b
cyc 4a + 2b

Which is equivalent to

ab2 + bc2 + ca2 3abc,

(ii)

clearly true by the AM - GM inequality. From (i) and (ii) follows that
2a2

b
c
1
a
+ 2 2
+ 2

2
2
+ b + 3 2b + c + 3 2c + a + 3 2

Example 2.86. Let a, b, c be positive real numbers. Prove that

()

ab
bc
ca
a+b+c
+
+

3a + 4b + 2c 3b + 4c + 2a 3c + 4a + 2b
9
Solution
The inequality is equivalent to
9ab
9bc
9ca
+
+
a + b + c.
3a + 4b + 2c 3b + 4c + 2a 3c + 4a + 2b
From Cauchy-Schwartz inequality it is known that
9ab
ab
ab
ab

+
+

3a + 4b + 2c a + b + c a + b + c a + 2b
Adding the other 2 similar inequalities we get that it suffices to show that
2(ab + bc + ca)
ab
bc
ca
+
+
+
a + b + c.
a+b+c
a + 2b b + 2c c + 2a
Because from the AM-GM inequality (a + b + c)2 3(ab + bc + ca), we conclude that
2(ab + bc + ca) 2(a + b + c)

a+b+c
3

So it is enough to prove that


ab
bc
ca
a+b+c
+
+

.
a + 2b b + 2c c + 2a
3
But from AM-GM inequality
(a + 2b)(b + 2a) 9ab

ab
b + 2a

a + 2b
9

Adding the other two similar inequalities we get that


bc
ca
b + 2a c + 2b a + 2c a + b + c
ab
+
+

+
+
=

a + 2b b + 2c c + 2a
9
9
9
3
The proof is complete.

()

75

2.2. Algebraic problems


n

Example 2.87. Let a1 , ..., an be positive real numbers such that

2
i=1 ai

Prove that

ai aj

1i<jn

1
= n 1.
+1

Solution
From the hypothesis and the Cauchy-Schwarz Inequality, we have
n
a2i
(ni=1 ai )
1
=

2
2
n + ni=1 a2i
i=1 ai + 1
i=1 ai + 1
2

1=n
n

Thus

( ai ) a2i n.
n

i=1

On the other hand, we have

i=1

2 ai aj = ( ai ) a2i
n

i=1

1i<jn

Therefore

i=1

ai aj

1i<jn

()

1
Equality holds when a1 = a2 = = an =

n1

Example 2.88. Let a, b, c 1 be real numbers such that a + b + c = 2abc. Prove that

3
3
3
3
(a + b + c)2 = ab 1 + bc 1 + ca 1.
(2.18)
First solution
We write the given equality as follows

1
1
1
ab 1 bc 1 ca 1
+ +
=2
+
+
= 1.
ab bc ca
ab
bc
ca

get

Applying H
olders inequality to the triples (a, b, c), (b, c, a), (

ab 1 bc 1 ca 1
,
,
), we
ab
bc
ca
1

ab 1 bc 1 ca 1 3
+
+
)
LHS(2.18) = (a + b + c) (b + c + a) (
ab
bc
ca
1
3

1
3

and we are done.

bc 1 3
ca 1 3
ab 1 3
(ab
) + (bc
) + (ca
)
ab
bc
ca

3
3
3
= ab 1 + bc 1 + ca 1 = RHS(2.18).

()

76

Chapter 2. Examples for practice


Second solution
a+bc
thus by the AM-GM inequality
We are given that 0 (ab 1)
2c

(a + b c) + a + b
3 ab(a + b c)
3
cyc
cyc

= 3 2abc(ab 1) = 3 (a + b + c)(ab 1).

a+b+c =
cyc

Therefore

cyc

3
3
(a + b + c)2 = ab 1 and we are done.

()

cyc

Example 2.89. Let a, b, c be positive real numbers such that


a2

1
1
1
+ 2 2
+ 2
1.
2
+ b + 1 b + c + 1 c + a2 + 1

Prove that ab + bc + ca 3.

Solution
By applying the Cauchy-Scharz inequlity, we have 4 or

Similarily, we obtain

Therefore

1
2 + a2
1
2 + b2

b2 + c2 + 1 (a + b + c)2 c2 + a2 + 1 (a + b + c)2
1

or
equivalent to

1
2 + c2

a2 + b2 + 1 (a + b + c)2

1
1
1
6 + a2 + b2 + c2
+
+

a2 + b2 + 1 b2 + c2 + 1 c2 + a2 + 1
(a + b + c)2
(a + b + c)2 6 + a2 + b2 + c2

a2 + b2 + c2 + 2(ab + bc + ca) 6 + a2 + b2 + c2 .

Thus ab + bc + ca 3 and we are done.

Example 2.90. Let a and b be positive real numbers. Prove that


a6 + b6 a4 + b4 a2 + b2

a4 + b4 a3 + b3 a + b
First solution
The above inequality is equivalent to
(a6 + b6 )(a3 + b3 )(a + b) (a4 + b4 )2(a2 + b2 ).

()

77

2.2. Algebraic problems

Using the Cauchy-Scharz inequality, we have


(a6 + b6 )(a2 + b2 ) (a4 + b4 )2 ; (a3 + b3 )(a + b) (a2 + b2 )2 .

Multiplying these inequalities, the conclusion follows. Equality occurs when a = b. ()


Second solution
After multiplying out the right side the inequality becomes
a6 + b6 a6 + a4 b2 + b4 a2 + b6
4
a4 + b4
a + a3 b + b3 a + b4
or

a6 + b6 a6 + b6 + a2 b2 (a2 + b2 )
4 4

a4 + b4
a + b + ab(a2 + b2 )

To simplify calculations let use make the following substitutions


a6 + b6 = x ; a4 + b4 = y ; ab = z ; a2 + b2 = t.
Our inequality transforms to

x x + z2t

y
y + zt
or after some algebra x yz. If we back substitute we obtain
which is exactly the same as

a6 + b6 (a4 + b4 )ab
(a5 b5 )(a b) = 0

and thus we are done. Equality occurs when a = b.

2.2.2

()

Senior problems

Example 2.91. Let x1 , . . . , xn , a, b > 0. Prove that the following inequality holds
x31
x3n
x1 + + xn
++

ax1 + bx2 )(ax2 + bx1 )


(ax1 + bxn )(axn + bx1 )
(a + b)2

First solution
We have

4x31 (x1 + x2 )2 (2x1 x2 ) = 2x31 3x21 x2 + x32 = (x1 x2 )2 (2x1 + x2 ) 0

with equality if and only if x1 = x2 . It follows that


x1 + + xn = (2x1 x2 ) + + (2xn x1 )

It then suffices to show that

4x31
4x3n
+

(x1 + x2 )2
(xn + x1 )2

1
4

(x1 + x2 )2 (a + b)2 (ax1 + bx2 )(ax2 + bx1 )

(2.19)

78

Chapter 2. Examples for practice

which is equivalent to (x1 x2 )2 (a b)2 0, clearly true with equality if and only if
x1 = x2 or a = b. The conclusion follows. Note that, since x1 = x2 = . . . = xn is necessary
and sufficient for equality in (2.19), then equality holds in the proposed inequality if and
only if x1 = x2 = = xn .
()
Second solution
Note that
(ax1 + bx2 )(ax2 + bx1 ) = (a2 + b2 )x1 x2 + ab(x21 + x22 )

Hence

(x2 + x2 )(a + b)2

x31
2x31

(ax1 + bx2 )(ax2 + bx1 ) (a + b)2 (x21 + x22 )

so we need to prove that

2x31
2x32
2x3n
x1 + + xn
+
+

+
x21 + x22 x22 + x23
x2n + x21
We suppose the contrary, that is to say, there exist x1, ..., xn positive real num- bers
such that
2x32
2x31
2x3n
< x1 + + xn
+
+

+
x21 + x22 x22 + x23
x2n + x21
by symmetry we have the following inequality
2x33
2x31
2x32
< x1 + + xn
+
+

+
x21 + x22 x22 + x23
x2n + x21
summing up we obtain
x31 + x32 x32 + x33
x3n + x31
< x1 + + xn
+
+

+
x21 + x22 x22 + x23
x2n + x21
contradiction since

x3 + y 3 x + y

x2 + y 2
2

So the original inequality is true, with equality if and only if x1 = x2 = . . . = xn .


Example 2.92. Let a and b be positive real numbers such that

Prove that a + b 2.

1
1
a 2b and 2a b
a
b

Solution
We start by squaring both inequalities,
a2 4ab + 4b2

1
1
and 4a2 4ab + b2
a
b

()

79

2.2. Algebraic problems

Next, we eliminate the fractions in each inequality,


a3 4a2 b + 4b2 a 1 and 4a2 b 4ab2 + b3 1.
Now add the two inequalities together,
a3 + b3 2.
By multiplying by 4 and applying H
olders Inequality, we have that,
8 (1 + 1)(1 + 1)(a3 + b3 ) (a + b)3

And finally, 2 a + b
Which is what we wanted to prove, so were done.

()

Example 2.93. Let a, b, c be real numbers such that a > b > c. Prove that for each real
number x the following inequality holds
1
(x a)4 (b c) = (a b)(b c)(a c)[(a b)2 + (b c)2 + (c a)2 ].
6
cyc

Solution
Note that the coefficients of x4 and x3 vanish. Therefore the left-hand side equals
6x2 a2 (b c) 4xa3 (b c) + a4 (b c).
cyc

Note that each of the coefficients of the polynomial on the left-hand side vanish when
a = b, b = c or c = a. Therefore (a b)(b c)(a c) divides each of them. It is easy to show
that
a2 (b c) = (a b)(b c)(a c),
cyc

a3 (b c) = (a b)(b c)(a c)(a + b + c),


cyc

a4 (b c) = (a b)(b c)(a c)(a2 + b2 + c2 + ab + bc + ca).


cyc

It results that the inequality is equivalent to

1
6x2 4x(a + b + c) + (a2 + b2 + c2 + ab + bc + ca) = (a2 + b2 + c2 ab bc ca),
3

or equivalently,

2
a+b+c 2
2
6x 4x(a + b + c) + (a + b + c) = 6 (x
) 0
3
3
2

and were done.

()

80

Chapter 2. Examples for practice

Example 2.94. Solve in nonzero real numbers the system of equations

First solution
Note first that

x4 y 4

x4 + 14x2 y 2 + y 4

121x 122y
4xy
122x + 121y
=
x2 + y 2

x4 + 14x2 y 2 + y 4 = 4(x2 + y 2 )2 3(x2 y 2 )2 = s4 s2 d2 + d4 ,

where we have defined x + y = s and x y = d, while


x4 y 4 = sd(x2 + y 2 ) =

(s2 + d2 )
sd(s2 + d2 )
, 4xy = s2 d2 , x2 + y 2 =

2
2

Therefore, the system may be rewritten as

We can then obtain

= 243d s,
sd(s2 + d2 )(s2 d2 )
4 2 2
4
2
2

(s s d + d )(s + d ) = 243s + d.

(243d s)(243s + d) = sd(s2 + d2 )(s2 d2 )(243s + d) = (s4 s2 d2 + d4 )(s2 + d2 )(243d s).

Since s2 + d2 > 0 (otherwise x = y = 0, in contradiction with the problem statement), it


follows that
243s2 d(s2 d2 ) + sd2 (s2 d2 ) = 243d(s4 s2 d2 + d4 ) s(s4 s2 d2 + d4 ),

which after simplification yields s5 = 243d5 , or s = 3d. Substitution in both equations


yields d6 = d, or since x y (if x = y 0 the LHS of the first equation would be zero, but
the RHS would not), we find that d5 = 1, ie s = 3 and d = 1 for x = 2, y = 1. These values
can be clearly shown to satisfy the system by plugging them into the given equations, and
no other solutions exist.
()
Second solution
Since
(x4 + 14x2 y 2 + y 4 ) (x2 + y 2 ) = 122x + 121y, 4xy (x4 y 4) = 121x 122y

and x, y 0 then

(x4 + 14x2 y 2 + y 4 ) (x2 + y 2 ) (xy)4xy (x4 y 4 ) (x+y) = (122x+121y)(xy)(121x122y)(x+y). ()

We have
() (x2 + y 2) (x4 + 14x2 y 2 + y 4) (x y) 4xy (x2 y 2 ) (x + y) = x2 + y 2

81

2.2. Algebraic problems

(x4 + 14x2 y 2 + y 4 ) (x y) 4xy (x2 y 2 ) (x + y) = 1


(x y)5 = 1 x y = 1.

Let t = x + y then

x2 y 2 = t, x2 + y 2 =
and the equation x4 y 4 =

Hence,

t1
t1
t2 + 1
, 4xy = t2 1, y =
, 121x 122y = 121
2
2
2

121x 122y
becomes
4xy

t (t4 1)
t1
= 121
t(t4 1) + t 1 = 242 t5 = 243 t = 3.
2
2

x y

x + y

=1
x = 2, y = 1.
=3

Example 2.95. Find all triples (x, y, z) of real numbers such that
First solution
We can write
hence
It follows

()

x2 + y 2 + z 2 + 1 = xy + yz + zx+ x 2y + z .

x2 + y 2 + z 2 + 1 = xy + yz + zx + x y + z y,

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

(x y)2 + (y z)2 + (z x)2 + 2 2x y + 2y z.

The last relation is equivalent to

(x y 1)2 + (y z 1)2 + (z x)2 0.

We get x y = 1, y z = 1 and x = z.
The desired triples (x, y, z) are (a, a 1, a), (a, a + 1, a), where a R.

()

Example 2.96. Let a, b, c be real numbers such that a < 3 and all zeros of the polynomial
x3 + ax2 + bx + c are negative real numbers. Prove that b + c 4.
First solution
Since all zeros of P (x) are negative real numbers, it follows that
P (x) = x3 + ax2 + bx + c = (x + )(x + )(x + )

for some positive real numbers , , , and

a = + + , b = + + , c = .

82

Chapter 2. Examples for practice


By AM-GM inequality
1

b 2
1
a
( ) c3
3
3
hence, a < 3 implies that b < 3 and c < 1, that is b + c < 4.

()

Second solution
Let p, q, r the absolute values of the zeroes of the polynomial. Vietes formula yields
a = p + q + r, b = pq + qr + pr, c = pqr

a < 3 implies by the AGM 3 > p + q + r = 3(pqr) 3 and then pqr < 1. Moreover
1

pq + qr + pr = (p + q + r) 3 < 3.
2

If b + c = 4 we would have 4 = pq + qr + pr + pqr < 3 + pqr < 4 contradiction.

()

Third solution
Since all zeros of the polynomial p(x) = x3 + ax2 + bx + c are negative real numbers,
then by Rolls theorem, (or simply by multiplicity for multiple roots,) we see that p (x) =
3x2 + 2ax + b, has also negative real zeros. In particular, its discriminant must be positive,
hence 3b a2 < 9, or b < 3.
On the other hand, if , and are the negative real zeros of p, then c = and
a = + + .
a 3
Thus, by the AM-GM inequality we see that c ( ) < 1. From c < 1 and b < 3 we
3
conclude that b + c < 4.
()
Example 2.97. Let a, b, c be positive real numbers. Prove that

1
1
1 1 1 1 2
1
+
+

( + + ) .
2a2 + bc 2b2 + ca 2c2 + ab 9 a b c

Solution
By the AM-GM Inequality, we have

This implies that

1
1
1
=

3
2a2 + bc a2 + a2 + bc 3a abc

1
1 1 1
1
1
1
( + + ).
+ 2
+ 2

3
+ bc 2b + ca 2c + ab 3 abc a b c
Thus, it is suficient to prove that
2a2

1 1 1 1 2
1
1 1 1
+
+
)

( + + ) .

(
3
9 a b c
3 abc a b c
But this last inequality is equivalent to
1 1 1
3

( + + )
3
a b c
abc

which follows directly from the AM-GM Inequality, so we are done.

()

83

2.2. Algebraic problems

Example 2.98. Let a, b, c be positive real numbers such that


(a2 + 1)(b2 + 1)(c2 + 1) (

1
a2 b2 c2

+ 1) = 2011.

Find the greatest possible value of max{a(b + c), b(c + a), c(a + b)}.
Solution
Assume that when a(b + c) reaches its maximum possible value (which by symmetry
in the variables will also be the maximum that we are looking for), the values of p = abc
and s = b + c are known. It remains thus to find the maximum value of a under these two
constraints; it would seem at first sight that bc would need to be minimized, since p = abc
is known, and a needs to be maximized; let us see that it is otherwise. Note first that
(b2 + 1)(c2 + 1) = s2 + b2 c2 2bc + 1 = s2 + (bc 1)2 ,
a2 + 1 =

2011
p2

s2 + (bc 1)2 p + 1

Thus the maximum of the RHS, hence of a2 and of a, is obtained when bc = 1, since
p, s are known and fixed. We may thus perform substitution bc = 1, and the relation given
in the problem statement becomes
(a2 + 1)2 s2 = 2011a2 .

Under this constraint, we need to maximize as, or equivalently, a2 s2 = x. Now, the previous
relation between a and s may be rewritten as
x2 (2011 2s2 )x + s4 = 0,

wherefrom the largest of both (clearly positive real) roots is

2011 2s2 + (2011 2s2 )2 4s4 2011 2s2 + 2011 2011 4s2
=

x=
2
2

It finally follows that, since bc = 1, and s needs to be minimized so that x = a2 s2 is


maximized, and s2 4bc = 4 by the AM-GM inequality, then the maximum is achieved
when b = c = 1, and the square of this maximum is

2003 + 2011 1995


,
x=
2
and finally,

2011 1995
2001 + 1995
=
max{a(b + c), b(c + a), c(a + b)} =
2
2

2001 + 1995
with equality if two of a, b, c are 1, and the other one is
. These values can
4
be easily shown to satisfy the relation given in the problem statement.
()
2003 +

84

Chapter 2. Examples for practice

Example 2.99. Solve the system of equations

1
1
1 7
x = y = z =
y
z
x 4
Solution

1 7 2
We have x = f (y), y = f (z), z = f (x) where f (x) = ( + ) is a strictly decreasing
x 4
function over the positive real numbers. Hence we have f (f (f (x))) = x and the same for
y and z.
Suppose that f (x) > x. Then we have f (f (x)) < f (x) and x = f (f (f (x))) > f (f (x)).
And, applying f again, we have f (x) < x, contradiction.
In a similar way we discard the case f (x) < x and we have f (x) = x, and the same for
y and z.
Solving f (x) = x we get only a real solution x = 4, so the solution of our system is
x = y = z = 4.
()

Example 2.100. Let a, b, c be positive real numbers such that a + b + c = 3. Prove


that
8(a2 + b2 + c2 ) 3(a + b)(b + c)(c + a).
Solution
By AM-GM inequality we have

2a + 2b + 2c 3 8
3(a + b)(b + c)(c + a) 3 (
) = (a + b + c)3 .
3
9

(2.20)

3
1
2
On the other hand, taking xi = a 3 , yi = b 3 , p = , q = 3, in the H
olders inequality
2

we get

yields

Since

1
p

p
q
xi yi ( xi ) ( yi )
n

k=1

k=1

1
q

k=1

1
23
2
1 2
1 2
1
a + b + c = a 3 a 3 + b 3 b 3 + c 3 c 3 ( a + b + c) (a2 + b2 + c2 ) 3 ,

2
(a + b + c)3 ( a + b + c) (a2 + b2 + c2 )


a + b + c = 3 from (2.21) it follows that

(a + b + c)3 9 (a2 + b2 + c2 )

(2.21)

(2.22)

From (2.20) and (2.22) we get the desired inequality. The equality holds if a = b = c = 1.
()
Example 2.101. Let a, b, c be positive real numbers such that a + b + c = 1. Prove that

3
(aa + ba + ca )ab + bb + cb (ac + bc + cc ) ( 3 a + b + 3 c) .

85

2.2. Algebraic problems

First solution
By H
olders Inequality
(aa + ba + ca ) 3 (ab + bb + cb) 3 (ac + bc + cc) 3 = a
1

a+b+c
3

+b

a+b+c
3

+c

a+b+c
3

= a
cyc

and by cubing both sides we get the desired result.

()

Second solution
By the symmetry of both the L.H.S. and R.H.S. we can assume a b c so that
(a, b, c) majorizes (1/3, 1/3, 1/3) :
1
2
a , a + b , a + b + c = 1.
3
3

The inequality is

3
ln (aa + ba + ca ) + ln (ab + bb + cb ) + ln (ac + bc + cc ) = 3 ln ( 3 a + b + 3 c)

and if we show that f (x) > 0, f (x) = ln (ax + bx + cx ), it follows by the majorization
inequality for convex functions that

which is our inequality.

1
1
1
f (a) + f (b) + f (c) = f ( ) + f ( ) + f ( )
3
3
3

f (x) = (ab)x ( ln a ln b)2 + (ac)x ( ln a ln c)2 + (bc)x ( ln b ln c)2 (ax + bx + cx)2 > 0.
The proof is completed.

()

Example 2.102. Let x1 , x2 , ..., xn be positive real numbers. Prove that

n1
x21 + x22 + + x2n
x1 + x2 + + xn n
(
) ( n x1 x2 xn )

n
n
Solution
By homogeneity, we may suppose x1 x2 xn = 1 and then prove
(x1 + x2 + + xn )2n = n2n1 S

where S = x21 + x22 + + x2n . Using AM-GM,

(x1 + x2 + + xn )2 = S + 2 xi xj
1i<jn

= S +2

hence it suffices to show that

2
n(n 1)
((x1 x2 xn )n1 ) n(n1) = S + n(n 1),
2

(S + n(n 1))n n2n1 S.

86

Chapter 2. Examples for practice


Now, by AM-GM again,
S + n(n 1) = S + n + n + + n n (Snn1 ) n
1

and so

(S + n(n 1))n nn Snn1 = n2n1 S,

as desired.

()

Example 2.103. Let a, b, c be distinct real numbers. Prove that


(

2
2
2
b
c
a
+ 1) + (
+ 1) + (
+ 1) 5.
ab
bc
ca

Solution
By expanding, the inequality becomes

On the other hand,

a 2
a
) + 2
2.
(
cyc a b
cyc a b

a(b c)(c a) + b(c a)(a b) + c(a b)(b c)


a
=
(a b)(b c)(c a)
cyc a b

It follows that

ab(c a) + bc(a b) + ca(b c)


b
a
+ 1 = (
)(
) + 1.
(a b)(b c)(c a)
bc
cyc a b

a 2
a
a
b
c 2
) + 2
=2+(
+
+
) 2
(
ab bc ca
cyc a b
cyc a b

as desired.
Equality occurs if and only if

2.2.3

a
b
c
+
+
= 0.
ab bc ca

()

Undergraduate problems

Example 2.104. Let x and y be positive real numbers. Prove that


xy y x (

x + y x+y
) .
2

First solution
From weighted AM-GM inequality we conclude that
xy y x (

xy + yx x+y
2xy x+y
x + y x+y
) =(
) (
) ,
x+y
x+y
2

where the last inequality again follows from AM-GM

2xy
x+y
x+y

xy

x+y
2
2

()

87

2.2. Algebraic problems

Second solution
Let us take a more indepth look at this inequality. Firstly, we will prove the stronger
inequality
x+y
xy y x (xy) 2 .
Indeed,

Since

xy y x (xy)

x+y
2

x xy
x2y y 2x xx+y y x+y xy y x xx y y ( ) 1.
y

(xy) 2
1

x + y x+y
) .
2
Third solution
By weighted AM-GM,

we obtain xy y x (

x+y
x+y
x + y x+y
(xy) 2 (
) ,
2
2

()

x
y
y
x
2xy
xx + y y x + y x
+y
=
x+y
x+y x+y
then original inequality immediately follows from the obtained stronger inequality

and the inequality

xy y x (

2xy x+y
)
x+y

2xy
x+y

. Thus we obtained the chain of inequalities


x+y
2
xy y x (

x+y
2xy x+y
x + y x+y
) (xy) 2 (
) .
x+y
2

Remark. Normalization of the inequality

by the condition x + y = 1 yields


xy y x 2xy

xy y x (

2xy x+y
)
x+y

1
1
xx y y xx (1 x)1x , x (0, 1)
2
2

which can proved by employing calculus. Let

then h (x) = ln

x
and
1x

h(x) = x ln x + (1 x) ln (1 x)

1
1
1
h (x) < 0 x (0, ) , h (x) > 0 x ( , 1) , h (x) = 0 x =
2
2
2

()

88

Chapter 2. Examples for practice


1
Thus min h(x) = h ( ) , i.e.,
x(0,1)
2

1
1
h(x) h ( ) xx (1 x)1x , x (0, 1).
2
2

()

Example 2.105. Let n be a given positive integer and let ak = 2 cos nk , k = 0, 1, ..., n 1.
2
Prove that
n1
(1)n1

(1 ak ) =
1 + a0
k=0
Solution
Note that

1 ak = (1 2 cos
=

1 + 2 cos

1 2 (1 + cos
1 + 2 cos

1 4 cos2

=
)
pi
2nk 1 + 2 cos
1 + 2 cos nk
2nk
2

2nk1

2nk

2nk

1 + 2 cos

2nk

2nk1

1 + 2 cos nk
2

1 + ak+1

1 + ak

Thus, the indicated product telescopes, and


(1 ak ) = (1)n

n1
k=0

2.2.4

1 + 2 cos (1)n+1 (1)n1


=
=

1 + a0
1 + a0
1 + a0

()

Olympiad problems

Example 2.106. Let a1 , a2 , , an be real numbers such that a1 + a2 + + an = 0. Prove


that for a 0,
a + a21 + a22 + + a2n m(a1 + a2 + + an ),

a
an
where m = 2
, if n is even, and m = 2
, if n is odd.
2
n
n 1
Solution
The first step is to dismiss the absolute value sign. We separate the sequence a1 , a2 , , an
into a sequence of non-negative real numbers x1 , x2 , , xk and a sequence of negative real
numbers y1 , y2 , , ynk . Denote zj = yj , j {1, 2, . . . , n k}, we have to prove that
x2i + zj2 + a m xi + m zj

Denote x =

nk

nk

i=1

j=1

i=1

j=1

1 nk
1 k
zi and z =
zj . Clearly,
k i=1
n k j=1

89

2.2. Algebraic problems

x2i
i=1

x1 + x2 + + xk 2
k(
) = kx2 .
k

zj2 (n k) (

nk
j=1

z1 + z2 + + znk 2
) = (n k)z 2 .
nk

After all, we would like to prove that

kx2 + (n k)z 2 + a m(kx + (n k)z).

From the the condition kx = (n k)z, as a1 + a2 + + an = 0, the above inequality


becomes
kx2 (1 +

k
) + a 2mkx.
nk

Using the AM-GM inequality we get

LHS 2 ax k (1 +

2k anx
k
)=

nk
k(n k)

n2
.
4
2
n 1
If n is odd, the maximum of k(n k) is
. The conclusion follows.
4
If n is even, the maximum of k(n k) is

()

Example 2.107. Let n be a positive integer. Prove that


2 2
1
(n 1) .
=
k
2
3
k=1 cos 2n
n

First solution
From De Moivres fomula and calculus one can show:
sin 2n
23 n(n2 1)
= (1)n+1 [2n cos
cos3 + ]
sin
3!

Then we can Define a function

f (x) = (1)n+1 2nx [1

22 (n2 1) 2 24 (n2 1) (n2 22 ) 4


x +
x + ]
3!
5!

k
In fact, f (x) is a polynomial of degree 2n 1 with the roots x = cos , 1 k 2n 1.
2n
1
1
1
Consider f ( ) and substitute y = 2 , then
x x
x
xf (x) = g(y) = y n1

22 (n2 1) n3 24 (n2 1) (n2 22 ) n5


y +
y +
3!
5!

90

Chapter 2. Examples for practice

k
This is a polynomial of degree n 1 in y whose roots are y = sec 2
with 1 k n 1.
2n
Using Vietes theorem we find that the sum of coefficients is
k
22 (n2 1) n1
2 2
(n 1) =
= sec 2 ,
3
3!
2n
i=1

and we are done.


()
Second solution
Note that for any polynomial P (x) = a0 xn +a1 xn1 ++an1 x+an , a0 0 with non-zero
roots x1 , x2 , . . . , xn we have
n
P (0)
1
=

(2.23)

P (0)
i=1 xi
Proof.
Let P (x) = a0 (x x1 )(x x2 )...(x xn ), then

n
n
P (x)
P (x)
1

=
(
ln
(x

x
))
=
(
ln
(x

x
))
=
(
ln
)
=
,

i
i
a0
P (x)
i=1
i=1
i=1 x xi
n

and plugging x = 0 the conclusion follows.

Tn+1
(x) sin(n + 1)
Let Un (x) =
=
be the Chebishev Polynomial of the
n+1
sin
Second Kind. Then Un (x) satisfies to recurrence

Un+1 (x) = 2xUn (x) Un1 (x), n N and U0 (x) = 1, U1 (x) = 2x.
sin n
k
Because
= 0 if and only if =
, n Z, we get Un1 (x) = 0 if and
sin
n
k
only if x = cos , k = 1, 2, . . . , n 1 and
n
2
(n 1)

),
Un1 (x) = 2n1 (x cos ) (x cos ) (x cos
n
n
n

as the coefficient of xn in Un (x) is 2n .


In particular,

U2n1 (x) = 22n1 (x cos


2n1
k=1

= 22n1 (x cos

k
)
2n

k n1
(2n k)
n n1
) (x cos ) (x cos
)
2n k=1
2n k=1
2n

= 22n1 x (x2 cos2


n1
k=1

U2n1 ( x)
Let Pn (x) =

, then
2 x

k
).
2n

Pn (x) = 4n1 x (x cos2


n1
k=1

k
).
2n

()

91

2.2. Algebraic problems

Note that U2n1 (x) can be Defined by the recurrence

U2n+1 (x) = 2 (2x2 1) U2n1 (x) U2n3 (x), with U1 (x) = 0, U1 (x) = 2x.

Since U2n1 (x) is divisible by 2x, then polynomial Pn (x) satisfy the recurrence
Pn+1 (x) = 2(2x 1)Pn (x) Pn1 (x), n N with P0 (x) = 0, P1 (x) = 1.

(2.24)

Thus applying (2.23) to the polynomial Pn (x) we obtain


n1

Pn (0)

Pn (0)

k
2n
In particularly, from (2.24) follows recurrence
k=1

cos2

Pn+1 (0) + 2Pn (0) + Pn1 (0) = 0, n N with P0 (0) = 0, P1 (0) = 1.


Pn (0)
Let bn =
, then (2.25) can be rewritten as
(1)n

(2.25)

bn+1 2bn + bn1 = 0, n N, b0 = 0, b1 = 1.

Since bn+1 bn = bn bn1 we have bn bn1 = 1 and nk=1 (bk bk1 ) = n.


Therefore bn b0 = n, implying bn = n.
From the other hand,

then

(x), with P0 (x) = 0, P1 (x) = 0,


(x) = 2(2x 1)Pn (x) + 4Pn (x) Pn1
Pn+1

(0) = 4Pn (0), with P0 (0) = 0, P1 (0) = 0.


(0) + 2Pn (0) + Pn1
Pn+1
P (0)
Pn (0)
Let an = n n then
= bn = n and (2.26) can be rewritten as
(1)
(1)n+1

an+1 2an + an1 = 4n, n N with a0 = a1 = 0.

2n (n2 1)
is particular solution of nonhomogeneous recurrence
3
2
2n (n 1)
+ n + , where = = 0, because a0 = a1 = 0.
(2.27), then an =
3
2n (n2 1)
and
Thus an =
3
Pn (0)
n1
1
an 2 2
(1)n
=
=
= (n 1)

k
Pn (0)
bn 3
k=1 cos2
2n
(1)n+1

(2.26)

(2.27)

Since sequence

()

Example 2.108. Let P (n) be the following statement: for all positive real numbers
x1 , x2 , . . . , xn such that x1 + x2 + + xn = n,
x3
x1
n
x2

+
++

x1 + 2x3
x2 + 2x4
xn + 2x2
3
Prove that P (n) is true for n 4 and false for n 9.

92

Chapter 2. Examples for practice

Solution
olders inequality,
Let S(x1 , x2 , . . . , xn ) be the left hand side of the inequality. Using H
we obtain
S 2 (x2 (x1 + 2x3 ) + + x1 (xn + 2x2 )) (x1 + x2 + + xn )3 = n3 .

On the other hand, we have

x2 (x1 + 2x3 ) + + x1 (xn + 2x2 ) = 3(x1 x2 + x2 x3 + + xn x1 ).

Using the fact that

x1 x2 + x2 x3 + + xn x1 n

whenever x1 + x2 + + xn = n and n 4. The last fact follows from the fact that
ab + bc + ca
and

(a + b + c)2
3

ab + bc + cd + da = (a + c)(b + d)

(a + b + c + d)2
4

The conclusion follows easily for n 4. Chosing x1 , x2 , x3 , x4 close to

n
and the other
4
n
variables equal and close to 0, one easily obtains that the expression is smaller than
3
for n 9. The conclusion follows.
()
Example 2.109.
Let a, b, c, d be real numbers such that

Prove that

(a2 + 1)(b2 + 1)(c2 + 1)(d2 + 1) = 16.

3 ab + bc + cd + da + ac + bd abcd 5.

Solution
Consider the complex number Z = (1 + ia)(1 + ib)(1 + ic)(1 + id). An easy calculation
yields:
Re(Z) = 1 (ab + bc + cd + da + ac + bd) + abcd and Z2 = (a2 + 1)(b2 + 1)(c2 + 1)(d2 + 1).
Now, the hypothesis gives Z = 4 so that the inequality Re(Z) Z writes as

that is,
as required.

(ab + bc + cd + da + ac + bd abcd) 1 4
3 ab + bc + cd + da + ac + bd abcd 5,

()

93

2.2. Algebraic problems

Example 2.110. Let a, b, c be positive real numbers such that abc = 1. Prove that
1

a5 (b + 2c)2

b5 (c + 2a)2

First solution
We will make use of the following lemma

c5 (a + 2b)2

Lemma 3. If x, y, z, a, b, c > 0 we have

x3 y 3 z 3 (x + y + z)3
+ +

a2 b2 c2 (a + b + c)2

We will also relax the condition abc = 1 to abc 1. Set


1
1
1
a = ,b = ,c =
x
y
z

Then, we have xyz 1 and the left hand side of the inequality is equal to
K = (xyz)2

x3

2
cyc (2y + z)

From the above lemma (and since xyz 1)we have

(x + y + z)3
x + y + z 3 3 xyz 1
K
=


9(x + y + z)2
9
9
3

()

Second solution
Since

1 3
( )
1
1
a
=
2 =
5
2
a (b + 2c)
2
2 2
1
1
a5 b2 c2 ( + )
( + )
c b
c b
1 1 1
then by replacing ( , , ) in original inequality with (a, b, c) we obtain the equivalent
a b c
inequality
a
1

3
cyc (2b + c)
with abc = 1. Note that

2
2b + c
a2
a
, hence
2b + c 3
9

a3
a
a
a2
2
2b + c
2

(
a

)
=

2
9
3 cyc 2b + c cyc 9
cyc (2b + c)
cyc (2b + c) 3

a a+b+c 2
2b + c
2
2
) =
(2b + c)
( a
3 cyc 3
9
3
9 cyc
cyc 9

3
a + b + c 3 abc 1
=

=
9
9
3

()

94

Chapter 2. Examples for practice

Example 2.111. Find all triples (x, y, z) of integers such that x2 y + y 2z + z 2 x = 20102
and xy 2 + yz 2 + zx2 = 2010.
Solution
Substract both relations and get (x y)(y z)(x z) = 2010 2011. If a = x y and
b = y z, we have
a b (a + b) = 2 3 5 67 2011,

primes decomposition. If P = a b, S = a + b, we get P S > 0, so both P, S are positive or


both are negative. In first case a = 2010, b = 1 and in second a = 2011, b = 1 and there
permutations.
Case 1: a = x y = 2010 = p, b = y z = 1.
We can write: y = z + 1, x = z + p + 1. From first relation we have
3z 3 + 3z 2 (p + 2) + z(p + 2)2 + 2p + 1 = 0.

(1)

From here, z(2p + 1) = 4021, prime number, so z {1, 1, 4021, 4021}. None of these
verifies (1)

Case 2: a = x y = 2011 = p 1, b = y z = 1.
We can write: y = z+1, x = zp. From first relation we have 3z 3 +3z 2 (1p)+z(p1)2 = 0.
We get z = 0 or 3z 2 + 3z(1 p) + (p 1)2 = 0, which has negative discriminant. From
z = 0, we have x = p = 2010, y = 1.
Finally all triples (x, y, z) are (2010, 1, 0), (1, 0, 2010), (0, 2010, 1).
()

Example 2.112. Let a, b, c be positive real numbers such that abc = 1. Prove that
a+b+1
b+c+1
c+a+1
(a + 1)(b + 1)(c + 1) + 1
+
+

2
3
2
3
2
3
a+b +c
b+c +a
c+a +b
a+b+c

First solution
From the Cauchy-Schwartz inequality and because abc = 1 we obtain that
(a + b2 + c3 )(a + 1 + ab) (a + b + c)2

So we get that

1 + a + ab
1

2
3
a+b +c
(a + b + c)2

a+b+1
(a + b + 1)(1 + a + ab)

2
3
a+b +c
(a + b + c)2

In a similar way we obtain the following inequalities

and

(b + c + 1)(1 + b + bc)
b+c+1

b + c2 + a3
(a + b + c)2

(c + a + 1)(1 + c + ca)
c+a+1

c + a2 + b3
(a + b + c)2

95

2.2. Algebraic problems

So it is enough only to prove that

(a + b + 1)(1 + a + ab) + (b + c + 1)(1 + b + bc) + (c + a + 1)(1 + c + ca) (a + 1)(b + 1)(c + 1) + 1

(a + b + c)2
a+b+c
The last one is equivalent to

(a + b + 1)(1 + a + ab) (a + b + c)(a + 1)(b + 1)(c + 1) + a + b + c, or


cyc

3 a+3+2 a2 +2 ab+ ab(a+b) abc a+3abc+ ab(a+b)+ a2 +2 ab+2 a,


cyc

cyc

cyc

cyc

cyc

cyc

cyc

cyc

cyc

which is true.

()

Second solution
By Cauchy-Schwarz Inequality, we have:

Similarly, we obtain
LHS
=
=

(a + b2 + c3 )(a + 1 + ab) (a + b + c)2

(a + b + 1)(a + 1 + ab) + (b + c + 1)(b + 1 + bc) + (c + a + 1)(c + 1 + ca)


(a + b + c)2

(a + b + c)(ab + bc + ca + a + b + c + 3)
(a + b + c)2
(a + 1)(b + 1)(c + 1) + 1
= RHS.
a+b+c

Equality holds when a = b = c = 1.

()
()

Example 2.113. Prove that for all real numbers x, y, z the following inequality holds
(x2 + xy + y 2)(y 2 + yz + z 2 )(z 2 + zx + x2 ) 3(x2 y + y 2 z + z 2 x)(xy 2 + yz 2 + zx2 ).

Solution
The inequality in question immediately follows from the identity

T = (x2 + xy + y 2)(y 2 + yz + z 2 )(z 2 + zx + x2 ) = 3(x2 y + y 2 z + z 2 x)(xy 2 + yz 2 + zx2 )


+ ((x y)(y z)(z x))2 .

What remains is to prove this identity. Of course, we can prove it by expanding, but
here is a more conceptual proof:

1+i 3
2
2
2
2
2
2
Denote a = x y + y z + z x and b = xy + yz + zx . We work in C. Let c =
.
2
Then, c3 = 1 and thus

3
2
2
2
2
2
2
2

(x + cy)(y + cz)(z + cx) = c + 1 xyz + c x y + y z + z x + c xy + yz + zx




=1+1=0

=a
=b

96

Chapter 2. Examples for practice


= ca + c2 b = c(a + cb).
The same computation with c replaced by

of c3 = 1) proves

1
1 3
everywhere (and using ( ) = 1 instead
c
c

1
1
1
1
1
(x + y) (y + z) (z + x) = (a + b) .
c
c
c
c
c
But any two complex numbers u and v satisfy
1
(u + cv) (u + v) = u2 + uv + v 2
c

(i)

1
1
1
(since (u + cv) (u + v) = u2 + (c + ) uv + v 2 and c + = 1 as we can easily see).
c
c
c
Hence, because T = (y 2 + yz + z 2 ) (z 2 + zx + x2 ) (x2 + xy + y 2 ), we have
1
1
1
T = (y + cz) (y + z) (z + cx) (z + x) (x + cy) (x + y)
c
c
c
1
1
1
= (x + cy)(y + cz)(z + cx) (x + y) (y + z) (z + x)
c
c
c
1
1
1
= c(a + cb) (a + b) = (a + cb) (a + b)
c
c
c
= a2 + ab + b2

(by (i))

(by (i))

= 3ab + (b a)2 = 3 (x2 y + y 2z + z 2 x) (xy 2 + yz 2 + zx2 ) + ((x y)(y z)(z x))2 .

Since a = x2 y + y 2z + z 2 x, b = xy 2 + yz 2 + zx2 , and a quick computation shows that


b a = xy 2 + yz 2 + zx2 x2 y + y 2 z + z 2 x = (x y)(y z)(z x)

and we are done.

()

Example 2.114. Let a, b, c and x, y, z be positive real numbers such that


3
3
a + b + 3 c = 3 m and x + y + z = n.
Prove that

a b c m
+ +
x y z n

Solution

A
B
C
3
Denote A = 3 a, B = b, C = 3 c, A+B +C = 3 m, and define u = , v = , w = .
y
x
z

A B C
Clearly n = + + , or the result that we need to prove is equivalent to showing
u v w
that, for all positive reals A, B, C, u, v, w, the following inequality holds:

A+B+C
Au2 + Bv 2 + Cw 2

A B C
A+B+C
+ +
u v w

97

2.2. Algebraic problems

This is the inequality between weighted quadratic and harmonic means of u, v, w and
respective weights A, B, C, which is well known to be always true, with equality holding
iff u = v = w, ie, if and only if

3
3
a
b 3c
= =
()
y
x
z
Example 2.115. Let a, b, c be positive real numbers. Prove that
8abc
a+b+c

+
4
3
(a + b)(b + c)(c + a)
abc

Solution
Defining a = x3 , b = y 3, c = z 3 we come to
4
Now

(x3 + y 3 )(y 3 + z 3 )(z 3 + x3 )

so we prove

Defining S =

x3 + y 3 + z 3
8(xyz)3
+ 3

xyz
(x + y 3 )(y 3 + z3)(z 3 + x3 )

(2(x3 + y 3 + z 3 ))3
27

x3 + y 3 + z 3
27(xyz)3
+ 3
4.
xyz
(x + y 3 + z 3 )3

(x3 + y 3 + z 3 )
we have
xyz

(2.28) S +

and we are done.

27
4 S 4 4S 3 + 27 0 (S 3)2 (S 2 + 2S + 3) 0
S3

Example 2.116. Let a, b, c be positive real numbers. Prove that


a2 b2 c2
2(a + b + c)3
+ + +a+b+c

b
c
a
3(ab + bc + ca)
Solution Denote LHS of the given inequality by T , we have
3T ab = 3 ( ab) (
cyc

cyc

= 3 ( a3 +
cyc

a2 b2 c2
+ + + a + b + c)
b
c
a

ab3 ca3 bc3


+
+
+ a2 c + c2 b + b2 a + a2 b + 3abc)
c
b
a
cyc

= 3 a3 + 3 (
cyc

ab3 ca3 bc3


+
+
) + 3 (a2 c + c2 b + b2 a) + 3 a2 b + 9abc
c
b
a
cyc

Thus
3

3T ab 2 ( a) = a3 3abc + 3 (
cyc

(2.28)

cyc

cyc

ab3 ca3 bc3


+
+
) 3 (a2 c + c2 b + b2 a)
c
b
a

()

98

Chapter 2. Examples for practice


Now by the AM-GM inequality we have

ab3 ca3
ab3 ca3
+
2

= 2a2 b
c
b
c
b

etc,. By adding these inequalities we have


ab3 ca3 bc3
+
+
a2 c + c2 b + b2 a.
c
b
a

And it remains to prove that a3 3abc which is known to be true.

()

cyc

Example 2.117. Let a, b, c, d be positive real numbers such that a2 + b2 + c2 + d2 = 1. Prove


that


1 a + 1 b + 1 c + 1 d a + b + c + d.
Solution

The inequality takes the form ( 1 a a) 0. We have


cyc

1 2a
1
1aa
1a a=
=
(1 2a)
1a+ a
1a+ a
2 cyc

where inthe last step we used the estimate (x + y)2 2(x2 + y 2 ) applied to the numbers

1 a, a. Hence, it suffices to prove that


(1 2a) 0 4 2(a + b + c + d) 0 4(a2 + b2 + c2 + d2 ) (a + b + c + d)2 ,
cyc

which is clearly true (for instance using Power Mean.)

()

Example 2.118. Let a, b, c, x, y, z 0. Prove that

(a2 + x2 )(b2 + y 2 )(c2 + z 2 ) (ayz + bzx + cxy xyz)2 .

First solution
The inequality is trivial if any of x, y, z equals 0. Suppose that xyz 0. Therefore,
dividing by (xyz)2 > 0 it follows that the inequality is equivalent to
a b c
(m2 + 1)(n2 + 1)(p2 + 1) (m + n + p 1)2 , where (m, n, p) = ( , , ) .
x y z

After expanding and rearranging some terms it follows that this inequality is equivalent
to
m2 n2 p2 + (m2 n2 + m + n) + (n2 p2 + n + p) + (p2 m2 + p + m) 2mn + 2np + 2pm.

From AM-GM it follows that m2 n2 + m + n 3mn 2mn, from where it is easy to


conclude the result.
()
Second solution

99

2.2. Algebraic problems

By expanding, we have
LHS = a2 b2 c2 + x2 b2 c2 + y 2 c2 a2 + z 2 a2 b2 + a2 y 2 z 2 + b2 z 2 x2 + c2 x2 y 2 + x2 y 2z 2 ,
RHS = x2 y 2 z 2 + 2xyz(abz + bcx + cay ayz bzx cxy).
The inequality becomes
a2 b2 c2 + x2 b2 c2 + y 2c2 a2 + z 2 a2 b2 + 2xyz(ayz + bzx + cxy) 2xyz(abz + bcx + cay).
By the AM-GM inequality, we have
x2 b2 c2 + xyz bzx + xyz cxy 3xyz xbc.
Adding two similar inequalities, we obtain
x2 b2 c2 + y 2 c2 a2 + z 2 a2 b2 + 2xyz(ayz + bzx + cxy) 2xyz(abz + bcx + cay),

and we are done. Equality holds if and only if a = b = c = x = y = z = 0.

Example 2.119. Let a, b, c be positive real numbers such that abc = 1. Prove that


3
3
a + b + 3 c 3 3(3 + a + b + c + ab + bc + ca).

()

First
solution

3
3
Set a = x, b = y, 3 c = z. We have then x, y, z > 0 and xyz = 1. What we have to
prove assumes the form
3 (3 + x3 + y 3 + z 3 + (xy)3 + (yz)3 + (zx)3 ) = (x + y + z)3 .

Now, recall Schurs inequality

A3 + B 3 + C 3 + 5ABC (A + B)(B + C)(C + A).

Setting A = xy, B = yz, C = zx, we find (xy)3 + (yz)3 + (zx)3 ) + 5 = (x + y)(y + z)(z + x),
since xyz = 1. Then it is enough to prove that
3(3 + x3 + y 3 + z 3 + (x + y)(y + z)(z + x) 5) x3 + y 3 + z 3 + 3(x + y)(y + z)(z + x),

or equivalently that x3 + y 3 + z 3 3, which is true by the AM-GM inequality.

()

Second solution
x
y
z
Setting a = , b = , c = . The inequality becomes
y
z
x

x
y
z
x y z y z x
3
+ 3
+ 3
3 3 (3 + + + + + + )
y
z
x
y z x x y z

By the Holder inequality, we have:

y
z
y
z
x
1 1 1
x
3
3 (3 + + + + + + ) = 3 3(x + y + z) ( + + )
y z x x y z
x y z

x 3 y 3 z
3
(
+
+
)
y
z
x

Hence we are done. Equality holds if and only if x = y = z or a = b = c = 1.

()

100

Chapter 2. Examples for practice

Example 2.120. Let a0 , a1 , ..., a6 be real numbers greater than 1. Prove that
a2 + 1
a2 + 1
a2 + 1
0
+ 1
++ 6
5.
a51 + a41 + 1
a52 + a42 + 1
a50 + a40 + 1

whenever

a3 + 1
a3 + 1
a3 + 1
+ 1
++ 6
9.
0
a51 + a41 + 1
a52 + a42 + 1
a50 + a40 + 1

Solution
It suffices to prove that

a3 + a20 + 2
a3 + a21 + 2
a3 + a26 + 2
0
+ 1
+ + 6
14.
a51 + a41 + 1
a52 + a42 + 1
a50 + a40 + 1

We observe that (a7 = a0 )

a3k + a2k + 2
=

5
ak+1 + a4k+1 + 1 k=1 a2k+1 + ak+1 + 1 a3k+1 ak+1 + 1
k=1
a3k + a2k + 2

which we rewrite as

k=1

namely

6
a3k a k +1
a2k + ak + 1

a2k+1 + ak+1 + 1 a3k+1 ak+1 + 1 k=1 a2k+1 + ak+1 + 1 a3k+1 ak+1 + 1

6
a3k a k +1
a2 + ak + 1
+ k
2l.7 = 14
2
ak+1 + ak+1 + 1 k=1 a3k+1 ak+1 + 1
k=1
6

the last inequality allowed by the AGM and we are done.

()

101

2.3. Geometric problems

2.3
2.3.1

Geometric problems
Junior problems

Example 2.121. Let ABC be a triangle with circumradius R. Prove that if the length of
one of the medians is equal to R, then the triangle is not acute. Characterize all triangles
for which the lengths of two medians are equal to R.
First solution
Let O be the circumcenter and M be the midpoint of the side BC. Without loss of
generality we have that a b c, we have

1 2
1 2
1 2
2b + 2c2 a2 ; mB =
2a + 2c2 b2 ; mC =
2b + 2a2 c2 ,
2
2
2
and we deduce that mA mB mC . On the other hand, if the triangle is acute angled,
then its circumcenter lies int the interior of the triangle. Note that mA > R, because
AOM is obtuse, and the equality does not occur. Thus triangle ABC is not acute
angled.
For the second part it is not difficult to see that if two medians in a triangle are equal,
then the triangle is isosceles, because
mA =

mB = mC

1 2
1 2
2a + 2c2 b2 ; mC =
2b + 2a2 c2 b = c.
2
2

Let the ABC be isosceles triangle with b = c. By the Law of Sines and the Law of
Cosines we have
b2 + c2 a2 2b2 a2
a2
, cos A =
=
sin A
2bc
2b2
and if mB = mC = R, we have
R=

1
m2B = R2 (2a2 + 2b2 b2 ) = R2 2a2 + b2 = 4R2
4

and finally
R2 =

a2
a2
=
=
2 sin2 A 4(1 cos 2A)

Finally, using (2.29) we get


2a2 + b2 =

yielding b = c =

(2.29)

a2
4b4
2
.

4R
=
4b2 a2
2b2 c2 2
)]
4 [1 (
2b2

4b4
7a2 b2 2a4 = 0 a2 (7b2 2a2 ) = 0,
4b2 a2

2
a, and we are done.
7
Second solution
Without loss of generality, let us assume that the length of the median from A equals
R. The square of the length of this median is given by

102

Chapter 2. Examples for practice

b2 + c2 a2 a2

=
+ bc cos A = R2 sin2 A + 4R2 sin B sin C cos A.
2
4
4
2
Equating this result to R and grouping terms in one side of the equality yields
cos A(cos A 4 sin B sin C) = 0.

One possible solution is that triangle ABC is right triangle at A, in which case the
midpoint of BC is also the circumcenter, and the median from A is a radius of the
circumcircle. Otherwise,
4 sin B sin C = cos(B + C) = cos B cos C + sin B sin C,
1
yielding tan Btan C =
3
Clearly, B and C cannot be simultaneously acute, and ABC is either rectangle or
obtuse.
If the lengths of two medians are equal to R, say ma = mb , then
b2 + c2 a2 c2 + a2 b2

=
,
2
4
2
4
yielding a = b, or ABC is isosceles at C. Since A = B, C is obtuse, and using the well
known identity
tan A + tan B + tan C = tan Atan Btan C,
1
tan A
we find 2tan A
=
and
3tan A
3
7
1
tan 2 A = tan 2 B = , tan 2 C =
7
9

2
tan

2
7
Using that sin2 =
, sin C =
, or the
we find that sin A = sin B =
2
4
4
1 + tan
lengths of two
medians
in a triangle are equal to R if and only if it is similar to a triangle
with sides 2, 2, 7.

Example 2.122. Consider a convex polygon A1 A2 ...An and a point P in its interior. Find
the least number of triangles Ai Aj Ak that contain P on their sides or in their interiors.

Solution
We prove that a point P may be found such that it is not contained in the interior or
on the sides of more than n 2 triangles. The result is true for n = 3, since P will be in the
interior of A1 A2 A3 only. If n = 4, denote by Q the point where diagonals A1 A3 and A2 A4
intersect. Clearly, each triangle Ai Aj Ak will have non void intersection with the interior of
A2 A3 Q if and only if one of its sides is A2 A3 , in which case A2 A3 Q is contained in it. Since
there are exactly n 2 different triangles A2 A3 Ak that contain P in their interior, and no
other triangle Ai Aj Ak may contain P on its sides or in its interior, the least number is no
larger than n 2.
The number cannot be less than n 2, we prove this by induction. The result is true
for the case n = 3. If the result is true for n 1 3, consider triangles A1 A2 A3 and A3 A4 A5
in an n-gon, n 4 (if n = 4, then A5 = A1 ).

103

2.3. Geometric problems

If n = 4, P is either on the common boundary A1 A3 of these triangles, or completely


outside one of them. If n 5, P cannot be simultaneously on the sides or in the interior of
both triangles, since they only have one common vertex A3 which cannot be P . Therefore,
either (n1)-gon A1 A3 A4 ...An , or (n1)-gon A1 A2 A3 A5 ...An , contains P in their interior.
Assume without loss of generality, and by hypothesis of induction n 3 triangles Ai Aj Ak
may be found that contain P on their sides or in their interiors, where i, j, k 2. Consider
now the partition of the n-sided polygon on triangles by drawing all diagonals A2 Ak .
Clearly, P is on the sides or in the interior of at least one of the triangles thus generated,
and this triangle is different to the n 3 previously considered, or the number of triangles
that contain P on their sides and in their interior is no less than n 2, and so this is the
least number.
Example 2.123. Let la , lb , lc be the lengths of the angle bisectors of a triangle. Prove the
following identity
sin BC
sin CA
sin AB
2
2
2
+
+
=0
(2.30)
lc
la
lb
where A, B, C are the angles of the triangle.
First solution
2R sin A sin B
Using the fact that lc =
, we get
cos ( AB
2 )
LHS(2.30) =

sin BC
sin CA
sin AB
2
2
2
+
+
2R sin A sin B 2R sin B sin C 2R sin C sin A
)
)
)
cos ( AB
cos ( BC
cos ( CA
2
2
2

1
1
1
sin(A B)
sin(B C)
sin(C A)
= 2
+ 2
+ 2
2R sin A sin B 2R sin B sin C 2R sin C sin A
1 sin(A B) sin(B C) sin(C A)
=
[
+
+
]
4R sin A sin B sin B sin C sin C sin A
1 cyc (sin A cos B cos A sin B) sin C
4R
sin A sin B sin C
0
1
= 0 = RHS(2.30).
=
4R sin A sin B sin C
=

Second solution
By Mollweides formula we have
AB
sin AB
a b sin A sin B 2 cos A+B
2 sin 2
2
=
=
=
c
sin C
2 sin C2 cos C2
cos C2

and lc =

2ab
C
cos , where a, b, c are the sides of the triangle; hence,
a+b
2
sin AB
a2 b2
2
=

lc
2abc

()

104

Chapter 2. Examples for practice

With this and the two similar results, we obtain

as desired.

sin BC
sin CA
sin AB
(a2 b2 ) + (b2 c2 ) + (c2 a2 )
2
2
2
+
+
=
= 0,
lc
la
lb
2abc

()

Example 2.124. In the Cartesian plane call a line good if it contains infinitely many
lattice points. Two lines intersect at a lattice point at an angle of 45o degrees. Prove that
if one of the lines is good, then so is the other.
First solution
Let L1 and L2 be lines that intersect at a lattice point at an angle of 45o. Further,
suppose that L1 is good. As L1 contains two lattice points, its slope must either be
undefined or rational. If the slope of L1 is undefined, then the slope of L2 is 1; in either
situation, L2 contains one lattice point and has rational slope so must therefore contain
infinitely many lattice points and is good. If, on the other hand, the slope of L1 is 1,
then L2 is either a horizontal line or a vertical line; again, in either situation, because
L2 is known to contain one lattice point it must therefore contain infinitely many lattice
points and is good. Finally, suppose the slope of L1 is rational but neither 1, and let
denote the angle of inclination of L1 . Then tan is rational and
tan ( 45o) =

tan tan 45o tan 1


=
1 tan tan 45o 1 tan

is also rational. Once again, L2 contains one lattice point and has rational slope so must
therefore contain infinitely many lattice points and is good.
()
Second solution Let us suppose that l1 and l2 are two lines that satisfy the conditions
stated in the hypothesis. Without loss of generality we may assume that l1 is a good
line, and that the coordinates of the lattice point at which those lines meet are (m, n).
The purported result clearly holds in any one of the following cases:
(a) The slope of line l1 is 1.
(b) The slope of line is l1 is -1.
(c) Line l1 is vertical.
(d) Line l1 is horizontal.
If line l1 falls into neither of those categories below, we infer that its slope is a rational
a
number of the form , where a, b Z {0}, a + b 0, and a b 0. Furthermore, the
b
hypothesis that lines l1 and l2 intersect at an angle of 45o imply that one and only one of
above relations holds
2 = 1 + 45o
(2.31)
2 = 1 45o ,

(2.32)

2 = 1 + 45o

where 1 and 2 are the elevation angles of lines l1 and l2 , respectively. Now, assuming
that (2.31) holds (similarly we can do for (2.32)), we get
tan 2 = tan (1 + 45o ) =

tan 1 + tan 45o


1 tan 1 tan 45o

105

2.3. Geometric problems

a
b

+1 a+b
=

1 ab b a

Hence, line l2 is represented by the equation

or equivalently,

y n = (tan 2 ) (x m) =

a+b
(x m)
ba

(b a)y (a + b)x = (b a)n (a + b)m.

(2.33)

Since gcd(ba, (a+b))(ba)n(a+b)m, the diophantine equation in (2.33) possesses


an infinite number of solutions in integers. Each one of these solutions corresponds with
a lattice point in l2 , and we are done.
()
Third solution
Let the two lines, say, l1 and l2 , intersect at a lattice point P (a, b) at an angle of 45o .
Without loss of generality we may assume that line l1 is good, i.e. l1 contains an infinite
number of lattice points.
Let Q(c, d) be an arbitrary lattice point on l1 . Construct a perpendicular on l1 passing
through Q such that it intersects l2 at R(c , d ). We show that R itself is a lattice point.
Note that triangle P QR is a right isosceles triangle, with QP = QR and P QR = 90o .
Now, consider the points on this plane as complex numbers. Recall that a complex number
a + ib when multiplied by ei rotates it by an angle in the counterclockwise direction.

We note that QR rotated by 90o in the counterclockwise direction coincides with QP .


Therefore we have

(c c + i(d d))ei 2 = a c + i(b d),

which implies d d + i(c c) = a c + i(b d). Solving for c and d , we obtain c = b + c d


and d = c + d a.
Now, since a, b, c and d are all integers, so are c and d , thus proving that R(c , d ) is
a lattice point.
Hence, if an arbitrary point Q(c, d) on line l1 is a lattice point, then so is R(c , d ) on
line l2 . This implies if l1 is good, then so is line l2 .
()

Example 2.125. Circles C1 (O1 , r) and C2 (O2 , R) are externally tangent. Tangent lines
from O1 to C2 intersect C2 at A and B, while tangent lines from O2 to C1 intersect C1 at
C and D. Let O1 AO2 C = {E} and O1 B O2 D = {F }. Prove that EF O1 O2 = AD BC.

First solution
Let M be the midpoint of O1 O2 and observe that A, B, C, D lie on the circunference
with center M and radius MO1 , as shown in Figure 2.1.
Figmr31
Let N = AD BC and note that N lies on O1 O2 because AD is the reflection of BC
across the line O1 O2 . Since O1 C = O1 D we haveCAO1 = O1 AD, so O1 A is the internal
bisector of CAN. By the same argument, since O2 A = O2 B, we have O2 CA = BCO2 ,
so O2 C is the internal bisector of ACN.

106

Chapter 2. Examples for practice

Figure 2.1:
Therefore E is the incentre of CNA, so CNE = ENA. Similarly F is the incenter
of BND, so DNF = F NB. Consequently we have
1
1
CNE = CNA = DNB = F NB
2
2

and this implies E, N, F are collinear. Thus EF O1 O2 = AD BC = {N} and the


claim is proved.
()
Second solution
We will show that the result holds in the more general case of any two circles such
that neither of them is contained in the other.

Note that O1 AO2 = O1 BO2 =


since O1 A is tangent to C2. In the same way
2

O1 CO2 = O1 DO2 = , so that the points A, C, O1 , D, B, O2 all lie on the circle of


2
diameter O1 O2 . Now applying Pascals theorem to the cyclic hexagon ADO2 CBO1 it
follows that P = AD BC, AO1 CO2 = E and BO1 DO2 = F are collinear. In other
words, AD, BC and EF concur.
Now it remains to observe that by symmetry O1 O2 is the perpendicular bisector of
AB and CD, and thus P lies on O1 O2 . Since we just proved that EF goes through P , it
follows that AD, BC, EF and O1 O2 concur, as we wanted.
()
Example 2.126. In the interior of a regular pentagon ABCDE consider the point M
such that triangle MDE is equilateral. Find the angles of triangle AMB.
Solution
We have that
MEA = AED MED = 108o 60o = 48o .

Morever, AE = EM implies that

EAM = EMA =

(180o 48o )
= 66o .
2

107

2.3. Geometric problems

By symmetry

ABM = MBC = 2ABC = 54o .

Finally
MAB = EAB DCM = 108o 66o = 42o , AMB = 180o MAB ABM = 84o .
()

Example 2.127. Let P be a point situated in the interior of an equilateral triangle ABC
and let A1 , B1 , C1 be the intersectionsof lines AP, BP, CP with sides BC, CA, AB, respectively. Find P such that
A1 B 2 + B1 C 2 + C1 A2 = AB12 + BC12 + CA21 .
Solution
Let us denote the areas of BP C, AP B, AP C as x, y, z respectively. we have,
BA1 AA1 B BP A1 AA1 B BP A1 AP B y
=
=
=
=
=
A1 C AA1 C CP A1 AA1 C CP A1 AP C z
ay
az
Let AB = BC = CA = a, then BA1 =
; and A1 C =
. Similarly we obtain four
y+z
y+z
other relations.

Figure 2.2:
Substituting these values in our first relation we obtain,
z2
x2
z2
x2
y2
y2
+
+
=
+
+
(y + z)2 (z + x)2 (x + y)2 (y + z)2 (z + x)2 (x + y)2

Which, on transposition, is equivalent to

y2 z2
z 2 x2
x2 y 2
+
+
= 0.
(y + z)2 (z + x)2 (x + y)2

108

Chapter 2. Examples for practice


xy
= 0, which rewrites into
cyc x + y
n

This relation can be simplified to

(x y)(z x)(z y)
= 0.
(x + y)(y + z)(z + x)

Hence the first relation actually holds true if and only if x = y or y = z or z = x. So, if
P is such that at least two of the triangles AP B, BP C, CP A have the same area,
then the first relation holds good. So, P must lie on any of the three medians of ABC.
()
= 90o and let d be a line passing trough
Example 2.128. Let ABC be a triangle with A
the incenter of the triangle and intersecting sides AB and AC in P and Q, respectively.
Find the minimum of AP AQ.
Solution
Let M, N be the projections of I on AC, AB respectively. We have IM = IN = r and
from the similarity of the triangles P MI, INQ we find
P M NQ = r 2 ,
where r is the inradius of triangle ABC. Then, we have
AP AQ = (AM + MP )(AN + NQ)
= AM AN + AM NQ + MP AN + MP NQ
= 2r 2 + r(NQ + MP ) 2r 2 + 2rpMP NQ = 2r 2 + 2r 2 = 4r 2 .

= 45o.
The sign of equality holds iff MP = NQ AP = AQ AP
Q = AQP

()

Example 2.129. Let ABC be an acute triangle and let MNP Q be a rectangle inscribed
in the triangle such that M, N BC, P AC, Q AB. Prove that
1
S(MNP Q) S(ABC).
2
First solution
Let x = NP and a = BC. Since P Q BC then QAP BAC and, therefore,
P Q ha x
a(ha x)
PQ =

=
CB
ha
ha
Hence,
S(MNP Q) =

Second solution

a
x + (ha x)
aha 1
a
x(ha x)
(
) =
= S(ABC).
ha
ha
2
4
2
2

()

109

2.3. Geometric problems

Since ABC is an acute triangle then the rectangle MNP Q is contained into the
triangle ABC. Let AH be the height from A to BC, then

Therefore

QM = P N = tAH, BM = tBH, CN = tCH for some t [0, 1].

S(MNP Q) = QM(BC BM CN) = t(1 t)AH BC = 2t(1 t)S(ABC).


The desired inequality follows by noting that
max 2t(1 t) =

t[0,1]

()

Example 2.130. Let ABCD be a square of center O. The parallel through O to AD


intersects AB and CD at M and N and a parallel to AB intersects diagonal AC at P .
Prove that
MN 4
4
OP + (
) = MP 2 NP 2 .
2

Solution
The parallel to AB intersects MN at T . It is clear that AMO and OT P are right
isosceles triangles. So, we can denote AM = MO = x and OT = T P = t.
Let MP = y, NP = z, and OP = u for convenience. So we have T N = x t. So we must
prove that u4 + x4 = y 2 z 2 . By Pythagoras, we have
y 2 z 2 = [(x + t)2 + t2 ] [(x t)2 + t2 ]

= (x2 + 2t2 + 2xt) (x2 + 2t2 2xt)


= (x2 + 2t2 )2 4x2 t2 = x4 + 4t4 .

From triangle OT P we have 2t2 = u2 then 4t4 = u4 and we are done.

()

Example 2.131. Let a, b, c be the side-lenghts of a triangle with the largest side c. Prove
that
ab(2c + a + b) a + b + c

(a + c)(b + c)
3
First solution
Without loss of generality assume that c a b. Consider the function

We have

and

f (a) =

f (x) =

a+b+x
1
1
ab (
+
).
3
a+x b+x

ab
1
1
+
) 0.
f (c) f (a) = (c a) ( +
3 2a(a + c) (b + a)(b + c)

1
1
2a + b b ab 1 1
5
2a + b
ab ( +
)=
( + ) = (a b) 0.
3
2a a + b
3
2 4 a b
12

(1)

110

Chapter 2. Examples for practice

Note that, the given inequality holds for any positive a, b, c with the largest c.
Second solution
Assume that c = max{a, b, c}, hence abc c3 , abc ac2 , abc bc2 and
a(c b) + b(c a) 0.

The original inequality can be written as

()

(i)

3abc
c2 + a(c b) + b(c a).
a+b+c
We have

c2 + c(a + b)
3abc
c(
) = c2 .
a+b+c
a+b+c
Combining (i) and (ii) the inequality is proved.

(ii)

()

Example 2.132. Let I be the incenter of triangle ABC and let A1 , B1 , C1 be the feet
of altitudes from vertices A, B, C. If IA1 = IB1 = IC1 , then prove that triangle ABC is
equilateral.
Solution
We start by noting that IA1 = IB1 = IC1 implies that I is the circumcenter of triangle
A1 B1 C1 whose circumcircle is of course the nine-point circle. Thus, I is the center of the
nine-point circle. Furthermore, Feuerbachs Theorem tells us that the nine-point circle
is internally tangent to the incircle. Since the nine-point circle and the incircle have the
same center and are internally tangent we can conclude that these circles have the same
radius. Moreover, if we let r be inradius and R be the circumradius of triangle ABC then
R
the condition that the radius of the nine-point circle, , is the same as the radius of the
2
incircle tells us that
R
= r R = 2r
2
which is the equality case of Eulers Triangle Inequality. But Eulers Triangle Inequality,
R = 2r, has equality if and only if triangle ABC is equilateral and so were done.
()
Example 2.133. Let ABC be a triangle with incenter I and let A1 , B1 , C1 be the symmetric points of I with respect to the midpoints of sides BC, CA, AB. If Ia , Ib , Ic denote the excenters corresponding to sides BC, CA, AB, respectively, prove that lines Ia A1 , Ib B1 , Ic C1
are concurent.
First solution
1
1
Consider the homothety H(I, ) with pole I and ratio . It is wellknown that the
2
2
midpoints of the segments IIa , IIb , IIc are in fact the midpoints of the arcs BC, CA, AB
not containing the vertices of the triangle; therefore, the lines Ia A1 , Ib B1 , Ic C1 are concurrent since the mediators of the triangle ABC are concurrent.
()
Second solution

111

2.3. Geometric problems

In exact trilinear coordinates, I = (r, r, r), r being the inradius, and the midpoint of
hb hc
BC is Ma = (0, , ) , where ha , hb , hc are the respective lengths of the altitudes from
2 2
A, B, C. Thus A1 = (r, hb r, hc r). Denoting by ra , rb , rc the exradii corresponding to
sides BC, CA, AB, clearly Ia = (ra , ra , ra ), or in trilinear coordinates (, , ), the points
on line Ia A1 satisfy
RR

RRRR
RR
R
RRR
+
+
1
1 RRRR =
0 = RR1

RR hb 2r hc 2r.
RR
RRRr hb r hc r RRR

or by cyclic permutations, the point that satisfies

(ha 2r)( + ) = (hb 2r)( + ) = (hc 2r)( + )

is simultaneously on lines Ia A1 , Ib B1 , Ic C1 . The conclusion follows.

()

Example 2.134. Let ABCD be a trapezoid (AB CD) with acute angles at vertices A
and B. Line BC and the tangent lines from A and E to the circle of center D tangent to
AB are concurrent at F. Prove that AC bisects the segment EF if and only if AF + EF =
AB.
Solution
We begin with the following preliminary result:
Lemma. Let ABC be an arbitrary triangle with incenter I centroid G. Then, the lines
IG and BC are parallel if and only if b + c = 2a.
Proof. Let la , ma be the bisector and median from A, respectively. We have that IG
is parallel to BC if and only if AI = AG. However, AI 2 = (p a)2 + r 2 by Pythagoras
theorem, where p denotes the semiperimeter and r the inradius of triangle ABC. Besides,
we have that
r2 =

(p a)(p b)(p c)
bc(a + b + c)(b + c a)
and la2 =
,
p
(b + c)2

so using 3AG = 2ma , we obtain


9 [(p a)2 +

4bc(a + b + c)(b + c a)
(p a)(p b)(p c)
]=
,
p
(b + c)2

which after several simple algebraic manipulations becomes (b + c 2a)(2a + 5b + 5c) = 0,


i.e. b + c = 2a.
Returning to the original problem we have that AF + EF = 2AE and note that G is
the midpoint of EF if and only if C is the centroid of triangle AEF, so by applying the
Lemma for this triangle, we arrive to desired conclusion.
()
Example 2.135. Give a straightedge and compass construction of a triangle ABC starting with its incenter I, the foot of the altitude from A, and the midpoint of the side BC.
Solution
We are given the foot of the altitude from A and the midpoint M of BC, so the line
determined by the two is precisely the line BC. Now, knowing the incenter we can draw

112

Chapter 2. Examples for practice

the perpendicular from I to BC and get the incircle of ABC. Recall the well-known fact
that if D is the tangency point of the incircle with the side BC, and D the antipode
of D with respect to the incircle, then the points A, D, X are collinear, where X is the
tangency point of the A-excircle with BC. However, in order to use this, we need to find
X. This is not a problem whatsoever, since MD = MX; so we have the construction of
X as the reflection of D in M. Now, just draw the lines XD and the altitude from A
(which we can draw since we have the foot of the altitude on BC and the line BC); they
intersect at the vertex A. Afterwards, just take the tangents from A to the incircle and
intersect them with BC ; this will give us the vertices B and C. Hence our construction
is complete.
()
Example 2.136. Given a nondegenerate triangle ABC, consider circles Ga , Gb , Gc with
diameters BC, CA, and AB, respectively. For which triangles ABC are Ga , Gb , Gc concurrent?
Solution
Let C and D be the intersections of Ga and Gb . As BDC = CDA = 90o (as D lies
on circles with diameter BC and CA), we get that BDA = 90o + 90o = 180o, and so
D AB. So suppose that Ga , Gb and Gc are concurrent. Then either C Gc or D Gc .
The first case gives a right triangle with right angle at C. The second case gives either
D = A or D = B, because D AB Gc = {A, B}. For D = A, we get a right angle at A,
and for D = B we get a right angle at B. The only triangles for which Ga , Gb and Gc are
concurrent, are therefore the right triangles. (One easily checks that these triangles satisfy
the conditions from the problem.)
()
Example 2.137. Consider a triangle ABC. Let D the orthogonal projection of A onto BC
and let E and F be points on lines AB and AC respectively such that ADE = ADF.
Prove that the lines AD, BF , and CE are concurrent.
Solution
We will prove a lemma first:
Lemma 4. If P is a point on the side BC of a triangleABC we have
P B AB sin P AB
=

P C AC sin P AC
Proof.
In triangles P AB and P AC, the law of sines gives
PB
AB
=
sin P AB sin AP B
AC
AC
PC
=
=

o
sin P AC sin(180 AP B) sin AP B
Dividing the above relations we get the desired result.
()
Coming back to the problem, let us denote x = ADE = ADF , as shown in figure.
By the Lemma we have
sin x
AE AD
=

EB BD sin(90o x)

113

2.3. Geometric problems

E
x x
B

D
Figure 2.3:
CF DC sin(90o x)
=

F A AD
sin x

Therefore
AE BD CF AD
sin x
BD DC sin(90o x)

=1
EB DC F A BD sin(90o x) DC AD
sin x
so, by Cevas theorem the lines AD, BF , and CE are concurrent and we are done.

()

Example 2.138. Let P be a point inside a triangle ABC and let da , db , dc be the distances
from point P to the triangles sides. Prove that
da h2a + db h2b + dc h2c = (da + db + dc )3 ,

where ha , hb , hc are the altitudes of the triangle.


Solution
First need the following
Lemma 5. If x, y, z, a, b, c > 0 we have

x3 y 3 z 3 (x + y + z)3
+ +

a2 b2 c2 (a + b + c)2

The proof follows from H


olders Inequality. Indeed, we have
1

x3 y 3 z 3 3
(a + b + c) ( 2 + 2 + 2 ) x + y + z.
a
b
c
2
3

and the result follows.


We now prove the initial inequality. Using the above lemma we have
d3b
d3a
d3c
da db dc
+
+
=
+
+
a2 b2 c2 (ada )2 (bdb )2 (cdc )2

(da + db + dc )3
(da + db + dc )3
=

2
(ada + bdb + cdc )
4S 2

since ada + bdb + cdc = 2S, where S denotes the area of the triangle ABC. The result
2S
2S
2S
,b =
,c =
.
()
follows upon substituting a =
ha
hb
hc

114

Chapter 2. Examples for practice

Example 2.139. Let A1 , A2 , ..., An be a regular n-gon inscribed in a circle of center O and
radius R. Prove that for each point M in the plane of the n-gon the following inequality
holds:
n
n
MAk (OM 2 + R2 ) 2 .
k=1

Solution
Let us work in the complex plane with O as the origin and without loss of generality,
2
n
)
R = 1. Let e = exp ( 2
n and let the complex numbers e, e , ..., e , x correspond to the points
A1 , A2 , ..., An , M respectively. Then our inequality is equivalent to
x ek
n

k=1

n
(x2 + 1) .

Since e, e2 , ..., en are the roots of z n 1 = 0, we have

x ek = xn 1 xn + 1,
n

k=1

by the triangle inequality. Hence it remains to show that

(xn + 1)2 (x2 + 1)n 2xn Cnk x2k


n1
k=1

which follows from AM-GM inequality since n 3 and

Cnk x2k nx2 + nx2n2 2nxn 2xn .

n1
k=1

Equality holds iff x = 0 i.e. when M = O.

()

Example 2.140. Let a1 , a2 , ..., an [0, 1] and be real numbers such that
a1 + a2 + + an = n + 1 .

For any permutation (bi )ni=1 of (ai )ni=1 prove that

a1 b1 + a2 b2 + + an bn n + 1 2 .

Solution
As a1 , ..., an [0, 1], we have = n + 1 (a1 + ... + an ) n + 1 n = 1. For every
i = 1, ..., n this implies that ai , bi 1 (where (bi ) is a permutation of (ai )), and
therefore ( ai )(1 bi ) 0. Summing these inequalities for every i = 1, ..., n, we find
0 n bi ai + ai bi .
n

i=1

i=1

i=1

Because a1 + ... + an = b1 + ... + bn = n + 1 , this yields


as desired.

a1 b1 + ... + an bn n + (n + 1 ) + (n + 1 ) = n + 1 2 ,

()

115

2.3. Geometric problems

Example 2.141. Let ABC be an isosceles triangle with A = 100o. Denote by BL the
angle bisector of angle ABC. Prove that AL + BL = BC.
First solution
Let D be a point on BL produced beyond L such that LD = LA and let E be a point
on BC such that LE bisects BLC.
Since ABC = BCA = 40o, we have ABL = LBE = 20o , BLA = 60o , BLC =
120o and BLE = ELC = 21BLC = 60o = DLC.
Thus, triangles ABL andEBL are congruent (angle-side-angle), giving LA = LE.
Therefore, LD = LE.
We also have that LC is the bisector of the vertical angle in isosceles triangle DLE.
Hence LC is actually the perpendicular bisector of the base DE, and accordingly it crosses
it at right angles at its midpoint.
Consequently, LCD = ECL = 40o andEDC = 90o LCD = 50o.
So we have
BDC = BDE + EDC = 30o + 50o = 40o + 40o = BCL + LCD = BCD,
making BCD isosceles with
BC = BD = BL + LD = BL + LA
and we are done.

()

Second solution
Clearly, B = C = 40o , or ABL = CBL = 20o , leading to ALB = 60o and
CLB = 120o . Using the Sine Law,
AL + BL sin ABL + sin BAL
=
AB
sin ALB
100o + 20o
100o 20o
cos
sin 20o + sin 100o
2
2
=
= 2 cos 40o
=
o
o
sin 60
sin 60
o
o
sin 100
sin A BC
sin 80
=
=
=

=
o
o
sin 40
sin 40
sin C AB
2 sin

The conclusion follows.

()

Third solution
First we note that CBA = BCA = 40o , hence LBA = 20o and BLA = 60o .
Let M be the point (on BC) symmetric of A with respect to BL.
Then BLM = BLA = 60o , hence also MLC = 60o . On the prolongation of BL
take P such that AL = LP. Since CLP = ALB = 60o and LP = LA = LM we see that
M and P are symmetrical with respect to AC, hence P CB = 2ACB = 80o .
Since CBP = 20o , we have CP B = 80o and BCP is isosceles, therefore
BC = BP = BL + LP = BL + AL.

()

116

Chapter 2. Examples for practice

Example 2.142. Prove that in each triangle


a2
R
1 b2 c2
= 4 ( + 1) .
( + )
r rb rc
rb rc
ra

First solution
We know the follwing facts:
ra =

S
S
S
abc
S
; rb =
; rc =
; r= ; R=

pa
pb
pc
p
4S

The left hand side of the equality may be simplified by using the above facts to obtain

L.H.S =
=
=
=
=
=

b2 p(p b) + c2 p(p c) a2 p(p b)(p c)


S2

(b2 + c2 a2 )p2 + p(b + c)(b2 bc c2 + a2 (b + c)p a2 bc)


S2

p(b + c a)(p b c) + bc((b + c)p a)


S2

2a2 (b2 + c2 + bc) a4 (b4 + c4 2b2 c2 ) + 2abc(b + c) 4a2 bc


4S 2
2(a2 b2 + b2 c2 + c2 a2 ) a4 b4 c4 + 2abc(b + c a)
4S 2

16S 2 + 4abc(p a) 4SR(p a)


4R
R
=
=4+
= 4 (1 + ) = R.H.S.
2
2
4S
S
ra
ra

and we are done.

()

Second solution
Let S and p be the area and semiperimeter of the triangle, respectively. Then we have
the well-known identities

S
abc
= rp = p(p a)(p b)(p c), and ra =

S=
4R
pa
and the respective cyclic ones. Using these the equality becomes
b2 p(p b) + c2 p(p c) a2 (p b)(p c) abc(p a) = 4S2
p
p
(a2 b + ab2 + b2 c + bc2 + c2 a + ca2 a3 b3 c3 2abc) = (b + c a)(c + a b)(a + b c),
2
2
which is true.
()
Example 2.143. Let ABCD be a quadrilateral whose diagonals are perpendicular. Denote
by 1 , 2 , 3 , 4 the centers of the nine-point circles of triangles ABC, BCD, CDA, DAB,
respectively. Prove that the diagonals of 1 2 3 4 intersect at the centroid of ABCD.

117

2.3. Geometric problems

First solution
Let MAB , MBC , MCD , MAD be the midpoints of the sides AB, BC, CD, DA, respectively. Since the nine-point circle of the triangle ABC passes through the midpoints of its
sides we have 1 belongs to the perpedicular bisector of MAB MBC . Similarly, 3 belongs
to the perpedicular bisector of MCD MDA . Since AC and BD are perpendicular we get
that MAB MBC MCD MAD is a rectangle. This implies that the line 1 3 is the midline
of opposite sides of this rectangle: MAB MBC and MCD MDA . Finally, the intersection of
the lines 1 3 and 2 4 coincides with the intersection of the diagonals of the rectangle
MAB MBC MCD MAD which is the centroid of ABCD.
()
Second solution
We proceed by coordinate geometry. Let the point of intersection of the diagonals be
(0, 0). Let A = (a, 0), B = (0, b), C = (c, 0), D = (0, d), then the centroid of ABCD,
G=(

a+c b+d
,
).
4
4

Now we will find the coordinates of 1 . Let H and O be the orthocenter and the
ac
circumcenter of triangle ABC. Then H = (0, ) and using that BH = 2OF where F is
b
a + c b2 + ac
the feet of the perpendicular to AC from O we obtain that O = (
,
) . It is well
2
2b
a + c b2 ac
,
).
known that 1 is the midpoint of HO, hence 1 = (
4
4b
Analogously,
2 = (

c2 bd b + d
a + c d2 ac
a2 bd b + d
,
) , 3 = (
,
) , 4 = (
,
).
4c
4
4
4d
4a
4

Clearly, 1 3 and 2 4 pass through G, and we are done.

()

Third solution
Let Mac and Mbd be the midpoints of the diagonals AC and BD. It is known that
the centroid G of ABCD coincides with the midpoint of Mac Mbd . Let Ox and Hx be the
cicrumcenter and the orthocenter of triangle Y ZT, {X, Y, Z, T } = {A, B, C, D}. Points Od
and Oc being cicrumcenters of triangle ABC and ACD, are on the same perpendicular
bisector of AC, yielding
Mac Od Ob AC.

(1)

The quadrilateral ABCD having the diagonals intersecting at a right angle implies
that BO and DO are heights in the triangles ABC and ACD. Thus Hd , Hb BD and
Hd Hb AC.

(2)

From (1) and (2) it follows that Od Ob Hd Hb . Last relation proves that Od Ob Hd Hb is a
trapezoid. The Euler circles center in a triangle is the midpoint of the segment determined
by the circumcenter and the orthocenter. It follows that d and b are the midpoints of
Od Hd and Ob Hb . Thus the line d b is the midline of this trapezoid and so passes through
G, the midpoint of Mac Mbd . Analogoulsly, we prove that G a c .
()

118

Chapter 2. Examples for practice

Example 2.144. Let ABC be a triangle and let Ia , Ib , Ic be its excenters. Denote by
Oa , Ob , Oc the circumcenters of triangles Ia BC, Ib AC, Ic AB. Prove that the area of triangle
Ia Ib Ic is twice the area of hexagon Oa COb AOc B.
First solution
It is well-known that the angle-bisectors of ABC are the altitudes of Ia Ib Ic with feet
A, B, C. Denote by H the orthocenter of Ia Ib Ic . Clearly HAIb C, HAIc B and HBIa C are
cyclics and so Ia Oa = Oa H, Ib Ob = Ob H and Ic Oc = Oc H. Then [Ia Oa C] = [Oa HC] and so
on. This leads to [Ia Ib Ic ] = 2[Oa CObAOc B].
()

Second solution
A
Let I denote the incenter of triangle ABC. We have BIa C = 90o which implies
2
BOa C = 180o A. Hence Oa , Ob , Oc lie on the circumcircle (O) of ABC. Again, since
BOa = COa , we conclude that Oa is the midpoint of minor arc BC of (O). Therefore
A, I, Oa , Ia all lie on a line. Now
BIC + BIa C = 90o +

A
A
+ 90o = 180o .
2
2

Thus I lies on the circumcircle (Oa ) of BIa C. Therefore Oa is the midpoint of diameter
IIa and hence 2[IBOa ] = [IBIa ], 2[ICOa ] = [ICIa ] so that 2[IBOaC] = [IBIa C].
Similarly 2[ICObA] = [ICIb A] and 2[IAOc B] = [IAIc B].
Adding, we conclude that 2[Oa COb AOc B] = [Ia Ib Ic ].
()

Example 2.145. In a triangle ABC, let a , b , c be the angles between medians and altitudes emerging from the same vertex. Prove that one of the numbers tan a , tan b , tan c
is the sum of the other two.
First solution
If we define a , b , c as oriented angles between medians and altitudes (let it be
counterclockwise orientation) then statement of problems becomes:
Prove that tan a + tan b + tan c = 0.

Since a = b cos C + c cos B and tan a =


obtain

a
2

c cos B
then, applying the Sine Theorem we
c sin B

tan a =

b cos C c cos B 2R sin B cos C 2R sin C cos B


=
c sin B
2R sin C sin B
sin B cos C sin C cos B
= cot C cot B
=
sin C sin B

therefore

tan a = (cot C cot B) = 0.


cyc

cyc

()

119

2.3. Geometric problems

Second solution
Let AM be the median and AD the altitude emerging from vertex A. It is obvious
that
MD

tan a =
AD
Recalling that b2 c2 = 2aMD, we obtain
tan a =

b2 c2 b2 c2
=
2aha
4S

where S is the area of triangle ABC. Similarly, we have


tan b =

c2 a2
a2 b2
; tan c =

4S
4S

Without loss of generality assume that a b c. We find then that


tan b = tan a + tan c .

2.3.2

Senior problems

()

Example 2.146. Let ABC be a triangle. Prove that


cos3 A + cos3 B + cos3 C + 5 cos A cos B cos C 1.
First solution Using the equality
cos2 A + cos2 B + cos2 C + 2 cos A cos B cos C = 1

the initial inequality becomes equivalent to

or

cos3 A + 3 cos A cos2 A


3 cos A cos2 A(1 cos A)

Now, by the AM-GM inequality, we have

3
2
cos
A(1

cos
A)

cos2 A (1 cos A)

Thus, it suffices to prove that

cos A (1 cos A)

(2.34)

When triangle ABC is obtuse, the above inequality is clearly true. So we will consider
the case it is acute. We have
(2.34) (cos A)(1 + cos A) (1 cos2 A)
8 (cos A) (cos2

A
) sin2 A
2

120

Chapter 2. Examples for practice


cos2

A
2

sin

sin A
cos A

A
A
cos
2
2
B
C
A
cot cot cot tan Atan Btan C
2
2
2
A
B
C
cot + cot + cot tan A + tan B + tan C.
2
2
2
Indeed, we have
tan A =

B+C
A
tan B + tan C
tan
= cot ,
2
2
2

and the equality holds if and only if triangle ABC is equilateral.

()

Second solution
Since cos2 A + cos2 B + cos2 C + 2 cos A cos B cos C = 1 we will prove that
cos2 A(1 cos A) = 3 cos A cos B cos C.
cyc

By the AM-GM Inequality we have

2
2
3 cos A(1 cos A)
cos
A(1

cos
A)

cyc

cyc

then it suffices to prove

3 3 cos2 A(1 cos A) 3 cos A cos B cos C.

(2.35)

cyc

Using that cos A =

A a2 (b c)2
b2 + c2 a2
and 2 sin2 =
, we get
2bc
2
2bc

(2.35) cos A cos B cos C (1 cos A) = cos A cos B cos C


cyc

cos A cos B cos C cos A cos B cos C = 8 sin2

cyc

b2 + c2 a2
a2 (b c)2

2bc
2bc
cyc

(b2 + c2 a2 ) (b + c a)2 .
cyc

A 2B 2C
sin
sin
2
2
2

cyc

Without loss of generality we can assume that b2 + c2 a2 > 0.


cyc

Then
b2 + c2 > a2 , ; c2 + a2 > b2 , a2 + b2 > c2 ,

(2.35.1)

121

2.3. Geometric problems

and, therefore,
(2.35.1) (b2 + c2 a2 ) (b + c a)4 .
2

Because

(2.36)

cyc

cyc

(b2 + c2 a2 ) = (b4 (c2 a2 ) )


2

cyc

cyc

and

2
4
(b + c a) = (b2 (c a) )

cyc

cyc

it is enough to prove b4 (c2 a2 ) (b2 (c a)2 )


We have
2

(b2 (c a)2 ) b4 + (c2 a2 ) = b4 2b2 (c a)2 + (c a)4 b4 + (c2 a2 )


2

= (c a)2 ((c + a)2 2b2 + (c a)2 )


= (c a)2 (2c2 + 2a2 2b2 )

= 2(c a)2 (c2 + a2 b2 ) 0,

and we are done.

()

Example 2.147. Let ma , mb , mc be the medians, ka , kb , kc the symmedians, r the inradius,


and R the circumradius of a triangle ABC. Prove that
3R ma mb mc
+
+
3.

2r
ka
kb
kc
Solution
Let K be the symmedian point of ABC and let BC = a, CA = b, AB = c. It is well
known that K is the barycentre of A, B, C with masses a2 , b2 , c2 , respectively. This gives
(b2 + c2 )A1 = b2 B + c2 C

where A1 is the point of intersection of the lines AK and BC and it follows that

(b2 + c2 )2 k 2 = (b2 + c2 )2 AA21 = (b2 AB + c2 AC)2


= b4 c2 + b2 c4 + b2 c2 (b2 + c2 a2 )
= b2 c2 (2b2 + 2c2 a2 )

that is, (b2 + c2 )2 ka2 = 4m2a b2 c2 . Thus,

m2a b2 + c2
and similarly,
=
ka2
2bc
m2b c2 + a2
m2c a2 + b2
=
and

=
kb2
2ca
kc2
2ab

As a result, the required inequalities become


3R b2 + c2 c2 + a2 a2 + b2

+
+
6.
r
bc
ca
ab

122

Chapter 2. Examples for practice

The right inequality rewrites as a(b2 + c2 ) + b(c2 + a2 ) + c(a2 + b2 ) 6abc, which holds
since b2 + c2 2bc, c2 + a2 2ca and a2 + b2 2ab.
As for the left inequality, it rewrites as
a2 (b + c) + b2 (c + a) + c2 (a + b) 6R2 (a + b + c)

Assuming that a b c, we have

a2

b2

c2

a2 (b + c) + b2 (c + a) + c2 (a + b)

(2.37)

(since abc = 2Rr(a + b + c)).


and b + c c + a a + b so that

(a2 + b2 + c2 )(2a + 2b + 2c)


3

(2.38)

by Chebyshevs inequality.
Now, let H and O be the orthocentre and circumcentre of ABC.

We know that OH = OA + OB + OC, from which we deduce OH 2 = 9R2 (a2 + b2 + c2 ).
Thus, a2 + b2 + c2 9R2 and taking (2.38) into account, we readily obtain (2.37). ()

Example 2.148. Let ABCD be a cyclic quadrilateral whose diagonals are perpendicular
to each other. For a point P on its circumscribed circle denote by P the line tangent to
the circle at P . Let U = A B, V = B C, W = C D, K = D A. Prove that UV W K is a
cyclic quadrilateral.
Solution
Let U1 , V1 , W1 , K1 be the midpoints of AB, BC, CD, DA, respectively. The Varignon
paral- lelogram U1 V1 W1 K1 of the quadrilateral ABCD is a rectangle (because AC BD),
hence U1 , V1 , W1 , K1 lie on a circle centered at the centre of the rectangle. Note that
O, U1 , U are collinear (on the perpendicular bisector of AB) and that AB is the polar of
U with respect to G. Similar results hold for V1 , W1 , K1 and it follows that the inverses
of U1 , V1 , W1 , K1 in the circle G are U, V, W, K, respectively, so that U, V, W, K all lie on
the inverse of the circle . Since U, V, W, K clearly cannot be collinear, the inverse of is
a circle and so UV W K is a cyclic quadrilateral.
Example 2.149. Let I be the incenter of triangle ABC. Prove that
AI BI CI = 8r 3 ,
where r is the inradius of triangle ABC.
Solution
Since

AI =

, BI =

, CI =

B
C
sin
sin
2
2
B
C 1
A
the inequality above is equivalent to sin sin sin , which is immediately true
2
2
2 8
because of the two well-known facts below:
A
sin
2

A
B
C
r
= sin sin sin and R 2r,
R
2
2
2
where R is the circumradius of triangle ABC.

()

123

2.3. Geometric problems

Figure 2.4:
Example 2.150. Let Ia be the excenter corresnponding to the side BC of triangle ABC.
Denote by A , B , C the tangency points of the excircle of center Ia with the sides BC, CA, AB,
respectively. Prove that the circumcircles of triangles AIa A , BIa B , CIa C have a common
point, different from Ia , situated on the line Ga Ia , where Ga is the centroid of triangle
A B C .
First solution
Let be the excircle. Since Ia A = Ia C an BA = BC , the line Ia B is the perpendicular bisector of A C and intersects A C in its midpoint B1 . Since A C is the polar
of B with respect to , the inversion in the circle exchanges B1 and B. Since B is
invariant under this inversion, the circumcircle of Ia BB inverts into the median B B1
of triangle A B C . Similarly, the circumcircles of Ia AA , Ia CC invert into the medians A A1 , C C1 . As a result, the three circumcircles all pass through Ia and through the
inverse of Ga (because Ga lies on the three medians A A1 , B B1 , C C1 ). The second result
follows from the fact that the inverse of Ga is on the line through Ia and Ga .
()
Second solution
Let D be the midpoint of B C . Now AB Ia B , while AIa B C where B D = C D
by symmetry around the internal bisector of angle A. Thus, triangles AB D and B Ia D
are similar, hence B D C D = B D 2 = Ia D AD , and the power of D with respect to
the circumcircles of A B C and AIa A is the same, or D lies on the radical axis of both
circles, which is median A D .
Let E be the midpoint of C A .BA Ia A , while BIa A C where A E = C E by
symmetry around the external bisector of angle B. Thus, triangles BE A and A E Ia are
similar, hence A E C E = A E 2 = BE Ia E , and median B E is the radical axis of the

124

Chapter 2. Examples for practice

Figure 2.5:
circumcircles of A B C and BIa B . Similarly, median C F (F is the midpoint of A B )
is the radical axis of the circumcircles of A B C and CIa C .
Clearly, the point Ga where the medians A D , B E and C F meet, has the same
power with respect to the four circumcircles; consider now the second point P where Ia Ga
meets the circumcircle of AIa A . Since Ia Ga is the radical axis of the circumcircles of
AIa A and BIa B because Ia , Ga have the same power with respect to both, then P also
has the same power with respect to both, but since it is on the circumcircle of AIa A , it is
also on the circumcircle of BIa B . Similarly, it is also on the circumcircle of CIa C . The
conclusion follows.
()
Example 2.151. Let ABC be at triangle. Find the locus of points X on line BC such
that
AB 2 + AC 2 = 2(AX 2 + BX 2 ).
Solution
By the median theorem,

AM 2 + BM 2 = AM 2 +

BC 2 AB 2 + AC 2
=
,
4
2

where M is the midpoint of side BC. Clearly X = M is one solution. Any other solution
is on the intersection of side BC and the circle with center the midpoint of AB passing
AB 2
, is the equation
through M, since it is well known that AX 2 + BX 2 = k, where k >
2
of a circle with center the midpoint of AB, and M is on that circle. There is therefore at
most one more solution to the equation, which is the symmetric of M with respect to the
perpendicular to BC through the center of the circle, ie the symmetric of M with respect
to the foot of the perpendicular from the midpoint of AB onto BC. There is exactly one

125

2.3. Geometric problems

solution, ie exactly one point X that satisfies the given condition, when the midpoint of
BC is also the foot of the perpendicular from the midpoint of AB onto BC, ie by Thales
= 90o .
theorem when C

Example 2.152. In triangle ABC, lines AA1 , BB1 , CC1 are concurrent at P , where
points A1 , B1 , C1 are situated on sides BC, CA, AB, respectively. Consider points A2 , B2 , C2
on segments B1 C1 , C1 A1 , A1 B1 , respectively. Prove that AA2 , BB2 , CC2 are concurrent if
and only if A1 A2 , B1 B2 , C1 C2 are concurrent.

Solution
Let D, E, F the second points where AA2 , BB2 , CC2 intersect lines BC, CA, AB, respectively. Applying the Sine Law to triangles ADB, ADC, C1 AA2 and B1 AA2 , we find:
BD =
C1 A2 =

AC sin CAD
AB sin BAD
, CD =
,
sin ADB
sin ADC

AB1 sin B1 AA2


AC1 sin C1 AA2
, B1 A2 =

sin AA2 C1
sin AA2 B1

Now, since
180o = ADB + ADC = AA2 C1 + AA2 B1 , BAD = C1 AA2

and CAD = B1 AA2 , we find

BD AB AB1 C1 A2

,
=

CD AC AC1 B1 A2

and similarly for its cyclic permutations. We then conclude that


BD CE AF
AB1 CA1 BC1 A1 B2 C1 A2 B1 C2

=(

)(
DC EA F B
B1 C A1 B C1 A B2 C1 A2 B1 C2 A1
By Cevas theorem, the first term in the RHS equals 1, and again by Cevas theorem,
AA2 , BB2 , CC2 are concurrent iff the LHS equals 1, iff the second term in the RHS equals
1, if and only if A1 A2 , B1 B2 , C1 C2 are concurrent. The conclusion follows.
()

2C.
Denote by D the foot of the altitude
Example 2.153. Let ABC be a triangle with B
AB

from A and by M be the midpoint of BC. Prove that DM


2
Solution
Clearly CD = b cos C and 2CM = a, or since by the Sine Law:
a = 2R sin A = b cos C + c cos B, we have
2DM = 2b cos C a = b cos C c cos B = 2R sin(B C) = 2R sin C = AB,

>B
C
C.
The conclusion follows, and equality is reached
where we have used that C
= 2C.

iff B
()

Example 2.154. Let ABC be a triangle with incenter I and circumcenter O and let M
be the midpoint of BC. The bisector of angle A intersects lines BC and OM at L and Q,
respectively. Prove that
AI LQ = IL IQ.

126

Chapter 2. Examples for practice

Solution (readers build their own figure).


It is a well-known that the bisector of CAB and the perpendicular bisector OM of
BC intersect on the circumcircle of ABC. In other words, Q is the midpoint of the arc
BC not containing A. Observe that BQ = IQ, since by angle chasing
QBI = QIB =

BCA

2
2

(in fact Q is the circumcenter of BIC). On the other hand, since the triangles BLQ and
ALC are similar we can write
BQ AC
=

LQ LC
Now repeated use of the angle bisector theorem in the triangles ABC and ABL yields
IQ BQ AC AB AI
=
=
=
=

LQ LQ LC LC IL
Therefore AI LQ = IQ IL, as desired.

()

Example 2.155. Let ABC be a triangle such that ABC > ACB and let P be an
exterior point in its plane such that
P B AB
=
P C AC

Prove that ACB + AP B + AP C = ABC.


Solution (readers build their own figure).
P B AB
Note that the relation
=
clearly defines an Apollonius circle with center on
P C AC
line BC, passes through A and through the point D where the internal bisector of angle
A intersects BC, leaving B inside and C outside because ABC > ACB. Note that
the powers of B, C with respect to are respectively PB , PC , such that
PB BD BD BA2 c2
=
=
=
PC CD CD CA2 b2
where D is the point diametrally opposite D in . Let now T, U be the second points
where P B, P C meet (the first one being clearly P in both cases). Note therefore that
b PB
b2 PB b2 PB
CT b BT
= 1,
=
=
CU c CU c CU P B c CU P C c PC
or CT = CU, and similarly BT = BU, ie BC is the perpendicular bisector of T U, which
are therefore symmetric with respect to BC. Therefore, if P is on the same half plane as
A,
AP B = AP T = 180o ADT = 180o ADB BDT
= 180o ADB BDU = 180o 2ADB ADU
= 180o 2ADB AP U = 180o 2ADB AP C,

127

2.3. Geometric problems

and similarly we find the same result if P is on the opposite half plane. In either case,
we find
A
AP B + AP C = 180o 2ADB = 180o 2 (180o B ) = 2B + A 180o = B C.
2
The conclusion follows.

()

and C.

Example 2.156. In triangle ABC let BB1 and CC1 be the angle bisectors of B
Prove that
A a
2bc
[(a + b + c) sin ] .
B1 C1
(a + b)(a + c)
2 2
Solution
First, note that by Ptolemys inequality for quadrilateral BCB1 C1 ,
B1 C1

BB1 CC1 BC1 CB1

BC
2ac
B
B
C p
A
Now, BB1 =
cos and that cos cos = sin , and these imply that
a+c
2
2
2 a
2
BB1 CC1 =
In addition, BC1 =

B
C 2abc(a + b + c)
A
4a2 bc
cos cos =
sin
(a + b)(a + c)
2
2
(a + b)(a + c)
2

ac
a2 bc
and this yields BC1 CB1 =
; therefore by just
a+b
(a + b)(a + c)
plugging in these identities in the inequality above, the desired conclusion follows. ()

Example 2.157. Prove that in any triangle,

ra 4R sin3 (

A
+ ).
3 6

Solution
We know that ra = R(1 cos A + cos B + cos C). Also,
4R sin3 (

A
A
+ ) = R [3 sin ( + ) cos A] .
3 6
3 6

Therefore, to establish the inequality we need to show that


1 + cos B + cos C 3 sin (

A
+ )
3 6

()

Now, in order to get(), observe that it is enough to prove the inequality for acute
angles B and C, since if one of them is obtuse or right, its cosine is nonpositive. On the

other hand, cos x is concave on the interval (0, ) , thus by Jensens inequality for concave
2
functions,
A
0+B +C
) = sin ( + )
()
1 + cos B + cos C cos (
3
3 6
This proves our inequality, as from () it follows that
ra 4R sin3 (

A
+ ).
3 6

()

128

Chapter 2. Examples for practice

Example 2.158. Let a, b, c be the side lengths and let s be the semiperimeter of a triangle
ABC. Prove that
a a b b c c
a
b
c
(s c) (s a) (s b) ( ) ( ) ( ) .
2
2
2
First solution
By the weighted AM-GM inequality we have

a+b+c
sa
sb
sc

a
+

b
+

c
sc
sa
sb
1
a
b
c

(
) (
) (
)
= a+b+c ,

a
b
c
a+b+c
2

that is equivalent to

a a b b c c
(s c)a (s a)b (s b)c ( ) ( ) ( ) .
2
2
2

()

Second solution
If the triangle is degenerate, then the LHS is identically zero, while the RHS is nonnegative, the inequality being trivially true. We need to consider thus only non-degenerate
triangles. Dividing by the RHS and taking logarithm, the inequality may be rewritten in
the following equivalent form:
a log (1 +

bc
ca
ab
) + b log (1 +
) + c log (1 +
) 0.
a
b
c

Clearly, g(x) = x log(1 + x) is zero for x = 0, while g (x) = 1

1
. Note that if
1+x
x > 0, then g(x) strictly increases, while if x < 0, then g(x) strictly decreases, or the
maximum of g(x) is 0, occurring iff x = 0. As a consequence, x log(1 + x) for all x > 1,
ba
with equality iff x = 0. The triangular inequality guarantees that 1 <
< 1 for nonc
bc
degenerate triangles, or a log 1 +
b c with equality iff b = c. Adding this inequality
a
to its cyclic permutations we obtain the proposed inequality. Equality holds if and only if,
either a = b = c, or the triangle is degenerate with two equal sides and one side of length
0.
()
Third solution
The inequality is equivalent to
(a + b c)a (b + c a)b (c + a b)c aa bb cc

It suffices to shows that


a ln

a+bc
b+ca
c+ab
+ b ln
+ c ln
0
a
b
c

Since the function f (x) = ln x is concave, by Jensen,s Inequality we get


LHS(2.39) (a + b + c) ln

a+bc+b+ca+c+ab
= (a + b + c) ln 1 = 0
a+b+c

(2.39)

129

2.3. Geometric problems

Hence we are done. The equality holds when a = b = c.

()

Fourth solution
Observe that the inequality we have to prove is equivalent to
(1 +

bc a
ca b
ab c
) (1 +
) (1 +
) 1.
a
b
c

Employing the well known inequality ex 1 + x and setting x =


e

bc
a

Analogously, we have
ca

1+

bc
we get
a

bc
bc a
ebc (1 +
)
a
a

ab c
ca b
ab
) ; e (1 +
) .
(1 +
b
c

Multiplying these three inequalities we get the desired result. Since ex = 1 + x x = 0,


it is clear that the equality holds iff a = b = c.
()
Example 2.159. Consider a point P inside a triangle ABC. Let AA1 , BB1 , CC1 be cevians through P . The midpoint M of BC different from A1 , and T is the intersection
of AA1 and B1 C1 . Prove that if the circumcircle of triangle BT C is tangent to the line

B1 C1 , then BT
M =A
1 T C.
Solution


Note first that BT
M =A
1 T C iff CT M = A1 T B since


BT
M + CT
M =A
1 T B + A1 T C = BT C,

or we may exchange B and C at will without altering the problem. Call S the point
where the tangent to the circumcircle of BT C meets BC. By Pappus harmonic theorem,
CA1 CS
=
, or since SA1 = SB + BA1 = SC CA1 wlog, then
A1 B SB
SA1 SM = SA1

SB + SC
= SB SC.
2

But the power of S with respect to the circumcircle of BT C is ST 2 = SB SC, or the


power of S with respect to the circumcircle of A1 T M is SA1 SM = ST 2 , and ST is the
tangent at T to the circumcircle of A1 T M. Therefore,


BT
M = ST
M ST
B = SA
1 T T CB = CA1 T T CA1 = A1 T C.

The conclusion follows.

()

Example 2.160. Let ABCD be a cyclic quadrilateral and let {U} = AB CD and
{V } = BC AD. The line that passes through V and is perpendicular to the angle bisector
of angle
AUD intersects UA and UD at X and Y , respectively. Prove that
AX DY = BX CY.

130

Chapter 2. Examples for practice

Solution
First, lets denote UC, CY, Y D, BU, XB, AF as a, b, c, d, e and f, respectively.Triangle
is perpendicular to XY.
UXY is isosceles, as bisector of U
Considering this fact, a + b = d + e. Let k be equal to a + b. Consequently, a = k b and
d = k e. As we have a cyclic quadrilateral, we consider the equation:
(k e)(k + f ) = (k b)(k + c)

(bc ef )

(c b + e f )
Now,we apply the Menelaus theorem for triangle AUD with a secant BV :

Moving k to the left side, we get: k =

AV b + c k e

= 1.
DV k b e + f

Then the same operation is done for AUD with a secant XV :


AV c k
= 1.
DV k f
Equilizing both parts gives us:
f b+c ke

= 1.
c kb e+f
Substituting the value of k:
f

b+c
(be)(b+f )
cb+ef

Factorization yields to:

(be)(e+c)
cb+ef

e+f

= 1 f (b + c)(c + e) = c(b + f )(e + f ).

(c f )(be cf ) = 0 (Y D AX)(XB CY Y D AX) = 0.

As Y D AX we get the desired answer:

XB CY = Y D AX.

()

Example 2.161. Prove that in any triangle with sidelenghts a, b, c the following inequality
holds:
b + c c + a a + b (b + c a)(c + a b)(a + b c)
+
+
+
7.
(2.40)
a
b
c
abc
First solution

a+b+c
Let s =
is semiperimeter of a triangle with sidelengths a, b, c. Then, due to
2
triangle inequalites a, b, c < s and setting x = s a, y = s b, z = s c we obtain
a = y + z, b = z + x, c = x + y, s = x + y + z, where x, y, z > 0.

131

2.3. Geometric problems

Thus, original inequality becomes


8xyz
2x + y + z
+
7
y+z
(y + z)(z + x)(x + y)
cyc

(2.40)

8xyz
s+x
+
7
(s x)(s y)(s z)
cyc s x

(s + x)(s y)(s z) + 8xyz 7(s x)(s y)(s z)


cyc

(s + x)(sx + yz) + 8xyz 7(xy + yz + zx xyz)


cyc

2s3 8s(xy + yz + zx) + 18xyz 0

(x + y + z)3 4(x + y + z)(xy + yz + zx) + 9xyz 0

x(x y)(x z) = 0.

(Schure Inequality)

cyc

and we are done.

()

Second solution
We make the well-known substitution a = x + y, b = y + z and c = z + x, where x, y, z > 0.
The inequality becomes equivalent to:
2x + y + z x + 2y + z x + y + 2z
8xyz
+
+
+
7.
y+z
z+x
x+y
(x + y)(y + z)(z + x)

After multiplying both sides by (x+y)(y +z)(z +x), the inequality becomes equivalent
with:
2(x3 + y 3 + z 3 ) + 6xyz = 2xyz xy(x + y) x(x y)(x z) 0
cyc

cyc

which is just Schurs inequality.

()

Third solution
The inequality to be proved may be written in the form

Let x =

b + c a c + a b a + b c (b + c a)(c + a b)(a + b c)
+
+
+
4.
a
b
c
abc

(2.41)

b+ca
c+ab
a+bc
,y =
,z =
; note that x, y, z are positive, since a, b, c
2
2
2
are sides of a triangle. Then a = y + z, b = z + x, c = x + y. We substitute these into (2.41)
and obtain, after multiplying both sides by (x + y)(y + z)(z + x),
x(z + x)(x + y) + y(y + z)(x + y) + z(y + z)(z + x) + 4xyz = 2(x + y)(y + z)(z + x)

which is equivalent to
x3 + y 3 + z 3 xy 2 yz 2 zx2 x2 y y 2 z z 2 x + 3xyz 0.

132

Chapter 2. Examples for practice

This, in turn, is equivalent to


x(x y)(x z) + y(y z)(y x) + z(z x)(z y) 0
which follows by Schurs inequality.
Equality occurs only if x = y = z, only if a = b = c, i.e., only if ABC is equilateral
triangle.
()
Example 2.162. Let ABC be a right triangle with A = 90o . Let D be an arbitrary point
on BC and let E be its reflection in the side AB. Denote by F and G the intersections of
AB with lines DE and CE, respectively. Let H be the projection of G onto BC and let I
be the intersection of HF and CE. Prove that G is the incenter of triangle AHI.
Solution
We will use the trivial fact that if a point P in the interior of XY Z is such that P lies
1
on the bisector of angle Y XZ and Y P Z = 90o + Y XZ then P is the incenter. Note
2
that G lies on segment AF and that AGHCis cyclic, so
AHG = ACG = ACE = CED = EDG = F DG.

If H lies between C and D then DHGF is cyclic, which implies that F DG = F HG.
If D lies between C and H, then HDGF is cyclic, which implies that F DG = F HG.
In both cases, F DG = F HG, so AHG = F HG, and this proves that G lies on the
internal bisector of angle AHF.
In order to prove that G is in the interior of triangle AHI it is enough to show that F
lies between H and I. If D lies between C and H, then angles GF H and CGF are both
obtuse, which implies that rays CE and HF intersect beyond E and F, respectively, as
we wanted to prove. If H lies between C and D, then DHGF is cyclic, so
GF H = GDH = 180o F DG C = 180o ACG C,

where we have used the fact that F DG = ACG, as we proved it before. Therefore,
GF H + CGF = (180o ACG C) + (90o + ACG) > 180o ,

which implies that rays CE and HF intersect beyond E and F, respectively, as we wanted
to prove.
This result and the fact that G lies in the internal bisector of angle AHF implies that G
1
lies on the internal bisector of angle AHI. Note that AGI = 90o +ACG = 90o + AHI,
2
and this completes the proof.
()
Example 2.163. Let ABC be a triangle and let X be the projection of A onto BC. The
circle with center A and radius AX intersects line AB at P and R and line AC at Q and
S such that P AB and Q AC. Let U = AB XS and V = AC XR. Prove that lines
BC, P Q, UV are concurrent.
First solution

133

2.3. Geometric problems

Since X, U, R, and X, V, S, are triplets of collinear points, Menelaus theorem ensures


that
CX BV AS
BX CU AR

= 1 and

= 1.
XC UA RB
XB V A SC
Call Y = BC UV. Therefore, again by Menelaus theorem,
BY UA BV BX2 AR CS c cos2 B 1 + sin C
=

Y C CU V A CX2 BR AS bcos2C 1 + sin B


c c sin B P B QA P B
=
=

,
=
b b sin C CQ CQ AP
where we have applied the Sine Law, and used that
AX = AP = AQ = AR = AS = b sin C = c sin B, BX = c cos B, and CX = b cos C,
where a, b, c are obviously the lengths of the sides opposing vertices A, B, C. Therefore,
by the reciprocal of Menelaus theorem, Y, P, Q are collinear, or BC, P Q, UV meet at Y.
()
Second solution
Denote by Y the intersection of BC and P Q and by Z the intersection of BC and
BY BZ
UV. It would suffice to prove that
=

Y C ZC
From Menelaus theorem for triangle ABC and secants P Q, UV, SX, RX we get:
BY BP AQ BP
=
=
;
Y C AP CQ CQ
BZ BU AV
=

;
ZC AU CV
BU SC BX
=

;
AU AS CX
AV AR BX
=

CV BR CX
Because AX BC, BC is tangent to the circle with radius AX, so:
BX 2 = BP BR and CX 2 = CQ CS.
Multiplying (iii) by (iv), using (i), (ii), and (v) we get:
BZ SC BX AR BX BX 2 SC BP BR SC BP BY
=

=
=
=
.
ZC AS CX BR CX CX 2 BR CQ CS BR CQ Y C

(i)

(ii)

(iii)

(iv)
(v)
()

Example 2.164. Let ABCD be a quadrilateral inscribed into a circle C(O, R) and let
(Oab ), (Obc ), (Ocd ), (Oad ) be the symmetric circles to C(O) with respect to AB, BC, CD, DA,
respectively. The pairs of circles (Oab ), (Oad ); (Oab ), (Obc ); (Obc ), (Ocd ); (Ocd ), (Oad ) intersect again at A1 , B1 , C1 , D1 . Prove that A1 , B1 , C1 , D1 lie on a circle of radius R.

134

Chapter 2. Examples for practice

Solution
We prove that the quadrilateral A1 B1 C1 D1 is congruent to ABCD and the conclusion
follows. Let M, N, P, Q be the reflections of O with respect to BC, AB, DA, and CD,
respectively. Then it is easy to see that A1 , B1 , C1 , D1 are reflections of A, B, C, D with
respect to NP, MN, QM and P Q respectively. We are going to show that AD1 QO is a
rhombus. Indeed, let X, Y be the midpoints of OP and QP respectively. Then X is the
midpoint of AD and Y is the midpoint of P Q. Since XY is median line in OP Q and
ADD1 , we get
AD1 XY OQ and AD1 = 2XY = OQ = R.
Hence AD1 QO is a parallelogram with AD1 = OQ = R. Since OA = R, AD1 QO is a
rhombus. Similarly, BC1 QO is a rhombus, hence BC1 = AD1 = R and BC1 OQ AD1 .
It follows that ABC1 D1 is a parallelogram, so C1 D1 = AB and C1 D1 AB. Working
analogously for other sides of the quadrilateral, we obatin the conclusion.
()

Example 2.165. Let ABC be a triangle and let Ma , Mb , Mc be the midpoints of sides
BC, CA, AB, respectively. Let the feet of the perpendiculars from vertices Mb , Mc in triangle AMb Mc be C2 and B1 ; the feet of the perpendiculars from vertices Ma , Mb in triangle
CMa Mb be B2 and A1 ; the feet of the perpendiculars from vertices Mc , Ma in triangle
BMa Mc be A2 and C1 . Prove that the perpendicular bisectors of B1 C2 , C1 A2 , and A1 B2
are concurrent.
Solution
Let A0 , B0 , C0 be the midpoints of Mb Mc , Ma Mc , Ma Mb and let a, b, c be the perpendicular bisectors of B1 C1 , C1 A2 , A1 B2 , respectively. Since triangle A0 B0 C0 is the complementary triangle of triangle Ma Mb Mc and triangle Ma Mb Mc is the complementray triangle of
ABC, triangles A0 B0 C0 and ABC are homothetic. They have the same centroid G and
at the same time G is the center of homothety with the ratio 1. Let O and O0 be the
centers of their circumcircles.
Thus
1
(1)
GO0 = GO.
4
Quadrilateral Mc Mb B1 C2 is cyclic, having center at A0 , the midpoint of Mb Mc . Thus
A0 B1 = A0 C2 , and triangle A0 B1 C2 is isosceles. This implies that A0 a.
On the other hand Mb Mc BC, thus OA B1 C2 and since a B1 C2 we get OA a.
This means that line a in triangle A0 B0 C0 is homologous to the radius OA of triangle
ABC and so O0 a. Analoguously it can be proved that b and c pass through O0 , the
circumcenter of triangle A0 B0 C0 .
Example 2.166. Let ABC be a triangle with altitudes BE and CF and let M be a
point on its circumcircle. Denote by P the intersection of MB and CF and by Q the
intersection of MC and BE. Prove that EF bisects the segment P Q at the midpoint of
P Q.
Solution
Denote by E1 and F1 intersections of the altitudes BE and CF with the circumcircle
of triangle ABC. Let H be orthocenter of the triangle and D the intersection of the lines
P Q and EF.

135

2.3. Geometric problems

Figure 2.6:
It is well known that E1 is reflection of H w.r.t. point E, so HE = EE1 and
F H BF
E1 HC = HE1 C. So EE1 C EHC F HB then
=
.
EE1 EC
F Q BF
=
.
But ABM = ACM, so F BQ ECP
EP EC
From (i) and (ii) we obtain that

(i)
(ii)

FQ FH FH
F Q EP
=
=

EP EE1 EH
F H EH
Then apply Menelaus Theorem to the triangle QHP and transversal F DE:
F Q HE P D
DQ

=1
= 1.
F H EP DQ
PD

()

Example 2.167. Consider a quadrilateral that is incribed in a circle and circumscribed


about a circle. Prove that the product of its diagonals is a constant.
First solution
Let ABCD be a bicentric quadrilateral and let O and I be the circumcenter and
incenter, respectively. Then we have
P OI and OP =

2R2 OI

R 2 + d2

2S
These two relations imply that the point P is fixed. We have AC BD =
, where
sin

is the angle between diagonals. On the other side we have S = abcd and thus

2 abcd
b
a
AC BD =
=2

cd
sin
sin sin

d
c
=2

AP CP
sin ABP sin DBC

= 4R AP CP .

136

Chapter 2. Examples for practice

But the value 4R AP CP is constant because both 4R and AP CP is constant(power


of the point P which is fixed).
()
Second solution
Let ABCD be a quadrilateral that is inscribed in a circle and circumscribed about a
cirle, then we have
AB = CD, BC = DA.
(I)
According to the Ptolemey theorem, we have

and using (I) we get that

which is constant.

AB CD + BC DA = AC BD
AB 2 + BC 2 = AC BD

()

Example 2.168. Let ABC be an acute triangle with altitudes BE and CF. Points Q and
CQ F R
=
Determine the locus
R lie on segments CE and BF, respectively, such that
QE QE
of the circumcenter of triangle AQR when Q and R vary.
First solution
BP BR EQ
=
=
Clearly, by Thales
Denote by P the point in segment BC such that
P C RF QC
theorem, P R CF AB, and P Q BE AC, or AP is a diameter of the circumcircle
of AQR. As R varies continuously from B to F, and hence Q varies continuously from E
to C, then P varies continuously from B to C, and the locus of the circumcenter of AQR
is clearly the segment joining the midpoints of AB and AC.
()
Second solution (See Figure 2.7).
If Q is a point of CE, the point R can be costructed in the following way:
through point Q draw a line parallel to BE to intersect BC at point U ;
through point U draw a line parallel to CF to intersect AB at point R.
From Thales theorem we have:
CQ CU CU F R
=
,
=
QE UB UB RB
so the point R satisfies the relation:
CQ F R
=

QE RB
The circle C with diameter AU contains Ha , Q, R because
AQU = ARU = AHa U = 90o .
Thus the circumcenter of AQU is the midpoint O1 of AU. This implies that the
required locus is the set of mid-points of the cevians AU, where U is a variable point

137

2.3. Geometric problems

Figure 2.7:
of BC. In other words the locus is the segment joining the midpoints M, N of the sides
AB, AC.
()
Third solution (See Figure 2.8).

Figure 2.8:

so

Let P, P BC, such that P R AB and P Q AC. Then P Q BE and P R CF,


BP EQ
BP BR
=
and =

P C RF
P C QC

138

Chapter 2. Examples for practice

But therefore from the given relation


BC BC
BP BP
=
=

PC P C
P C P C

So P C = P C and P P .
Quadrilateral ARP Q is cyclic (because P RA = P QA = 90o ), so the circumcenter
of the triangle ARQ is midpoint of the segment AP. When Q and R vary, P moves on
BC and ints midpoint always lies on the midline of the triangle ABC.
()

2.3.3

Olympiad problems

Example 2.169. Let ABCD be a quadrilateral and let P be a point in its interior.
Denote by K, L, M, N the orthogonal projections of P onto lines AB, BC, CD, DA, and by
Ha , Hb , Hc , Hd the orthocenters of triangles AKN, BKL, CLM, DMN, respectively. Prove
that Ha , Hb , Hc , Hd are the vertices of a parallelogram.
First solution
We have P K AB and NHa AK, so P K NHa . We also have P NAD, KHa
AN, so P N KHa . Hence, we deduce that the quadrilateral P KHa N is a parallelogram.
Similarly, the quadrilateral P LHb K is also a parallelogram. It implies that NHa is parallel and equal to LHb (both are parallel and equal to P K). Thus means NHa Hb L is a
parallelogram. Similarly, NHd Hc L is a parallelogram. Therefore we conclude that Ha Hb
and Hc Hd are parallel and equal. Hence, Ha , Hb , Hc , Hd are the vertices of a parallelogram,
as desired.
()
Second solution
Note that Ha NKP and Ha KNP , hence Ha KP N is a parallelogram. This implies
that Ha K = P N and Ha KP N. Analogously, it follows that Hd M = P N and Hd MP N.
Thus Ha K = Hd M and Ha KHdM.
Similarly, Hb K = Hc M and Hb KHc M. Therefore, triangles Ha HbK and Hd Hc M are
congruent and with all of its correspondent sides parallel. In particular, Ha Hb is parallel
to Hd Hc . Analogously, we prove that Ha Hd is parallel to Hb Hc and this completes the
proof.
()
Third solution
Let us prove that Ha Hb NL. Denote by Oa , Ob , Oc , Od the circumcircles of triangles
AKN, BKL, CLM, and DMN, respectively, and by Ga , Gb , Gc , Gd the gravity gravity of
these triangles, respectively.
Using following properties:
G is between H and O and OH = 3OG.
Oa is midpoint of AP and Ob is midpoint of BP
we get

Ha Hb = Ha Oa + Oa A + AB + BOb + ObHb


= 3Ga Oa + Oa A + AB + BOb + 3Ob Gb

139

2.3. Geometric problems



Ob B + Ob K + Ob L
AOa + KOa + NOa
+ Oa A + AB + BOb + 3
=3
3
3

= AOa + KOa + NOa + Oa A + AB + BOb + Ob B + Ob K + Ob L


AOa + Oa A KP + KA NP + NA AB + AB
=
+
+
+
2
2
2
2

BOb + Ob B P K + BK P L + BL
+
+
+
2
2
2

KP + P K KA + AB + BK NP + P L NA + AB + BL
=
+
+
+
2
2
2
2

NL NL
=
+
= NL.
2
2

Analogously, Hc Hd NL Ha Hb and Ha Hc KM Hb Hd , hence Ha Hb Hc Hd is a


parallelogram.
()
Example 2.170. Let H be the orthocenter of an acute triangle ABC, and let A , B , C
be the midpoints of sides BC, CA, AB. Denote by A1 and A2 the intersections of circle
C(A , A H) with side BC. In the same way we define points B1 , B2 and C1 , C2 , respectively. Prove that points A1 , A2 , B1 , B2 , C1 , C2 are concyclic.
First solution

Figure 2.9:
Considering the power of A1 with respect to the circumcircle C(O, R) of ABC, we
obtain

BC 2
(2.42)
OA2 R2 = A1 C A1 B = A1 A2
4

140

Chapter 2. Examples for practice

Let A, B, C denote the angles of the triangle. If K is the orthogonal projection of C


onto AB, we clearly have BCK = 90o B, hence the law of cosines in triangle CHA
yields
A H 2 = A C 2 + CH 2 2A C CH cos(90o B) =

BC 2
+ 4OC 2 2BC OC sin B (2.43)
4

where we have used the well-known CH = 2OC .


Observing that OC = R cos C (since C is acute), BC = 2R sin A and A H = A1 A ,
(2.42) and (2.43) readily give
OA21 = R2 + 4R2 cos C(cos C sin A sin B).

Now, cos C sin A sin B = cos(A + B) sin A sin B = cos A cos B so that
OA21 = R2 (1 4 cos A cos B cos C).

Owing to the symmetry of the result, we see that

OA1 = OA2 = OB1 = OB2 = OC1 = OC2


and A1 , A2 , B1 , B2 , C1 , C2 are all on the same circle (with centre O).

()

Second solution

Figure 2.10:
Let Ca , Cb , Cc be the circles with centers A , B , C and radii A H, B H, C H respectively. Denote by K the second intersection point of Cb and Cc , besides H.
Since B C BC and AH BC we have AH B C . Thus AH is the radical axis of
Cb and Cc , so K AH. The power of a point theorem implies
AB1 AB2 = AH AK = AC1 AC2

141

2.3. Geometric problems

so the points B1 , B2 , C1 , C2 are concyclic. Since the axes of B1 B2 and C1 C2 intersect at


the circumcenter O of ABC, it follows that B1 , B2 , C1 , C2 lies on the circle with center
O and radius OB1.
Similarly, we can prove that B1 , B2 , A1 , A2 lies on the circle with center O and radius
OB1 . Therefore A1 , A2 , B1 , B2 , C1 , C2 are concyclic and the proof is complete.
()
Example 2.171. Let a, b, c be the sidelengths of a triangle. Prove that

abc
abc
abc
+
+
a + b + c.
a + b + c
ab+c
a+bc
First solution
We change variables
1
1
1
a = (x + y z), b = (y + z x), c = (z + x y),
2
2
2
or y = a + b, z = b + c, x = a + c.
This sets the inequality as

1
1
1
(x + y)(y + z)(z + x) ( + + ) 2 2(x + y + z)
y
x
z

Squaring we get

2
1
(x + y)(y + z)(z + x) ( + ) 8(x + y + z)2
xy
cyc 2

By AM - GM

(x + y)

whence
xy
2

1
4
(x + y)(y + z)(z + x) ( +
) 8(x + y + z)2
x+y
cyc 2

By clearing the denominators we get

x3 y 2 x3 yz

cym

cym

which follows also by the AM - GM since


done.

(x3 y 2 + x3 z 2 )
x3 yz, and cyclic and we are
2
()

Second solution
Applying H
olders inequality it follows that

a2 (a + b + c)
abc
(
) ( a) .

bc
cyc a + b + c cyc
cyc
It is enough to prove that

a2 (a + b + c)

bc
cyc

a
cyc

142

Chapter 2. Examples for practice

Its easy to verify that the last inequality is equivalent to Schurs inequality
a2 (a b)(a c) 0,
cyc

and were done.

()

Example 2.172. On side BC of triangle ABC consider m points, on CAn points, and
on ABs points. Join the points from the sides AB and AC with the points on side BC.
Determine the maximum number of the points of intersection situated in the interior of
triangle ABC.
First solution
Denote respectively by M, N, S the set of points on sides BC, CA, AB. Since the
problem statement says nothing about the locations of points on these sides, we will
assume that no three lines intersect at a point inside the triangle; otherwise, for each one
of the lines P Q (with P M and Q N S) that concurs at a given point with two
or more other lines, we can move ever so slightly P on BC until it passes through no
point where two other lines concur, thus increasing the number of intersection points (line
P Q will now meet each one of the lines with which it concurred at different intersection
points, instead of at one single intersection point).
Note now that we can establish a bijection between the pairs of pairs of points, one
pair of points taken from M, another taken from N S, and the number of intersection
points in the interior of ABC. Indeed, exactly two lines pass through any intersection
point inside the triangle. Each one of these two lines passes through one point on BC,
and another point either on CA or AB. Note also that the two points on BC cannot
coincide, neither can the points on CA, AB, otherwise either the lines would be the same,
or their intersection point would be on the border, not in the interior, of ABC. We may
thus associate exactly one pair of pairs of points to each intersection point.
Reciprocally, consider any pair of points P, Q M, and any pair of points X, Y N S.
Note that P QX is a triangle contained entirely inside ABC, except for segment P Q and
vertexX, who are on its border. Therefore, Y is outside P QX, since it is a point on the
border of ABC, not on BC, and distinct from X. Similarly, X is outside P QY, P is outside
QXY , and Q is outside P XY , or P, Q, X, Y are the vertices of a convex quadrilateral.
Note that P QnXY is clearly on line BC, while out of P X, QY and P Y, QX, two will be
sides of the quadrilateral and will thus meet outside ABC, and two will be their diagonals
and will meet in the interior of ABC ; no other intersection points of any two lines defined
by these pairs of points may be in the interior of ABC. We may thus associate exactly
one intersection point in the interior of ABC to each such pair of pairs of points.
2 C 2 , this is also the maximum
Since the number of pairs of pairs of points is Cm
n+s
number of intersection points in the interior of ABC.
()
Second solution
First, we notice that the maximum number of points of intersection can be obtained
as soon as the intersections occur between no more than two segments. So we just count
all the possible intersections between the segments. To achieve this we first count the
intersections between the sides BC and CA, then between BC and AB and finally between
2 points, between BC and AB there
AB and CA. Between BC and CA there are Cn2 Cm
2 points, and between AB and CA there are snC 2 points.
are Cs2 Cm
m

143

2.3. Geometric problems

Hence the maximum number of the points of intersection we are looking for is given
by the sum of these three numbers:
2
(Cn2 + Cs2 + sn) .
Cm

()

Example 2.173. Let P be point situated in the interior of a circle. Two variable perpendicular lines through P intersect the circle at A and B. Find the locus of the midpoint of
the segment AB.
First solution
We can assume, without loss of generality, that P = t [0, 1] and the circle C = {z = 1}.
Let A = z = x + iy C then B = w = si(z P ) + P C with some s > 0. Hence
1 = w2 = (t sy)2 + s2 (x t)2

The midpoint of the segment AB is given by M =

In fact, by (2.44),

P
M =
2

(2.44)

(A + B)
. Now we verify that
2

2 P 2
.
2

P 2
) = (x sy)2 + (s(x t) + y)2 = x2 + y 2 + 1 t2 = 2 t2 .
2

2 P 2
P
and radius
.
()
Hence the required locus is a circle with center
2
2
In the general setting, if the circle C has
center at P0 and radius R then the locus is
2R2 P P0
(P0 + P )
and radius

a circle with center


2
2
(2M

Second solution
Let ABCD be a quadrilateral and let M and N be the midpoints of sides AB and
CD, respectively. Using the Median Theorem it is easy to prove that the following relation
holds :
AC 2 + BD 2 + BC 2 + DA2 = AB 2 + CD2 + 4MN 2 .

Let M be the midpoint of the segment AB and let N be the midpoint of the segment OP , where O is the center of the given circle. Applying the relation above in the
quadrilateral ABP O we obtain
AP 2 + R2 + BP 2 + R2 = AB 2 + OP 2 + 4MN 2 .

It is clear that AP 2 + BP 2 = AB 2 , hence we get 4MN 2 = 2R2 OP 2, that is


NM =

1 2
2R OP 2.
2

Since the point N is fixed, it follows that the desired locus is the circle of center N
1 2
2R OP 2.
()
and radius
2

144

Chapter 2. Examples for practice

Example 2.174. Prove that any convex quadrilateral can be dissected into n 6 cyclic
quadrilaterals.
Solution
Any convex quadrilateral is dissected into two triangles by either of its diagonals; any
concave quadrilateral is dissected into two triangles by exactly one of its diagonals; any
crossed quadri- lateral is already formed by two triangles joined at one vertex, and where
two of the sides of each triangle are on the straight line containing two of the sides of the
other.
In triangle ABC, let I be the incenter and D, E, F the points where the incircle
touches respectively sides BC, CA, AB. Clearly, ABC my be dissected into three cyclic
quadrilaterals AEIF, BF ID, CDIE.
In triangle ABC, wlog acute at C, consider the circumcenter O, and take a point O
on the perpendicular bisector of AB that is closer to AB than O. The circle with center
O through A, B leaves C outside, hence it must intersect the interior of segments AC, BC
at E, D, or ABDE is cyclic.
We may then proceed as follows: write n = 3 + 3u + v, where u 1 is an integer
and v {0, 1, 2}. Dissect (any) quadrilateral ABCD in two triangles, then dissect one of
them into three cyclic quadrilaterals. If v 0, dissect the other triangle into one cyclic
quadrilateral and one triangle, and if v = 2, dissect again this latter triangle into one
cyclic quadrilateral and one triangle. After this procedure, we have dissected the original
quadrilateral into 3 + v cyclic quadrilaterals (3, 4, 5 respectively for v = 0, 1, 2) and one
triangle. Dissect now this triangle into u triangles (for example dividing one of its sides
in u equal parts and joining each point of division with the opposite vertex), and dissect
now each one of these u triangles into three cyclic quadrilaterals. We have thus dissected
the original quadrilateral into 3 + v + 3u = n cyclic quadrilaterals.
()
Example 2.175. Let ABC be a triangle and let A1 be a point on the side BC. Starting
with A1 construct reflections in one of the angle bisectors of triangle such that the next
point lies on the other side of the triangle. The process is done in one direction: either
clockwise or counterclockwise. Thus at the first step we construct an isosceles triangle
A1 CB1 with point B1 lying on AC. At the second step we construct an isosceles triangle
B1 AC1 with point C1 on AB. In fact we get a sequence of points A1 , B1 , C1 , A2 , ....
(a) Prove that the process terminates in six steps, that is A1 = A3 .
(b) Prove that A1 , A2 , B1 , B2 , C1 , C2 lie on the same circle.
Solution
Let RM N denote the reflection in the line MN and let I be the incentre of ABC.
(a) As the product of three opposite isometries (reversing the orientation), the isometry R = RBI RAI RCI is opposite as well and since R(I) = I, R must be a reflection in
a line l. Since R(A1 ) = A2 , l must be the perpendicular bisector of the line segment A1 A2
in the general case when A1 A2 (and IA1 if A1 = A2 ). Thus, R = Rl and
RBI RAI RCI RBI RAI RCI = Rl Rl = Id

where Id denotes the identity of the plane. As a result, A3 = Id (A1 ) = A1 .


(b) Since l is the the perpendicular bisector of A1 A2 , we have IB1 = IA1 = IA2 = IC1 .

145

2.3. Geometric problems

Figure 2.11:
Similarly, if l denote the perpendicular bisector of B1 B2 , we have Rl = RCI RBI RAI
and so IC2 = IB2 = IB1 .
In conclusion, IA1 = IA2 = IB1 = IB2 = IC1 = IC2 and the six points A1 , A2 , B1 , B2 , C1 ,
and C2 all lie on a circle with centre I.
()
Example 2.176. Let R and r be the circumradius and the inradius of a triangle ABC
with the lengths of sides a, b, c. Prove that
2 2(
cyc

Solution
Note that
2 2(
cyc

a 2 r
)
b+c
R

a 2
r
a 2 r
) 6 2(
) 4+
b+c
R
R
cyc b + c
2 (3 (
cyc

(b + c)2 a2
r
4+
2
(b + c)
R
cyc

2
Since cos A + cos B + cos C = 1 +

a 2
r
) )4+
b+c
R

1
1
(b + c)2 a2
r
and

and
= 1 + cos A then
R
(b + c)2 4bc
2bc

(b + c)2 a2
(b + c)2 a2
r

2
= 2 (1 + cos A) = 4 +

2
(b + c)
2bc
R
cyc
cyc
cyc

Remark. Let la , lb , lc be angle bisectors of a triangle ABC. Noting that


(b + c)2 a2 ala2
=
(b + c)2
abc

()

146

Chapter 2. Examples for practice

we can rewrite original inequality in such form


ala2 + blb2 + clc2
ala2
r
r
ala2
4 + 2
4+
r(4R + r).
2
R
R
a+b+c
cyc 4Rrp
cyc abc
Example 2.177. Prove that in any convex quadrilateral ABCD,
cos

AB
BC
C D
DA
1
+ cos
+ cos
+ cos
2 + (sin A + sin B + sin C + sin D).
4
4
4
4
2

Solution
We can write

2 + sin A + sin B = 2 + 2 sin

A+B
AB
AB
cos
2 + 2 cos
2
2
2
AB
AB
4 cos
,
= 4 cos2
4
4

with equality iff A+B = 180o and simultaneously A = B, ie, equality holds iff A = B = 90o .
Adding the cyclic permutations of both sides of this resulting inequality, we obtain the
proposed inequality with both sides multiplied by 4. The conclusion follows, equality holds
in the proposed inequality iff ABCD is a rectangle.
()
Example 2.178. Given a convex polygon A1 A2 ...An , n 4, denote by Ri the radius of the
circumcircle of triangle Ai1 Ai Ai+1 , where i = 2, 3, ..., n and An+1 is the vertex A1 . Given
that R2 = R3 = = Rn , prove that the polygon A1 A2 ...An is cyclic.
Solution
That the polygon A1 A2 ...An is convex means that Ai1 Ai Ai+1 is obtuse - if you take
the smallest of the two angles formed at the vertex. Therefore, the circumcenter Oi of
Ai1 Ai Ai+1 is exterior to it. In fact, all circumcenters Oi lie in a zone Z that is exterior
to all triangles Ai1 Ai Ai+1 (where i = 2, 3, ..., n and An+1 is the vertex A1 ) and interior to
the polygon A1 A2 ...An . We have then
O2 A1 = O2 A2 = O2 A3 = R, and O3 A2 = O3 A3 = O3 A4 = R, etc.
We see that O2 is the intersection of two arcs of circle of radius R, centered at A2 and
A3 ; and that O3 is determined in exactly the same way. Therefore, since O2 and O3 lie on
the same side of A2 A3 , O2 and O3 must coincide. Repeating this reasoning, we find that
all circumcenters must coincide in a unique circumcenter O common to all vertices. The
circle C(O, R) is the circumcircle of the polygon; therefore, the latter is cyclic.
()
Example 2.179. Consider a triangle ABC and a point P in its interior. Lines P A, P B, P C
intersect BC, CA, AB at A1 , B1 , C1 , respectively. Prove that
BA1 CB1 AC1 3
+
+
=
BC
CA
AB 2
if and only if at least two of the triangles P AB, P BC, P CA have the same area.

147

2.3. Geometric problems

Solution
Denote the area of triangle XY Z by XY Z. Note that we have
P AB
BA1 ABA1 P BA1 ABA1 P BA1
=
=
=
=
BC
ABC
P BC
ABC P BC
P AB + P AC

Denoting P AB = x, P BC = y, P CA = z we have, from the given condition that


x
y
z
3
+
+
=
z+x x+y y+z 2

This also implies

and so we obtain

x
y
3
z
+
+
=
z+x x+y y+z 2
(
cyc

x
x

)=0
x+y z+x

Which, on simple calculations, is equivalent to with


(x y)(z y)(z x)
=0
(x + y)(y + z)(z + x)

Which is true if and only if x = y or y = z or z = x; i.e when two of triangles


P AB, P AC, P BC have the same area.
()
Example 2.180. A non-isosceles acute triangle ABC is given. Let O, I, H be the circumcenter, the incenter, and the orthocenter of the triangle ABC, respectively. Prove that
> 135o
OIH
Solution
It is relatively well known that

OI 2 = R(R 2r); OH 2 = 9R2 (a2 + b2 + c2 ); IH 2 = 4R2 + 2r 2

a2 + b2 + c2

where as usual R, r denote the circumradius and inradius of ABC with sidelengths a, b, c.
The first two relations may be found through the respective powers of I and H with respect
to the circumcircle, the second one requiring some algebra using the Cosine Law and
expressions of the area of the triangle. The third one may be deduced therefrom, applying
the median theorem to triange OIH, since the nine-point center N is the midpoint of OH
and IN = R r because the incircle and nine-point circle are tangent at the Feuerbach
point. Note that OH 2 IH 2 OI 2 = IH 2 + 2r(R 2r) > 0 because R > 2r and IH 2 > 0,
approaching equality when ABC approaches an equilateral triangle, or OIH will be obtuse
at I.
We may also write after some algebra a2 + b2 + c2 = 8R2 + 8R2 cos A cos B cos C, or since
ABC is acute, a2 + b2 + c2 = 8R2 + 4d for some d > 0. Therefore, OH 2 = R2 4d and
IH 2 = 2r 2 2d, or
OH 2 IH 2 OI 2
Rr r 2 d

=
2OI IH
2 r 2 d R(R 2r)

148

Chapter 2. Examples for practice

1
Assume that this quantity is smaller than or equal to . Then,
2
d2 + (R2 + 2r 2 )d + r 4 < 0,

< 1 = cos 135o , and OIH


> 135o .
clearly impossible. Hence cos OIH
2

()

Example 2.181. Let ABC be a triangle with sidelengths a, b, c and medians ma , mb , mc .


Prove that
1 2 2 2
7(a + b + c ) + 2(ab + bc + ca).
ma + mb + mc
2
Solution
Since

(2(a2 + c2 ) b2 ) (2(a2 + b2 ) c2 )

= 4a4 + 2a2 (b2 + c2 ) + 5b2 c2 2b4 2c4

= 4a4 + 2a2 (b2 + c2 ) + b2 c2 2(b2 c2 )2

= (2a2 + bc)2 2 ((b2 c2 )2 a2 (b c)2 )

= (2a2 + bc)2 2(b c)2 (b + c a)(b + c + a)

(2a2 + bc)2 = 2a2 + bc

4mb mc =

and 4 (m2a + m2b + m2c ) = 3 a2 then denote ma + mb + mc by T, we have


cyc

4T 2 = 3 a2 + 8 mb mc 3 a2 + 2 (2a2 + bc) = 7 a2 + 2 bc.


cyc

cyc

From this inferred

ma + mb + mc

cyc

cyc

cyc

cyc

1 2 2 2
7(a + b + c ) + 2(ab + bc + ca).
2

()

= 20o and side-lengths a, b, c satisExample 2.182. Prove that an acute triangle with A
fying

3
a3 + b3 + c3 3abc = min{b, c}
is isosceles.

sin 80o
sin 160o 1
=

Let

=
; note
8 sin 20o 8
sin 20o
1
, thus, since
that the previous identity yields = 4 cos 20o cos 40o =
2 cos 80o
Solution First, note that cos 20o cos 40o cos 80o =

= 4 cos 20o cos 40o = 1 + 2 cos 20o.

It follows that 3 = 2(1 cos 20o ) = 4 sin 210o, i.e. 2 ( 3) = 1,that is


3 32 + 1 = 0. At this point, we assume without loss of generality that b c.

2.3. Geometric problems

149

Accordingly, the equality from the hypothesis can be rewritten as a3 3abc + b3 = 0,


c
b
which after dividing both sides by a3 becomes x3 3xy + 1 = 0, where x = and y =
a
a
C,
so 90o B
80o C
(as the angles B
and
Now, since b c we have x y and B
o
both add up to 160o . Hence, by the Law of Sines, x sin 80 = . On the other hand,
C
sin 20o
3
2
> 2 and x x 3x + 1 is strictly increasing in (2, +); therefore
0 = x3 3xy + 1 x3 3x2 + 1 3 32 + 1 = 0,

and so it follows that x = y = . This proves that ABC is isosceles, as desired.

()

Example 2.183. In a convex quadrilateral ABCD, AC BD = {E}, AB CD = {F }, and


EF intersects the sides AD and BC at X and Y. Let M and N be the midpoints of AD
and BC, respectively. Prove that quadrilateral BCMX is cyclic if and only if quadrilateral
ADNY is cyclic.
Solution
Applying Menelaus theorem to triangles CDG, ACD and ACG, we find
CF DA GB
AX DF CE
AE CB GD

=1;

=1;

= 1.
F D AG BC
XD F C EA
EC BG DA
Multiplying these three equalities, we find
2GA GD
GA AX GX GA
=
=
, or GX =
,
GD DX GD GX
GA + GD

and since M is the midpoint of AD, 2GM = GA + GD, yielding GX GM = GA GD.


Note therefore that BCMX is cyclic iff GX GM = GB GC iff GB GC = GA GD
iff ABCD is cyclic. Similarly, we find that ADNY is cyclic iff ABCD is cylcic.
The conclusion follows.
()
Example 2.184. Consider a non-isosceles acute triangle ABC such that AB 2 + AC 2 =
2BC 2 . Let H and O be the orthocenter and the circumcenter of triangle ABC, respectively.
Let M be the midpoint of BC and let D be the intersection of MH with the circumcircle
of triangle ABC such that H lies between M and D. Prove that AD, BC, and the Euler
line of triangle ABC are concurrent.
Solution
It is well known that the intersection N of MH with the circumcircle is the same as
the intersection of AO with the circumcircle. Hence AN is a diameter of the circumcircle
and AD MH. Let P Q be the intersection of AD with BC. We have that AH MQ
and MH AQ. Therefore H is the orthocenter of triangle AMQ, so QH AM.

If we prove that OH AM we are done. Let OA be


a , OB be b , OC be
c.

2
2
2
2
It is clear that a = b = c = r . Then:


a + b +
c
b +
c

6OH AM = 6
a
3
2

= 2
a2 b 2
c 2+
a
c +
a b 2 b
c =
a
c +
a b 2 b
c

150

Chapter 2. Examples for practice

= b 2+
c 2
a2+
a
c +
a b 2 b
c

= 2( b
c ) (
a
c )2 (
a b )2


= 2BC 2 AC 2 AB 2 = 2BC 2 AC 2 AB 2 = 0

()

Example 2.185. Let ABCD be a cyclic quadrilateral inscribed in the circle C(O, R) and
let E be the intersection of its diagonals. Suppose P is the point inside ABCD such that
triangle ABP is directly similar to triangle CDP . Prove that OP P E.
Solution
P A P B AB
Since
=
=
, point P is the intersection of two distinct Apollonius circles
P C P D CD
constructed by taking the ratios of the distances to A and C for one, B and D for the
other. If AB = CD, the circles degenerate to straight lines that meet only at one point.
Otherwise, assuming wlog that AB < CD, then the centers of both circles are on the rays
CA, DB from C, B, but not on segments CA or DB respectively. Since both points where
the circles meet are symmetric with respect to the line joining the centers of both circles,
and this line is outside ABCD, then both points cannot be in ABCD simultaneously,
and P is therefore unique.
If AB CD, then ABCD is an isosceles trapezoid, and P = E, or line P E cannot
be defined. Assume henceforth then that AB and CD are not parallel, and call F =
AB CD. Assume furthermore wlog that BC < DA (if BC = DA then ABCD would be
an isosceles trapezium and AB CD, which we are assuming not to be true). Obviously,
line EF contains all points Q such that the distances from Q to lines AB and CD,
d(Q, AB) AB
respectively d(Q, AB) and d(Q, CD), satisfy
=
, since it contains E and
d(Q, CD) CD
passes through the intersection of both lines, and trivially AEB and DEC are similar, or
the altitudes from E to AB and CD are proportional to the lengths of the sides AB and
CD. Therefore, since AP B and CP D are similar, the altitudes from D to AB and CD
are also proportional to AB and CD, or P EF.
We will now show that P is the point the circumcircles of ABE and CDE, and line
EF, meet. Call first P the second point where the circumcircle of ABE and line EF
meet. The power of F with respect to the circumcircle of ABE (which is also the power
of F with respect to the circumcircle of ABCD is then F E F P = F A F B = F C F D.
Therefore, CDP E is also cyclic, and P is also on the circumcircle of CDE. Now, since
ABEP and CDP E are cyclic, then P AB = BEF = P ED = P CD, and similarly
ABP = AEP = CEF = CDP, or indeed P AB and P CD are similar. Note finally
that, if the circumcircles of ABE and CDE where tangent, then
ABE = BEF = DEF = DCE = ABE,
and ABE and CDE are isosceles and similar, or AB CD.
Call now A0 , B0 , C0 , D0 the second points where P D, P C, P B, P A meet the circumcircle of ABCD. Trivially, ACB0 = ECP = EDP = BDA0 , or AB0 = BA0 ,
and similarly AC0 = CA0 , AD0 = DA0 , or AA0 BB0 , AA0 CC0 and AA0 D0 D are isosceles
trapezii, and AA0 BB0 CC0 DD0 . Trivially, the diagonals AD0 and DA0 of AA0 D0 D

151

2.3. Geometric problems

meet at P, which is then in the common perpendicular bisector of AA0 , BB0 , CC0 , DD0 ,
which trivially passes also through O, and A0 B0 C0 D0 is the result of taking the reflection of ABCD with respect to OP. Therefore, OP is the internal bisector of angles
AP A0 , BP B0 , CP C0 and DP D0. Now, BP E = BAE = CDE = CP E, and
P E is the internal bisector of angle BP C = BP B0 , or P E is the external bisector
of angles BP B0 and CP C0 , and hence perpendicular to their internal bisectors, ie, to
OP. The proof is completed.
()
Remark. Note that this solution includes also the way to construct point P, i.e., the
second point where the circumcircles of ABE and CDE meet. If both circles are tangent,
then as shown P = E, and ABCD is an isosceles trapezium with AB CD.
Example 2.186. Let ABC be an acute triangle. Prove that
tan A + tan B + tan C

where p and r are semiperimeter and inradius of triangle ABC, respectively.


Solution
In addition to the usual trig formulas, we will make use of the following two known
results:
T = tan A + tan B + tan C = tan Atan Btan C
and

sin 2A + sin 2B + sin 2C = 4 sin A sin B sin C.

Now,
sin A sin B sin C 1 sin 2A + sin 2B + sin 2C
=
cos A cos B cos C 4
cos A cos B cos C
1
sin A
sin B
sin C
= (
+
+
)
2 cos B cos C cos C cos A cos A cos B

T=

sin B
sin C
sin A
+
+
cos(B C) cos A cos(C A) cos B cos(A B) cos C

sin A
sin B
sin C
+
+
1 cos A 1 cos B 1 cos C
B
C pa pb pc p
A
+
+
=
= cot + cot + cot =
2
2
2
r
r
r
r
and the result follows.

()

Example 2.187. Let C be a circle with center O and let A be a fixed point outside C.
Choose points B and C on C, with AB = AC, such that AO is a symmedian, but not a
median, in triangle ABC. Prove that the circumcircle of triangle ABC passes through a
second fixed point.

152

Chapter 2. Examples for practice

Solution
Let P, Q be the second points where AB, AC respectively intersect C, let C0 be the
circumcircle of ABC with center O0 , let M be the midpoint of BC, and let N the second
point where the internal bisector of angle BAC intersects the circumcircle of ABC.
Since BC is a chord both in C and C0 , its midpoint M clearly lies on line OO0. The point
N is also clearly the midpoint of arc BC, or it is also on line OO0. Hence, M, N, O, O0
are collinear.
Claim: P Q is a diameter of C.
Proof : Triangles ABC and AQP are clearly similar, hence the internal bisector of
angles BAC and QAP is the same. Since AB, AC are the respectively symmetric
lines of AQ, AP with respect to this internal bisector, and AO is a symmedian in triangle
BAC, then it is a median in triangle AP Q. Assume now that P Q is a chord of C that is
not a diameter. Since AO passes through its midpoint and through the center of C, then
AO is the perpendicular bisector of P Q, and since it is also a median, AP Q is isosceles
in A, and so is ABC. We reach a contradiction, hence P Q is a diameter.
()
Example 2.188. Let ABC be a triangle with sides a, b, c and corresponding angles A, B, C.
Prove that the equality A = 3B implies the inequality (a2 b2 )(a b) = bc2 , and determine
whether the converse also holds.
Solution:
By the extended law of sines, a = 2R sin A, b = 2R sin B, c = 2R sin C, where R is the
circumradius of ABC. Thus,
(a2 b2 )(a b) = 8R3 (sin2 A sin2 B)(sin A sin B)
= 8R3 (sin2 3B sin2 B)(sin 3B sin B)
= 8R3 (sin 3B sin B)2 (sin 3B + sin B)
= 8R3 (8 cos2 2B sin2 B sin2 B cos B)
= 8R3 (sin2 (180o 4B))(sin B)
= 8R3 (sin2 C)(sin B) = bc2 .

The converse is false in general; we can also have A = 3B 360o, e.g. for
A = 15o , B = 125o , C = 40o.

()

153

2.4. Analysic problems

2.4
2.4.1

Analysic problems
Junior problems

Example 2.189. Let a0 = a1 = 1 and


an+1 = 1 +

a21 a22 a23


a2
+ + ++ n
a0 a1 a2
an1

for n 1. Find an in closed form.

Solution
Let us prove by induction that an = n!. The result is not difficult to verify for n = 0, 1.
Assume that ak = k! for k n 1. Now
ak+1 = (1 +

By the induction hypothesis,

a2
a2
a21 a22 a23
+ + + + k1 ) + k
a0 a1 a2
ak2
ak1

ak+1 = ak +
which is equivalent to
ak+1 =

a2k
ak
=
(ak1 + ak )
ak1 ak1

k!
((k 1)! + k!) = k(k 1)!(k + 1) = (k + 1)!.
(k 1)!

Thus an = n! for n 1.

()

Example 2.190. Prove that if the polynomial P R[x] has n distinct real zeros, then for
any R the polynomial Q(x) = xP (x) + P (x) has at least n 1 distinct real zeros.
Solution
Let be a1 < a2 < < an be the real roots of P and

P (x) = T (x) (x aj )
n

j=1

where T R[x] has no real roots. Therefore

Q(ai ) = ai P (ai ) + P (ai ) = P (ai ) = T (ai ) (ai aj )


ji

and its clear that in the last product has ni1 negative factors. Since T (x) has constant
sign, it follows that Q(ai )Q(ai+1 ) < 0 for i = 1, ..., n1, and by Intermediate Value Theorem
there exists at least a zero of Q in (ai , ai+1 ). This means that Q has at least n 1 distinct
real zeros.
()

Example 2.191. Let f R R be a function such that f (x) + f (x + y) is a rational


number for all real numbers x and all y > 0. Prove that f (x) is a rational number for all
real numbers x.

154

Chapter 2. Examples for practice

Solution
For each real number x, consider a real number y > 0 and define u, v, w in the following
way
u = f (x) + f (x + y), v = f (x y) + f (x), w = f (x y) + f (x + y)

Since x = (x y) + y and x + y = (x y) + 2y, the numbers u, v, w are rational by


hypothesis. Therefore
1
f (x) = (u + v w)
2
is a rational number and we are done.
()
Example 2.192. Let m 1 and f [m, +) [1, +), f (x) = x2 2mx + m2 + 1.
(a) Prove that f is bijective;
(b) Solve the equation f (x) = f 1 (x);

(c) Solve the equation x2 2mx + m2 + 1 = m + x 1.

Solution
(a) The parabola f (x) = (x m)2 + 1 is symmetric with respect to the axis x = m
thus it is bijective on its domain [m, +). A more detailed proof about injectivity is the
following

f (x1 ) > f (x) (x1 m)2 + 1 > (x m)2 + 1 x1 m > x m x1 m > x m x1 > x

From the calculation above we deduce that the function is increasing as well thus its
image is a subset of the interval [f (m), +) = [1, +). As for the surjectivity we must
solve for any y = 1 the equation

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

(b) The equation f (x) = f 1 (x) may occur only on the bisector of the first and third
quadrant y = x thus we must solve

1 + 2m 4m 3
2
2
2
(x m) + 1 = x x (2m + 1)x + m + 1 = 0 x =
.
2

1 + 2m + 4m 3
The root x =
satisfies x > 1 for any m 1.
2

(c) x2 2mx + m2 + 1 = m + x 1 is the equation f (x) = f 1 (x) whose solution is


precisely that found in (b).

Indeed the inversion of y = (x m)2 + 1 yields x = m + x 1.


()

Example 2.193. Let P R[x] be a nonconstant polynomial and let f R R be a


function with the intermediate value property such that P f is continuous. Prove that f
is continuous.
Solution
Let x be a real number and consider a sequence xn that converges to x. Then the
continuity of P f implies that P (f (xn )) converges to P (f (x)) and so f (xn ) is bounded
(as a nonconstant polynomial is a proper map). We claim that f (xn ) converges. If not,
there are two real numbers a < b and two subsequences yn and zn of xn such that f (yn )

155

2.4. Analysic problems

converges to a and f (zn ) converges to b. Now, take any c (a, b). Then for large enough
n we have f (yn ) < c < f (zn ) and, since f has the intermediate value property, there
exists tn between yn and zn such that c = f (tn ). But since yn and zn converge to x, so
does tn . Moreover, P (c) = P (f (tn )) must converge to P (f (x)) by continuity of P f .
Therefore we find that P (c) = P (f (x)) for all c (a, b), which implies that P is constant,
a contradiction.
So, the sequence f (xn ) converges for all choices of a convergent sequence xn . This
immediately implies that if xn tends to x, then f (xn ) tends to f (x), since one can consider
the sequence x1 , x, x2 , x, ..., which still converges to x (and for which the sequence obtained
after applying f has infinitely many terms equal to f (x)). This obviously implies the
continuity of f .
()
Remark. All we used about P is that it is a proper map and it is non constant on
any nontrivial interval. So, by replacing P with any map g satisfying these two properties,
the conclusion still holds. An interesting question is to find all maps g with the following
property: if f has the intermediate value property and g f is continuous, then f is
continuous.
Example 2.194. A sequence (an )n2 of real numbers greater than 1 satisfies the relation

(n + 1)!
an =
1 +
1
1

)
2 (a2 ) (an1
a2
an1
for all n > 2. Prove that if ak = k for some k 2, then an = n for all n 2.
Solution
1
1
) then for n > 2
Letbn = (a2 ) (an1
a2
an1

(n + 1)!
(n + 1)!
an = 1 +
a2n 1 =
,
2bn1
2bn1

and, since

we have
Letting bn1 =

1 a1n 1
bn
= an
=
,
bn1
an
an

(i)

(n + 1)!
(n + 1)!
bn
=
bn =

bn1 2bn1 an
2an

n!
in (i) for n > 2 give us a2n 1 = (n + 1)an1 . Thus,
2an1
a2n+1 = 1 + (n + 2)an , n 2.

Let ak = k for some k 2. Then an = n for any n k. Indeed, since ak = k and in


supposition an = n, n k we obtain an+1 = 1 + (n + 2)an = 1 + (n + 2)n = (n + 1)2 then by
induction an = n for any n k
If k > 2 then for any 2 < n = k from supposition an = n follows
an1 =

a2n 1 n2 1
=
= n 1.
n+1
n+1
Thus, by induction an = n for any 2 n k. Finally, we obtain an = n, n 2.

()

156

2.4.2

Chapter 2. Examples for practice

Senior problems

Example 2.195. Let k be a nonzero real number. Find all functions f R R such that
f (xy) + f (yz) + f (zx) k[f (x)f (yz) + f (y)f (zx) + f (z)f (xy)]

for all x, y, z R.

3
,
4k

Solution
1
Taking x = y = z = 0, the condition becomes (2kf (0) 1)2 0, or f (0) =
.
2k
1
Take x = y = z = 1, the condition becomes again (2kf (1) 1)2 0, or f (1) =
.
2k
1
Take now y = z = 0, the condition becomes f (x)
.
2k
Take finally x = y and z = 1, the condition becomes
2kf (x2 ) + 8kf (x) 3 + 8k 2 f 2 (x).

But 2kf (x2 ) 1, or (2kf (x) 1)2 0, yielding f (x) =

1
for all real x.
2k

()

1
Example 2.196. Find all continuous functions f on [0, 1] such that f (x) = c if x [0, ]
2
1
and f (x) = f (2x 1) if x ( , 1] , where c is a given constant.
2
Solution
We will show that for every positive integer n

f (x) = c, x [0, 1

1
].
2n

We will induct on n. For n = 1 it is the condition of the problem and therefore is


1
1
true. Now suppose it is true for n. Let us prove it for n + 1. If x [ , 1 n+1 ], then
2
2
1
2x 1 [0, 1 n ] (by induction hypothesis) f (2x 1) = c f (x) = c. Thus f (x) = c
2
if x [0, 1). Because f is continuous we have
f (1) = f ( lim (1
x

Then f (x) = c for x [0, 1].

1
1
)) = lim f (1 n ) = lim c = c.
n
x
x
2
2

()

Example 2.197. Prove that there are sequences (xk )k1 and (yk )k1 of positive rational
numbers such that for all positive integers n and k,

n
1+ 5
(xk + yk 5) = Fkn1 + Fkn
,
2
where (Fm )m1 is the Fibonacci sequence.

157

2.4. Analysic problems

Solution
1
1
Take xk = Lk and yk = Fk where Ln is the n-th Lucas number. Since
2
2
k
k
k
k
1+ 5
1 1+ 5
1 5
1 5
Fk = (
) (
) and Lk = (
) +(
)
2
2
2
2

5
it follows that

kn

n
1+ 5
1+ 5
(xk + yk 5) = (
) = Fkn1 + Fkn
.
2
2

()

Example 2.198. Let f N [0, +) be a function satisfying the following conditions:


(a) f (100) = 10;
1
1
1
(b)
+
++
= f (n + 1), for all nonnegative
f (0) + f (1) f (1) + f (2)
f (n) + f (n + 1)
integers n.
Find f (n) in closed form.
Solution
By the condition (b) we have that

that is

f (n + 1) f (n) =

1
, n N,
f (n) + f (n + 1)

(f (n + 1)) = 1 + (f (n)) = 2 + (f (n 1)) = = n + 1 + (f (0)) .


2

Letting n = 99, by (a)

100 = (f (100)) = 100 + (f (0))


2

that is f (0) = 0 and finally (note that f (n) 0 by hypothesis) f (n) =

Example 2.199. Consider the polynomial

P (x) =
n

1
xk ,
n
+
k
+
1
k=0

with n 1. Prove that the equation P (x2 ) = (P (x))2 has no real roots.
Solution
Suppose there exist a real root t to the equation. Since
P (t2 )

1
> 0,
n+1

it follows that P (t2 ) = (P (t))2 > 0. From Cauchy-Schwarz we get

n
n
1
1
1
(
) (
t2k ) (
tk )
n
+
k
+
1
n
+
k
+
1
n
+
k
+
1
k=0
k=0
k=0
n

n.

()

158

Chapter 2. Examples for practice

which implies that

1
1.
k=0 n + k + 1
n

However, we have

1
1
< (n + 1)
= 1.
n+1
k=0 n + k + 1
n

a contradiction. It follows that the equation P (x2 ) = (P (x))2 has no real roots.

Example 2.200. The sequence (xn )n1 is defined by


Prove that lim nxn = 2.

()

x1 < 0, xn+1 = exn 1, n 1.

Solution
By induction it can be proved that xn < 0 for all n 1. On the other hand, since
ex 1 x for all x R, we get that xn+1 = exn 1 xn , and hence, the sequence increases.
It follows that x1 < xn < 0 for all n 1, and hence, the sequence converges. If l = lim xn ,
then passing to the limit as n in the recurrence relation we obtain that l = el 1
from which it follows that l = 0. We calculate lim nxn by using Cesaro-Stolz lemma. We
have, since
ex 1
x2
lim
= 1 and lim x
= 2,
x0
x0 e 1 x
x
that
xn+1 xn
n
1
= lim
lim nxn = lim 1 = lim 1
1
n xn+1 xn
n
n
n
xn
xn xn+1
x2
exn 1
= (1) 1 2 = 2.
lim xn n
n
xn n e 1 xn

= lim
and the problem is solved.

()

Example 2.201. Find all functions f [0, 2] (0, 1] that are differentiable at the origin
and satisfy f (2x) = 2f 2 (x) 1, for all x [0, 1].
Solution
Let g(x) = arccos f (x) for all x [0, 2]. Then for x [0, 1],

cos g(2x) = f (2x) = 2 cos2 g(x) 1 = cos(2g(x)),

thus g(2x) = 2g(x). Hence, for any x [0, 2],

x
x
x
g(x) = 2g ( ) = 4g ( ) = = 2n g ( n ) for all n 1.
2
4
2
x
g( n)
2
Since g is differentiable at 0, we have lim
x = k for some constant k.
n
2n
Therefore, g(x) = kx for all x [0, 2]. Considering the range of f, we conclude that

f (x) = cos(kx) with < k < .


4
4
Finally, it is easy to verify that all such functions f do satisfy the conditions.
()

159

2.4. Analysic problems

2.4.3

Undergraduate problems

Example 2.202. A polynomial p R[X] is called a mirror if p(x) = p(x). Let


f R[X] and consider polynomials p, q R[X] such that p(x) p (x) = f (x), and q(x) +
q (x) = f (x). Prove that p+q is a mirror polynomial if and only if f is a mirror polynomial.

Solution
It is well known that a polynomial p R[X] has nonzero coefficients only for terms
with even degree of x if and only if p(x) = p(x) for all x; we call such a polynomial an
even polynomial (or polynomial with even symmetry). Similarly, a polynomial p R[X]
has nonzero coefficients only for terms with odd degree of x if and only if p(x) = p(x) for
all x; we call such a polynomial an odd polynomial (or polynomial with odd symmetry).
Lemma 6. p is a mirror polynomial if and only if it is either odd or even.

Proof. if p is either odd or even, it is clearly a mirror. If p is a mirror, then either


p(x) = p(x) for an infinitude of values of x, or p(x) = p(x) for an infinite of values of
x. In either case, either finite-degree polynomial p(x) p(x) or finite-degree polynomial
p(x) + p(x) has an infinite number of real roots, and needs to be thus identically zero,
ie, either p(x) p(x) = 0 for all x (and p is even), or p(x) + p(x) = 0 for all x (and p is
odd), or both (and p is identically zero).
We prove our statement using induction on the degree n of f , which is by definition
equal to the degree of p and q, since the degree of p and q is less than the degree of p
and q, unless p and q, and therefore also f , are constant. By the previous argument, it
is also clearly true that the highest degree of x has the same coefficient in f, p, q. When
p(x) + q(x)
are constant, thus even, thus mirrors. When n = 1,
n = 0, f (x) = p(x) = q(x) =
2
write without loss of generality f (x) = a1 x + a0 with a1 0. Then p(x) = a1 x + a0 + a1 and
q(x) = a1 x + a0 a1 , and p(x) + q(x) = 2f (x) = 2a1 x + 2a0 . Since neither f nor p + q may
be even, and f is odd if and only if a0 = 0 or p + q is odd, then f is a mirror polynomial
if and only if p + q is a mirror polynomial.
Assume now that the proposed result is true for all polynomials of degree less than
n 2, and write without loss of generality f (x) = ak xk where an 0.
n

k=0

Define now polynomials r, s, f, p, q as follows:

where

f (x) = f (x) an xn , p(x) = p(x) r(x), q(x) = q(x) s(x),

Note that

r(x) = an Cnk k!xnk ; s(x) = an Cnk (1)k k!xnk .


n

k=0

k=0

r (x) = an Cnk (n k)k!xnk1 = an Cnl l!xnl = r(x) an xn ;


n

k=0

k=1

s (x) = an Cnl (1)l l!xnl = an xn s(x),


n

k=1

160

Chapter 2. Examples for practice

where we have performed the substitution k = l 1. With this definitions, it is clear


that the degree of f, p, q is less than the degree of f, p, q, while
p(x) p(x) = p(x) p(x) r(x) + r(x) = f (x) an xn = f (x),
q(x) + q(x) = q(x) + q(x) s(x) s(x) = f (x) an xn = f (x),
p(x) + q(x) = p(x) + q(x) an Cnk (1 (1)k ) k!xnk .
n

k=0

From the First two of the last relations, we find that f, p, q satisfy the conditions
given in the problem, and from the third that the only terms that are different in p + q
with respect to p + q are those whose degree has the same parity as n. Therefore, f is a
mirror if and only if f is even or odd, if and only if f is even or odd (we obtain f
by making 0 one coefficient in f which has the same parity of all other nonzero coefficients in f ), if and only if p + q is even or odd (by hypothesis of induction), if and
only if p + q is even or odd (because p + q is obtained by modifying only coefficients
of p+q that have the same parity as n), if and only if p+q is a mirror, and we are done. ()
Second solution
The condition p(x) = p(x) is equivalent to p2 (x) = p2 (x) i.e. (p(x)p(x))(p(x)+
p(x)) = 0. This can happen only when one of the two factors is identically 0, so either
p(x) = p(x), or p(x) = p(x). By comparing the coefficients of the two polynomials,
this can happen if and only if all monomials appearing if f are either of even degree, or
of odd degree. Thus f is a mirror polynomial if and only if f (x) = g(x2 ) or f (x) = xg(x2 )
for some polynomial g.
The next idea is that p and q can be exhibited in a rather explicit form. For example,
p must be unique, because if p1 p1 = p2 p2 , then (p1 p2 ) = (p1 p2 ) and a polynomial
equals its derivative if and only if it is identicallyzero. Then, clearly the sum
p = f (x) + f (x) + f (x) + ...

is finite, as the higher-order derivatives of a polynomials are all eventually zero. Moreover,
p p = (f + f + ...) (f + f + ...) = f.

Thus we have found p, and analogously q = f f +f ... and hence p+q = 2(f +f +...)
(the sum of all derivatives of even order).
We are left to prove that f is a mirror polynomial if and only if f + f + ... is a mirror
polynomial.
If f is a mirror polynomial, then all monomials appearing in f have either even degree, or odd. Since differentiating twice preserves the parity of the degree, all monomials
appearing in f , ... are also all even or odd (according to whether f is even or odd), and
so the sum of all these polynomials has all monomials of even degree, or odd degree, thus
p + q is a mirror polynomial. For the converse, observe that 2f = (p + q) (p + q) . If p + q
is mirror polynomial, then (p + q) is a mirror polynomial of the same type, and hence so
is their diference 2f .

161

2.4. Analysic problems

Remark. It can be proved directly that 2f = (p + q) (p + q) , by differentiating the


initial relations and manipulating, and by the method exposed above of computing p and
q, we can conclude that p + q = 2(f + f + ...).
()
Example 2.203. Let n be an even integer. Evaluate
lim [

x1

First solution
Since n is even we have that
lim [

x1

1
n (xn + 1)

]
2
n
(x 1) (x 1) (x + 1)2

n (xn + 1)
1
n (xn + 1)
1

]
=
lim
[

] = lim T.
2
n
2
2
n
x1 (x 1) (x 1)
x1
(x 1) (x 1) (x + 1)
(x 1)2

Note that for x = 1 we have that


T=
=
=

j
n (xn + 1) (x + 1) n1
j=0 x
j
(x 1)2 (x + 1) n1
j=0 x

n1
j=0 (xn + 1 xj xnj )
=
j
(x 1)2 (x + 1) n1
j=0 x

n1
n1
j=0 (xj 1) (xnj 1) j=1 (xj 1) (xnj 1)
=
j
j
(x 1)2 (x + 1) n1
(x 1)2 (x + 1) n1
j=0 x
j=0 x

j1 i
nj1 k
(x 1)2 n1
x )
j=1 (i=0 x ) (k=0

It follows that

j
(x 1)2 (x + 1) n1
j=0 x

j1
nj1
n1
j=1 (i=0 xi ) (k=0 xk )
=
.
j
(x + 1) n1
j=0 x

n1 ( j1 i ) ( nj1 k )
j=1 i=0 x k=0 x n1
j(n j)
= j=1
lim T = lim

j
x1
x1
2n
(x + 1) n1

j=0 x

and we are done.


Second solution
Let

n(n 1) n(n 1)(2n 1)

n2 1
2
6
=

2n
12

()
fn (x) =

(x2

1
n(xn + 1)

n
1)(x 1) (x + 1)2

Let y = 1 x then, x 1 y 0, and since n is even,


fn (x) =
=

n ((1 + y)n + 1)
1
2
2
n
((1 + y) 1) ((1 + y) 1) y
n (2 + y nk=1 Cnk y k1)
1
2
n
k1
k
2
y (y + 2) (k=1 Cn y ) y

n (2 + y nk=1 Cnk y k1) (y + 2) (nk=1 Cnk y k1)


y 2 (y + 2) (nk=1 Cnk y k1 )

162

Chapter 2. Examples for practice


=
=
=

Therefore

2n + ((n 1)y 2) (n + Cn2 y + Cn3 y 2 + o(y 2 ))


2ny 2 + o(y 2 )

2n + n(n 1)y + Cn2 (n 1)y 2 2n n(n 1)y 2Cn3 y 2 + o(y 2 )


2ny 2 + o(y 2 )

Cn2 (n 1)y 2 2Cn3 y 2 + o(y 2 )


.
2ny 2 + o(y 2 )
lim fn (x) = lim [

x1

y0

Cn2 (n 1)y 2 2Cn3 y 2 + o(y 2 )


]
2ny 2 + o(y 2 )

(n 1)Cn2 2Cn3 n2 1
=
.
2n
12

()

Example 2.204. Let f [0, +) R be a continuous function such that f (0) = 0 and
f (2x) f (x) + x for all x 0. Prove that f (x) x for all x [0, +).
Solution
Note that

x
x
x
x x
x
x x x
f (x) f ( ) + f ( ) + + f ( ) + + + . . .
2
2
4
4 2
8
8 4 2

and after n steps we have

f (x) f (

The limit n +yields


f (x) lim (f (
n+

n
x
x
)
+

n
k
2
k=1 2

n
n
x
x
x
x
)
+
)
=
lim
f
(
)
+
lim
=0+x=x

n
k
n
k
n
n
2
2
k=1 2
k=1 2

where we have used the continuity of f (x) and f (0) = 0 for writing
lim f (

and we are done.

x
x
)
=
f
(
lim
) = f (0) = 0
n 2n
2n

()

Example 2.205. Sequences (xn )n1 and (yn )n1 are defined by x1 = 2, y1 = 1, and xn+1 =
x2n + 1, yn+1 = xn yn for all n. Prove that for all n 1
xn 651
<

yn 250

163

2.4. Analysic problems

Solution
The inequality holds for 1 n 4 because x1 /y1 = 2, x2 /y2 = 5/2, x3 /y3 = 13/5, and
x4 /y4 = 677/260. Moreover, for n 5
xn x2n1 + 1 xn1 1 x4 n 1 677 1
=
=
+
=
+
<
+
yn xn1 yn1 yn1 yn y4 k=5 yk 260 k=5 yk

Hence it suffices to show that

1 651 677
1

250 260 6500


k=5 yk

We have that for n 1

xn x2n1 x4n2 x8n3 22

n1

and,
yn = xn1 yn1 22

n2

This means that yn 22

yn1 22

n2

n1

+2n3

yn1 22

n2

+2n3 ++1

23n = 8n for n 5, and

1
1
1
1
1
k =
=
<

4
78
28672 6500
k=5 8
k=5 yk

yn1 = 22

n1

()

Example 2.206. Let f R R be a strictly increasing invertible function such that for
all x R, f (x) + f 1 (x) = ex 1 for all x R. Prove that f has at most one fixed point.
Solution
If x0 is a fixed point then f (x0 ) = f 1 (x0 ) = x0 and

f (x0 ) + f 1 (x0 ) = 2x0 = ex0 1

which has two solutions: 0 and a some c > 0. So, if f has more than one fixed point then it
has just these two fixed points: 0 and c. Since f is a strictly increasing invertible function
then f 1 is strictly increasing too.
Take x < 0 then f (x) < f (0) = 0, f 1 (x) < f 1 (0) = 0. Moreover f (x) < x otherwise
f (x) > x and x > f 1 (x). In both cases
ex 1 = f (x) + f 1 (x) < x + 0 = x

which is a contradiction because for x < 0 we have that x < ex 1. Therefore c is the only
possible fixed point.
()
Example 2.207. Find all functions f [0, ) [0, ) such that
(a) f is multiplicative
(b) lim f (x) exists, is finite, and different from 0.
x

164

Chapter 2. Examples for practice

Solution
Let lim f (t) = c 0 and let a > 0, then
t

c = lim f (at) = lim f (a)f (t) = f (a) lim f (t) = f (a)c


t

which implies that f (a) = 1. Hence f is identically equal to 1 in (0, +), whereas f (0)
can assume any nonnegative real number. Its trivial to check that such functions verify
the assumptions.
()

Example 2.208. Let n be a positive integer and let


f (x) = xn+8 10xn+6 + 2xn+4 10xn+2 + xn + x3 10x + 1.

Evaluate f ( 2 + 3) .

First solution

Let x = 2 + 3. Then x2 = 5 + 2 6 and more manipulation gives x4 10x2 + 1 = 0.


The polynomial to be evaluated can be written
f (x) = (xn + xn+4 )(x4 10x2 + 1) + x3 10x + 1.

The first term vanishes and we evaluate the remaining trinomial. Using the binomial

3
theorem ( 2 + 3) = 11 2 + 9 3. Finally




3
( 2 + 3) 10 ( 2 + 3) + 1 = 11 2 + 9 3 10 ( 2 + 3) + 1 = 2 3 + 1.

()

Second solution

Note that 2 + 3 is a root of the polynomial




x4 10x2 + 1 = (x 2 3) (x + 2 3) (x + 2 + 3) (x 2 + 3) ,

Because

f (x) = xn+8 10xn+6 + 2xn+4 10xn+2 + xn + x3 10x + 1

= (x4 10x2 + 1) (xn+4 + xn ) + x3 10x + 1


1
1
= (x4 10x2 + 1) (xn+4 + xn + ) + 1
x
x


1
then f ( 2 + 3) = 1 = 2 3 + 1.
2+ 3

()

Example 2.209. Let f [0, +) R be a function such that f (x1 ) + f (x2 ) 2f (x1 + x2 )
for all x1 , x2 0. Prove that
f (x1 ) + f (x2 ) + + f (xn ) nf (x1 + x2 + + xn )

for all x1 , x2 , ..., xn 0.

165

2.4. Analysic problems

Solution
We want to show that

f (xi ) mf ( xi )
m

i=1

i=1

for m 1. This holds trivially for m = 1, and we are given that this holds for m = 2;
assume now that this holds for m = n 1 (with n 3). Then for 1 j n we have, where
unless otherwise indicated all summation indices range from 1 to n, the following.
f (xi ) f (xj ) + (n 1)f ( xi )
n

i=1

i=1,ij

Adding these n inequalities gives

n f (xi ) f (xj ) + (n 1) f ( xi )
n

i=1

j=1

j=1

i=1,ij

Subtracting out the common sum and then dividing by (n 1) gives


f (xi ) f ( xi )
n

i=1

j=1

i=1,ij

We also have, again for 1 j n, the following.

(A)

f (xj ) + f ( xi ) 2f ( xi )
ij

Adding these n inequalities gives

f (xj ) + f ( xi ) 2nf ( xi ) .
j

ij

(B)

Adding inequalities (A) and (B) gives

2 f (xj ) + f ( xi ) 2nf ( xi ) + f ( xi ) .
i

ij

ij

Cancelling the common term on both sides and dividing by 2 we have


f (xj ) nf ( xi ) .
j

which establishes the result for m = n; therefore by induction the result holds for all m 1
as was to be proved. Note by the way that only f (x1 )+f (x2 ) = 2f (x1 +x2 ) for all xi in the
domain of f was used above; that is the result also holds for a wider class of functions.
()
Example 2.210. If P (x) is a polynomial of degree m 1, and P (n), P (n+1), ..., P (n+m)
are integers for some integer n, then m!P (x) is a polynomial with integral coefficients.

166

Chapter 2. Examples for practice

Solution
The result is clearly true for m = 1, since if P (x) = ax + b, then P (n + 1) P (n) = a
must be an integer if P (n), P (n + 1) are integers, hence b = P (n) an must be an integer
too.
Assume that the result is true for a given m, and for any polynomial P (x) of degree
m + 1 with coefficient a 0 for xm+1 , define Q(x) = P (x + 1) P (x). Clearly Q(x) is a
polynomial with degree m and coefficient (m + 1)a for xm .
Note that if P (n), P (n + 1), ..., P (n + m + 1) are integers, so are Q(n), Q(n + 1), ...,
Q(n + m), or applying the hypothesis of induction, (m + 1)!a is an integer.
Note therefore that R(x) = P (x) a(x n)(x n 1)...(x n m) is a polynomial of
mth degree, such that P (x) = R(x) are integers for x = n, n + 1, ..., or by hypothesis of
induction, the coefficients of m!R(x) are integers, hence the coefficients of
(m + 1)!P (x) = (m + 1)!R(x) (m + 1)!a(x n)(x n 1)...(x n m)

are integers too. The conclusion follows.

()

Example 2.211. Prove that there is no polynomial P R[x] of degree n 1 such that
P (x) Q for all x R Q.

First solution
If there were such q polynomial than we could build an injection f RQ {0, 1, 2, ..., n}
Q in the following way: take some t R Q. Let P (t) = y Q. The equation P (x) = y
has k n solutions. Let them be t1 < t2 < ... < tk . Clearly t = ti for some 1 i k n.
Define f (t) = (i, y). It is clear why this function is injective. The set {0, 1, 2, ..., n} Q is
countable, hence Imf must be countable too. Then g R Q Imf, g(x) = f (x) is a
bijection, so g 1 exists, hence R Q is countable, impossible.
()
Second solution
We will prove the statement of problem using induction on the degree n = 1. Suppose
that P (x) = ax + b, where a, b R and a 0, such that P (x) Q for all x R Q. Since
x
x
x + 1, R Q and P (x + 1), P ( ) Q then
2
2
x
a = P (x + 1) P (x) Q and b = 2P ( ) P (x) Q.
2

P (x) b
Q and that contradicts that x R Q.
Hence, x =
a
Let n 2. Suppose that the statement of problem holds for polynomials of degree
m {1, 2, ..., n 1} we should to prove that there is no polynomial P R[x] of degree n
such that P (x) Q for all x R Q. Suppose the opposite
P (x) = a0 xn + a1 xn1 + + an ,

where a0 0, holds P (x) Q for all x R Q.


Since x + 1 R Q then P (x + 1) Q and for P1 (x) = P (x + 1) P (x) holds
1 degP1 (x) < n, P1 (x) Q for any x R Q. Thus we get a contradiction with earlier asumption of the induction, and so we are done.
()

167

2.4. Analysic problems

Third solution
Since R Q is uncountable and Q is countable there is a rational number q Q such
that P (x) = q for an infinite number of x R Q. This contradicts the fact that the
polynomial P, which is not constant, has at most n 1 real solutions.
()
Example 2.212. Let be the plane domain consisting of all interior and boundary points
of a rectangle ABCD, whose sides have lengths a and b.
Define f R, f (P ) = P A + P B + P C + P D. Find the range of f.
Solution
Clearly P A + P C AC, with equality iff P is in segment AC, and similarly
P B + P D BD, with equality iff P is in segment BD. Therefore,

f (P ) AC + BD 2 a2 + b2 ,

with equality iff P is the center of rectangle ABCD.


Assume that the maximum of f occurs for some point P in the interior of ABCD,
and consider the ellipses, passing through P, E1 with foci A and B, and E2 with foci C
and D. Both ellipses intersect at P inside ABCD. Consider one of the points Q where
E1 intersects the perimeter of ABCD. Clearly, Q is outside E2 , or QC + QD > P C + P D,
while QA+QB = P A+P B because P and Q belong to E1 . Since Q , then f (Q) > f (P ),
and the maximum of f cannot occur in the interior of ABCD.
Wlog, P such that f (P ) is maximum, is on AB or on AD.
In the first case, P A + P B = AB, while P C + P D is maximum for the case of the
largest ellipse with foci C and D that may be constructed with some intersection point
in segment AB.
Clearly, this happens when P = A or P = B, for P C + P D = AD + AC. The result is
the same in the
second case by analogous reasoning. Restoring generality, the maximum
of f is a + b + a2 + b2 , attained when P is one of the vertices of ABCD.
As P moves continuously from
the center of ABCD
to one of its vertices, f varies

2
2
2
continously, or the range of f is [2 a + b , a + b + a + b2 ], where as stated above the
maximum occurs at the vertices of ABCD, and the minimum occurs at its center. ()
Example 2.213. Find all monic polynomials P and Q, with real coefficients, such that
P (1) + P (2) + + P (n) = Q(1 + 2 + 3 + + n), for all n 1.
Solution
We have the following choices for P (x) and Q(x)
(a) P (x) = Q(x) = x
(b) P (x) = x3 + bx and Q(x) = x2 + bx for some real b.
Proof of the above statement follows:

Lemma 1: If P (x) and Q(x) satisfy the given condition we must have
P (x) = Q (

x(x 1)
x(x + 1)
) Q(
).
2
2

168

Chapter 2. Examples for practice

Proof: Define,
R(x) = Q (

x(x + 1)
(x + 1)(x + 2)
) Q(
) P (x + 1).
2
2

For any positive integer n we have


R(n) = Q (

(n + 1)(n + 2)
n(n + 1)
) Q(
) P (n + 1)
2
2

= Q(1 + 2 + + (n + 1)) Q(1 + 2 + + n) P (n + 1)

= (P (1) + P (2) + + P (n + 1)) (P (1) + P (2) + + P (n)) P (n + 1)

= 0.

So, R(x) vanishes at all integers. Now R(x) is clearly a polynomial, and hence can
have only finitely many roots if it is not constant, so the the above is possible iff R(x) is
(x + 1)(x + 2)
x(x + 1)
identically 0. Hence, P (x + 1) = Q (
) Q(
) or equivalently
2
2
P (x) = Q (

x(x 1)
x(x + 1)
) Q(
).
2
2

()

Lemma 2: If P (x) and Q(x) satisfy the given condition we must have Q(0) = 0.
x(x + 1)
x(x 1)
Proof: We have P (x) = Q (
) Q(
) . Hence, P (1) = Q(1) Q(0),
2
2
but from the given condition P (1) = Q(1), hence we must have Q(0) = 0.
()

Lemma 3: If P (x) and Q(x) satisfy the given condition the degree of Q is either 1 or 2.
Proof: Let Q(x) = xn + an1 xn1 + + a0 . Then
P (x) = Q (
= [(

x(x + 1)
x(x 1)
) Q(
)
2
2
n

n1

x(x 1)
x(x + 1)
x(x + 1)
) (
) ] + an1 [(
)
2
2
2
+ + a1 x

n1

x(x 1)
)
2

]+

nx2n1
The leading term of the P (x) above is easily seen to be n1 , and, since P (x) is monic,
2
n
we must have n1 = 1.
2
This condition is clearly true for n {1, 2} and note for n 3 we have 2n1 > n, hence
the only possible values for n are 1 and 2.
()

169

2.4. Analysic problems

Since, Q has to be monic, Q(0) = 0 and degree of Q is 1 or 2, hence we must either


have Q(x) = x or Q(x) = x2 + bx, and the corresponding values of P from Lemma 1 have
to be P (x) = x and
2

x(x + 1)
x(x + 1)
x(x 1)
x(x 1)
P (x) = (
) + b(
)(
) b(
) = x3 + bx
2
2
2
2

respectively.
We now prove sufficiency, it is trivial that P (x) = Q(x) = x satisfy the conditions in
the problem. In case Q(x) = x2 + bx and P (x) = x3 + bx, using the well known result that
the sum of cubes of first n positive integers is the square of their sum, we get
P (1) + + P (n) = 13 + + n3 + b(1 + + n)
= (1 + + n)2 + b(1 + + n)
= Q(1 + + n).

Hence proved.

()

Example 2.214. Let f (0, +) [0, +) be a bounded function. Prove that if

x
1

f ( ) = 0 and lim (f (x) 2f (2x)2 ) = 0,


lim f (x)
x0
x0
2
2

then lim f (x) = 0.


x0

First solution
1
f (x) then by condition lim (x) = 0
Let (x) = f (x)2f 2 (2x) and (x) = f (2x)
x0
2
and

1
t
1

f (x)) = lim f (t)


f ( ) = 0
lim (x) = lim (f (2x)
x0
x0
t0
2
2
2

where t = 2x. Since


1
f (2x)
f (x) = (x) f (2x)2pf (x) = (x) f (x) = 4f 2 (2x)8f (2x)(x)+4 2 (x)
2
and 2f (x) = 4f 2 (2x) + 2(x) then

f (x) = 4f 2 (2x) + 2(x) (4f 2 (2x) 8f (2x)(x) + 4 2(x))


= 8f (2x)(x) 4 2 (x) + 2(x).

Since f (x) is bounded and lim (x) = lim (x) = 0 then


x0

x0

lim 8f (2x)(x) = 0, lim (4 2 (x)) = 0, lim 2(x) = 0,


x0

and, therefore, lim f (x) = 0.


x0

x0

x0

()

170

Chapter 2. Examples for practice

Second solution
Substituting x by 2x in the first condition, we obtain
lim [f (2x)
x0

1
f (x)] = 0
2

Since 2f (2x)+ f (x) is bounded by hypothesis, multiplying the limit by this function
we obtain
1
lim (2f (2x)2 f (x)) = 0
x0
2
Adding the second condition yields the proposed result.
()

2.4.4

Olympiad problems

Example 2.215. Let (an )n0 and (bn )n0 be sequences defined by

an+3 = an+2 + 2an+1 + an , n = 0, 1, ..., a0 = 1, a1 = 2, a2 = 3

and

bn+3 = bn+2 + 2bn+1 + bn , n = 0, 1, ..., b0 = 3, b1 = 2, b2 = 1.

How many integers do the sequences have in common?

Solution
Clearly a3 = b3 = 8, while a4 = 16, a5 = 35, a6 = 75, and b4 = 12, b5 = 29, b6 = 61. Note
that for n = 4, 5, 6, an > bn > an1 , or by trivial induction, for any n 3,
an+3 = an+2 + 2an+1 + an > bn+3 = bn+2 + 2bn+1 + bn > an+1 + 2an + an1 = an+2 .

Therefore, since both sequences are clearly strictly increasing, no bn for n 4 may
appear in (an ), and the only values that appear in both sequences are {1, 2, 3, 8}, while
the only ns for which an = bn are n = 1 and n = 3 with a1 = b1 = 2 and a3 = b3 = 8.
()

Example 2.216. Let p and q be odd primes such that q (p 1) and let a1 , a2 , ..., an be
distinct integers such that q(ai aj ) for all pairs (i, j). Prove that
P (x) = (x a1 )(x a2 )...(x an ) p,

is irreducible in Z[x] for n 2.

Solution
If n = 2, and P (x) is not irreducible in Z[x], and since P (x) is monic,
then P (x) = (x r)(x s) for integers r, s. Then, P (a1 ) = (a1 r)(a1 s) = p and
P (a2 ) = (a2 r)(a2 s) = p, where wlog a1 r = 1 and a1 s = p since we may exchange
r and s, and we may exchange x by x without altering the problem. Clearly a2 r 1
because a1 and a2 are distinct. If a2 r = 1, then a1 a2 = 2, not divisible by any
odd prime p. Hence a2 r = p or a2 r = p. In the first case, a1 a2 = 1 p, hence
q(a2 a1 ) = p 1, absurd, or a2 r = p and a2 s = 1. Substitution yields r = a1 1 = a2 p
and s = a1 p = a2 1. Note that r s = p 1 = 1 p, yielding p = 1, absurd, hence P (x) is
irreducible for n = 2. We will assume in the rest of the problem that n 2.

171

2.4. Analysic problems

Assume that P (x)is not irreducible in Z[x].


Then, polynomials Q(x), R(x) Z[x] exist such that Q(x)R(x) = P (x), where wlog
n
. Clearly, Q(ai )R(ai ) = P (ai ) = p
1 deg(Q(x)) deg(R(x)), ie, 1 deg(Q(x))
2
for i = 1, 2, ..., n, where Q(ai ) and R(ai ) are integers, or Q(ai ) {1, 1, p, p} for all
i = 1, 2..., n. Assume now that Q(ai ) takes only two values k1 , k2 {1, 1, p, p} when i
n
takes all possible values between 1 and n. There are thus at least values of i for which
2
n
n
wlog Q(ai ) = k1 . Therefore, Q(x) k1 , which has degree at most , has at least roots,
2
2
or it is identically zero, and Q(x) = k1 , absurd. Therefore, Q(ai ) takes at least three diff
erent values when i takes all possible values between 1 and n. We conclude that distinct
i, j {1, 2, ..., n} exist such that, either Q(ai ) = p and Q(aj ) = 1, or Q(ai ) = 1 and
Q(aj ) = p. Therefore, q(ai aj )(Q(ai ) Q(aj )) = p 1, contradiction.
The result follows.
()
Example 2.217. Let P (x) = a0 xn + a1 xn1 + + an , an 0, be a polynomial with real
coefficients and have n roots such that there is an m with

am
> Cnm .
an

Prove that the polynomial P has at least a zero with the absolute value less than 1.
First solution
If x1 , x2 , ..., xn are the roots of P (x) = 0, by the Vietes formulae
an
am
= (1)m x1 x2 xm ;
= (1)n nx1 x2 xn .
a0
a0

hence

and

am (1)m x1 x2 xm
(1)nm
=
=

an
(1)n nx1 x2 xn
x1 x2 xnm

(1)nm
am
(1)nm
= > Cnm .

x1 x2 xnm
x1 x2 xnm
an
If = min {xi } we have

1in

nm

Cnnm

1
> Cnm = Cnnm > 0
x1 x2 xnm

and then < 1. The proof is completed.


Second solution
The zeros of the polynomial
are {

Q(x) = an xn + an1 xn1 + + a0

1
, k = 1, ..., n} (note that wk 0 because an 0). By Vietas formula
wk

n
1
am
=
an
IJnm kI wk

()

172

Chapter 2. Examples for practice

where Jnm is the set of all subsets of {1, 2, ..., n} such that Jnm = n m. If all zeros
1
1 and for any integer
of P has the absolute value greater or equal than 1 then
wk
m 0...n 1
am
1 = Cnnm = Cnm
an
IJnm
and this contradicts the hypothesis.

()

Example 2.218. Let p be a prime. Prove that p(x) = xp + (p 1)! is irreducible in Z[x].

Solution
Let q be the greatest prime less than p. Note that the coefficients of xp1 , xp2 , ..., x0
are divisible by q. If q 2 does not divide (p1)!, then we are done by Eisensteins Criterion.
Consider the factors of (p 1)! = 1 2 ... q ... (p 1). Clearly, 1 2 ... (q 1) is not divisible
by q. It remains to prove that (q + 1)(q + 2)...(p 1) is not divisible by q. By Bertrands
Postulate, there is at least one prime between q and 2q, p is one such prime. Hence, there
is no multiple of q among (q + 1), (q + 2), ..., (p 1). Therefore q 2 does not divide (p 1)!,
and our proof is complete.
()
Example 2.219. Suppose f R+ R+ is a decreasing function such that for all x, y R+ ,
f (x + y) + f (f (x) + f (y)) = f (f (x + f (y))) + f (y + f (x)).

Prove that f (f (x)) = x.


Solution:
Putting y = x gives

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

Replacing x with f (x) gives

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

Subtracting these two equations gives

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

If f (f (x)) > x, the left side of this equation is negative, so


f (f (x) + f (f (x)) > f (x + f (x))

and f (x) + f (f (x)) < x + f (x), a contradiction. A similar contradiction occurs if


f (f (x)) < x. Thus f (f (x)) = x as desired.
()

Example 2.220. For which a does there exist a nonconstant function f R R such
that
f (a(x + y)) = f (x) + f (y)?

2.4. Analysic problems

173

Solution:
ax
gives
For a = 1, we may set f (x) = x. For any other a, putting y =
1a
f (y) = f (x) + f (y), so f (x) = 0 for all x, which is not allowed. So only a = 1 works.

()

Example 2.221. Let P (x) be a quadratic polynomial with nonnegative coefficients. Show
that for any real numbers x and y, we have the inequality
(P (xy)) = P (x2 )P (y 2).

()

(a0 x2n + + a0 )(a0 y 2n + + a0 ) = (a0 xn y n + + a0 )2

()

Solution:
This actually holds for any polynomial with nonnegative coefficients.
If P (x) = a0 xn + + an x0 , then
by the Cauchy-Schwarz inequality.

174

2.5

Chapter 2. Examples for practice

Problems of Other Topics

2.5.1

Junior problems

Example 2.222. Consider n, (n 6) circles of radius r < 1 that are pairwise tangent
and all tangent to a circle of radius 1. Find r.
Solution
We will assume that the problem statement means that n circles or radius r are all
tangent to a given circle of radius 1, and each circle of radius r is tangent to another two
circles of radius r, since it is well known that at most four circles can be tangent to each
other, and at most three of them would have the same radius, since either one of the four
circles contains the other three inside them, or one of them is in the gap formed by the
other three.
With this assumption, clearly all centers of the circles with radii r are at a distance
1 + r from the center of the circle with radius 1, and at a distance 2r from their closest
neighbours of radius r, and by cyclic symmetry, these n centers are at the vertices of a
regular n-gon, with sidelength 2r and circumradius 1 + r, hence considering the isosceles
triangle with equal sides of length 1 + r and sidelength 2r for the other side, where the
180o
360o
, we have r = (1 + r) sin
, or equivalently
different angle is
n
n
180o
n
r=

180o
1 sin
n
sin

Example 2.223. Evaluate


T = 12 + 22 + 32 42 52 + 62 + 72 + 82 92 102 + 20102 + 20112 ,
where each three consecutive signs + are followed by two signs .
Solution
Notice that

Hence

(5k 4)2 + (5k 3)2 + (5k 2)2 (5k 1)2 (5k)2 = 25k 2 80k + 28.
T = 20112 + 25 k 2 80 k + 28 1
= 20112 + 25

402

402

402

k=1

k=1

k=1

402 403 805


40 402 403 + 28 402 = 57736735.
6

Example 2.224. Consider the sequences are given


(a) (an )nN 1, 1, 2, 1, 2, 3, 1, 2, 3, 4, ..., 1, 2, 3, ..., p 1, p, ...
(b) (bn )nN 1, 2, 1, 3, 2, 1, 4, 3, 2, 1, ..., p, p 1, p 2, ..., 2, 1, ...
How many of the first 2012 terms of these sequences are equal?

()

175

2.5. Problems of Other Topics

Solution
For every k N , there exists a unique nk N such that
1
1
nk (nk 1) < k nk (nk + 1).
2
2

Its not hard to prove that

1
1
ak = k nk (nk 1) and bk = nk (nk + 1) + 1 k,
2
2

for all k N . Equal terms of the sequences satisfy ak = bk for a k N .


Using the above formulae about ak and bk , this is equivalent with 2k = n2k + 1. We
conclude that ak = bk if and only if there exists a positive integer n such that 2k = n2 + 1.
Clearly, n must be odd, and so there should exist a m such that 2k = (2m + 1)2 + 1,
and thus k = 2m2 + 2m + 1.
We find that the only equal terms are those with index
212 +21+1 = 5, 222 +22+1 = 13, 25, ..., 2312 +231+1 = 1985, 2322 +232+1 = 2113 > 2012.
The total number of equal terms is therefore 31.

n2 n, if 4 divides n2 n
Example 2.225. Let an =
2

n n , otherwise.
Evaluate S = a1 + a2 + ... + a2012 .

()

First solution
It is well known that 4 divides n2 n if and only if n = 0(mod4) or n = 1(mod4). Hence
the sum is
S = ((12 1) + (2 22 )) + ((3 32 ) + (42 4)) + + ((2007 20072) + (20082 2008)).
But n2 n + (n + 1) (n + 1)2 = 2n and so

S = 21+23++22011 = 2((1+3)+(5+7)++(2005+2007)+(2009+2011)) = 2012. ()

Second solution
S = an = (a4k+1 + a4k+2 + a4k+3 + a4k+4 )
2012

502

k=1

k=0

= [(4k + 1)2 (4k + 1) + 4k + 2 (4k + 2)2 + 4k + 3 (4k + 3)2 +


502

k=0

+(4k + 4)2 (4k + 4)]

= [16k 2 + 4k 16k 2 12k 2 16k 2 20k 6 + 16k 2 + 28k + 12]


502

k=0

= 4 = 4 503 = 2012.
502

k=0

()

176

Chapter 2. Examples for practice

Example 2.226. Mister Tien has a box with n not necessarily equal matches. He is able
to construct with them a cyclic n-gon. Mister Tien then constructs other cyclic n-gons
with these matches. Prove that all of them have the same area.
First solution
Two cyclic n-gons P and P1 with the same sides, but not necessarly in the same
order, are inscribed in circles with the same radius and therefore they have the same area
(because it is equal to the sum of the areas of the n isosceles triangles whose bases are
the sides and the two other sides are equal to the radius).
Assume by contradiction that the radius R1 of the circle of P1 is larger than the radius
R of the circle of P . On the sides of P1 we construct the arcs of the circle in which
the P is inscribed (the arcs do not overlap because P1 is convex). We get a closed curve
of lenght 2R which contains an area larger than R2 . Since among all closed curves
of the same lenght, the circle is the one with the largest area, we have a contradiction.
Note that when Mister Tien constructs a cyclic n-gon, then keeping the same circle, he can
swap any two adjacent sides and therefore he can obtain any permutation of the sides. ()
Second solution
Let a1 , a2 , ..., an be the n matches. We consider an arbitrary cyclic n-gon of radius R
constructed with these matches. Let i be the central angle that corresponds to the ai .
a2
By the Law of Cosines we have that a2 = 2R2 2R2 cos i . Thus cos i = 1 i 2
2R
a2i
a2i
and we obtain i = arccos (1
) . Let fi (x) = arccos (1 2 ) i = f (R) and let
2R2
2x
g(x) = f1 (x) + f2 (x) + + fn (x), we will prove that g(x) is strictly decreasing.
Indeed, if x < y,
a2i
a2i
a2i
a2i
<
1

arccos
(1

)
>
arccos
(1

) fi (x) > fi (y).


1 2
2x
2y 2
2x2
2y 2

It suffices to add over all of the i. The following equation has an unique solution:

()

g(x) = 2,

since g(x) is continuos and strictly decreasing, the solution is x = R. It follows that all
cyclic n-gons constructed by Mister Tien have the same radius and the area of them is
1
1
1
S = R2 sin 1 + R2 sin 2 + + R2 sin n .
2
2
2

()

Third solution
Note that the triangle flipping operation illustrated by the figure is area pre- serving
(since that part of the n-gon below AC is unchanged, and the that part above, namely the
triangle ABC, still has the same area after being flipped). Because with this operation
we can construct all possible cyclic n-gons, we get that all of them have the same area.
Example 2.227. Characterize triangles with sidelengths in arithmetical progression and
lengths of medians also in arithmetical progression.

177

2.5. Problems of Other Topics

Figure 2.12:
First solution
We prove that only equilateral triangles have the desired property. Let ABC be a
triangle with sides a, b, c and the corresponding medians ma , mb , mc . Assume that a b c.
Then we have ma mb mc . Since the sides and the medians form arithmetic progressions
we have
2b = a + c, 2mb = ma + mc .
(i)
3
It is a known fact that m2a + m2b + m2c = (a2 + b2 + c2 ). and we also know that for reals
4
x, y, z we have the inequality 3(x2 + y 2 + z 2 ) (x + y + z)2 . Hence, we have
1
3 2 2 2
(a + b + c ) = m2a + m2b + m2c (ma + mb + mc )2 .
4
4

From this, using (i) and the relation 4m2b = 2a2 + 2c2 b2 , we obtain the following chain
of inequalities
9 2 2 2
(a + b + c ) (3mb )2 a2 + b2 + c2 4m2 a2 + b2 + c2 2a2 + 2c2 b2 .
4

Hence, 2b2 a2 + c2 (a + c)2 2a2 + 2c2 (a c)2 0.


This means that a = c. Therefore, we have a = b = c and triangle is equilateral.

()

Second solution
We use the habitual notation in a triangle. Without loss of generality a b c. By the
well known Apollonius-formulas (application of Stewarts theorem) for the medians of a
triangle we have
1 2
ma =
2b + 2c2 a2 .
2
and the analoguous ones. We can then deduce that mc mb ma . By the properties of
arithmetic progression we have that
a + c = 2b ;

[1] and ma + mc = 2mb . [2]

By the relations [1] and [2] we have

2b2 + 2c2 a2 + 2a2 + 2b2 c2 = 2 2a2 + 2c2 b2

178

Chapter 2. Examples for practice

which equivalent

a2 + 4b2 + 2 (2b2 + 2c2 a2 )(2a2 + 2b2 c2 ) = 4(2a2 + 2c2 b2 ).

Doing all the subsequent calculations will yeild

c 4
c 3
c 2
c
5 ( ) 8 ( ) + 6 ( ) 8 ( ) + 5 = 0.
a
a
a
a

Then the polynomial P (x) = 5x4 8x3 + 6x2 8x + 5 = (x 1)2 (5x2 + 2x + 5) and the quadratic
c
equation 5x2 + 2x + 5 has no real roots. Then = 1 which means a = b = c and the triangle is
a
equilateral. Finally, it is clear that equilateral triangle satisfies the problem and we are done.
()

Example 2.228. Khoa and Tien play the following game: there are 22 cards labeled 1
through 22. Khoa chooses one of them and places it on a table. Tien then places one of
the remaining cards at the right of the one placed by Khoa such that the sum of the two
numbers on the cards is a perfect square. Khoa then places one of the remaining cards
such that the sum of the numbers on the last two cards played is a perfect square, and so
on. The game ends when all the cards were played or no more card can be placed on the
table. The winner is the one who played the last card. Does Khoa have a winning strategy?
First solution
The winning strategy for Khoa is to choose the card labeled 2 in the first step. Note
that the perfect squares in the play are: 4, 9, 16, 25, 36. We consider the equation 2+m = n2
to obtain that Tien just can choose 7 or 14 in the second step.
If Tien chosses 7, then Khoa chosses 18. Tien cannot play since the equation 18+m = n2
is not solvable, because 7 was chosen before. Thus Khoa is winner (2 - 7 - 18).
If Tien chosses 14, then Khoa have another winning chain (2 - 14 - 11 - 5 - 20 -16 9 - 7 - 18). At each step Tien have no choice other than choosing the card shown in the
chain.
So Khoa has a winning strategy and we are done.
()
Second solution
We claim that Khoa has a winning strategy.
On his first move, let Khoa choose 22. On his subsequent moves, let Khoa choose the
maximum number available to him, such that the sum of his number and the previous
number is a perfect square. It is clear that Tien has two choices for her first move, namely
3 and 14. Let us consider these two cases separately.
Case 1. Tien chooses 14. Thus Tien has only one choice in each of her subsequent
moves if Khoa sticks to his strategy. The sequence of moves are the following: 22, 14, 11,
5, 20, 16, 9, 7, 18. Once Khoa places 18, Tien has to either place 7, or 18, both of which
are impossible.
Case 2. Tien chooses 3. The sequence of moves runs 22, 3, 13, 12, 4. Tien can now
either place 5 or 21.
Case 2. I. Tien plays 5. Then the sequence of moves continues as follows: 5,
20, 16, 9, 7, 18. Once again we reach a state when Tien can make no further move.

179

2.5. Problems of Other Topics

Case 2. II. Tien plays 21. Then the sequence of moves continues as follows:
21, 15, and Tien can now play either 1 or 10.
Case 2.II.1 Tien plays 1. The sequence continues thus: 1, 8, 17, 19, 6, 10. Now
Tien can play either 15 or 6, both of which are not possible as they have been played
earlier.
Case 2.II.2 Tien plays 10. The sequence continues as follows: 10, 6, 19, 17, 8,
1. Now Tien has to play either 3, 8, or 15, none of which is possible as they have been
played earlier.
Thus Khoa can always force a win by sticking to this strategy.
()

5
13

Example 2.229. The squares in the figure


on the left side are labeled 1 through 16 such
that the sum of the numbers in each row and
each column is the same.The positions of 1,
5, and 13 are given. Prove that there is only
one possibility for the number in the darkened
square and find this number.

First solution
Let x the number in the darkened square, and c the sum in each row and each column.
We obtain that
(1 + 2 + + 16) + (1 + 5 + 13) + x = 4c.

That is to say 155 + x = 4c, but 155 3 (mod 4) so x 1 (mod 4) and x could be 1, 5,
9, 13. From this we deduce that x = 9 because 1, 5, 13 are given.
()
Second solution Let x the number in the darkened square, and s the sum in each
row and each column. Clearly, 4s x 1 5 13 = 1 + 2 + + 16 = 56 is a multiple of 4, or
x must have a remainder of 1 when divided by 4. But all integers not exceeding 16 with
remainder 1 modulus 4 have been used except for 9. So x = 9 is the only possible value.
()
Example 2.230. Evaluate
2011
2010
6
5
4
3
.
+ 2009C2012
+ 2008C2012
4C2012
+ 3C2012
2C2012
L = C2012

First solution
From the Binomial Theorem we can write that
2012
2011
3
2
1
0
= (1 + 1)2012 = 0.
C2012
+ C2012
+ C2012
C2012
+ C2012
C2012

Using (1) we get that

2012
2011
3
2
1
0
)
C2012
+ C2012
+ C2012
C2012
+ C2012
L = L + 2 (C2012

2011
3
2
2012
1
0
).
) + (2C2012
+ 2011C2012
+ 3C2012
2C2012
+ 2C2012
= (2C2012

k1
Since kCnk = nCn1
, we can rewrite (2), as

2010
2009
3
2
1
)
+ C2011
+ C2011
C2011
+ C2011
L = 2 + 2012 2 2 + 2012 (C2011

(1)

(2)

(3)

180

Chapter 2. Examples for practice

But as we have that Cnk = Cnnk , (3) can be rewrite as

L = 4 + 4024 + 2008 0 = 4020.

()

(1 + x)2012
Then expanding this out, taking the derivax2
tive, and evaluating the resulting expression at x = 1, we find (since f (1) = 0) that the
sum given in the problem statement is equal to 2012 + 2010 2 = 4020.
()
Second Solution Let f (x) =

2.5.2

Senior problems

Example 2.231. Lan has a pair of scales that display the weight in grams. At step n she
cuts a square of side n from a very large laminated sheet and places it on one of the two
scales. A square of side 1 weighs 1 gram.
(a) Prove that for each integer g Lan can place the laminated squares on the scales
such that after a certain number of steps the difference between the aggregate weights on
the two scales is g grams.
(b) Find the least number of steps necessary to reach a difference of 2010 grams.
Solution
(a) Clearly the weight added to one scale at step n is n2 . Note first that for any integer
n, we have
n2 (n + 1)2 (n + 2)2 + (n + 3)2 = 4,

ie, if by step n1 we have managed to obtain a difference of g, by step n+3 we can obtain
differences g + 4 and g 4, by placing on one scale the squares with sides n, n + 3, and on
the other scale the squares with sides n+1, n+2, choosing which squares go on which scale
according to whether we want the difference to increase or decrease. It therefore suffices to
show that we may obtain differences 1, 2, 3 (difference 0 is present at the initial condition
and may be clearly obtained again after 8 steps, 4 steps to add 4, and 4 steps to substract
4).
Difference 1 is easily obtained after 1 step, and difference 3 after 2 steps, by placing
the squares of sides 1 and 2 in opposite scales (with respective weights 1 and 4). Finally,
difference 2 is obtained by placing in opposite scales the squares with sides 1, 2, 3 (total
weight 14 grams), and the square with side 4 (weight 16 grams). The conclusion to part
(a) follows.
()
(b) It is well known (or easily checked by induction) that
12 + 22 + + n2 =
Since

n(n + 1)(2n + 1)

12 + 22 + + 172 = 1785 < 2010 < 2109 = 12 + 22 + + 182 ,

we need a minimum of 18 steps. In 18 steps the task is impossible, since there are 9 odd
perfect squares among the first 18 perfect squares, or the difference of weight between
both scales will be odd, because one will contain an even number of odd weights and the

181

2.5. Problems of Other Topics

other an odd number of odd weights. A difference of 2010 may be however obtained after
19 steps, since
12 + 122 + + 192 2(152 + 22 + 12 ) =

19 20 39
2(225 + 4 + 1) = 2470 460 = 2010,
6

ie a difference of 2010 grams may be obtained by placing on one of the scales all
squares with sides 3 to 19 inclusive, except for the square with side 15, which is placed on
the other scale together with the squares with sides 1, 2. The minimum is thus 19 steps.
()
Example 2.232. Let a, b, c, d be the complex numbers corresponding to the vertices A, B,
C, D of a convex quadrilateral ABCD. Given that ac = ac, bd = bd and a + b + c + d = 0,
prove that ABCD is a parallelogram.
Solution
Denote a = a (cos + i sin ), a = a being a nonnegative real, and express similarly
b, c, d with respective radii b , c , d and angles , , . Since ac = ac, we conclude that ac
is real, yielding either = or = + . Similarly, = or = + . Note that, if = ,
whereas if = + ,

a + c = (a + c ) cos + i(a + c ) sin ,

a + c = (a c ) cos + i(a c ) sin ,

and similarly for b + d. In either case, a + c is either 0 or collinear with 0, a and c, while
b + d is also either 0 or collinear with 0, b and d. If a + c and b + d are both nonzero, they
are both collinear with 0 since they add up to 0, hence a, b, c, d are collinear, absurd. It
follows that at least one of a + c, b + d, is zero, and clearly
a + c = b + d = 0, or a = c , b = d , = + , = + .
Clearly 0 is the midpoint of diagonals AC and BD, it follows that ABCD is a parallelogram.
()
Example 2.233. On each vertex of the regular hexagon A1 A2 A3 A4 A5 A6 we place a rod.
On each rod we have ai rings, where ai corresponds to the vertex Ai . Taking a ring from
any three adjacent rods we can create chains of three rings. What is the maximum number
of such chains that we can create?
Solution
If by adjacent rods, we understand consecutive rods, we have the following possibilities
for three adjacent vertices:
A1 A2 A3 , A2 A3 A4 , A3 A4 A5 , A4 A5 A6 , A5 A6 A1 , A6 A1 A2 .
Hence, the maximum number of chains is
M = a1 a2 a3 + a2 a3 a4 + a3 a4 a5 + a4 a5 a6 + a5 a6 a1 + a6 a1 a2 .

()

182

Chapter 2. Examples for practice

Example 2.234. Prove that for different choices of signs + and the expression
1 2 3 (4n + 1),

yields all odd positive integers less than or equal to (2n + 1)(4n + 1).
First solution
If we take all signs positive we have

1 + 2 + 3 + + 4n + (4n + 1) = (2n + 1)(4n + 1) = S.

Consider now what happens if on the left hand side we change the sign of exactly
one summand: if we replace 1 by 1 this subtracts 2 from S; if we replace 2 by 2 this
subtracts 4 from S; and so on, up to replacing (4n + 1) by (4n + 1) which subtracts
(8n + 2) from S. Clearly this process can be repeated, leaving the rightmost terms with
the signs changed alone on the next iteration (that is for the first iteration all terms are
initially positive, for the second iteration the (4n + 1) term starts out negative, for the
third iteration the 4n and (4n + 1) terms start out negative, and so on). This gives a
decreasing sequence of consecutive odd positive integers starting with S and ending with
S, and this sequence.
()
Second solution
Clearly, the result is true for n = 1 since 15 = 5 + 4 + 3 + 2 + 1, 13 = 5 + 4 + 3 + 2 1,
11 = 5 + 4 + 3 2 + 1, 9 = 5 + 4 3 + 2 1, 7 = 5 4 + 3 + 2 + 1, 5 = 5 4 + 3 + 2 1,

3 = 54+32+1 and 1 = 543+2+1. Assume now that the result is true for n1. Then,
since (4n + 1) + 4n + (4n 1) + (4n 2) = 16n 2, adding the sum of the remaining 4n 3
elements with all combinations of signs that produce all positive integers between 1 and
(2n1)(4n3), we obtain all odd numbers between 16n1 and 8n2 +6n+1 = (4n+1)(2n+1).
Moreover, subtracting from 16n 2 the same sums of the remaining 4n 3 elements
with the same combinations of signs, we obtain all odd integers between 16n 3 and
8n2 + 10n 3 = (2n 1)(4n 3). Clearly, this last integer is negative for any positive
integer n, so all odd numbers between 1 and (2n + 1)(4n + 1) have thus been generated.
The conclusion follows.
()

Example 2.235. Zeroes are written at every vertex of a regular n-gon. Every minute,
Mister Tien picks a vertex, adds 2 to the number written at that vertex, and subtracts 1
from the numbers written at the two adjacent vertices. Prove that, no matter how long
Tien plays, he will never be able to achieve a configuration in which a 1 is written at one
vertex, a 1 is written at another, and a zero is written everywhere else.
Solution
Number the vertices from 1 to n counterclockwise, and denote xi the number of times
that a vertex has been picked before arriving to a given configuration (ie, to a given
distribution of values in the vertices of the n-gon), and use cyclic notation such that
xn+i = xi . Clearly, the amount assigned to vertex i is 2xi xi1 xi+1 .
Note that we may thus assume wlog that min{xi } = 0, since we may subtract min{xi }
from each xi , resulting in the same configuration since each 2xi xi1 xi+1 does not

183

2.5. Problems of Other Topics

change. Assume that in the ending configuration, x1 = 1, and that xk = 0 for some k 1.
Clearly, since 2xk xk1 xk+1 is equal to either 0 or 1, xk1 and xk+1 are non-positive,
ie xk1 = xk+1 = 0, and 0 is assigned to the k th vertex. Note that we may travel from the
k th vertex to the vertex that has 1 assigned, clockwise or counterclockwise, without going
through the vertex that has 1 assigned. By trivial induction forward or backward, we
conclude that xi = 0 for the vertex that has 1 assigned, absurd since the value that has
1 assigned would have a non-positive value. Hence x1 = 0, and x2 + xn = 1, or wlog by
symmetry since we may number the vertices clockwise instead of counterclockwise, x2 = 1
and xn = 0.
Note therefore that xn1 = 0 or the nth vertex would have a negative value assigned.
Now, traveling backward from the nth vertex, we must encounter the vertex that has 1
assigned. By trivial backwards induction, and since 0 is assigned to all vertices from the
nth down to the vertex that has 1 assigned, we conclude that xi = 0 for the vertex that
has 1 assigned, or its value is non-positive, contradiction. The conclusion follows.
()
Example 2.236. Let x, y, and z be complex numbers such that
(y + z)(x y)(x z) = (z + x)(y z)(y x) = (x + y)(z x)(z y) = 1.

Determine all possible values of L = (y + z)(z + x)(x + y).

Solution
First, we note that x y = 0, since if x y = 0, then 1 = (y + z)(x y)(x z) = 0. Thus
we can divide both sides of the equation (y + z)(x y)(x z) = (z + x)(y z)(y x) by
xy
(y + z)(x z) + (z + x)(y z) = 0 xy + xz zy z 2 + yz + yx z 2 zx = 0 xy = z 2 .
If x = 0, then since x y, x z 0, y, z are non-zero. Then

yz(y z) = (z + x)(y z)(y x) = (x + y)(z x)(z y) = yz(z y),

so since yz = 0, then y z = 0, which is a contradiction. Thus x, y, z are non-zero. Since


y z x
x z
xy = z 2 , we obtain = . Similarly, zx = y 2 , and z = y. Let = = = . Clearly
z y
x y z
y z x
3 = ( ) ( ) ( ) = 1,
x y z

but since 1 (if = 1, then x y = 0), it follows that


2 + + 1 =

3 1
= 0.
1

Thus we have y = x, z = y = x 2 .
Substituting back into the equation (y + z)(x y)(x z) = 1

(x + 2 x)(x x)(x 2 x) = 1 x3 ( + 2 )(1 )2 (1 + ) = 1.

Using the identity 2 + + 1 = 0, we obtain

+ 2 = 1, (1 )2 = 3, 1 + = 2 ,

184

Chapter 2. Examples for practice

so:

x3 (1)(3)( 2 ) = 1 x3 (3) = 1 x3 =

Now, we can compute L = (x + y)(y + z)(z + x)

1
3

L = (x + x)(x + 2x)(x + 2 x) = x3 (1 + )()(1 + )(1 + 2 )


1
1
1
( 2 )()( 2 )() = ( 6 ) =
=
3
3
3

1
Thus, the only possible value of (x + y)(y + z)(z + x) is
3

n
4
Example 2.237. Let an = 2 + 4, n = 2, 3, 4, ... Prove that

1
1
1
1
4
L =
+ +
+
= 8.
a5 a6 a12 a20

()

Solution
We observe that

a5 = 2 4 + 2 5 ; a6 = 2 4 + 2 3 ; a12 = 2 4 + 2 3 ; a20 = 2 4 + 2 10 .
1

Let u = 2 60 and observe that


1

a5 = u15 + u24 = u15 (u9 + 1) ; a6 = u15 + u20 = u15 (u5 + 1) ;

Now,

a12 = u15 + u10 = u10 (u5 + 1) ; a20 = u15 + u6 = u6 (u9 + 1).

L=

u15 (u9

u15 (u5

+ 1)
+ 1)

2
4
= 15 = 8, as desired.
u

2.5.3

u10 (u5

+ 1)

u6 (u9

+ 1)
()

Undergraduate problems

Example 2.238. What is the maximum number of points of intersection that can appear
after drawing in a plane l lines, c circles, and e ellipses?
Solution
the intersection of two circles yields 2 points, so the contribution is 2Cc2 ;
the intersection of a line with a circle or an ellipse yields 2 points, so the contribution
is 2l(c + e);
the intersection of an ellipse with a circle or another ellipse yields 4 points, so the
contri- bution is 4Ce2 + 4ec;
the intersection of two incindent lines yields 1 point so the contribution is Cl2 .
Therefore the final formula is
2Cc2 + 2l(c + e) + 4Ce2 + 4ec + Cl2 .
It is easy to find a configuration of l lines, c circles, and e ellipses with such a number
of intersection points.
()

185

2.5. Problems of Other Topics

Example 2.239. Evaluate

3n2 1
3
2
n=2 (n n)

Solution
Since

1
1
1
1
3k 2 1
= 2 +
2+
3
2
2
(k k)
2k
2(k 1)
2k
2(k + 1)2

we have that

thus the

3k 2 1
1
1
1
1
= 2+

3
2
2
2 2n
2(n + 1)
8
k=2 (k k)
n

Sn =

1
1
1
1
3
3n2 1
= lim Sn = lim ( 2 +
)=
3
2
2
n
n
2 2n
2(n + 1)
8
8
n=2 (n n)

()

Example 2.240. Points on the real axis are colored red and blue. We know there exists
a function f R R+ such that if x, y have distinct color then min{f (x), f (y)} x y.
Prove that every open interval contains a monochromatic open interval.
Solution
We suppose the contrary: there exist an open interval (a, b) such that no contains any
monochromatic open interval. Let r0 (a, b) red (this point exist because in other case the
interval (a, b) is blue; contradiction). In the neighborhood Vr0 = (r0 , r0 + ) with > 0
exists infinitely many blue points, because in other case we will have b1 , ..., bn the only
blue points, so we consider b = min{r0 b1 , ..., r0 bn } and set Wr0 = (r0 2b , r0 + 2b ); the
latter is an monochromatic (red) open interval, contained in (a, b), contradiction. Now we
consider Vr0 = (r0 0 , r0 + 0 ) with 0 < 0 < f (r0 ), let b1 a fixed blue point in this interval;
so we have min{f (r0 ), f (b1 )} = r0 b1 < 0 and hence min{f (r0 ), f (b1 )} = f (b1 ) < 0 .
Now we consider Wb1 = (b1 1 , b1 + 1 ) such that 0 < 1 < f (b1 ) and r0 Wb1 . By analogy
we obtain that f (r0 ) < 1 . Hence f (r0 ) < 1 < f (b1 ) < 0 < f (r0 ) contradiction!
We are done.
()

2.5.4

Olympiad problems

Example 2.241. A circle is divided into n equal sectors. We color the sectors in n 1
colors using each of the colors at least once. How many such colorings are there?
Solution
There exists at least an color repeated by pigeonhole principle, and this color appear
exactly two times, because in other case we have n k sectors remaining and n 2 colors,
impossible, since all colors are used. Let c1 , . . . , cn the n colors, we can enumerate all
colorings counting these that involve each pair, (c1 c1 ), . . . , (cn cn ) because if we have
a coloring with two pairs (ci ci ), (cj cj ) then the n 4 sectors remaining can not be
colored using n 3 colors. By each pair we obtain Cn2 (n 2)! colorings, so in total we have
n!(n 1)
colorings.
()
2

186

Chapter 2. Examples for practice

Example 2.242. Prove that if a 7 7 square board is covered by 38 dominoes such that
each domino covers exactly two squares of the board, then it is possible to remove one
domino after which the remaining 37 cover the board.
Solution
Consider a graph with 49 vertices, representing each vertex one of the squares in
the board, where two vertices are joined by and edge iff their corresponding squares are
covered by the same domino. Assume that the board can be covered by 38 dominoes
so that no matter which domino is removed, at least one square becomes uncovered.
Therefore, no edge in the graph may join two vertices, such that both of them are ends
of other edges. Consider two vertices V1 and V2 , joined by an edge. Clearly, one of them
(wlog V2 ) cannot have any other edges. If V1 has other vertices joined to it through
edges, these must also have no other edges connecting them to further vertices in the
graph, or all points that may be reached from V1 through edges of the graph form a
tree with root V1 and leaves V2 , V3 , ..., no other vertices involved. Clearly, no vertex can
be joined to more than 4 other vertices, otherwise by Dirichlets principle, two dominoes
would cover the same two squares, and one of them could be removed leaving all 49 squares
covered. Therefore, the graph may be decomposed in disjoint subgraphs, each one of them
with k vertices, such that k 1 of them are joined by an edge to the remaining vertex,
no other vertices or edges are present in each subgraph, ie, denoting nk the number of
subgraphs with k vertices (where clearly k = 2, 3, 4, 5) we have 2n2 + 3n3 + 4n4 + 5n5 = 49
and n2 + 2n3 + 3n4 + 4n5 = 38. Assume that n5 = 4 d where d 0, hence n2 + n3 + n4 = 7 + d,
whereas n2 + 2n3 + 3n4 = 22 + 4d > 3n2 + 3n3 + 3n4 , absurd, hence n5 5. Now, on the board,
a graph with k = 5 would represent a cross-shaped pentamino with one square joined to
its four neighbours, hence no more than 2 squares on each side of the board, and none
of its corners, can be covered by a domino involved in each one of these n5 subgraphs. It
follows that the n5 subgraphs cover no more than 33 squares, ie n5 6.
Assume that n5 = 5, then 2n2 + 3n3 + 4n4 = 24 and n2 + 2n3 + 3n4 = 18. Note now
that 0 = 3 24 4 18 = 2n2 + n3 , and since n2 , n3 0, we have n2 = n3 = 0, n4 = 6. Note
however that each corned must be covered by one of the n4 subgraphs because it cannot
be covered by one of the n5 subgraphs, and each n4 subgraph is a T -shaped tetramino,
hence each one of the four corners of the board must be covered by one of the two squares
forming the horizontal bar of the T , leaving one prisoner square, limiting on one side
with the side of the board, and on two sides with sides of this T -shaped tetramino. This
square must therefore be covered, next to each corner square, by another one of the n4
subgraphs, hence n4 8, contradiction. Therefore n5 = 6.
Assume finally that n5 = 6, then 2n2 + 3n3 + 4n4 = 19 and n2 + 2n3 + 3n4 = 14, or
1 = 3 19 4 14 = 2n2 +n3 , or since n2 , n3 0, it follows that n2 = 0, n3 = 1, and n4 = 4. As in
the case of n5 = 5, it follows that one of the four corners of the board must be covered by
the n3 subgraph, while the other three must be covered by n4 subgraphs, yielding n4 6
(two of the n4 subgraphs must be placed on or close to each corner), contradiction again.
The conclusion follows.
It is possible however to cover the board with n5 = 3 pentaminoes, n4 = 7 tetraminoes
and n3 = 2 trominoes (leaving n2 = 0), or the board can be covered by 37 dominoes, such
that no domino can be removed without leaving uncovered squares.
()
Example 2.243. Mister Tien jumps on the real axis, from the origin towards point (1,

187

2.5. Problems of Other Topics

1
times its distance to the point (1, 0), where
pn
prime (p1 = 2, p2 = 3, p3 = 5, ...). Can Mister Tien reach point (1, 0)?

0) such that the length of the nth jump is


pn is the nth

First solution
Let xn be the distance to the point (1, 0) after the nth jump. Hence x0 = 1 and for
n > 0, xn is a strictly decreasing sequence such that
xn = xn1

It is well known that

n
1
xn1
= (1 ) .
pn
pk
k=1

xn = (1
n

k=1

1
1
)
= 0,
pk
(1)

so, after a finite number of jups, Mister Tien can arrive arbitrarily close to (1, 0) but it
will never reach it.
()
Second solution
pn 1
times the distance
After the nth jump, the distance from (1, 0) will be at least
pn
before the n th jump, hence always positive. Mister Tien never reaches point (1, 0). ()

Example 2.244. Let G be a graph with n 5 vertices. The edges of G are colored in two
colors such that there are no monochromatic cycles C3 and C5. Prove that there are no
3
more than n2 edges in the graph.
8
Solution
3
Assume that the number of edges E is greater than n2 . By Turans Theorem,
8
n2
1
3
(1
)
E > n2 =
8
2
51

implies that there is a K5 in G. The 10 edges of K5 are colored in two colors, and the
degree of each vertex is 4. If a vertex has at least three monochromatic edges then there
is a monochromatic C3 and we have a contradiction. Hence any vertex has exactly two
edges of one color and two edges of the other one. After a few diagrams it is easy to see
that also in this case K5 contains a monochromatic C3 or a monochromatic C5 .
()
Example 2.245. Numbers 1 through 24 are written on a board. At any time, numbers
a, b, c may be replaced by
2b + 2c a 2c + 2a b 2a + 2b c
,
,

3
3
3
Can a number greater than 70 appear on the board?
Solution

188

Chapter 2. Examples for practice

Let Sn = {x1 , x2 , . . . , x24 } be the numbers written on the board after n iterations.
For any three real numbers a, b, c we have
(n)

(n)

(n)

2b + 2c a 2
2c + 2a b 2
2a + 2b c 2
) +(
) +(
) = a2 + b2 + c2
3
3
3
Therefore the function
(

I(n) = max{a2 + b2 + c2 a, b, c Sn }

is an invariant, i.e. I(n) does not change during the whole process. Since
I(1) = 222 + 232 + 242 = 1589

a number greater than 70 can not appear on the board (after n steps) because we would
have I(n) 702 = 4900 > 1589
This is a contradiction and the result follows.
()
Example 2.246. A triple of different subsets Si , Sj , Sk of a set with n elements is called
a triangle. Define its perimeter by
(Si Sj ) (Sj Sk ) (Sk Si ) .

Prove that the number of triangles with perimeter n is

1 n1
(2 1) (2n 1) .
3

Solution
Let us call Sm , m = 1, 2, ..., n the elements of the n-element set S out of which subsets
S1 , S2 and S3 are chosen. Each triangle may be represented by a n 3 matrix, where
element in position (m, l) equals 1 if sm is in Sl , 0 otherwise, for l = 1, 2, 3. The conditions
of the problem require that:
1) no two columns are equal, otherwise the corresponding subsets would be equal, and
2) each row contains at least 2 1s, otherwise an element of S would not be in any of
the intersections, and the cardinal of the union would be less than n. Provided that these
two conditions are met, it is obvious that the sets are distinct, and that the cardinal of
the union equals the cardinal n of S.
Therefore, each row can take 4 different values, (0, 1, 1), (1, 0, 1), (1, 1, 0) and (1, 1, 1),
for a total of 4n possible combinations. However, some are not allowed, since they would
result in two columns being equal. In fact, columns 1 and 2 are equal if and only if the
only values taken by the rows are (1, 1, 0) and (1, 1, 1), out of which there are a total of
2n possible combinations. We may calculate in the same way the number of permutations
that we need to discard in order to avoid the combinations such that columns 1 and 3
are equal, and columns 2 and 3 are equal. Note however that, out of the 2n combinations
such that two columns are equal, one corresponds to the entire matrix being 1s, and will
appear thus in the three calculations, or the total number of combinations that are not
allowed out of the 4n is 3 (2n 1) + 1 = 3 (2n 2) , for a total of (4n 3) (2n + 2) allowed
combinations.
Finally, note that in counting these combinations, we have counted each triangle six
times, since a permutation of the three columns leaves the triangle unchanged but produces
a different matrix. The total number of triangles with perimeter n is then
4n 3 2n + 2 2((2n1 ) 3 2n1 + 1 (2 2n1 1) (2n1 1)
=
=

6
3
3
2

()

189

2.5. Problems of Other Topics

Example 2.247. Prove that among any ten points located in a circle of diameter 5, there
exist two at distance less than 2 from each other.
Solution
Divide the circle into nine pieces: a circle of radius 1 concentric with the given circle,
and the intersection of the remainder with each of eight equal sectors. Then one checks
that two points within one piece have distance at most 2.
()
Example 2.248. Show that the numbers from 1 to 16 can be written in a line, but not
in a circle, so that the sum of any two adjacent numbers is a perfect square.
Solution
If the numbers were in a circle with 16 next to x and y, then
16 + 1 16 + x, 16 + y 16 + 15,

forcing 16 + x = 16 + y = 25, a contradiction. They may be arranged in a line as follows:


16, 9, 7, 2, 14, 11, 5, 4, 12, 13, 3, 6, 10, 15, 1, 8.

()

Example 2.249. The sides of the acute triangle ABC are diagonals of the squares
K1 , K2 , K3 . Prove that the area of ABC is covered by the three squares.
Solution
Let I be the incenter of ABC. Since the triangle is acute, IAB, IBA < 45o , so the
triangle IAB is covered by the square with diagonal AB, and likewise for IBC and ICA.
()
Example 2.250. The numbers from 1 to 37 are written in a line so that each number
divides the sum of the previous numbers. If the first number is 37 and the second number
is 1, what is the third number?
Solution
The last number x must divide the sum of all of the numbers, which is 37 19; thus
x = 19 and the third number, being a divisor of 38 other than 1 or 19, must be 2.
()
Example 2.251. A regular 1997-gon is divided by nonintersecting diagonals into triangles. Prove that at least one of the triangles is acute.
Solution
The circumcircle of the 1997-gon is also the circumcircle of each triangle; since the
center of the circle does not lie on any of the diagonals, it must lie inside one of the
triangles, which then must be acute.
()
Example 2.252. On a chalkboard are written the numbers from 1 to 1000. Two players
take turns erasing a number from the board. The game ends when two numbers remain:
the first player wins if the sum of these numbers is divisible by 3, the second player wins
otherwise. Which player has a winning strategy?
Solution
The second player has a winning strategy: if the first player erases x, the second erases
1001 x. Thus the last two numbers will add up to 1001.
()

190

Chapter 2. Examples for practice

Example 2.253. 300 apples are given, no one of which weighs more than 3 times any
other. Show that the apples may be divided into groups of 4 such that no group weighs
11
times any other group.
more than
2
Solution
Sort the apples into increasing order by weight, and pair off the heaviest and lightest
apples, then the next heaviest and next lightest, and so on. Note that no pair weighs more
than twice any other; if a, d and b, c are two groups with a b c d, then
a + d 4a 2b + 2c, b + c 3a + d 2a + 2d.

Now pairing the heaviest and lightest pairs gives foursomes, none weighing more than
3
times any other; if e f g h are pairs, then
2
3
3
()
e + h 3e (f + g), f + g 2e + h (e + h).
2
2
Example 2.254. The vertices of triangle ABC lie inside a square K. Show that if the
triangle is rotated 180o about its centroid, at least one vertex remains inside the square.
Solution
Suppose the square has vertices as (0, 0), (0, 1), (1, 1), (1, 0), and without loss of gen1
erality suppose the centroid is at (x, y) with x, y . There must be at least one vertex of
2
the triangle on the same side as (0, 0) of the line through (2x, 0) and (0, 2y); the rotation
of this vertex remains inside the square.
()

Example 2.255. The members of Congress form various overlapping factions such that
given any two (not necessarily distinct) factions A and B, the complement of A B is
also a faction. Show that for any two factions A and B, A B is also a faction.
Solution By putting A = B, we see the complement of any faction is a faction. Thus
for any factions A and B, the complement of A B is a faction, so A B is also.
()
Example 2.256. Show that if 1 < a < b < c, then

loga (loga b) + logb (logb c) + logc (logc a) > 0.

Solution:
Since loga b > 1, loga loga b > logb loga b. Since logc a < 1, logc logc a > logb logc a. Thus the
left side of the given inequality exceeds

Example 2.257.

logb (loga b logb c logc a) = 0.

()

Given a convex polygon M invariant


under a 90o rotation, show that there exist two

circles, the ratio of whose radii is 2, one containing M and the other contained in M.
Solution
Let O be the center of the rotation and A1 a vertex at maximum distance R from O. If
A1 goes to A2 under the rotation, A2 to A3 , A3 to A4 , and A4 to A1 , then A1 A2 A3 A4 is a
R
square with center O contained entirely in M. Thus the circle with radius is contained
2
in the square and thus in M, and the circle with radius R contains M.
()

191

2.5. Problems of Other Topics

Example 2.258. Do there exist real numbers b and c such that each of the equations
x2 + bx + c = 0 and 2x2 + (b + 1)x + c + 1 = 0 have two integer roots?

Solution
No. Suppose they exist. Then b + 1 and c + 1 are even integers, so b and c are odd and
2
b 4c 5 (mod 8) is not a square, a contradiction.
()

Example 2.259. A class consists of 33 students. Each student is asked how many other
students in the class have his first name, and how many have his last name. It turns out
that each number from 0 to 10 occurs among the answers. Show that there are two students
in the class with the same first and last name.

Solution
Consider groups of students with the same first name, and groups of students with
the same last name. Each student belongs to two groups, and by assumption there are
groups of size 1, ..., 11; but these numbers add up to 66 = 2 33, so there is one group of
each size from 1 to 11 and no other groups.
Suppose the group of 11 is a group of students with the same first name. There are at
most 10 groups by last name, so two students in the group of 11 must also have the same
last name.
()
Example 2.260. Of the quadratic trinomials x2 + px + q where p, q are integers and
1 p, q 2012, which are there more of: those having integer roots or those not having
real roots?
Solution
There are more not having real roots. If m n are integer roots of x2 + ax + b = 0, then
m + n = a and mn = b. Assuming 1 a, b 2012, we must have 2012 m, n < 0. But
now the polynomial x2 nx + mn also has integer coefficients between 1 and 2012, and
= n2 4mn = n(n 4m) < 0,
so it has no real roots. Since one can also write down a polynomial having no real roots
not of this form (e.g. x2 3x + 5), there are strictly more polynomials with no real roots.
()
Example 2.261. From an initial triangle A0 B0 C0 a sequence A1 B1 C1 , A2 B2 C2 , ... is
formed such that at each stage, Ak+1 , Bk+1 , Ck+1 are the points where the incircle of Ak Bk Ck
touches the sides Bk Ck , Ck Ak , Ak Bk , respectively.
(a) Express Ak+1 Bk+1 Ck+1 in terms of Ak Bk Ck .
(b) Deduce that as k , Ak Bk Ck 60o .
Solution
(a) We have Ak Bk+1 = Ak Ck+1 by equal tangents so triangle Ak Bk+1 Ck+1 is isosceles
Ak
Ck
with Ak Bk+1 Ck+1 = 90o
. Similarly, Ck Bk+1 Ak+1 = 90o
.
2
2
Adding up angles at Bk+1 , we have
Bk+1 =

Bk
(Ak + Ck )
= 90o
.
2
2

192

Chapter 2. Examples for practice

(b) We have Bk+1 60o = 90o

obviously lim Bk = 60o .


k

Bk
(Bk 60o )
(B0 60o)
60o =
so Bk 60o =
, and
2
2
(2)k
()

Example 2.262. For each parabola y = x2 + px + q meeting the coordinate axis in three
distinct points, a circle through these points is drawn. Show that all of the circles pass
through a single point.
Solution. All such circles pass through the point (0, 1). If (0, q), (r1 , 0), (r2 , 0) are
the three points, then r1 + r2 = p. If (x a)2 + (y b)2 = r 2 is the circle, we then have
p
a = , and
2
p 2
1
1 2
p + (q b)2 = (r1 ) + b2 = (r1 r2 )2 + b2 or q 2 2qb = q
4
2
4

which gives b =

()

q+1
, and so the reflection of (0, q) across the horizontal diameter is (0, 1).
2

Example 2.263. A rectangular grid is colored in checkerboard fashion, and each cell
contains an integer. It is given that the sum of the numbers in each row and the sum of
the numbers in each column is even. Prove that the sum of all numbers in black cells is
even.
Solution. Say the colors are red and black, with the top left square being red. (Since
the sum of all of the numbers is even, it suffices to show the sum of the numbers in the
red squares is even.) The sum of the first (from the top), third, etc., rows and the first
(from the left), third, etc., columns equals the sum of all of the black squares plus twice
the sum of some of the red squares. Since this sum is even, the sum of the numbers in the
black squares is even.
()
49
Example 2.264. On a blackboard are written the numbers
with k = 1, 2, ..., 97. At
k
each step, two numbers a, b are erased and 2ab a b + 1 is written in their place. After 96
steps, a single number remains on the blackboard. Determine all possible such numbers.
Solution. The only such number is 1. Note that 2(2abab+1)1 = (2a1)(2b1),
so the product of 2a 1 over the numbers a on the board never changes. Thus the final
number N must satisfy

and so N = 1.

2N 1 = (2

49
97
96
1
49
1) (2
1) = ( ) ( )( ) = 1
1
97
1
2
97

()

Example 2.265. In a community of more than six people, each member exchanges letters
with precisely three other members of the community. Prove that the community can be
divided into two nonempty groups so that each member exchanges letters with at least two
members of the group he belongs to.

193

2.5. Problems of Other Topics

Solution. Let n be the number of people. Consider a graph whose vertices correspond
to the people, with edges between people who exchange letters. Since each vertex has
degree greater than 1, a cycle must exist. Find a cycle of minimal length and let it have x
people. Put the people in the cycle into group A and the others into group B. Note that
each person in A corresponds with at least 2 other people in A.
If x 5, then A and B satisfy the condition of the problem. If a member of B exchanged
x
letters with two people in A, he and at most + 1 people in A would form a cycle,
2
contradicting our choice of the shortest cycle.
(n + 1)
If x
, which in particular holds for x 4, the following algorithm produces
2
satisfactory groups: as long as there exists a person in B corresponding with at least two
people in A, transfer him into A. It is clear that this gives what we need provided that
B does not end up empty. However, in the original groups, there were at most x edges
between A and B, and each transfer reduces this number by at least 1. So B ends up with
at least n 2x > 0 members, and the proof is complete.
()
Example 2.266. Is it possible to place 100 solid balls in space so that no two of them
have a common interior point, and each of them touches at least one-third of the others?

Solution. Find a sphere S with minimum radius r, and form the sphere X with the
same center but with three times the radius. Any sphere with radius r and touching S
4r 3
is enclosed within X, so the intersection of X and any such sphere is
. Any sphere
3
with a radius larger than r and touching S completely encloses a sphere with radius r
touching S at the same point; thus, the intersection of X and any sphere touching S is
4r 3
. Since the volume of X is 27 times this volume, at most 26 spheres can
at least
3
touch S, which is less than 33, one-third of the other spheres. Therefore, it is not possible
to have such an arrangement.
()
Example 2.267. For any set A of positive integers, let nA denote the number of triples
(x, y, z) of elements of A such that x < y and x + y = z. Find the maximum value of nA
given that A contains seven distinct elements.
Solution. The maximum of 9 is achieved by A = {1, 2, 3, 4, 5, 6, 7}. To see that 9 is a
maximum, consider any 7 numbers
a<b<c<d<e<f <g

how many times each can serve as the middle term; i.e., for each y A how many x, z A
are there such that x < y < z, x + y = z. The answer is 0, 1, 2, 3, 2, 1, 0 times for
a, b, c, d, e, f, g, respectively, giving n(A) 9.
()

Example 2.268. Around a circle are written 2012 zeroes and one 1. The only permitted
operation is to choose a number and change its two neighbors, from 0 to 1 and vice versa.
Is it possible to change all of the numbers to 1? And what if we started with 2013 zeroes?
Solution. This is possible with 2012 zeroes, but not with 2013 zeroes. For 2012
zeroes, group them into 503 groups of 4, then select the second and third zeroes in each
group. As for 2013 zeroes, note that the parity of the sum of the numbers does not change
under the operation, so it cannot go from even (the initial position) to odd (the desired
final position).
()

194

Chapter 2. Examples for practice

Example 2.269. Show that one cannot draw two triangles of area 1 inside a circle of
radius 1 so that the triangles have no common point.
Solution. Consider the problem of drawing a triangle of maximum area in a semicircle
of radius 1. We may assume this triangle has its base along the diameter of the semicircle
(translate it towards the diameter until one vertex hits the diameter, then rotate around
that point until a second vertex hits the diameter), so its base is at most 2 and its height
at most 1. In other words, the triangle has area at most 1, with equality only for the
isosceles right triangle with base along the diameter.
We conclude that two triangles of area 1 are drawn in a circle of radius 1, they each
contain the center of the circle, and so have a common point.
()
Example 2.270. A 4 4 square is divided into 1 1 squares. A secret number is written
into each small square. All that is known is that the sum of the numbers in each row,
each column, and each of the diagonals equals 1. Is it possible to determine from this
information the sum of the numbers in the four corners, and the sum of the numbers in
the four central squares? And if so, what are these sums?
Solution. Let x be the sum of the numbers in the four central squares, y the sum of
the numbers in the four corners, and z the sum of the eight remaining squares. Since the
diagonals contain the four central squares and the four corners, we have x + y = 2. Adding
the second and third rows and columns gives 2x + z = 4; adding the first and fourth rows
and columns gives 2y + z = 4. Hence x = y = 1.
()
Example 2.271. The incircle of a triangle is projected onto each of the sides. Prove that
the six endpoints of the projections are concyclic.

Solution. Each endpoint has distance r 2 from the incenter, where r is the inradius.
So they are concyclic.
()
a + b ab
Example 2.272. Let a b be integers. Prove that
1.
ab

Solution. If a, b are of the same sign, then the fraction has absolute value greater
than 1 and the exponent is positive, so the result is greater than 1. If a, b are of opposite
sign, then the fraction has absolute value less than 1 but the exponent is negative, so
again the result is greater than 1.
()
Example 2.273. The number N is the product of k diferent primes (k 3). Two players
play the following game. In turn, they write composite divisors of N on a blackboard. One
may not write N. Also, there may never appear two coprime numbers or two numbers,
one of which divides the other. The first player unable to move loses. Does the first player
or the second player have a winning strategy?
Solution. The first player has a winning strategy: first write pq for p, q distinct
primes. Every subsequent number must be of the form pn or qn for some n > 1 relatively
prime to pq. If the second player writes pn, the first player then writes qn, and vice versa.
()
Example 2.274. Prove that for x, y, z = 2, (y 3 + x)(z 3 + y)(x3 + z) 125xyz.

2.5. Problems of Other Topics

195

Solution. The left side is at least (4y + x)(4z + y)(4x + y). By weighted AM-GM,
1
4
()
4y + x 5y 5 x 5 ; this and the two analogous inequalities imply the claim.
Example 2.275. Given a convex 50-gon with vertices at lattice points, what is the maximum number of diagonals which can lie on grid lines?
Solution. There can be at most 24 diagonals parallel to any given line (as we move
the line to pass over the vertices, diagonals correspond to pairs passed over at the same
time, but the first and last vertices encountered cannot belong to diagonals). Thus the
maximum is 48, which is easily obtained by making a hexagon with 4 diagonals along grid
lines, then laying rectangles on top of it and subdividing the grid as needed.
()
Example 2.276. The number 9999 (with 1997 nines) is written on a blackboard. Each
minute, one number written on the blackboard is factored into two factors and erased, each
factor is (independently) increased or diminished by 2, and the resulting two numbers are
written. Is it possible that at some point all of the numbers on the blackboard equal 9?
Solution. No, there always is a number congruent to 3 modulo 4: factoring such
a number gives one factor congruent to 1 modulo 4, and changing that by 2 in either
direction gives a number congruent to 3 modulo 4.
()

Chapter 3
Exercises for training
3.1
3.1.1

Exercises
Exercises from Mathematic Reflextion

Exercise 3.1. (O69MR1 - 2008.p41) Find all integers a, b, c for which there is a positive
integer n such that
n

a + bi 3
(
) = c + i 3.
2
Answer.
If n = 1, then we get a = 2c, b = 2, where c is any integer.
If n = 2, then the solutions are (a, b, c) = (1, 1, 1), (1, 1, 1), (2, 0, 1), (2, 0, 1).
The equation is not solvable for n 3.
()

Exercise 3.2. (O70MR1 - 2008.p45) In triangle ABC let Ma , Mb , Mc be the midpoints of


BC, CA, AB, respec- tively. The incircle (I) of triangle ABC touches the sides BC, AC, AB
at points A, B, C. The line r1 is the reflection of line BC in AI, and line r2 is the perpendicular from A to IMa . Denote by Xa the intersection of r1 and r2 , and Define Xb
and Xc analogously. Prove that Xa , Xb , Xc lie on a line that is tangent to the incircle of
triangle ABC.
Exercise 3.3. (O72MR1 - 2008.p52) For n 2, let Sn be the set of divisors of all polynomials of degree n with coefficients in {1, 0, 1}. Let C(n) be the greatest coefficient of
a polynomial with integer coefficients that belongs to Sn . Prove that there is a positive
integer k such that for all n > k,
n2007 < C(n) < 2n .

Exercise 3.4. (J146MR1 - 2010.p3) Let A1 A2 A3 A4 A5 be a convex pentagon and let


X A1 A2 , Y A2 A3 , Z A3 A4 , U A4 A5 , V A5 A1
be points such that A1 Z, A2 U, A3 V, A4 X, A5 Y intersect at P . Prove that
A1 X A2 Y A3 Z A4 U A5 V
=1
A2 X A3 Y A4 Z A5 U A1 V
196

197

3.1. Exercises

Exercise 3.5. (J149MR1 - 2010.p6) Let ABCD be a quadrilateral with A 60o . Prove
that
AC 2 < 2 (BC 2 + CD2 ) .
Hint: Using to Ptolemys inequality and the Cosine Law.

Exercise 3.6. (S148MR1 - 2010.p14) Let n be a positive integer and let a, b, c be real
numbers such that a2 b c2 . Find all real numbers x1 , . . . , xn , y1 , . . . , yn for which
x21 + + x2n + b (y12 + + yn2 ) = c.

Exercise 3.7. (S149MR1 - 2010.p15) Prove that in any acute triangle ABC,
1
r 2
r
r
(1 + ) cos A cos B cos C
(1 ) .
2
R
2R
R

Exercise 3.8. (S150MR1 - 2010.p17) Let A1 A2 A3 A4 be a quadrilateral inscribed in a


circle C(O, R) and circum- scribed about a circle (I, r). Denote by Ri the radius of the
circle tangent to Ai Ai+1 and tangent to the extension of the sides Ai1 Ai and Ai+1 Ai+2 .
Prove that the sum R1 +R2 +R3 +R4 does not depend on the position of points A1 , A2 , A3 , A4 .
Exercise 3.9. (O148MR1 - 2010.p38) Let ABC be a triangle and let A1 , A2 be the intersections of the trisectors of angle A with the circumcircle of ABC. Define analogously
points B1 , B2 , C1 , C2 . Let A3 be the intersection of lines B1 B2 and C1 C2 . Define analogously B3 and C3 . Prove that the incenters and circumcenters of triangles ABC and
A3 B3 C3 are collinear.
Exercise 3.10. (O150MR1 - 2010.p41) Let n be a positive integer, e0 , ..., en1 be the nth
roots of unity, and a, b be complex numbers. Evaluate the product
P = (a + be2k )
n1
k=0

n
n 2

(a 2 (b) 2 )
Answer: P = n n

a + b

if n is even
if n is odd.

()

Exercise 3.11. (J186MR1-2011.p6) Let ABC be a right triangle with AC = 3 and BC = 4


and let the median AA1 and the angle bisector BB1 intersect at O. A line through O
intersects hypotenuse AB at M and AC at N . Prove that
MB NC 4


MA NA 9
Hint: We can to prove the inequality by using analytic geometry.
a2n1
Exercise 3.12. (S183MR1-2011.p10) Let a0 (0, 1) and an = an1
for n 1. Prove
2
that for all n 1,

1 n+1+ n
n 1

<
<
2 an a0
2

198

Chapter 3. Exercises for training

Exercise 3.13. (S184MR1-2011.p11) Let Hn =


eHn >

1 1
1
+ + + , n 2. Prove that
2 3
n

n
n! 2Hn .

Exercise 3.14. (S185MR1-2011.p12) Let A1 , A2 , A3 be non-collinear points on parabola


x2 = 4py, p > 0, and let B1 = l2 l3 , B2 = l3 l1 , B3 = l1 l2 where l1 , l2 , l3 are tangents to the
[A1 A2 A3 ]
parabola at points A1 , A2 , A3 , respectively. Prove that
is a constant and find its
[B1 B2 B3 ]
value.
Answer.
Exercise 3.15. (O183MR1-2011.p31) Evaluate
T = tan 4
2010
k=1

Answer. T = 2011

[A1 A2 A3 ]
= 2.
[B1 B2 B3 ]

()

2011

20113 4 2011 + 3
, using the de Moivre formula. ()
3

Exercise 3.16. (O184MR1-2011.p33) Points A, B, C, D lie on a line in this order. Using


a straight edge and a compas construct parallel lines a and b through A and B, and
parallel lines c and d through C and D, such that their points of intersection are vertices
of a rhombus.
(a2 + 1)2
Exercise 3.17. (S179MR6-2010.p16) Find all positive integers a and b for which
ab 1
is a positive integer.
Answer:
All the possible pairs of positive integers are of the form (F4n1 , F4n+3 ), (L2n , L2n+2 )
(for all integers n) and (dn , dn+1 ) , and its permutations. Numbers dn by defining as:
dn =

1
(( 2 + 1)2n + ( 2 1)2n ) .
2

()

Exercise 3.18. (U176MR6-2010.p21) In the space, consider the set of points (a, b, c)
where a, b, c {0, 1, 2}. Find the maximum number of non-collinear points contained in
the set.
Answer. It is possible to have 16 non-collinear points, taking for example the
following ones:
(0, 0, 0), (0, 1, 0), (1, 0, 0), (1, 2, 0), (2, 1, 0), (2, 2, 0), (0, 0, 1), (0, 2, 1),

(2, 0, 1), (2, 2, 1), (0, 1, 2), (0, 2, 2), (1, 0, 2), (1, 2, 2), (2, 0, 2), (2, 1, 2).

Exercise 3.19. (O180MR6-2010.p33) Let p be a prime. Prove that each positive integer
2
p
2p
2p
n p, p2 divides (Cn+p
) Cn+2p
Cn+p
.

199

3.1. Exercises

Exercise 3.20. (J174MR5-2010.p7) The incircle of triangle ABC touches sides BC, CA, AB
at D, E, F , respectively. Let K be a point on side BC and let M be the point on the line
segment AK such that AM = AE = AF . Denote by L and N the incenters of triangles
ABK and ACK, respectively. Prove that K is the foot of the altitude from A if and only
if DLMN is a square.
Hint: First, prove two following lemmas:
Corollary 3.1. The points D, K lie on the circle with diameter LN .
Corollary 3.2. If M is the second intersection point of AK with the circle circumscribed
to DKNL, then DM LN and AM = AE = AF.
Exercise 3.21. (S169MR5-2010.p11) Let k > 1 be an odd integer such that ak +bk = ck +dk
for some positive integers a, b, c, d.
ak + bk
Prove that
is not a prime.
a+b

Exercise 3.22. (U170MR5-2010.p19) Sequences (xn )n1 , (yn )n1 are defined as follows:
x1 = a, y1 = b, with a b 0, and
xn+1 = max{xn yn , xn + yn }, yn+1 = min{xn yn , xn + yn },

for all n 1. Prove that

lim xn = lim yn = .

Exercise 3.23. (U174MR5-2010.p24) Let p be a prime. A linear recurrence of degree n


in Fp is a sequence {ak }k0 in Fp satisfying a relation of the form
ai+n = cn1 ai+n1 + + c1 ai+1 + c0 ai for all i 0,

where c0 , c1 , ..., cn1 Fp and c0 0.


(a) What is the maximal possible period of a linear recurrence of degree n in Fp ?
(b) How many distinct linear recurrences of degree n have this maximal period?
Answer. (a) pn 1 ;

(b)

(pn 1)(pn 1)
.
n

()

Exercise 3.24. (O171MR5-2010.p27) Prove that in any convex quadrilateral ABCD,


sin (

A
B
C
D
+ 60o) + sin ( + 60o ) + sin ( + 60o) + sin ( + 60o)
3
3
3
3
1
(8 + sin A + sin B + sin C + sin D).
3

Exercise 3.25. (O166MR4-2010.p29) The incircle s of triangle ABC with incenter I is


tangent to sides BC and AC at points A1 and B1 , respectively. Points A2 and B2 are
diametrically opposite to A1 and B1 in s. Let A3 and B3 be the intersection points of AA2
with BC and BB2 with AC, respectively. Let M be the midpoint of side AC and let N
be the midpoint of A1 A3 . Line MI meets BB1 in T and line AT meets BC in P . Let
Q (BC), R be the intersection of lines AB and QB1 and NR AC = {S}. Prove that
[AS] = 2[SM] if and only if [BP ] = [P Q].

200

Chapter 3. Exercises for training

Exercise 3.26. (S161MR32010p11).


Let ABC be a triangle inscribed in a circle of center O and radius R. If dA , dB , dC are
the distances from O to the sides of the triangle, prove that
R3 (d2A + d2B + d2C ) R 2dA dB dC = 0.

Exercise 3.27. (U158MR32010p15) Let (an )n0 be a sequence with a0 > 0 and
an+1 = an +

(a) Prove that lim an = +.


n
an
(b) Find L = lim
n
n

for n = 0, 1, ....

1
an

Hint: As for (b) we employ the Cesaro-Stolz theorem:

Theorem 3.1. Let (an ) and (bn ) be two real sequences, such that bn is positive, increasing
and unbounded. Then
an
an+1 an
= L lim
= L.
lim
n bn
n bn+1 bn
Exercise 3.28. (U161MR32010p20). Let f (0, +) (0, +) be a function satisfying
f (f (x)) = x2 for all x (0, +). Determine the function f if it is differentiable at x = 1.
Answer. f (x) = x

or f (x) = x

for all positive real x. ()

Exercise 3.29. (O160MR32010p25). Let a1 , a2 , ...an , be a sequence of positive integers,


such that for each prime p there are infinitely many terms in the sequence that are divisible
by p. Prove that every positive rational number less than 1 can be represented as
b1
b2
bn
+
++
,
a1 a1 a2
a1 a2 an

where b1 , b2 , ...bn are integers such that 0 bi ai 1, i = 1, ..., n.


Exercise 3.30. (O162MR32010p28). In a convex hexagon ABCDEF, AB DE, BC
EF, CD F A and AB + DE = BC + EF = CD + F A. Denote the midpoints of sides
1
AB, BC, DE, EF by A1 , B1 , D1 , E1 , respectively. Prove that D
1 OE1 = DEF , where O
2
is the point of intersection of segments A1 D1 and B1 E1 .
Exercise 3.31. (J152MR2-2010p2). Let a, b, c > 0. Prove that the following inequality
holds
a+b
b+c
c+a
2(ab + bc + ca) 13
+
+
+

a + b + 2c b + c + 2a c + a + 2b 3(a2 + b2 + c2 )
6

Exercise 3.32. (J153MR2-2010p3). Find all integers n such that n2 + 2010n is a perfect
square.
Answer. The values of n are then 506018, 169344, 102010, and
57122, 34680, 21218, 12250, 8576, 7776, 3618, 3362, 2680, 2144, 2010,
0, 134, 670, 1352, 1608, 5766, 6566, 10240, 19208, 32670, 55112, 100000, 167334, 504008.

201

3.1. Exercises

Exercise 3.33. (S152MR2-2010p9). Let k 2 be an integer and let m, n 2 be relatively


prime integers. Prove that the equation
m
m
m
xm
1 + x2 + + xk = xk+1

has infinitely many solutions in distinct positive integers.


Exercise 3.34. (S153MR2-2010p10). Let X be a point interior to a convex quadrilateral
ABCD. Denote by P, Q, R, S the orthogonal projections of X onto AB, BC, CD, DA,
respectively. Prove that

if and only if

P A AB + RC CD =

1
(AD 2 + BC 2 )
2

QB BC + SD DA =

1
(AB 2 + CD2 ) .
2

Exercise 3.35. (S154MR2-2010p11). Let k 2 be an integer and let n1 , ..., nk be positive


integers. Prove that there are no rational numbers x1 , ..., xk , y1 , ..., yk such that

(x1 + y1 2)2n1 + + (xk + yk 2)2nk = 5 + 4 2.


Exercise 3.36. (O155MR2-2010p31). Prove that the equation
x2 + y 3 = 4z 6
is not solvable in integers.
Exercise 3.37. (O156MR2-2010p33). In a cyclic quadrilateral ABCD with AB = AD
points M, N lie on the sides BC and CD, respectively so that MN = BM + DN. Lines
AM and AN meet the circumcircle of ABCD again at points P and Q, respectively. Prove
that the orthocenter of the triangle AP Q lies on the segment MN.
Exercise 3.38. (J189MR2-2011p3)
Find all primes q1 , q2 , q3 , q4 , q5 such that q14 + q24 + q34 + q44 + q54 is the product of two
consecutive even integers.
Answer. q1 = q2 = q3 = q4 = q5 = 2. ()
Exercise 3.39. (J190MR2-2011p4) Points A1 , B1 , C1 are chosen on sides BC, CA, AB
of triangle ABC such that lines AA1 , BB1 , CC1 are concurrent at M and
AM BM CM

= 2012.
MA1 MB1 MC1
Evaluate

T=

BM
CM
AM
+
+

MA1 MB1 MC1


Answer. T = 2010. ()

202

Chapter 3. Exercises for training

Exercise 3.40. (S191MR2-2011p12). Prove that for any positive integer k the sequence
( (k + n2 ))n1 is unbounded, where (m) denotes the number of divisors of m.

Exercise 3.41. (S192MR2-2011p13). Let p, R, r and ra , rb , rc be the semiperimeter, circumradius, inradius, and exradii of a triangle ABC. Prove that

2
p
ra + rb + rc
p
R
r

Exercise 3.42. (S194MR3-2011p8 ). Let p be a prime of the form 4k + 3 and let n be


a positive integer. Prove that for each integer m there are integers a and b such that
n
n
a2 + b2 m (mod p).
Exercise 3.43. (S197MR3-2011p11). Let (Fn )n0 be the Fibonacci sequence.
Prove that for any prime p 3, p divides F2p Fp .

Exercise 3.44. (S198MR3-2011p12). Let x, y, z be positive real numbers such that


(x 2)(y 2)(z 2) xyz 2.

y
z
3
x
+
+

Prove that
x+y+z
x5 + y 3 + z
y5 + z3 + x
z 5 + x3 + y

Exercise 3.45. (O195MR3-2011p23). Let O, I, H be the circumcenter, incenter, and orthocenter of a triangle ABC , and let D be an interior point to triangle ABC such that
BC DA = CA DB = AB DC. Prove that A, B, D, O, I, H are concyclic if and only if
= 60o .
C

Exercise 3.46. (J201 MR4-2011p3). Let ABC be an isosceles triangle with AB = AC.
Point D lies on side AC such that CBD = 3ABD. If
1
1
1
+
=
,
AB BD BC

find A.
Answer. A =

()
9

Exercise 3.47. (S203MR4-2011p11). Let ABC be a triangle, and P a point not lying on
its sides. Call XY Z the cevian triangle of P with respect to ABC and consider the points
Ya , Za of intersection BC with the parallel lines to AX through Y and Z, respectively.
Prove that AX, Y Za , Ya Z concur in a point Q that satisfies the cross-ratio
(AXP Q) =

AP

AX

Exercise 3.48. (S204MR4-2011p12). Find all positive integers k and n such that k n 1
and n are divisible by precisely the same primes.
Answer. n = 1, k = 2 and n = 2, k = 3. ()

203

3.1. Exercises

Exercise 3.49. (U199MR4-2011p14). Prove that in any triangle ABC,

B+C
C +A 3 3 p
A+B
+ cot
+ cot

+
3 3 cot
4
4
4
2
2r

where p and r denote the semiperimeter and the inradius of triangle ABC, respectively.
Exercise 3.50. (U203MR4-2011p18). Let P be a polynomial of degree 5, with real coefficients, all whose zeros are real. Prove that for each real number a that is not a zero of P
or P there is a real number b such that
b2 P (a) + 4bP (a) + 5P (a) = 0.

Exercise 3.51. (U204MR4-2011p21). Let A1 A2 ...An be a convex polygon and let P be a


point in its interior. Prove that
min P Ai Ai+1 =

i1..n



2 n

Exercise 3.52. (O201MR4-2011p24). Let ABC be a triangle with circumcenter O, and


let perpendiculars at B, C to BC, CA intersect the sidelines CA, AB at E, F, respectively.
Prove that the perpendiculars to OB and OC at F and E, respectively intersect at a point
L lying on the altitude AD, satisfying DL = LA sin2 A.
Exercise 3.53. (O203MR4-2011p26). Let M be an arbitrary point on the circumcircle
of triangle ABC and let the tangents from this point to the incircle of the triangle meet
the sideline BC at X1 , and X2 . Prove that the second intersection of the circumcircle
of triangle MX1 X2 with the circumcircle of ABC(different from M ) coincides with the
tangency point of the circumcircle with mixtilinear incircle in angle A (As usual, the Amixtrilinear incircle names the circle tanget to AB, AC and to the circumcircle of ABC
internally).
Exercise 3.54. (J110MR2-2009 p3). Let (n) and (n) denote the number of divisors
of n and the number of positive integers less than or equal to n that are relatively prime
to n, respectively. Find all n such that (n) = 6 and 3(n) = 7!.
Answer. n = 281 32 = 2529 and n = 41 72 = 2009. ()

Exercise 3.55. (J112MR2-2009 p7). Let a, b, c be integers such that gcd(a, b, c) = 1 and
ab + bc + ca = 0. Prove that a + b + c can be expressed in the form x2 + xy + y 2 , where x and
y are integers.
have

Answer. With a = du and b = dv, then a = u(u + v), b = v(u + v) and c = uv. We

a + b + c = u2 + uv + v 2 = (u v)2 + (u v)v + v 2 = (u v)2 + (u v)u + u2 .

Exercise 3.56. (J114MR2-2009 p10). Let p be a prime. Find all solutions to the equation
a + b c d = p, where a, b, c, d are positive integers such that ab = cd.

204

Chapter 3. Exercises for training


Answer. We have four types of solutions:
1) a = s(t 1), b = t(s + p), c = st, d = (t 1)(s + p);
2) a = s(t p), b = t(s + 1), c = st, d = (t p)(s + 1);
3) a = s(t + 1), b = t(s p), c = st, d = (s p)(t + 1);
4) a = s(t + p), b = t(s 1), c = st, d = (s 1)(t + p),

where s and t are any non-zero integers.

()

Exercise 3.57. (S110MR2-2009 p15). Let X be a point on the side BC of a triangle


ABC. The parallel through X to AB meets CA at V and the parallel through X to AC
meets AB at W. Let D = BV XW and E = CW XV . Prove that DE is parallel to
BC and
1
1
1
=
+

DE BX CX
Exercise 3.58. (S114MR2-2009 p22). Consider triangle ABC with angle bisectors AA1 , BB1 ,
CC1 . Denote by U the intersection of AA1 and B1 C1 . Let V be the projection from U to
BC. Let W be the intersection of the angle bisectors of BC1 V and CB1 V . Prove that
A, V, W are collinear.
Exercise 3.59. (O110MR2-2009 p41). Hexagon A1 A2 A3 A4 A5 A6 is inscribed in a circle
C(O, R) and at the same time circumscribed about a circle (I, r). Prove that if
1
1
1
1
1
1
+
+
=
+
+
A1 A2 A3 A4 A5 A6 A2 A3 A4 A5 A6 A1
then one of its diagonals coincides with OI .
Exercise 3.60. (O112MR2-2009 p50). Let a, b, cbe real positive numbers. Prove that
a3 + abc b3 + abc c3 + abc 3 a3 + b3 + c3
+
+

(b + c)2 (c + a)2 (a + b)2 2 a2 + b2 + c2

Exercise 3.61. (J118MR3-2009 p6). Prove that for each integer n 3 there are n pairwise
distinct positive integers such that each of them divides the sum of the remaining n 1.
Exercise 3.62. (J119MR3-2009 p7). Let , , be angles of a triangle. Prove that
cos3

sin
+ cos3 sin
+ cos3 sin
= 0.
2
2
2
2
2
2

Exercise 3.63. (S115MR3-2009 p10). Prove that for each positive integer n, 2009n can
be written as a sum of six nonzero perfect squares.
Exercise 3.64. (S116MR3-2009 p11). Points P and Q lie on segment BC with P between
B and Q. Suppose that BP, P Q, and QC form a geometric progression in some order.
Prove that there is a point A in the plane such that AP and AQ are the trisectors of angle
BAC if and only if P Q is less then BP and CQ.

205

3.1. Exercises

Exercise 3.65. (S117MR3-2009 p12). Let a, b, c be positive real numbers. Prove that
1
1
3abc
5
1
+
+
+

a + b b + c c + a 2(ab + bc + ca)2 a + b + c
Exercise 3.66. (S120MR3-2009 p16). Let P be a point interior to a triangle ABC and
let da , db , dc be the distances from P to the sides of the triangle. Prove that
4 AP BP CP
BP
CP
AP
+
+

(da + db )(db + dc )(dc + da ) db + dc da + dc da + db

Exercise 3.67. (U115MR3-2009 p18). Let an = 2

1
, n = 1, 2, ... Prove that

n2 + n4 + 14

a1 + a2 + + a119

is an integer.

Exercise 3.68. (U117MR3-2009 p21). Let n be an integer greater than 1 and let x1 , x2 , ..., xn
be positive real numbers such that x1 + x2 + + xn = n. Prove that
n

k=1

and find all equality cases.

n2

1
xk

2
n + 1 nxk + (n 1)xk n 1

Answer. Equality occurs if and only if x = 1. ()

Exercise 3.69. (J132MR4-2009 p8). Consider a regular hexagon A1 A2 A3 A4 A5 A6 with


center O. In how many different ways up to rotation can one color regions Ai OAi+1 (take
i mod 6) in n colors?
Answer. The total number of inequivalent colorings is
1 6
(n + n3 + 2n2 + 2n) .
6

Exercise 3.70. (S127MR4-2009 p8).


Let x, y, z be positive real numbers such that x2 + y 2 + z 2 3. Prove that

()

y3
z3
+
1.
+
z 2 + x2 + 7
y2 + z2 + 7
x2 + y 2 + 7
x3

Exercise 3.71. (O129MR4-2009 p26). Let ABC be a triangle and let points P and Q
lie on sides AB and AC, respectively. Let M and N be the midpoints of BP and CQ,
respectively. Prove that the centers of the nine-point circles of triangles ABC, AP Q, and
AMN are collinear.
Exercise 3.72. (J136MR5-2009 p10). Let a, b, c be the sides, ma , mb , mc the medians,
ha , hb , hc the altitudes, and la , lb , lc the angle bisectors of a triangle ABC. Prove that the
diameter of the circumcircle of triangle ABC is equal to

m2a h2a
la2

ha
la2 h2a

206

Chapter 3. Exercises for training

Exercise 3.73. (J137MR5-2009 p12). Let ABC be a triangle and let tangents to the
circumcircle at A, B, C intersect BC, AC, AB at points A1 , B1 , C1 , respectively. Prove that
1
1
1
1
1
1
+
+
= 2 max {
,
,
}.
AA1 BB1 CC1
AA1 BB1 CC1

Exercise 3.74. (J138MR5-2009 p15). Let a, b, c be positive real numbers. Prove that
b2

a
b
c
a+b+c
+ 2 2+ 2 2

2
+c
a +c
a +b
2

Exercise 3.75. (S134MR5-2009 p19). Find all triples (x, y, z) of integers satisfying the
system of equations

x + y = 5z

= 5z 2 + 1.

xy

Answer. All integer solutions are

(x, y, z) = (L2n , L2n+2 , F2n+1 ) or (x, y, z) = (L2n+2 , L2n , F2n+1 )

where Lj and Fj denote the j th Lucas number and the j th Fibonacci number, respectively.
()
Exercise 3.76. (O134MR5-2009 p36). Let p be a prime and let n be an integer, n > 4.
Prove that if a is an integer that is not divisible by p, then the polynomial axn px2 +px+p2
is irreducible in Z[x].
Exercise 3.77. (J126MR2-20092p10). Let a, b, c be positive real numbers. Prove that
3(a2 b2 + b2 c2 + c2 a2 )(a2 + b2 + c2 ) (a2 + ab + b2 )(b2 + bc + c2 )(c2 + ca + a2 ).

Exercise 3.78. (S124MR3-20092p16). Let ABC be a triangle with midpoints Ma , Mb , Mc


and let X, Y, Z be the points of tangency of the incircle of triangle Ma Mb Mc with Mb Mc ,
Mc Ma , Ma Mb , respectively.
a) Prove that the lines AX, BY, CZ are concurrent at some point P.
b) If AA1 , BB1, CC1 are cevians through P , then the perimeter of triangle A1 B1 C1 is
greater than or equal to the semiperimeter of triangle ABC.
Exercise 3.79. (S125MR3-20092p17). Find all pairs (p, q) of positive integers that satisfy
p
1
1
< 2
q
q
2

Exercise 3.80. (S126MR3-20092p21). Let a, b, c be positive real numbers. Prove that

a2 (b2 + c2 )
b2 (c2 + a2 )
c2 (a2 + b2 )
+
+
a + b + c.
a2 + bc
b2 + ca
c2 + ab

Exercise 3.81. (O121MR3-20092p36). Let a, b, c be positive real numbers. Prove that

ab(a + b) + bc(b + c) + ca(c + a) (a + b)(b + c)(c + a) + 2abc.

207

3.1. Exercises

Exercise 3.82. (O123MR3-20092p39). Let ABC be a triangle and let A1 , A2 , A3 be the


points of tangency of its incircle with the triangles sides. Medians A1 M, B1 N, C1 P
in triangle A1 B1 C1 intersect at A2 , B2 , C2 , respectively. Prove that AA2 , BB2 , CC2 are
concurrent at the isogonal conjugate of the Gergonne point.
Exercise 3.83. (O126MR3-20092p39). Let ABC be a scalene triangle and let Ca be the
A-mixtilinear incircle (the circle tangent to sides AB, AC and internally tangent to the
circumcircle G of triangle ABC). Denote by A1 the tangency point of Ca with G and let
A2 be the diametrically opposed point of A1 with respect to Ca . Similarly, define B2 and
C2 . Prove that lines AA2 , BB2 , CC2 are concurrent.
Exercise 3.84. (J74MR2-2008 p4). A triangle has altitudes ha , hb , hc and inradius r.
Prove that
3 ha 2r hb 2r hc 2r 3

+
+
<
5 ha + 2r hb + 2r hc + 2r 2

Exercise 3.85. (S73MR2-2008p17). The zeros of the polynomial P (x) = x3 + x2 + ax + b


are all real and negative. Prove that 4a 9b 1.
Hint: Use Vietes formulae and Schurs inequality.

()

Exercise 3.86. O84MR3-2008p45). Let ABCD be a cyclic quadrilateral and let P be the
intersection of its diagonals. Consider the angle bisectors of the angles AP B, BP C, CP D,
DP A. They intersect the sides AB, BC, CD, DA at Pab , Pbc , Pcd , Pda , respectively and the
extensions of the same sides at Qab , Qbc , Qcd , Qda , respectively. Prove that the midpoints
of Pab Qab , Pbc Qbc , Pcd Qcd , Pda Qda are collinear.
Exercise 3.87. (S93MR4-2008p17). Let n be
an integer greater than 1 and let x1 , x2 , ..., xn
be nonnegative real numbers whose sum is 2.
Determine the maximum, as a function of n, of
x21
x22
x2n
+
+

1 + x21 1 + x22
1 + x2n
Answer. The maximum value is

2
when x1 = 2 and x2 = x3 = ... = xn = 0. ()
3

3
Exercise 3.88.
(U93MR4-2008p29). Let x0 (0, 1] and xn+1 = xn arcsin(sin xn ), n 0.
Evaluate lim nxn .
n

Answer. lim

Exercise 3.89. (O96MR4-2008p48). Let p and q be primes, q p.


p
Prove that pq divides Cp+q
Cqp 1.
Hint: Prove
Lemma. If p is prime, then
Cp1 ; Cp2 ; Cp3 ; ; Cpp1
are divisible by p.

1
nxn = ()
2

208

Chapter 3. Exercises for training

Exercise 3.90. (S102MR5-2008 p24). Consider triangle ABC with circumcenter O and
incenter I. Let E and F be the points of tangency of the incircle with AC and AB,
respectively. Prove that EF, BC, OI are concurrent if and only if r 2 = rb rc , where ra , rb , rc
are the radii of the excircles.
Exercise 3.91. (J103MR6-2008p1). The numbers 1, 2, ..., 9are randomly arranged on a
circle. Prove that there are adjacent numbers whose sum is at least 16.
Exercise 3.92. (J104MR6-2008p2). Let a, b, c be positive real numbers such that abc = 1.
Prove that
a2 + b2
b2 + c2
c2 + a2
a+b
b+c
c+a
+
+

+
+

2
2
2
2
2
a + b + 1 b + c + 1 c + a + 1 a + b + 1 b + c + 1 c + a2 + 1
Exercise 3.93. (J105MR6-2008p4). Let A1 A2 ...An be a polygon that is inscribed in a
circle C(O, R) and at the same time circumscribed about a circle (I, r). The points of
tangency of A1 A2 ...An with form another polygon B1 B2 ...Bn . Prove that
P (A1 A2 ...An ) R

P (B1 B2 ...Bn ) r

where P (S) stands for the perimeter of figure S.

Exercise 3.94. (J106MR6-2008p6). Prove that among any four positive real numbers
there are two, say a and b, such that
1
ab + 1 a b.
3

Exercise 3.95. (J107MR6-2008p7). Find all quadruples (a, b, c, d) of positive integers


such that
1
1
1
1
(1 + ) (1 + ) (1 + ) (1 + ) = 5.
a
b
c
d
Answer. The solutions are
(24, 5, 1, 1), (14, 6, 1, 1), (9, 8, 1, 1), (9, 2, 2, 1), (4, 3, 2, 1).

Exercise 3.96. (S103MR6-2008p10). Let x1 , x2 , ..., xn be positive real numbers. Prove that

n
x1 + x2 + + xn +
(n + 1) n x1 x2 xn ..
1
1
1
+
++
x1 x2
xn
Exercise 3.97. (S104MR6-2008p12). A set of four points in the plane is said to be nice
if one can draw four circles centered at these points such that each circle is externally
tangent to the other three. Given a triangle ABC with orthocenter H, incenter I, and
excenters Ia , Ib , Ic , prove that {A, B, C, H} and {I, Ia , Ib , Ic } are nice if and only if triangle
ABC is equilateral.
Exercise 3.98. (S105MR6-2008p14). Let P be a point in the interior of a triangle ABC
and let da db dc be distances from P to the triangles sides. Prove that

max{AP, BP, CP } d2a + d2b + db dc + d2c .

209

3.1. Exercises

Exercise 3.99. (S108MR6-2008p22). In triangle ABC let D, E, F be the feet of the altitudes from vertices A, B, C. Denote by P and Q the feet of the perpendiculars from D
onto AB and AC, respectively. Let R = BE DP, S = CF DQ, M = BQ CP, and
N = RQ P S. Prove that M, N, and H are collinear, where H is the orthocenter of
triangle ABC.
Exercise 3.100. (U103MR6-2008p24). Let a1 , a2 , ..., an be positive real numbers such that
a1 + a2 + + an n.
Prove that

a1a1 a2a2 anan 1.


1

Exercise 3.101. (U106MR6-2008p28). Let x be a positive real number. Prove that


xx 1 = ex1 (x 1).

Exercise 3.102. (O103MR6-2008p34). Let a, b, c be positive real numbers, and abc = 1.


Prove that

3
(1 + a)(1 + b)(1 + c) 4 4(1 + a + b + c).
Exercise 3.103. (O104MR6-2008p38). In a convex quadrilateral ABCD let K, L, M, N
be the midpoints of sides AB, BC, CD, DA, respectively. Line KM meets diagonals AC
and BD at P and Q, respectively, and line LN meets diagonals AC and BD at R and
S, respectively. Prove that if AP P C = BQ QD, then AR RC = BS SD.

Exercise 3.104. (O105MR6-2008p40). Let P (t) be a polynomial with integer coefficients


such that P (1) = P (1). Prove that there is a polynomial Q(x, y), with integer coefficients
such that
P (t) = Q(t2 1, t(t2 1)).
Exercise 3.105. (O106MR6-2008p42). A polynomial with integer coefficients is called
good if it can be represented as a sum of cubes of several polynomials in x with integer
coefficients. For example, 9x3 3x2 + 3x + 7 = (x 1)3 + (2x)3 + 23 is good.
a) Is 3x7 + 3x good?
b) Is 3x2008 + 3x7 + 3x good?
Answer. a) 3x7 + 3x is good!

b) 3x2008 + 3x7 + 3x is not good!

Exercise 3.106. (O107MR6-2008p44). Let p1 , p2 , p3 be distinct primes and let n be a


positive integer. Find the number of functions f {1, 2, ..., 2n} {p1 , p2 , p3 } for which
f (1)f (2)f (2n) is a perfect square.
Answer.

32n + 3
()
4

210

3.1.2

Chapter 3. Exercises for training

Exercises from OP from Around the World

Exercise 3.107. (OPfAtW9798-p3). Solve the system for x, y real:

(x 1)(y 2 + 6) = y(x2 + 1)

2
2

(y 1)(x + 6) = x(y + 1).

Answer. (x, y) are (2, 2), (3, 3), (2, 3), and (3, 2). ()

Exercise 3.108. (OPfAtW9798-p). Consider the sequence of positive integers which satisfies
an = a2n1 + a2n2 + a2n3 for all n 3.
Prove that if ak = 2011 then k 3.

Exercise 3.109. (OPfAtW9798-p3). Let k be a positive integer. The sequence an is defined by a1 = 1, and an is the nth positive integer greater than an1 which is congruent to
n modulo k. Find an in closed form.
Answer. an =

n(2 + (n 1)k)

()

Exercise 3.110. (OPfAtW9798-p7). Find all real numbers m such that the equation
(x2 2mx 4(m2 + 1))(x2 4x 2m(m2 + 1)) = 0

has exactly three different roots.

Answer. m = 3. ()
3
Exercise 3.111. (OPfAtW9798-p9). Let f (x) = x2 2ax a2 . Find all values of a
4
such that
f (x) 1 for all x [0, 1].

1
2
Answer. a
()
2
4
Exercise 3.112. (OPfAtW9798-p12). Let ABCD be a trapezoid (AB CD) and choose
F on the segment AB such that DF = CF. Let E be the intersection of AC and BD,
and let O1 , O2 be the circumcenters of ADF, BCF. Prove that the lines EF and O1 O2 are
perpendicular.
Exercise 3.113. (OPfAtW9798-p15). For any real number b, let f (b) denote the maximum of the function
2
+ b
sin x +
3 + sin x
over all x R. Find the minimum of f (b) over all b R.

Answer. The minimum value is

3
()
4

211

3.1. Exercises

Exercise 3.114. (OPfAtW9798-p15). Let ABCD be a convex quadrilateral such that


DAB = ABC = BCD.
Let H and O denote the orthocenter and circumcenter of the triangle ABC.
Prove that H, O, D are collinear.
Exercise 3.115. (OPfAtW9798-p17). Find all natural numbers a, b, c such that the roots
of the equations
x2 2ax + b = 0
x2 2bx + c = 0
x2 2cx + a = 0.
are natural numbers.
Answer. (a, b, c) = (1, 1, 1). ()

Exercise 3.116. (OPfAtW9798-p18). Prove that the equation


x2 + y 2 + z 2 + 3(x + y + z) + 5 = 0
has no solutions in rational numbers.

Exercise 3.117. (OPfAtW9798-p19). Find all continuous functions f R R such that


for all x R,
1
f (x) = f (x2 + ) .
4

Answer. f (x) = c = const R. ()

Exercise 3.118. (OPfAtW9798-p19). Find the number of nonempty subsets of {1, 2, ..., n}
which do not contain two consecutive numbers.

Answer. F1 = 1, F2 = 2, and Fn is the nth Fibonacci number. ()

Exercise 3.119. (OPfAtW9798-p19). For any natural number n 2, consider the polynomial
Pn (x) = Cn2 + Cn5 x + Cn8 x2 + + Cn3k+2 xk
n2
where k = [
].
3
(a) Prove that Pn+3 (x) = 3Pn+2 (x) 3Pn+1 (x) + (x + 1)Pn (x).
n1
(b) Find all integers a such that 3[ 2 ] divides Pn (a3 ) for all n 3.
(a)

k
Hint: Use Cnk1 + Cnk = Cn+1
.

(b)

Answer. = 1 (mod 3).

()

Exercise 3.120. (OPfAtW9798-p21). Let m, n be natural numbers and m + i = ai b2i for


i = 1, 2, ..., n, where ai and bi are natural numbers and ai is squarefree. Find all values of
n for which there exists m such that a1 + a2 + + an = 12.
Answer. (m, n) = (98, 2), (3, 3).

()

212

Chapter 3. Exercises for training

Exercise 3.121. (OPfAtW9798-p22). Let a, b, c be positive numbers such that abc = 1.


Prove that
1
1
1
1
1
1
+
+

+
+

1+a+b 1+b+c 1+c+a 2+a 2+b 2+c


Exercise 3.122. (OPfAtW9798-p22). Let ABC be a triangle and M, N the feet of the
angle bisectors of B, C, respectively. Let D be the intersection of the ray MN with the
circumcircle of ABC. Prove that
1
1
1
=
+

BD AD CD
Exercise 3.123. (OPfAtW9798-p24). Prove that
1
1 3 5 2010 1
<
<

2012 2 4 6 2011 44
Exercise 3.124. (OPfAtW9798-p25). Let O be a point inside a parallelogram ABCD
such that AOB + COD = . Prove that OBC = ODC.

Exercise 3.125. (OPfAtW9798-p36). For each natural number n = 2, determine the


largest possible value of the expression
Vn = sin x1 cos x2 + sin x2 cos x3 + + sin xn cos x1 ,
where x1 , x2 , ..., xn are arbitrary real numbers.
Answer. max Vn =

n
when x1 = = xn = ()
2
4

Exercise 3.126. (OPfAtW9798-p40). Determine all primes p for which the system

has a solution in integers x, y.

p + 1 = 2x2
2
2

p + 1 = 2y .
Answer. p = 7, (x, y) = (2, 5). ()

Exercise 3.127. (OPfAtW9798-p43). Find all real solutions of the system of equations

x3

3
y

= 2y 1
= 2z 1
= 2x 1.

Answer. The solutions are

1 + 5 1 5
,
}.
()
x = y = z = t, t {1,
2
2

213

3.1. Exercises

Exercise 3.128. (OPfAtW9798-p43). Define the functions


f (x) = x5 + 5x4 + 5x3 + 5x2 + 1 ;
g(x) = x5 + 5x4 + 3x3 5x2 1.

Find all prime numbers p for which there exists a natural number 0 x < p, such that
both f (x) and g(x) are divisible by p, and for each such p, find all such x.

Answer. p = 5, 17. If p = 5, x = 4. If p = 17, x = 13. ()

Exercise 3.129. (OPfAtW9798-p). Let f (0, +) R be a function such that


(a) f is strictly increasing;
(b) f (x) >

1
for all x > 0;
x

1
(c) f (x)f (f (x) + ) = 1 for all x > 0. Find f (1).
x

1 5
Answer. f (1) =
()
2

Exercise 3.130. (OPfAtW9798-p45). Let P be a polynomial with integer coefficients


having at least 13 distinct integer roots. Show that if n Z is not a root of P, then
P (n) 7(6!)2 , and give an example where equality is achieved.

Answer. Equality is satisfied, for example, when x = 0 and


P (x) = (x + 1)(x 1)(x + 2)(x 2)...(x + 7).
()

Exercise 3.131. (OPfAtW9798-p47). Let a, b, c be the sides, ma , mb , mc the lengths of


the altitudes, and da , db , dc the distances from the vertices to the orthocenter in an acute
triangle. Prove that
a2 + b2 + c2
ma da + mb db + mc dc =

Exercise 3.132. (OPfAtW9798-p48). Let R be the circumradius of triangle ABC, and


let G and H be its centroid and orthocenter, respectively. Let F be the midpoint of GH.
Show that AF 2 + BF 2 + CF 2 = 3R2 .
Exercise 3.133. (OPfAtW9798-p49). Find all solutions in integers of the equation
x3 + (x + 1)3 + (x + 2)3 + + (x + 7)3 = y 3 .

Answer. The solutions are (2, 6), (3, 4), (4, 4), (5, 6).

()

Exercise 3.134. (OPfAtW9798-p59). Let f be a real-valued function such that for any
real x,
(a) f (10 + x) = f (10 x);
(b) f (20 + x) = f (20 x).
Prove that f is odd (f (x) = f (x)) and periodic ( T > 0 such that f (x+T ) = f (x)).

214

Chapter 3. Exercises for training

Exercise 3.135. (OPfAtW9798-p62). Let a, b, c be positive real numbers. Prove the


inequality
(c + a b)2
(a + b c)2 3
(b + c a)2
+
+
,
(b + c)2 + a2 (c + a)2 + b2 (a + b)2 + c2 5
and determine when equality holds.

Exercise 3.136. (OPfAtW9798-p65). Let f N N be a function satisfying


(a) For every n N, f (n + f (n)) = f (n).
(b) For some n0 N, f (n0 ) = 1.
Show that f (n) = 1 for all n N.

Exercise 3.137. (OPfAtW9798-p66). Find all integers x, y, z satisfying


x2 + y 2 + z 2 2xyz = 0.

Answer. x = y = z = 0. ()

Exercise 3.138. (OPfAtW9798-p73). The positive integers x1 , ..., x7 satisfy the conditions

Compute x7 .

x6 = 144, xn+3 = xn+2 (xn+1 + xn ),

n = 1, 2, 3, 4.
Answer. x7 = 3456. ()

Exercise 3.139. (OPfAtW9798-p74). Solve the following system of equations in real numbers x, y, z

=1
3(x2 + y 2 + z 2 )
2 2
2
2
2
2
3

x y + y z + z x = xyz(x + y + z) .
1 1 1
1 1 1
Answer. (x, y, z) = ( , , ) and (x, y, z) = ( , , ) . ()
3 3 3
3 3 3

Exercise 3.140. (OPfAtW9798-p76). Given a convex pentagon ABCDE with DC = DE

AF AE
and BCD = DEA = , let F be the point on segment AB such that
=
. Show
2
BF BC
that
F CE = F DE and F EC = BDC.

3.2. Problems of Hanoi Open Mathematical Olympiad

3.2
3.2.1

215

Problems of Hanoi Open Mathematical Olympiad


Hanoi Open Mathematical Olympiad 2006

Junior Section, Sunday, 9 April 2006


Question 1. What is the last two digits of the number
(11 + 12 + 13 + + 2006)2 ?

Question 2. Find the last two digits of the sum

200511 + 200512 + + 20052006 .


Question 3. Find the number of different positive integer triples (x, y, z) satisfying
the equations
x2 + y z = 100 and x + y 2 z = 124.
Question 4. Suppose x and y are two real numbers such that
x + y xy = 155 and x2 + y 2 = 325.

Find the value of x3 y 3.

Question 5. Suppose n is a positive integer and 3 arbitrary numbers are choosen


from the set {1, 2, 3, ..., 3n + 1} with their sum equal to 3n + 1. What is the largest possible
product of those 3 numbers?
Question 6. The figure ABCDEF is a regular hexagon. Find all points M belonging
to the hexagon such that
Area of triangle MAC = Area of triangle MCD.
Question 7. On the circle (O) of radius 15cm are given 2 points A, B. The altitude
OH of the triangle OAB intersect (O) at C . What is AC if AB = 16cm?

Question 8. In ABC, P Q//BC where P and Q are points on AB and AC respectively. The lines P C and QB intersect at G. It is also given EF //BC, where G EF,
E AB and F AC with P Q = a and EF = b. Find value of BC.
Question 9. What is the smallest possible value of
x2 + y 2 x y xy?
Senior Section, Sunday, 9 April 2006
Question 1. What is the last three digits of the sum
11! + 12! + 13! + + 2006!

216

Chapter 3. Exercises for training

Question 2. Find the last three digits of the sum


200511 + 200512 + + 20052006 .
Question 3. Suppose that alogb c + blogc a = m.
Find the value of clogb a + alogc b ?
Question 4. Which is larger
2

,2

1+ 1

and 3?

Question5. The figure ABCDEF is a regular hexagon. Find all points M belonging
to the hexagon such that
Area of triangle MAC = Area of triangle MCD.
Question 6. On the circle of radius 30cm are given 2 points A, B with AB = 16cm
and C is a midpoint of AB. What is the perpendicular distance from C to the circle?
Question 7. In ABC, P Q//BC where P and Q are points on AB and AC respectively. The lines P C and QB intersect at G. It is also given EF //BC, where G EF,
E AB and F AC with P Q = a and EF = b. Find value of BC.
Question 8. Find all polynomials P (x) such that
1
1
P (x) + P ( ) = x + ,
x
x

x 0.

Question 9. Let x, y, z be real numbers such that x2 + y 2 + z 2 = 1. Find the largest


possible value of
x3 + y 3 + z 3 xyz?

3.2.2

Hanoi Open Mathematical Olympiad 2007

Junior Section, Sunday, 15 April 2007


Question 1. What is the last two digits of the number
(3 + 7 + 11 + + 2007)2 ?

(A) 01; (B) 11; (C) 23; (D) 37; (E) None of the above.
Question 2. What is largest positive integer n satisfying the following inequality:
n2006 < 72007 ?
(A) 7; (B) 8; (C) 9; (D) 10; (E) 11.

3.2. Problems of Hanoi Open Mathematical Olympiad

217

Question 3. Which of the following is a possible number of diagonals of a convex


polygon?
(A) 02; (B) 21; (C) 32; (D) 54; (E) 63.
Question 4. Let m and n denote the number of digits in 22007 and 52007 when expressed in base 10. What is the sum m + n?
(A) 2004; (B) 2005; (C) 2006; (D) 2007; (E) 2008.
1
Question 5. Let be given an open interval (; ) with =
. Determine the
2007
a
a maximum number of irreducible fractions in (; ) with 1 b 2007?
b
(A) 1002; (B) 1003; (C) 1004; (D) 1005; (E) 1006.
Question 6. In triangle ABC, BAC = 600 , ACB = 900 and D is on BC . If AD
bisects BAC and CD = 3cm. Then DB is
(A) 3; (B) 4; (C) 5; (D) 6; (E) 7.
Question 7. Nine points, no three of which lie on the same straight line, are located
inside an equilateral triangle of side 4. Provethat some three of these points are vertices
of a triangle whose area is not greater than 3.
Question 8. Let a, b, c be positive integers. Prove that
(c + a b)2
(a + b c)2 3
(b + c a)2
+
+

(b + c)2 + a2 (c + a)2 + b2 (a + b)2 + c2 5

Question 9. A triangle is said to be the Heron triangle if it has integer sides and
integer area. In a Heron triangle, the sides a, b, c satisfy the equation b = a(a c). Prove
that the triangle is isosceles.
Question 10. Let a, b, c be positive real numbers such that
1
1
1
+
+
1.
bc ca ab
Prove

a
b
c
+
+
1.
bc ca ab
Question 11. How many possible values are there for the sum a + b + c + d if a, b, c, d
are positive integers and abcd = 2007.
Question 12. Calculate the sum
5
5
5
+
++

2 7 7 12
2002 2007
Question 13. Let be given triangle ABC . Find all points M such that
area of MAB= area of MAC.

218

Chapter 3. Exercises for training

Question 14. How many ordered pairs of integers (x, y) satisfy the equation
2x2 + y 2 + xy = 2(x + y)?

Question 15. Let p = abc be the 3-digit prime number. Prove that the equation
ax2 + bx + c = 0

has no rational roots.


Senior Section, Sunday, 15 April 2007
Question 1. What is the last two digits of the number
(112 + 152 + 192 + + 20072 )

(A) 01; (B) 21; (C) 31; (D) 41; (E) None of the above.

Question 2. Which is largest positive integer n satisfying the following inequal- ity:
n2007 > (2007)n .

(A) 1; (B) 2; (C) 3; (D) 4; (E) None of the above.


Question 3. Find the number of different positive integer triples (x, y, z) satsfying
the equations
x + y z = 1 and x2 + y 2 z 2 = 1.
(A) 1; (B) 2; (C) 3; (D) 4; (E) None of the above.
Question 4. List the numbers


2, 3 3, 4 4, 5 5 and 6 6 in order from greatest to least.

Question 5. Suppose that A, B, C, D are points on a circle, AB is the diameter, CD


is perpendicular to AB and meets AB at E, AB and CD are integers and AE EB = 3.
Find AE?
Question 6. Let P (x) = x3 + ax2 + bx + 1 andP (x) 1 for all x such that x 1.
Prove that a + b 5.

Question 7. Find all sequences of integers x1 , x2 , ..., xn , ... such that ij divides xi + xj
for any two distinct positive integers i and j.

Question 8. Let ABC be an equilateral triangle. For a point M inside ABC , let
D, E, F be the feet of the perpendiculars from M onto BC, CA, AB, respectively. Find
the locus of all such points M for which F DE is a right angle.
Question 9. Let a1 , a2 , ..., a2007 be real numbers such that

219

3.2. Problems of Hanoi Open Mathematical Olympiad

a1 + a2 + + a2007 (2007)2 and a21 + a22 + + a22007 (2007)3 1.

Prove that ak [2006; 2008] for all k {1, 2, ..., 2007}.

Question 10. What is the smallest possible value of


x2 + 2y 2 x 2y xy?
Question 11. Find all polynomials P (x) satisfying the equation
(2x 1)P (x) = (x 1)P (2x),

x.

Question 12. Calculate the sum

1
1
1
+
++

2 7 12 7 12 17
1997 2002 2007
Question 13. Let ABC be an acute-angle triangle with BC > CA. Let O, H and F
be the circumcenter, orthocentre and the foot of its altitude CH , respectively. Suppose
that the perpendicular to OF at F meet the side CA at P . Prove F HP = BAC.
Question 14. How many ordered pairs of integers (x, y) satisfy the equation
x2 + y 2 + xy = 4(x + y)?

Question 15. Let p = abcd be the 4-digit prime number. Prove that the equation
ax3 + bx2 + cx + d = 0

has no rational roots.

3.2.3

Hanoi Open Mathematical Olympiad 2008

Junior Section, Sunday, 30 March 2008


Question 1. How many integers from 1 to 2008 have the sum of their digits divisible
by 5 ?
Question 2. How many integers belong to (a, 2008a), where a (a > 0) is given.
Question 3. Find the coefficient of x in the expansion of

(1 + x)(1 2x)(1 + 3x)(1 4x)(1 2008x).

Question 4. Find all pairs (m, n) of positive integers such that


m2 + n2 = 3(m + n).

220

Chapter 3. Exercises for training

Question 5. Suppose x, y, z, t are real numbers such that

x + y + z + t

x y + z + t

x + y z + t

x + y + z t

Prove that x2 + y 2 + z 2 + t2 1.

1
1
1
1.

Question 6. Let P (x) be a polynomial such that


Find P (x2 + 1)?

P (x2 1) = x4 3x2 + 3.

Question 7. The figure ABCDE is a convex pentagon. Find the sum


DAC + EBD + ACE + BDA + CEB?
Question 8. The sides of a rhombus have length a and the area is S. What is the
length of the shorter diagonal?
Question 9. Let be given a right-angled triangle ABC with A = 900 , AB = c,
AC = b. Let E AC and F AB such that AEF = ABC and AF E = ACB. Denote by P BC and Q BC such that EP BC and F Q BC. Determine EP +EF +P Q?
Question 10. Let a, b, c [1, 3] and satisfy the following conditions
max{a, b, c} 2, a + b + c = 5.

What is the smallest possible value of a2 + b2 + c2 ?


Senior Section, Sunday, 30 March 2008
Question 1. How many integers are there in (b, 2008b], where b (b > 0) is given.
Question 2. Find all pairs (m, n) of positive integers such that
m2 + 2n2 = 3(m + 2n).

Question 3. Show that the equation


x2 + 8z = 3 + 2y 2
has no solutions of positive integers x, y and z.
Question 4. Prove that there exists an infinite number of relatively prime pairs (m, n)
of positive integers such that the equation
x3 nx + mn = 0

3.2. Problems of Hanoi Open Mathematical Olympiad

221

has three distint integer roots.


Question 5. Find all polynomials P (x) of degree 1 such that
max P (x) min P (x) = b a, a, b R

axb

where a < b.

axb

Question 6. Let a, b, c [1, 3] and satisfy the following conditions


max{a, b, c} 2, a + b + c = 5.

What is the smallest possible value of a2 + b2 + c2 ?


Question 7. Find all triples (a, b, c) of consecutive odd positive integers such that
a < b < c and a2 + b2 + c2 is a four digit number with all digits equal.
Question 8. Consider a convex quadrilateral ABCD. Let O be the intersection of
AC and BD; M, N be the centroid of AOB and COD and P, Q be orthocenter of BOC
and DOA, respectively. Prove that MN P Q.
Question 9. Consider a triangle ABC . For every point M BC we define N CA
and P AB such that AP MN is a parallelogram. Let O be the intersection of BN and
CP . Find M BC such that P MO = OMN.
Question 10. Let be given a right-angled triangle ABC with A = 900 , AB = c,
AC = b. Let E AC and F AB such that AEF = ABC and AF E = ACB. Denote
by P BC and Q BC such that EP BC and F Q BC. Determine EP + EF + F Q?

3.2.4

Hanoi Open Mathematical Olympiad 2009

Junior Section, Sunday, 29 March 2009


Question 1. What is the last two digits of the number
1000 1001 + 1001 1002 + 1002 1003 + + 2008 2009?
(A) 25; (B) 41; (C) 36; (D) 54; (E) None of the above.
Question 2. Which is largest positive integer n satisfying the inequality
1
1
1
1
6
+
+
++
<
12 23 34
n(n + 1) 7
(A) 3; (B) 4; (C) 5; (D) 6; (E) None of the above.

222

Chapter 3. Exercises for training

Question 3. How many positive integer roots of the inequality 1 <

there in (10, 10).


(A) 15; (B) 16; (C) 17; (D) 18; (E) None of the above.

x1
< 2 are
x+1

Question 4. How many triples (a, b, c) where a, b, c {1, 2, 3, 4, 5, 6} and a < b < c
such that the number abc + (7 a)(7 b)(7 c) is divisible by 7.
(A) 15; (B) 17; (C) 19; (D) 21; (E) None of the above.
Question 5. Show that there is a natural number n such that the number a = n! ends
exacly in 2009 zeros.

Question 6. Let a, b, c be positive integers with no common factor and satisfy the
conditions
1 1 1
+ =
a b c
Prove that a + b is a square.
Question 7. Suppose that a = 2b + 19, where b = 210n+1 . Prove that a is divisible by
23 for any positive integer n.
Question 8. Prove that m7 m is divisible by 42 for any positive integer m.
Question 9. Suppose that 4 real numbers a, b, c, d satisfy the conditions
a2 + b2 = c2 + d2 = 4 and ac + bd = 2

Find the set of all possible values the number M = ab + cd can take.
Question 10. Let a, b be positive integers such that a + b = 99. Find the smallest and
the greatest values of the following product P = ab.
Question 11. Find all integers x, y such that x2 + y 2 = (2xy + 1)2 .

Question 12. Find all the pairs of the positive integers such that the product of the
numbers of any pair plus the half of one of the numbers plus one third of the other number
is three times less than 15.
Question 13. Let be given ABC with area (ABC) = 60cm2. Let R, S lie in BC
such that BR = RS = SC and P, Q be midpoints of AB and AC, respectively. Suppose
that P S intersects QR at T. Evaluate area (P QT ).

Question 14. Let ABCbe an acute-angled triangle with AB = 4 and CD be the


altitude through C with CD = 3. Find the distance between the midpoints of AD and
BC.

223

3.2. Problems of Hanoi Open Mathematical Olympiad

Senior Section, Sunday, 29 March 2009


Question 1. What is the last two digits of the number
1000 1001 + 1001 1002 + 1002 1003 + + 2008 2009?
(A) 25; (B) 41; (C) 36; (D) 54; (E) None of the above.
Question 2. Which is largest positive integer n satisfying the inequality
1
1
1
6
1
+
+
++
<
12 23 34
n(n + 1) 7
(A) 3; (B) 4; (C) 5; (D) 6; (E) None of the above.
Question 3. How many integral roots of the inequality 1 <

(10, 10).
(A) 15; (B) 16; (C) 17; (D) 18; (E) None of the above.

x1
< 2 are there in
x+1

Question 4. How many triples (a, b, c) where a, b, c {1, 2, 3, 4, 5, 6} and a < b < c
such that the number abc + (7 a)(7 b)(7 c) is divisible by 7.
(A) 15; (B) 17; (C) 19; (D) 21; (E) None of the above.
Question 5. Suppose that a = 2b + 19, where b = 210n+1 . Prove that a is divisible by
23 for any positive integer n.
Question 6. Determine all positive integral pairs (u, v) for which
5u2 + 6uv + 7v 2 = 2009.

Question 7. Prove that for every positive integer n there exists a positive integer m
such that the last n digists in decimal representation of m3 are equal to 8.
Question 8. Give an example of a triangle whose all sides and altitudes are positive
integers.
Question 9. Given a triangle ABC with BC = 5, CA = 4, AB = 3 and the points
E, F, G lie on the sides BC, CA, AB, respectively, so that EF is parallel to AB and area
(EF G) = 1. Find the minimum value of the perimeter of triangle EF G.
Question 10. Find all integers x, y, z satisfying the system

x + y + z
3
3
3

x + y + z

=8
= 8.

Question 11. Let be given three positive numbers p, q and r. Suppose that 4 real
numbers a, b, c, d satisfy the conditions

224

Chapter 3. Exercises for training

a2 + b2 = p

2
c + d2 = q

ac + bd = r.

Find the set of all possible values the number M = ab + cd can take.
Question 12. Let a, b, c, d be positive integers such that a + b + c + d = 99. Find the
smallest and the greatest values of the following product P = abcd.
Question 13.Given an acute-angled triangle ABC with area S, let points A, B, C be
located as follows: A is the point where altitude from A on BC meets the outwards facing
semicirle drawn on BC as diameter. Points B, C are located similarly. Evaluate the sum
T = (area BCA)2 + (area CAB)2 + (area ABC)2 .

Question 14. Find all the pairs of the positive integers such that the product of the
numbers of any pair plus the half of one of the numbers plus one third of the other number
is 7 times less than 2009.

3.2.5

Hanoi Open Mathematical Olympiad 2010. Senior Section

Sunday, 28 March 2010

08h45-11h45

Important:
Answer all 10 questions.
Enter your answers on the answer sheet provided.
For the multiple choice questions, enter only the letters (A, B, C, D or E) corresponding to the correct answers in the answer sheet.
No calculators are allowed.
Question 1 . The number of integers n [2000, 2010] such that 22n + 2n + 5 is divisible
by 7, is
(A) : 0; ; (B): 1; (C) : 2; (D) : 3; (E) : None of the above.
Question 2. The last 5 digits of the number 52010 are
(A) : 65625; (B) : 45625; (C) : 25625; (D) : 15625; (E) : None of the above.
Question 3. How many real numbers a (1, 9) such that the corresponding number
1
a is an integer.
a
(A) : 0; (B) : 1; (C) : 8; (D) : 9; (E) : None of the above.

225

3.2. Problems of Hanoi Open Mathematical Olympiad

Question 4. Each box in a 2 2 table can be colored black or white. How many
different colorings of the table are there?
Question 5. Determine all positive integer a such that the equation
2x2 210x + a = 0
has two prime roots, i.e. both roots are prime numbers.
Question 6. Let a, b be the roots of the equation x2 px + q = 0 and let c, d be the
roots of the equation x2 rx+s = 0, where p, q, r, s are some positive real numbers. Suppose
that
2(abc + bcd + cda + dab)
M=
p2 + q 2 + r 2 + s2
is an integer. Determine a, b, c, d.
Question 7. Let P be the common point of 3 internal bisectors of a given ABC. The
line passing through P and perpendicular to CP intersects AC and BC at M and N,
AM
respectively. If AP = 3cm, BP = 4cm, compute the value of
?
BN
Question 8. If n and n3 + 2n2 + 2n + 4 are both perfect squares, find n?
Question 9. Let x, y be the positive integers such that 3x2 + x = 4y 2 + y. Prove that
x y is a perfect integer.
Question 10. Find the maximum value of
T=

y
z
x
+
+
,
2x + y 2y + z 2z + x

x, y, z > 0.

226

3.3
3.3.1

Chapter 3. Exercises for training

Singapore Open Mathematical Olympiad 2009


Junior Section

Tuesday, 2 June 2009


0930-1200 hrs
Important:
Answer ALL 35 questions.
Enter your answer sheet provided.
For the multiple choice questions, enter your answer on the answer sheet by shading
the bubble containing the letter (A, B, C, D or E) corresponding to the correct answer
For the other short questions, write your answer in the answer sheet
No steps are needed to justify your answer
Each question carries 1 mark
No calculators are allowed.
Part A. Multiple Choice Questions
Question 1. Let C1 and C2 be distinct circles of radius 7 cm that are in the same
plane and tangenr to each other. Find the number of circles of radius 26 cm in this plane
that are tangent to both C1 and C2 .
(A) 2 ; (B) 4 ; (C) 6 ; (D) 8 ; (E) non of the above
Question 2. In the diagram below, the radius of quadrant ODA is 4 and the radius of
quadrant OBC is 8. Given that COD = 30o , find the area of the shades region ABCD.
(A) 12p ; (B) 13p ; (C) 15p ; (D) 16p ; (E) non of the above
Question 3. Let k bea real number.
Find the maximum value of k such that the

following inequality holds: x 2 + 7 x k.


(A)

5 ; (B) 3 ; (C) 2 + 3 ; (D) 10 ; (E) 2 3

Question 4. Three circles of radius 20 are arranged with their respective centres
A, B and C in a row. If the line W Z is tangent to the third circle, find the langth of XY.
(A) 30 ; (B) 32 ; (C) 34 ; (D) 36 ; (E) 38
Question 5. Given that x and y are both negative integers satisfying the equation
10x
, find the maximum value of y.
y=
10 x
(A) -10 ; (B) -9 ; (C) -6 ; (D) -5 ; (E) non of the above
Question 6. The sequence an satisfy an = an1 + n2 and a0 = 2009. Find a50 .
(A) 42434 ; (B) 42925 ; (C) 44934 ; (D) 45029 ; (E) 45359

3.3. Singapore Open Mathematical Olympiad 2009

227

Question 7. Coins of the same size are arranged on a very large table (the infinite
plane) such that each coin touches six other coins. Find the percetage of the plane that
is covered by the coins.

50
20
(A) % ; (B) % ; (C) 16 3% ; (D) 17 3% ; (E) 18 3%
3
3
Question 8. Given that x and y are real numbers satisfying the following equations:

x + xy + y = 2 + 3 2 and x2 + y 2 = 6,

find the value of x + y + 1.


(A) 1 +

3 ; (B) 2 3 ; (C) 2 + 3 ; (D) 3 2 ; (E) 3 + 2

Question 9. Given that y = (x 16)(x 14)(x + 14)(x + 16), find the minimum value
of y.
(A) 896 ; (B) 897 ; (C) 898 ; (D) 899 ; (E) 900

Question 10. The number of positive integral solutions (a, b, c, d) satisfying


1 1 1 1
+ + + =1
a b c d

with the condition that a < b < c < d is

(A) 6 ; (B) 7 ; (C) 8 ; (D) 9 ; (E) 10

Part B. Short Questions

Question 11. There are two models of LCD television on sale. One is a 20 inch
standard model while the other is a 20 inch widescreen model. The ratio of length to
the height of the standard mode is 4 : 3, while that of the widescreen model is 16 : 9.
Television screens are measured by the length of their diagonals, so both models have the
same diagonal length of 20 inches. If the ratio of the area of the standard model to that
of the widescreen model is A 300, find the value of A.
Question 12. The diagram below shows a pentagon (made up of region A and region
B) and a rectangle (made up of region B and region C ) that overlap. The overlapped
3
2
region B is
of the pentagon and
of the rectangle. If the ratio of region A of the
16
9
m
pentagon to region C of the rectangle is
in its lowest term, find the value of m + n.
n
Question 13. 2009 students are taking a test which comprises ten true or fales questions. Find the minimum number of answer scripts required to guarantee two scripts with
at least nine identical answers.
Question 14. The number of ways to arrange 5 boys and 6 giels in a row such that
girls can be adjacent to other girls but boys cannot be adjacent to other boys is 6! k.

228

Chapter 3. Exercises for training

Find the value of k.


Question 15. ABC is a right-angled triangle with BAC = 900 . A square is constructed on the side AB and BC as shown. The area of the square ABDE is 8cm2 and
the area of the square BCF G is 26cm2 . Find the area of triangle DBG in cm2 .
Question 16. The sum of
1
1
1
1
1
+
+
++
+
234 345 456
13 14 15 14 15 16
m
in its lowest terms. Find the value of m + n.
n
1
1
= b+
2 and a b + 2 0, find the value of
Question 17. Given that a +
a+1
b1
ab a + b.
is

Question 18. If x + x + 5y = 2 and y y + x = 7, find the value of x + y + 2009.

Question 19. Let p and q represent two consecutive prime number. For some fixed
integer n, the set {n1, 3n19, 385n, 7n45} represents {p, 2p, q, 2q}, but not necessarily
in that order. Find the value of n.

Question 20. Find the number of ordered pairs of positive intergers (x, y) that satisfy
the equation

x y + y x + 2009xy 2009x 2009y 2009 = 0.


Question 21. Find the intergers part of

1
1
1
1
1
1
1
+
+
+
+
+
+
2003 2004 2005 2006 2007 2008 2009
Question 22. Given the rectangle ABLJ, where the area of ACD, BCEF, DEIJ
and F GH are 22cm2 , 500cm2 and 22cm2 respectively. Find the area of HIK in cm2 .

3
3
Question 23. Evaluate 77 20 13 + 77 + 20 13.

Question 24. Find the number of integers in the set {1, 2, 3, ..., 2009} whose sum of
the digits is 11.
Question 25. Given that
x + (1 + x)2 + (1 + x)3 + + (1 + x)n = a0 + a1 x + a2 x2 + + an xn ,

where each ar is an integer, r = 0, 1, 2, ..., n. Find the value of n such that


a0 + a2 + a3 + a4 + + an2 + an1 = 60

n(n + 1)

229

3.3. Singapore Open Mathematical Olympiad 2009

Question 26. In the diagram, OAB is a triangle with AOB = 900 and OB = 13cm.
P and Q are 2 points on AB such that 26AP = 22P Q = 11QB. If the vertical height of
P Q = 4cm, find the area of the triangle OP Q in cm2 .
Question 27. Let x1 , x2 , x3 , x4 denote the four roots of the equation
x4 + kx2 + 90x 2009 = 0.

If x1 x2 = 49, Find the value of k.

Question 28. Three sides OAB, OAC and OBC of tetrahedron OABC are rightangled triangles, i.e. AOB = AOC = BOC = 900. Given that OA = 7, OB = 2 and
OC = 6, find the value of
(Area of OAB)2 +(Area of OAC)2 +(Area of OBC)2 +(Area of ABC)2 .
n 10
is a non-zero reQuestion 29. Find the least positive integer n for which
9n + 11
ducible fraction.
Question 30. Find the value of the smallest positive integer m such that the equation

has only integer solutions.

x2 + 2(m + 5)x + (100m + 9) = 0

Question 31. In a triangle ABC, the length of the altitudes AD and BE are 4 and
12 respectively. Find the largest possible integer calue for the length of third altitude CF.
Question 32. A four-digit number consists of two distinct pairs of repeated digit (for
example 2211, 2626 and 7007). Find the total number of such possible number that are
divisible by 7 or 101 but not both.
Question 33. m and n are two positive integer satisfying 1 m n 40. Find the
number of pairs of (m, n) such that their produc mn is divisible by 33.
Question 34. Using the digits 0, 1, 2, 3, and 4, find the number of 13-digit sequences
that can be written so that the diffenrence between any two consecutive digits is 1.

Question 35. m and n are two positive integers of reverse order (for example 123
and 321) such that mn = 1446921630. Find the value of m + n.

3.3.2

Senior Section

Tuesday, 2 June 2009

0930-1200 hrs

230

Chapter 3. Exercises for training

Important:
Answer ALL 35 questions.
Enter your answer sheet provided.
For the multiple choice questions, enter your answer on the answer sheet by shading
the bubble containing the letter (A, B, C, D or E) corresponding to the correct answer
For the other short questions, write your answer in the answer sheet
No steps are needed to justify your answer
Each question carries 1 mark
No calculators are allowed.
Part A. Multiple Choice Questions
Question 1. Suppose that P is a plane and A and B are two points on the plane P.
If the distance between A and B is 33 cm, how many lines are there in the plane such
that the distance between each line and A is 7 cm and the distance between each line and
B is 26 cm, respectively.
(A) 1 ; (B) 2 ; (C) 3 ; (D) 4 ; (E) infinitely many.

Question 2. Let y = (17 x)(19 x)(19 + x)(17 + x), where x is a real number. Find
the smallest posible value of y.
(A) 1296 ; (B) 1295 ; (C) 1294 ; (D) 1293 ; (E) 1292

Question 3. If two real numbers a and b are randomly chosen from the interval (0, 1),
find the probability that the equation x2 ax + b = 0 has real roots.
(A)

1
5
3
1
1
; (B)
; (C)
; (D) ; (E)
8
16
16
4
3

Question 4. If x and y are real numbers for which x + x + 5y = 2 and y y + x = 7,


find the value of x + y.
(A) 3 ; (B) 1 ; (C) 1 ; (D) 3 ; (E) 5

Question 5. In a triangle ABC, sin A =


(A)

3
5
and cos B =
Find the value of cos C.
5
13

16
56
16
56
56
16
56
or
; (B)
; (C)
; (D)
; (E)
or
65
65
65
65
65
65
65

Question 6. The area of a triangle ABC is 40cm2 . Points D, E and F are on sides
AB, BC and CA, respectively, as shown in the figure below. If AD = 3cm, DB = 5cm,
and the area of triangle ABE is equal to the area of quadrilateral DBEF, find the area
of triangle AEC in cm2 .
(A) 11 ; (B) 12 ; (C) 13 ; (D) 14 ; (E) 15

Question 7. Find the value of

3
4
22
+
++

1! + 2! + 3! 2! + 3! + 4!
20! + 21! + 22!

3.3. Singapore Open Mathematical Olympiad 2009

231

1
1
1
1
1
1
1
1
; (B)
; (C)
; (D) 1
; (E)
24!
2 23!
2 22!
22!
2 24!
Question 8. There are eight envolopes numbered 1 to 8. Find the number of ways
in which 4 identical red buttons and 4 identical blue buttons can be put in the envolopes
such that each envolope contains exactly one button, and the sum of the nimbers on the
envolopes containing the blue buttons.
(A) 1

(A) 35 ; (B) 34 ; (C) 32 ; (D) 31 ; (E) 62

Question 9. Determine the number of acute-angled triangles (i.e., all angles are less
than 90o ) in which all angles (in degrees) are positive integers and the largest angle is
three times the smallest angle.
(A) 3 ; (B) 4 ; (C) 5 ; (D) 6 ; (E) 7

Question 10. Let ABCD be a quadrilateral inscribed in a circle with diameter AC,
and let E be the foot of perpendicular from D onto AB, as shown in the figure below. If
AD = DC and the area of quadrilateral ABCD is 24cm2 , find the length of DE in cm.

(A) 3 2 ; (B) 2 6 ; (C) 2 7 ; (D) 4 2 ; (E) 6

Part B. Short Questions

Question 11. Find the number of positive divisors of (20083 +(3 2008 2009)+1))2.
Question 12. Suppose that a, b and c are real numbers greater than 1.
1
1
1
+
+

Find the value of


c
a
1 + loga2 b ( a ) 1 + logb2 c ( b ) 1 + logc2 a ( cb )

Question 13. Find the remainder when

is divided by 2009.

(1! 1) + (2! 2) + (3! 3) + + (286! 286)

1
1
Question 14. Find the value of (25 + 10 5) 3 + (25 10 5) 3

1 + 2009
Find the value of (a3 503a 500)10 .
Question 15. Let a =
2

Question 16. ABC is a triangle and D is a point on side BC. Point E is on side AB
such that DE is the angle bisector of ADB, and point F is on side AC such that DF
AE BD CF
is the angle bisector of ADC. Find the value of

EB DC F A
Question 17. Find the value of
(cot 25o 1)(cot 24o 1)(cot 23o 1)(cot 22o 1)(cot 21o 1)(cot 20o 1).

Question 18. Find the number of 2-element subset {a, b} of {1, 2, 3, ..., 99, 100} such
that ab + a + b is a multiple of 7.

232

Chapter 3. Exercises for training

Question 19. Let x be real number such that x2 15x+1 = 0. Find the value of x4 +

x4

Question 20. ABC is a triangle with AB = 10cm, BC = 40cm. Points D and E lie
on side AC and point F on side BC such that EF is parallel to AB and DF is parallel
to EB. Given that BE is an angle bisector of ABC and that AD = 13.5cm, find the
length of CD in cm.
Question 21. Let S = {1, 2, 3, ..., 64, 65}. Determine the number of ordered triples
(x, y, z) such that x, y, z S, x < z and y < z.
Question 22. Given that an+1 =

find the value of

an1
, where n = 1, 2, 3, ..., and a0 = a1 = 1,
1 + nan1 an

a199 a200

Question 23. ABC is a triangle with AB = 5cm, BC = 13cm and AC = 10cm. Points
area of AP Q 1
P and Q lie on sides AB and AC respectively such that
=
area of ABC 4
Given that the least posible length of P Q is kcm, find the value of k.
Question 24. If x, y are real numbers such that x + y + z = 9 and xy + yz + zx = 24,
find the largest possible value of z.
Question 25. Find the number of 0 1 binary sequences formed by six 0 s and 1 s
such that no three 0 s are together. For example, 110010100101 is such a sequence but
101011000101 and 110101100001 are not.
Question 26. If

cos 1000
= tanx, find x.
1 4 sin 250 cos 250 cos 500

Question 27. Find the number of positive integers x, where x 9, such that
log x
9

9
x2
< 6 + log3 .
3
x

Question 28. Let n be the positive integer such that


1
1
1
1

+
++
=
9 11 + 11 9 11 13 + 13 11
n n + 2 + (n + 2) n 9

Find the value of n.

Question 29. ABCD is a rectangle, E is the midpoint of AD and F is the midpoint


of CE. If the area of triangle BDF is 12cm2 , find the earea of rectangle ABCD in cm2 .
Question 30. In each of the following 6-digit positive integers:
555555, 555333, 818811, 300388,

3.3. Singapore Open Mathematical Olympiad 2009

233

every digit in the number appears at least twice. Find the number of such 6-digit positive
integers.
Question 31. Let x and y be positive integers such that 27x + 35y 945. Find the
largest posible value of xy.
Question 32. Determine the coefficient of x29 in the expansion (1 + x5 + x7 + x9 )16 .

Question 33. For n = 1, 2, 3, ..., let an = n2 + 100, and let dn denote the greatest common divisor of an and an+1 . Find the maximun value of dn as n ranges over all positive
integers.
Question 34. Using the digits 1, 2, 3, 4, 5, 6, 7, 8, we can form 8! (= 40320) 8-digit
numbers in which the eight digits are all distinct. For 1 k 40320, let ak denote the k th
number if these numbers are arranged in increasing order:
12345678, 12345687, 12345768, ..., 87654321;
that is, a1 = 12345678, a2 = 12345687, ..., a40320 = 87654321. Find a2009 a2008 .
100
].
x
Here [c] denotes the greatest integer less than or equal to c. Find the largest possible
value of 2a2 3b2 .
Question 35. Let x be a positive integer, and write a = [log10 x] and b = [log10

Bibliography
[1] Nguyen Van Mau. Hanoi Open Mathematical Olympiad, Problems and Solutions. HMS.
Hanoi, 2009.
[2] Dusan Djukic- Vladimir Jankovic- Ivan Matic-Nikola Petrovic, The IMO Compendium: A
Collection of Problems Suggested for The International Mathematical Olympiads, 1959-2004.
Springer, 2006.
[3] Titu Andreescu and Zuming Feng, 102 Combinatorial Problems from the Training of the
USA IMO Team. Bikhauser, 2002.
[4] Titu Andreescu and Zuming Feng, A path to combinatorics for undergraduates counting
strategies. Bikhauser, 2004.
[5] Titu Andreescu, Razvan Gelca, Mathematical Olympiad Challenges. Bikhauser, 2000.
[6] Titu Andreescu and Zuming Feng, Mathematical Olympiads 1998 - 2000 Problems and Solutions From Around the World. MMA 2002.
[7] Arthur Engel, Problem - Solving Strategies. Springer, 1998.
[8] Loren C. Larson, Problem - Solving Through Problems. Springer, 1983.
[9] Walter Mientka et al., Mathematical Olympiads 1996 - 1998, Problems and Solutions From
Around the World, MMA 1997, 1998.
[10] Aczel, J.: 1966, Lectures on Functional Equations and Their Applications.New York,
Birkhauser.
[11] Gleason, A. M., Greenwood, R. E. and Kelly, L. M.: 1980, The William Lowell Putnam
Mathematical Competition. Problems and Solutions: 1965- 1984. MMA .
[12] Kedlaya, K. S., Poonen, B., and Vakil, R.: 2002, The William Lowell Putnam Mathematical
Competition 1985-2000. Problems, Solutions and Commentary. MMA .
[13] A. Engel, Problem-Solving Strategies, Springer Verlag, 1998
[14] Websites of mathemathics:

http://diendantoanhoc.net
http://mathnfriend.net
www.kalva.co.uk
http://www.ams.org/
http://www.math.ac.vn/
http://math.ca/crux/
http://kvant.mccme.ru/

234

http://mathforum.org
http://www.math.com/
http://www.bymath.com/
http://lib.mexmat.ru/
http://hms.org.vn/hms/
http://sms.math.nus.edu.sg/
http://thesaurus.maths.org/

Anda mungkin juga menyukai