Вы находитесь на странице: 1из 44

MATHEMATICAL

COMPETITIONS
IN CROATIA

Zagreb, 2017

Croatian Mathematical Society


Mathematical competitions in Croatia in 2017

In Croatia, children undergo twelve years of formal education – from 1st to 8th grade
in primary school and from 9th to 12th grade in secondary school. Mathematical com-
petitions are held for students from 4th to 12th grade. Competitions at the secondary
school level have two categories – category A (for students who attend mathematical
grammar schools) and category B (for all other secondary school students).
In the year 2017, around 17000 students participated in the first round of the compet-
itions, more than 13000 of them from primary schools. The City competitions were
held on 26th January. All tests at this level consist of five easier problems and two
harder problems, as well as tests for category B at County competitions. All other
tests at County competitions and all the tests at National competition consist of five
problems.
On County competitions held on 28th February, 1994 primary school students and
1141 secondary school students (546 of them in category A) participated. The best
88 students from primary school (from 5th to 8th grade), and 175 secondary school
students (98 in category A and 77 in category B) were invited to the National com-
petition in Primošten (from 3rd to 5th April).
After the National competition, the top 23 students from category A were invited to
the Croatian Mathematical Olympiad, a three-round team selection phase. The first
two tests took place on 22nd and 23rd April, and separate final selection tests for the
IMO and MEMO teams on 6th May. All tests consisted of four olympiad-style prob-
lems from four different areas: algebra, combinatorics, geometry and number theory.
After the Croatian Mathematical Olympiad, the teams for the 58th International
Mathematical Olympiad (IMO) in Brazil and the 11th Middle European Mathematical
Olympiad (MEMO) in Lithuania were selected:

IMO MEMO
1 6
Adrian Beker (grade 12) Tea Arvaj (grade 11)
Lugo Mihovilić 2 (grade 12) Aleksandra-Saša Božović 7 (grade 11)
Petar Nizić–Nikolac 1 (grade 11) Noel Lakić 5 (grade 9)
Lukas Novak 3 (grade 12) Ivan Sinčić 8 (grade 11)
Borna Šimić 4 (grade 11) Luka Šimek 1 (grade 10)
Marin Varivoda 5 (grade 10) Tadej Petar Tukara 1 (grade 11)

1
XV. gimnazija, Zagreb, 2 Gimnazija Lucijana Vranjanina, Zagreb, 3 Gimnazija Josipa
Slavenskog Čakovec, Čakovec, 4 Gimnazija “Matija Mesić”, Slavonski Brod, 5 Gimnazija
Franje Petrića, Zadar, 6 III. gimnazija, Osijek, 7 Prva gimnazija Varaždin, Varaždin,
8
Gimnazija Andrije Mohorovičića, Rijeka.

i
Contents

Mathematical competitions in Croatia in 2017 i

1 First round – City competition 1

2 Second round – County competition 4

3 Final round – National competition 7


3.1 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

4 Croatian Mathematical Olympiad 23


4.1 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

In this booklet we present the problems from mathematical competitions for


category A in Croatia in 2017. The solutions are given for the problems of the Final
round and the Croatian Mathematical Olympiad.
Problems are numerated according to secondary school classes, e.g. 2.4. is the 4th
problem for the 2nd class of secondary school (10th grade).
We have tried to acknowledge the authors or the sources of the problems wherever it
was possible.

ii
1 First round – City competition

1.1. In a lake, there are two kinds of fish: red and yellow. Of all the fish in the lake,
two fifths are yellow, while the others are red. Three quarters of all the yellow
fish are female. If the total number of female fish equals the total number of male
fish, what is the percentage of red male fish in the lake? (The Netherlands 2015)

1.2. Let S be the centre of the circle k with radius 1. Vertices A and B of the square
ABCD belong to the circle k, and its side CD passes through the point S. Find
the length of the side of the square ABCD.

1.3. Determine the digits a and b if the number a2017b is divisible by 72.
(Hong Kong 1990)

1.4. An 8 × 8 × 8 cube consists of 512 small grey cubes. Three sides of the big cube
are painted in white, and the other three sides are painted in red. If each of the
eight small cubes in the vertices of the big cube has at least one white and at
least one red side, how many small cubes have at least one red and at least one
white side? (Russia)

1.5. Let x and y be distinct real numbers such that


x + 4 = (y − 2)2 and y + 4 = (x − 2)2 .
Determine x2 + y 2 .

1.6. Let ABCD be a rectangle such that |AB| : |AD| = 2 : 3 and let E be a
point on the side AD such that |AE| = |AB|. Point F lies on the ray AB so
that triangle AF E and quadrilateral CDEF have the same areas. Determine
|AB| : |BF |. (Matija Bašić)

1.7. Prove that among any three positive integers we can choose two, say a and b, so
that the number a3 b − ab3 is a multiple of 10. (AwesomeMath, AMY 2011–2012)

∗∗∗

2.1. Find all complex numbers z such that z 3 = z. (Mathematics Stack Exchange)

2.2. Determine all real numbers a such that the equation


x2 − (5 − a)x + a2 − 11a − 46 = 0
has two real solutions, one of which is less than 2, and the other greater than 2.
(Art of Problem Solving, Problem c6h1224109)

1
2 First round – City competition

2.3. Let a and b be positive integers such that 8a2 + 1 = b2 . Prove that ab is divisible
by 3. (Željko Hanjš)

2.4. Let AB and CD be two diameters of the circle k with centre S and let ^BAD =
28◦ . The circle centred at A passing through the point S meets the circle k at
E and F (D and F are on the same side of AB). Find ^CF S. (Matija Bašić)

2.5. Let n be a positive integer. Vertices of a regular 2n-gon are painted red and
blue in an alternating fashion. All the sides and diagonals of the 2n-gon are
drawn. If the number of segments with ends of the same colour is 3192, find the
number of segments with ends of different colours. (Mea Bombardelli)

2.6. Let ABCD be a trapezium with parallel sides AB and CD, and let P be the
intersection of its diagonals. If the area of the trapezium equals 25, and the area
of the triangle CDP equals 9, find the area of the triangle ABP . (Stipe Vidak)

2.7. A cutting of an 8 × 8 chessboard into rectangles is called diversified if the


following two conditions hold:

• Each rectangle contains the same number of black and white cells.
• No two rectangles have the same total number of cells.

Find the largest n such that a diversified cutting into n rectangles exists.
(Ratko Višak)

∗∗∗

3.1. Find all rational numbers a and b such that



sin 75◦ · cos 15◦ = a + b. (Nikola Adžaga)

3.2. Let x and y be positive real numbers such that


2
2x = 16y and log√2017 x + log2017 y > 0.

Prove that y > 12 . (Kristina Ana Škreb)

3.3. A number written only with digits 2 and 3 is called happy. Therefore, the happy
numbers are 2, 3, 22, 23. . . Determine the 2050th happy number.
(Sphere Online Judge: TSHOW1 – Amusing numbers)

3.4. Find all pairs (m, n) of positive integers which satisfy the equation

mn2 = 100(n + 1). (Ukraine 2009)


First round – City competition 3

3.5. Let ABC be an acute-angled triangle with ^BAC = 75◦ . Let P be the midpoint
of the side BC, and let M and N be the feet of the altitudes from vertices B
and C respectively. Determine the angle ^M P N .
(Kristina Ana Škreb, Nikola Adžaga)

3.6. Let ABC be an acute-angled triangle and let BCD, ACE and ABF be equi-
lateral triangles drawn on the outside of triangle ABC. Let M be the midpoint
of the segment √ BD, and let O be the centre of the triangle ACE. Prove that
|AM | : |OF | = 23 .

3.7. Twenty points with integer coordinates are given in the coordinate plane so
that no three of them are collinear. Prove that there exists a triangle with
vertices chosen among the given points whose centroid is also a point with
integer coordinates. (Kristina Ana Škreb)

∗∗∗

4.1. Find all positive integers b such that 11 · 22 · 33 = 13310 holds in base b.
(Stipe Vidak)

4.2. Sequences (xn ) and (yn ) are given recursively:


x1 = 3, y1 = 1,
xn+1 = 3xn + yn , for all n ∈ N;
yn+1 = xn + 3yn , for all n ∈ N.
Prove that x22017 − 2
y2017 =82017
. (Miljen Mikić)

4.3. In the rectangle with vertices A(−1, 0), B(1, 0), C(1, 1), D(−1, 1) a point T
is randomly chosen. Find the probability that T is closer to A than to the
point P (2, 1). (Mathematics Stack Exchange)

4.4. Find the smallest multiple of 84 whose decimal representation contains only
digits 6 and 7. (Hong Kong)

4.5. Find the number of nine-digit numbers with digits 1, 2, 3, 4, 5, 6, 7, 8 and 9,


such that no three consecutive digits equal 123, 246 or 678. (Matija Bašić)

4.6. Let z1 and z2 be complex numbers such that |z1 | = |z2 | = 1, and let a and b be
real numbers such that a + b = 1. Prove
1
|az1 + bz2 | > |z1 + z2 |.
2
(T. Andreescu, D. Andrica, Complex Numbers from A to. . . Z)

4.7. Let A, B and C be points on the curve (hyperbola) with equation xy = 1. Prove
that the orthocentre of the triangle ABC also belongs to that curve.
(Mathematics Stack Exchange)
2 Second round – County competition

1.1. Determine the sum


1 1 1
√ √ + √ √ + ··· + √ √ .
2 1+1 2 3 2+2 3 100 99 + 99 100

1.2. Gargamel caught N Smurfs and distributed them into three bags. After moving
Papa Smurf from the first bag to the second, Grouchy Smurf from the second
bag to the third, and Smurfette from the third bag to the first one, the average
Smurfs’ height in the first bag decreased by 8 millimetres, while the average
Smurfs’ heights in the second and the third bag increased by 5 and 8 millimetres,
respectively.
If there were nine Smurfs in the first bag, determine N . (Stipe Vidak)

1.3. Find all three-digit positive integers n for which the numbers n and n2 coincide
in the last three digits. (AIME 2014)

1.4. Points M and N lie on the sides BC and CD (respectively) of the square ABCD
so that ^BM A = ^N M C = 60◦ . Determine ^M AN . (Ukraine 2013)

1.5. Karlo and Lovro play the following game. Karlo cuts the paper of dimensions
9 × 9 into rectangles of integer dimensions having at least one side of length 1.
Lovro then chooses a positive integer k ∈ {1, . . . , 9}, after which Karlo gives
him as many coins as the total area of 1 × k and k × 1 cut rectangles is. Lovro
chooses k so that he gets as many coins from Karlo as possible, while Karlo
wants to give as few coins to Lovro as possible.
Find the minimum possible number of coins that Karlo gives to Lovro.
(Ukraine 2013)

∗∗∗

2.1. Let p be a prime number. Determine all pairs (a, b) of integers such that

p(a − 2) = a(b − 1). (Greece 2013)

2.2. Determine all complex numbers z for which the ratio of the imaginary part of
the fifth power of z to the fifth power of the imaginary part of z is the smallest
possible. (Revista de Matematică din Timis, oara 1984)

4
Second round – County competition 5

2.3. Determine all triples (x, y, z) of positive real numbers that satisfy
3bxc − {y} + {z} = 20.3
3byc + 5bzc − {x} = 15.1
{y} + {z} = 0.9.

For a real number t, btc denotes the largest integer not greater than t, while {t}
denotes its fractional part, i.e. {t} = t − btc. For example, if t = 15.1, then
btc = 15 and {t} = 0.1. (Unione Matematica Italiana)

2.4. Let ABC be an acute-angled triangle such that |AC| > |AB|, and let O be its
circumcentre. The angle bisector of ^BAC meets the side BC at point D. The
line through B perpendicular to the line AD intersects the line AO at point E.
Prove that points A, B, D and E all lie on the same circle.

2.5. What is the maximum number of elements that a subset of {1, 2, 3, . . . , 2017}
can have so that for any two distinct elements a and b of that subset, the number
a + b is not divisible by a − b? (Hong Kong 1992)

∗∗∗

3.1. In a triangle ABC the interior bisector at vertex C meets the side AB at point D.
ab
Let a and b be the lengths of the sides BC and AC respectively. If |CD| = ,
a+b
determine ^ACB.
2
3.2. Does there exist a positive integer m such that 2m − 4 is divisible by 7?
(Ukraine 2013)

3.3. Determine the product (1 + tan 1◦ ) (1 + tan 2◦ ) · · · (1 + tan 45◦ ).


(T. Andreescu, Z. Feng, 103 Trigonometry Problems)

3.4. A tetrahedron is given, having one edge of length 3 and all other edges of
length 2. Find the volume of that tetrahedron. (Stipe Vidak)

3.5. What is the maximum number of elements that a finite set S can have so that
among any three elements of S there exist two distinct whose sum is an element
of S as well? (Russia 2000)

∗∗∗

4.1. There are seven people in a room. Four of them each know exactly one other
person, while the other three each know exactly two people. All acquaintances
are mutual. What is the probability that two randomly chosen people do not
know each other? (Graduate Management Admission Test)
6 Second round – County competition

4.2. Determine the number of positive integers c 6 1 000 000, that can be expressed
as c = a2 + 3b2 − 4ab for some non-zero integers a and b. (Ilko Brnetić)

4.3. A sequence a0 , a1 , a2 , . . . of positive real numbers is given such that

a1 = 1 − a0 , an+1 = 1 − an (1 − an ) for all n > 1.

Prove that
1 1 1
Å ã
a0 a1 · · · an + + ··· + =1
a0 a1 an
holds for every positive integer n. (China 2009)

4.4. Let ABC be an acute-angled triangle. The tangents at A and B to its circum-
circle intersect at M . The line through M parallel to the side BC intersects the
side CA at point N . Prove that |BN | = |CN | holds. (Stipe Vidak)

4.5. There are 3000 points marked on a circle, and there is a grasshopper standing
on one of these points. With every jump, the grasshopper jumps over one or two
marked points (in the clockwise direction) and lands on the next marked point.
Determine the smallest possible number of jumps made by the grasshopper if it
landed on every marked point at least once and then returned to the starting
point. (Austria 2016)
3 Final round – National competition

1.1. If a and b are positive integers, then a.b is a decimal number obtained by writing
the number a, then the decimal point and then the number b. For example, if
a = 20 and b = 17, then a.b = 20.17 and b.a = 17.2.
Determine all pairs (a, b) of positive integers such that a.b · b.a = 13.
(Belarus 2013)

1.2. Let a and b be integers of different parity. Prove that there exists an integer c
such that the numbers ab + c, a + c and b + c are squares of integers.
(Mathematics Stack Exchange)

1.3. If x, y, z and w are real numbers such that


x y z w
+ + + = 1,
y+z+w z+w+x w+x+y x+y+z
find
x2 y2 z2 w2
+ + + .
y+z+w z+w+x w+x+y x+y+z
(Ukraine 2013)

1.4. Let ABC be an acute-angled triangle. Let point B 0 be the reflection of B


across the line AC, and point C 0 the reflection of C across the line AB. Circles
circumscribed to triangles ABB 0 and ACC 0 intersect at points A and P . Prove
that the circumcentre of triangle ABC lies on the line AP . (Russia 2005)

1.5. The unit squares of an N ×N board are coloured black and white so that squares
that share a side have different colours, and so that at least one corner square
is coloured black. In each step we choose a 2 × 2 square and change the colour
of all four unit squares inside that square, so that white unit squares become
black, black become grey and grey become white.
Determine all positive integers N > 1 for which it is possible, using a finite
number of steps, to achieve that all unit squares that were originally black
become white, and all unit squares that were originally white become black.
(Russia 2012)

∗∗∗

2.1. If x, y, z and w are real numbers such that


x2 + y 2 + z 2 + w2 + x + 3y + 5z + 7w = 4,
determine the largest possible value of x + y + z + w. (Israel 2012)

7
8 Final round – National competition

2.2. Let S and T be points inside the triangle ABC. The distance from S to lines
AB, BC and CA is 10, 7 and 4, respectively. The distance from T to these lines
is 4, 10 and 16, respectively.
Determine the radius of the incircle of triangle ABC. (Stipe Vidak)

2.3. Let a and b be positive integers such that a > b and

a − b = 5b2 − 4a2 .

Prove that a − b is a square of some positive integer. (Gazeta Matematică 2016)

2.4. A triangle ABC is given. Circle k touches BC from outside the triangle at point
K, and the extensions of lines AB and AC over points B and C at points L
and M , respectively. The circle with diameter BC intersects segment LM at
points P and Q so that point P lies between L and Q.
Prove that lines BP and CQ intersect at the centre of circle k. (Stipe Vidak)

2.5. A city has M streets and N squares, where M and N are positive integers such
that M > N . Each street connects two squares and does not go through any
other squares.
The citizens wish to change the appearance of the city. This year, each street
will be coloured for the first time, in either red or blue. It is also determined
that each subsequent year one square will be chosen and all the streets leading
to that square will simultaneously change colour from blue to red and vice versa.
Prove that the citizens can choose the initial colours of streets so that it can
never happen that all streets have the same colour. (Russia 1994)

∗∗∗

3.1. Determine the maximum value of the expression

sin x sin y sin z + cos x cos y cos z

where x, y and z are real numbers. (Russia)

3.2. Let a and b be positive integers of different parity. Prove that (a + 3b)(5a + 7b)
cannot be a perfect square. (Thailand 2015)

3.3. Find all polynomials P with real coefficients such that

P (x2 ) + 2P (x) = (P (x))2 + 2

holds for every real number x. (Kristina Ana Škreb, Nikola Adžaga)
Final round – National competition 9

3.4. A triangle ABC is given, with altitudes AD, BE and CF , and orthocentre H.
The segments EF and AD intersect at point G. The segment AK is the dia-
meter of the circumcircle of triangle ABC and it intersects BC at point M .
Prove that the lines GM and HK are parallel. (Go Geometry)

3.5. Let R be a positive integer less than 2017. Exactly R vertices of a regular
2017-gon are red, while the remaining ones are blue. Prove that the number of
isosceles triangles having all three vertices of the same colour does not depend
on the arrangement of the blue and the red vertices. (Russia 1994)

∗∗∗

4.1. Let c and d be positive divisors of a positive integer n. If c > d, prove that
d2
c>d+ . (Russia 2011)
n

4.2. Determine all functions f : R → R such that


f (x + f (y)) = f (f (y)) + 2xf (y) + x2
holds for all real numbers x and y. (Tonći Kokan)

4.3. We call a point P inside a triangle ABC marvelous if exactly 27 rays can be
drawn from it, intersecting the sides of ABC such that the triangle is divided into
27 smaller triangles of equal areas. Determine the total number of marvelous
points inside a given triangle ABC. (Naboj)

4.4. An acute-angled triangle ABC is given, such that |AB| > |AC|. Let O be
the centre of the circumcircle, and let OQ be a diameter of the circumcircle of
triangle BOC. The line parallel to BC passing through A intersects the line
CQ at M , while the line parallel to CQ and passing through A intersects the
line BC at N . Denote by T the intersection of AQ and M N .
Show that T lies on the circumcircle of triangle BOC. (Ukraine 2005)

4.5. On a 2017 × 2017 board, some of the squares are occupied by a single ladybird;
the rest of the squares are empty. The ladybirds move, never leaving the board,
according to the following principles. Every second, each ladybird moves to a
neighbouring square. The moves are horizontal (to the square immediately to
the right or to the left of the current square), or vertical (to the square above or
below the one currently occupied). A ladybird which makes a horizontal move
must move vertically in its next move. Similarly, a ladybird which makes a
vertical move must move horizontally in its next move.
Determine the smallest number of ladybirds such that, regardless of their initial
positions and their chosen paths, we may be sure that two of them will eventually
find themselves in the same square, at the same moment.
(P. Soberón, Problem-Solving Methods in Combinatorics)
10 Final round – National competition

3.1 Solutions

1.1. If (a, b) is a solution, then so is (b, a) – and vice versa. Therefore, we may assume
that a > b.
If a > 10, then necessarily b = 1, and the only possibilities for a are 10, 11 and 12.
We can verify that in those cases the product is not 13.
Therefore, numbers a and b are both smaller than 10, i.e. they are digits. The condi-
tion is then equivalent to
ab · ba = 1300.
Since the last digit of the product is 0, assuming a > b the only option is a = 5,
b = 2. We can check that 5.2 · 2.5 = 13, i.e. those numbers satisfy the condition of
the problem.
Therefore, the solutions are (a, b) = (2, 5) and (a, b) = (5, 2).

1.2. For arbitrary a and b of different parity define


1 + a2 + b2 − 2a − 2b − 2ab
c= .
4
Note that c is an integer, since a and b are of different parity (for example, if a is even
and b is odd, then 1 − 2b + b2 = (1 − b)2 and a2 − 2a − 2ab are divisible by 4).
Then
ã2
1 + a2 + b2 + 2a − 2b − 2ab 1+a−b
Å
a+c= = ,
4 2
ã2
1 + a2 + b2 − 2a + 2b − 2ab 1−a+b
Å
b+c= = ,
4 2
ã2
1 + a2 + b2 − 2a − 2b + 2ab 1−a−b
Å
ab + c = = .
4 2
Since a and b are of different parity, the numbers 1 + a − b, 1 − a + b and 1 − a − b
are even, which completes the proof.

1.3. If we multiply the condition by x + y + z + w, we get:


x2 + x(y + z + w) y 2 + y(x + z + w) z 2 + z(x + y + w) w2 + w(x + y + z)
+ + +
y+z+w z+w+x w+x+y x+y+z
= x + y + z + w,
i.e.
x2 y2 z2 w2
+x+ +y+ +z+ + w = x + y + z + w.
y+z+w z+w+x w+x+y x+y+z
It follows that
x2 y2 z2 w2
+ + + = 0.
y+z+w z+w+x w+x+y x+y+z
Final round – National competition 11

1.4. Denote α = ^BAC, β = ^CBA and γ = ^ACB, and let O be the circumcentre
of triangle ABC.

C′
A

B′

B C

A′
P

Point P is on the circle circumscribed to triangle ACC 0 , so ^AP C = ^AC 0 C since


they are subtended by AC. Because of the reflection, we have ^AC 0 C = 90◦ −
^BAC 0 = 90◦ − α.
Analogously, by observing the circle circumscribed to triangle ABB 0 , we can conclude

^AP B = ^AB 0 B = 90◦ − α.

Therefore, ^CP B = ^AP C + ^AP B = 180◦ − 2α.


Since AC 0 P C is a cyclic quadrilateral, we have

^CP C 0 = 180◦ − ^CAB − ^BAC 0 = 180◦ − 2α.

Therefore, ^CP B = ^CP C 0 , so points P , B and C 0 are collinear.


Let A0 be the point diametrically opposite point A on the circle circumscribed to
triangle ACC 0 . Since AB is the bisector of segment CC 0 , point A0 lies on line AB.
Therefore, ^A0 BP = ^ABC 0 = β. By Thales’ theorem we have ^AP A0 = 90◦ , so

^BP A0 = 90◦ − ^AP B = α.

Therefore, ^AA0 P = 180◦ − α − β = γ and ^A0 AP = 90◦ − ^AA0 P = 90◦ − γ, i.e.


^BAP = 90◦ − γ.
On the other hand, ^BOA = 2γ and ^BAO = 90◦ − γ, so we can conclude that A,
O and P lie on the same line. This completes the proof.
12 Final round – National competition

1.5. We claim that the sought numbers are all multiples of 3.


Note that for N = 3 it is possible to achieve that black unit squares become white,
and vice versa, by choosing each of the four 2 × 2 squares exactly twice.

=⇒ =⇒ =⇒ =⇒

=⇒ =⇒ =⇒ =⇒

Moreover, for a 3 × 3 board whose corner and central squares are white, and the
remaining four squares are black, we can change the colour of black squares to white
and vice versa by choosing each of the four 2 × 2 squares once.

=⇒ =⇒ =⇒ =⇒

If N is divisible by 3, we can divide the board into disjoint 3×3 boards. As previously
shown, we can conclude that for such N all black unit squares can be coloured white,
and vice versa. Let N = 3K + L, for L ∈ {1, 2} and K ∈ N.
We claim that, if 3 does not divide N , it is not possible to change the colour of all
black squares to white, and vice versa.
Note that a black unit square will become white if and only if the number of steps
in which we change the colour of that square gives remainder 2 when divided by 3.
Analogously, a white unit square will become black if and only if the number of steps
in which we change the colour of that square gives remainder 1 when divided by 3.
Moreover, note that any 2 × 2 square needs not be chosen more than twice, since the
colours of the unit squares which we obtain after 3q + r steps are the same as the
colours obtained after r steps, for r ∈ {0, 1, 2} and q ∈ N.
Observe only the first two rows of the N × N board and assume that the first unit
square in the first row is black. The 2×2 square in the first two colums must be chosen
twice. After that, the second unit square in the first row is grey, so the 2 × 2 square
in the second and third column must be chosen twice. By doing this, we achieve that
the second unit square is black, and the third one is white. It follows that we must
not choose the 2 × 2 square in the third and fourth row. Analogously, we conclude
that the 2 × 2 square in the fourth and fifth column must be chosen once, the one
in the fifth and sixth row must be chosen once, and the one in the sixth and seventh
column must not be chosen. In this way we can conclude exactly what must be done
Final round – National competition 13

with each 2 × 2 square in the top 2 × 3K part of the board, i.e. the 2 × 2 squares must
be chosen, in order,

2, 2, 0, 1, 1, 0, 2, 2, 0, . . .

times, where this sequence is periodical with the period 6.


Regardless of whether L = 1 or L = 2, it is unambiguously determined how many
times we need to choose the 2 × 2 square in the last two columns in order to achieve
that the next to last unit square in the first row changes its colour from black to white
or vice versa. However, that number is different from the number of times we would
need to chose that same 2 × 2 square in order to change the colour of the last unit
square from black to white or vice versa. This completes the proof.

∗∗∗

2.1. By completing the squares of sums, the given equation can be written as

1 2 3 2 5 2 7 2
Å ã Å ã Å ã Å ã
x+ + y+ + z+ + w+ = 25.
2 2 2 2

By the inequality of arithmetic and quadratic means, we have


(x + 12 ) + (y + 32 ) + (z + 52 ) + (w + 72 )
  4
(x + 12 )2 + (y + 32 )2 + (z + 25 )2 + (w + 72 )2
6 .
4
We can conclude that (x + 12 ) + (y + 32 ) + (z + 52 ) + (w + 72 ) 6 10, i.e.

x + y + z + w 6 2.

Note that the value 2 can indeed be attained, e.g. for x = 2, y = 1, z = 0, w = −1.
Therefore, the largest possible value of x + y + z + w is 2.

2.2. Denote a = |BC|, b = |CA|, c = |AB|. Let P be the area, s = 21 (a + b + c) the


semiperimeter, and r the radius of the incircle of triangle ABC.
We can divide triangle ABC into three smaller triangles by connecting point S to
vertices A, B and C.

T S

B C
14 Final round – National competition

By adding the areas of triangles ABS, BCS and CAS, we get the area of triangle
ABC, so we have
2P = 10c + 7a + 4b.
Analogously, by observing point T we get

2P = 4c + 10a + 16b.

If we multiply the first equality by 2 and add it to the second equality, we get

6P = 24(a + b + c) = 48s, i.e. P = 8s.

Since P = rs, we have r = 8.

2.3. The given equality can be written more conveniently as:

(a − b)(1 + 4(a + b)) = b2 .

It suffices to show that the bracketed expressions are coprime, since that implies that
a − b and 1 + 4(a + b) are squares of positive integers.
Denote by d the greatest common divisor of a − b and 1 + 4(a + b). We have

d = M (a − b, 1 + 4(a + b)) = M (a − b, 1 + 4(a + b) − 4(a − b)) = M (a − b, 1 + 8b).

From this, it follows that d divides 1 + 8b.


From the initial equality, we have that d also divides b2 .
Since M (1 + 8b, b2 ) = 1, we conclude that d = 1, i.e. a − b and 1 + 4(a + b) are coprime,
which completes the proof.

2.4. Denote α = ^CAB, β = ^ABC, γ = ^BCA.

B K C

L
P′ = P M
Q′ = Q

k
Final round – National competition 15

The centre of circle k is the intersection of the bisectors of angles ^CBL and ^BCM .
Therefore, it suffices to show that those bisectors intersect line LM at points P and
Q respectively.
Denote by P 0 and Q0 the intersections of line LM with bisectors of angles ^KBL and
^KCM , respectively. We have to show that P 0 = P and Q0 = Q.
Triangles KBP 0 and LBP 0 are congruent because they have a common side BP 0 , and
|KB| = |LB|, ^KBP 0 = ^LBP 0 . It follows that ^P 0 KB = ^BLP = ^ALM .
Triangle ALM is isosceles, so ^AM P 0 = ^AM L = ^ALM = ^P 0 KB = 180◦ −
^P 0 KC. From this it follows that quadrilateral M CKP 0 is cyclic.
Since |M C| = |KC|, by the inscribed angle theorem we have ^M P 0 C = ^KP 0 C.
Since we also have ^KP 0 B = ^LP 0 B, it follows that

180◦ = ^M P 0 C + ^KP 0 C + ^KP 0 B + ^LP 0 B = 2(^KP 0 C + ^KP 0 B) = 2^CP 0 B,

therefore the angle ^CP 0 B is right. From this we can conclude that point P 0 lies on
the circle with diameter BC.
Analogously, we can show that Q0 lies on the circle with diameter BC.
Now we have P 0 , Q0 ∈ {P, Q}, since both P 0 and Q0 are the intersections of line LM
and the circle with diameter BC. This is possible only if P = P 0 and Q = Q0 ;
otherwise the centre of circle k would be on the same side of line LM as point A.
This completes the proof.

2.5. Let us call one arrangement of colours on the streets a colouring, and one se-
lection of a square and the change of colours of all streets leading to that square a
transformation.
Note that if we can get from one colouring to another by a sequence of transformations,
then we can also get from the second colouring to the first one using that same
sequence of transformations.
Therefore, the initial claim is equivalent to claiming that, if we start from a colouring
where all streets have the same colour, there exists a colouring that cannot be achieved
by any sequence of transformations.
In any sequence of transformations, we can assume that each square either does not
appear or appears exactly once. Namely, if we perform an even number of trans-
formations on some square, the result will be the same as if we did not perform any
transformations, and an odd number of transformations will have the same result as
if there was exactly one transformation.
Furthermore, note that in the sequence of transformations which are performed to get
from one colouring to another, the order of the transformations is not important, i.e.
it is only important to determine the set of squares included in these transformations.
The number of different sets of squares is 2N , since for each of the N squares we can
decide whether it is included in the set or not.
Note that the set where all squares are included leads to the colouring which is
identical to the initial colouring, since we changed the colour of each street exactly
16 Final round – National competition

twice. The same result can obviously also be obtained if we do not include any square
in the set. Since those two sets lead to the same colouring, the number of different
colourings that can be achieved from any initial colouring is not greater than 2N − 1.
We have two possible initial colourings (one where all the streets are red, and one
where all the streets are blue), so the total number of colourings that can be achieved
frome an initial colouring is not greater than 2 · (2N − 1) = 2N +1 − 2.
Since there are M streets, and each street can be coloured in one of two possible
colours (red or blue), the total number of possible colourings is 2M .
Since M > N , we have 2M > 2N +1 > 2N +1 − 2, so we can conclude that there exists
a colouring that cannot be achieved from the inital monochromatic colourings. This
completes the proof.

∗∗∗

3.1. Since | sin z| 6 1 and | cos z| 6 1, using the triangle inequality one gets

sin x sin y sin z + cos x cos y cos z 6 | sin x sin y sin z + cos x cos y cos z|
6 | sin x|| sin y|| sin z| + | cos x|| cos y|| cos z|
6 | sin x|| sin y| + | cos x|| cos y|. (1)

There exist real numbers x0 and y 0 such that

sin x0 = | sin x|, cos x0 = | cos x|, sin y 0 = | sin y| and cos y 0 = | cos y|. (2)

Therefore, (1) and (2) imply

sin x sin y sin z + cos x cos y cos z 6 sin x0 sin y 0 + cos x0 cos y 0 = cos(x0 − y 0 ) 6 1.

In the above inequality, equality is attained when x = y = z = 0, so the maximum


value of sin x sin y sin z + cos x cos y cos z equals 1.

3.2. By considering the given expression modulo 8, we get

(a + 3b)(5a + 7b) ≡ 5a2 + 22ab + 21b2 ≡ 5a2 − 10ab + 5b2 = 5(a − b)2 (mod 8).

Let us assume the contrary, that (a + 3b)(5a + 7b) = k 2 for some positive integer k.
Then
k 2 ≡ 5(a − b)2 (mod 8).
Since a and b are of different parity, a − b is odd, so that (a − b)2 gives remainder 1
when divided by 8.
This implies k 2 ≡ 5 (mod 8). However, 5 is not a quadratic residue modulo 8.
We have arrived at a contradiction, so we conclude that the initial assumption was
false. Therefore, (a + 3b)(5a + 7b) is not a perfect square.
Final round – National competition 17

3.3. The original equation is equivalent to P (x2 ) − 1 = P 2 (x) − 2P (x) + 1, i.e.

P (x2 ) − 1 = (P (x) − 1)2 .

If we define Q(x) := P (x) − 1, the above equation becomes

Q(x2 ) = Q2 (x). (3)

If Q(x) is a constant polynomial, (3) implies that it has to be 0 or 1. On the other


hand, if the degree of Q is a positive integer n, we can write Q(x) = an xn + R(x),
where R is a polynomial of degree r < n. Plugging this into (3), we get

an x2n + R(x2 ) = a2n x2n + 2an xn R(x) + R2 (x).

We conclude an = 1 and R(x2 ) = 2an xn R(x) + R2 (x).


Notice that the left-hand side of the above equality is of degree 2r, while the right-
hand side is of degree n + r. Since r < n, it follows that r = 0. Therefore, Q(x) = xn .
The solutions of the inital equation are P (x) = 1, P (x) = 2 and P (x) = xn + 1, for
any positive integer n.

3.4. Let us denote β = ^ABC, γ = ^BCA.

F G
E

B M D C

Since BF C is a right-angled triangle, we have ^BCF = 90◦ − β. As CDH is also a


right-angled triangle, we get ^DHC = β.
It follows that ^AHF = β (because the angles ^DHC and ^AHF are vertically
opposite).
As ABKC is a cyclic quadrilateral, inscribed angles which subtend AC are equal to
^AKC = ^ABC = β. We have that ^ACK is a right angle because it subtends
18 Final round – National competition

the diameter AK (using Thales’ theorem). We now infer that the triangles AF H and
ACK are similar (because we know that ^AF H and ^ACK are right angles, whereas
^AKC = ^AHF = β). From this similarity, we get
|AF | |AC|
= . (4)
|AH| |AK|

We also know that AF HE is a cyclic quadrilateral, because ^AF H + ^HEA =


90◦ + 90◦ = 180◦ .
Therefore, ^F EA = ^F HA = β holds, because both these angles subtend the chord
AF . Thus, triangle AF E shares an angle (^BAC = ^F AE) with triangle ABC.
Since we also have ^F EA = β, it follows that AEF and ABC are similar, so that
|AE| |AB|
= . (5)
|AF | |AC|

The quadrilateral ABKC being cyclic, we know that the inscribed angles which sub-
tend AB are equal to ^BKA = ^BCA = γ.
In triangle AF E we have ^AF E = γ (by similarity with triangle ABC), so its
complementary angle equals ^HF E = 90◦ − γ. Notice that ^HAE = ^HF E =
90◦ −γ, because these are both inscribed angles inside the cyclic quadrilateral AF HE.
It follows that the triangles AGE and AM B are similar, so that
|AG| |AM |
= . (6)
|AE| |AB|

|AG| |AM |
By multiplying (4), (5) and (6) we obtain the equality of ratios and .
|AH| |AK|
Finally, using the converse to the intercept theorem, we conclude that the lines GM
and HK are parallel.

3.5. First, notice that none of the triangles formed by taking three vertices of the
given polygon is equilateral. If there were one, we would have the same number of
vertices of the original polygon between any two of the triangle’s vertices. This would
imply that 3 divides 2014, which is false.
Let us now prove that each segment (whose endpoints are vertices of the given poly-
gon) belongs to exactly 3 different isosceles triangles.

even number of vertices

odd number of vertices


Final round – National competition 19

Consider an arbitrary segment. Its endpoints divide the circumcircle of the polygon
into two arcs containing 2015 vertices. One of the arcs contains an odd number,
and the other contains an even number of vertices. Therefore, there exists a unique
isosceles triangle with the chosen segment as its base (the third vertex is the midpoint
of the arc containing an odd number of vertices). In a similar fashion, we infer that
the chosen segment is a base in exactly two isosceles triangles – there are two possible
choices of the third segment, both on the longer arc.
Let B be the number of blue points, so that B + R = 2017. Let db , dr , ds , be the
number of blue (both endpoints blue), red, and two-colour segments, respectively.
Furthermore, let tb , tr , tsb , tsr be the number of isosceles triangles with the vertices
coloured all blue, all red, two blue – one red, and two red – one blue, respectively.
From the fact that each segment belongs to three isosceles triangles, we get:

3db = 3tb + tsb ,


3dr = 3tr + tsr ,
3ds = 2tsb + 2tsr .

This implies 3db + 3dr − 23 ds = 3tb + 3tr , i.e.

1 1
tb + tr = db + dr − ds = · (B(B − 1) + R(R − 1) − BR),
2 2
which depends only on the number of the red and blue points, but not on their
arrangement.

∗∗∗

n n n n
4.1. Consider and . These are also positive divisors of n which satisfy > .
d c d c
We now have
n n
− > 1,
d c
from which
dc d2
c−d> >
n n
follows.

4.2. Considering x = −f (y) we get f (0) = f (f (y)) − (f (y))2 . Denoting f (0) by a


and substituting y = 0 we arrive at f (a) = a2 + a. Now we return to the original
equality and plug in y = 0:

f (x + a) = f (a) + 2xa + x2 = x2 + 2xa + a2 + a = (x + a)2 + a.

Thus, all the possible solutions are of the form f (x) = x2 + a, where a is a real
constant. A routine check shows that every function of this form is indeed a solution
to the original equation.
20 Final round – National competition

4.3. Let P be a marvelous point of ABC and let (without loss of generality) the area
of ABC be 27. We first note that among the 27 rays, there must always be P A, P B
and P C; otherwise not all the shapes determined by the rays will be triangles.
Each of the 27 smaller triangles is contained within (or is equal to) one of the triangles
P AB, P BC, P CA. Let P AB contain m, and P BC contain n smaller triangles.
Numbers m and n are elements of {1, 2, . . . , 25} and they satisfy m + n 6 26.
We now consider the function which assigns the pair (m, n) to each marvelous point
of ABC. We claim that it is a bijection from the set of marvelous points to the set
of pairs of positive integers (m, n) which satisfy m + n 6 26.
We first show that for each pair (m, n) we have at most one marvelous point. Let P
be a marvelous point assigned to (m, n). This means that P AB is divided into m
triangles, while P BC is divided into n triangles, each of whose areas equal 1. The m
2m
triangles contained within P AB have a common altitude whose length equals |AB| , so
we know that point P lies on the line p which is parallel to AB, and whose distance
2m
to AB equals |AB| . Analogously, we infer that point P lies on the line q which is
2n
parallel to BC, and whose distance to BC equals |BC| . Since p and q are uniquely
determined by m and n, the fact that their intersection contains at most one point
shows that there is at most one point to which the pair (m, n) is assigned.

P p

A B

We now show that each pair (m, n) is indeed assigned to a marvelous point. Given
m and n, consider the point P defined as the intersection of lines p and q, as in
the previous paragraph. Drawing rays from point P , we can divide triangles P AB
and P BC into m and n smaller triangles (of area equal to 1), respectively. Finally,
triangle P AC has area 27 − m − n, so we can divide it into 27 − m − n triangles of
area 1 by drawing rays originating in P . Thus, we can divide the triangle ABC into
27 smaller triangles of equal areas by drawing rays originating in P . By definition,
this means that P is a marvelous point to which the pair (m, n) is assigned.
We have yet to determine the number of ways in which the pair (m, n) can be chosen.
For each m ∈ {1, 2, . . . , 25}, we can choose n in exactly 26 − m ways. Therefore, the
solution is
25
X 25 · 26
(26 − m) = = 325.
m=1
2
Final round – National competition 21

4.4. Let the line AQ intersect the circumcircle of BOC at point Q and T 0 . Notice
that OQ is a diameter of this circle, so ^OT 0 Q = 90◦ .
Let P be the midpoint of the segment AC. Since O is the centre of the circumcircle
of ABC, we have ^OP A = 90◦ .
From here, we get ^OT 0 A = 180◦ − ^OT 0 Q = 90◦ = ^OP A, which implies that the
quadrilateral AOT 0 P is cyclic.

A M

P
O
T′ = T

B N C

Notice that point P lies on the line M N (moreover, P is the midpoint of M N ) because
AN CM is a parallelogram. Thus,
1
^AT 0 M = ^AT 0 P = ^AOP = ^AOC = ^ABC.
2

As in the previous solution, we show that the quadrilateral ABN T 0 is cyclic and that
the points M , N and T 0 are collinear. This implies T ≡ T 0 , which concludes the
proof.

4.5. We claim that the required number is 20162 + 1. Let us first show that we can
find an arrangement of 20162 ladybirds and chose their paths so that no two occupy
the same square at the same moment in time. We place the ladybirds in the lower
left 2016 × 2016 square of the board and let them all move in the same manner: up,
right, down, left, up, right. . . We see that no two ladybirds will meet after the first 4
22 Final round – National competition

seconds. Since we arrive at the initial position after that, we see that no two ladybirds
will ever occupy the same square at the same time.

We now show that, if the board contains 20162 + 1 ladybirds, a collision must occur,
regardless of the initial arrangement and the ladybirds’ paths.
We label the squares with one of the four labels, A, B, C and D, so that the squares in
odd rows alternate between labels A and B, while the squares in even rows alternate
between C and D. We will call a square which has been labelled by A an A-square.

A B A B A B A B A

C D C D C . . .D C D C

A B A B A B A B A

C D C D C D C D C
...

...
.
..

A B A B A B A B A

C D C D C D C D C
...
A B A B A B A B A

Two observations are crucial for our solution. A ladybird occupying a B-square or
a C-square will after one second move to an A-square or a D-square. Similarly, a
ladybird which occupies an A-square will after two seconds be sitting in a D-square.
The board contains 20162 + 1 ladybirds, so we can assume that at least 1008 · 2016 + 1
of them occupy an A- or a D-square. If the opposite were true, we would have at least
1008 · 2016 + 1 on a B-square or a C-square, so that after one second we would arrive
at the desired situation. Since the number of D-squares equals 10082 , the A-squares
contain at least 10082 + 1 ladybirds.
All the ladybirds which are now occupying the A-squares will after two seconds move
to D-squares. This means that there will be at least 10082 + 1 ladybirds on 10082
D-squares, so that at least two of them will be in the same square.
4 Croatian Mathematical Olympiad

Day 1

A1. Find the smallest real constant C such that for any positive real numbers a1 , a2 ,
a3 , a4 and a5 (not necessarily distinct), one can always choose distinct subscripts
i, j, k and l such that
ai
− ak 6 C.

a (IMO Shortlist 2016)
j al

C1. Prove that it is possible to colour each positive integer with one of three colours
so that the following conditions are satisfied:
i) For each n ∈ N0 , all positive integers x such that 2n 6 x < 2n+1 have the
same colour.
ii) There are no positive integers x, y and z of the same colour (except x =
y = z = 2) such that x + y = z 2 . (B. Green, S. Lindqvist, arXiv:1608.08374)

G1. Let ABC be a triangle such that |AB| < |BC|, and let I be its incentre. Let M
be the midpoint of the segment AC, and N midpoint of the arc AC ¯ of the
circumcircle of that triangle containing point B. Prove that
^IM A = ^IN B. (Russia 2004)

N1. Prove that there are no positive integers m and n such that
5m3 = 27n4 − 2n2 + n. (Nikola Adžaga)

∗∗∗

Day 2

A2. Determine all functions f : R → R such that


xf (x) − yf (y) = (x − y)(f (x + y) − xy)
holds for all real numbers x and y. (Tonći Kokan)

C2. In an organization there are three committees. Each person belongs to exactly
one committee. For any two persons belonging to different committees, in the
third committee there are exactly 10 people that both persons know and exactly
10 people that both persons do not know. All acquaintances are mutual. How
many people are there in all three committees? (Russia 2008)

23
24 Croatian Mathematical Olympiad

G2. Let M be a point in the interior of the triangle ABC. The line AM intersects the
circumcircle of the triangle M BC once more at D, the line BM intersects the
circumcircle of the triangle M CA once more at E, and the line CM intersects
the circumcircle of the triangle M AB once more at F . Prove the inequality
|AD| |BE| |CF | 9
+ + > . (Tajikistan 2014)
|M D| |M E| |M F | 2

N2. Let τ (n) be the number of positive divisors of n. Let τ1 (n) be the number
of positive divisors of n which give remainder 1 when divided by 3. Find all
possible integral values of the fraction
τ (10n)
. (IMO Shortlist 2016)
τ1 (10n)

∗∗∗

Final test for MEMO team selection

A3. Let a, b and c be positive real numbers. Prove that


3 4a + b 4b + c 4c + a
< + + < 9. (Japan 2009)
2 a + 4b b + 4c c + 4a

C3. A crazy bishop is an unorthodox chess piece which turns towards one of the four
diagonally contiguous cells and attacks all cells directly in front, to the left, and
to the right of it (just like a regular chess bishop which cannot see behind his
back). We say that two cells of a board are diagonally contiguous if they have
exactly one vertex in common.
Determine the largest positive integer N with the following property: It is
possible to put N crazy bishops on an 8 × 8 gaming table so that none of them
is attacked by another crazy bishop. (Russia 2013)

G3. Let AD be the altitude of an acute-angled triangle ABC. On the line AD there
are distinct points E and F such that |DE| = |DF | and the point E is inside
the triangle ABC. The circumcircle of the triangle BEF meets segments BC
and AB again at points K and M , respectively. The circumcircle of the triangle
CEF meets segments BC and CA again at points L and N , respectively.
Prove that the lines AD, KM and LN are concurrent. (Go Geometry)

N3. Determine all positive integers n for which there exist positive integers a and b
such that
(n2 + 2)a = (2n − 1)b . (Ukraine 2013)

∗∗∗
Croatian Mathematical Olympiad 25

Final test for IMO team selection

A4. Prove that for all positive real numbers a, b and c the following inequality holds
 
a b c ab + bc + ca 5
+ + + 2 2 2
> . (Ukraine)
b+c c+a a+b a +b +c 2

C4. In some archipelago, there are 2017 islands labelled 1, 2, . . . , 2017. Two agencies,
Red dragon and Blue eye, are planning to establish ship lines between certain
islands. Exactly one agency will operate between each two islands, with a ship
line from the island labelled with a smaller number to the island labelled with
a larger number.
An arrangement of ship lines is good if there are no two islands, labelled A
and B, where A < B, such that it is possible to go from A to B in two ways:
one, using only the ships owned by Red dragon, and the other, using only ships
owned by Blue eye.
Determine the number of good arrangements of ship lines. (Russia 2005)

G4. Let ABC be a triangle with |AB| = |AC| > |BC| and let I be its incentre. The
line BI meets segment AC at point D, and the line through D perpendicular
to AC meets AI at point E. Prove that J, the reflection of I in AC, lies on the
circumcircle of triangle BDE. (IMO Shortlist 2016)

N4. Denote by N the set of all positive integers. Find all functions f : N → N such
that for all positive integers a and b holds that

f (a) + f (b) − ab | af (a) + bf (b). (IMO Shortlist 2016)

4.1 Solutions

Day 1

A1. The answer is C = 21 .


Consider the numbers 1, 2, 2, 2, n, where n > 5. The fractions formed by two of these
numbers in ascending order are
1 2 1 n
< < < 1 < 2 < < n.
n n 2 2
The smallest two fractions cannot be obtained simultaneously. Therefore, the min-
1 2 1 2
imum value of the difference is − . This shows that C > − for every positive
2 n 2 n
1
integer n > 5. Hence C > .
2
26 Croatian Mathematical Olympiad

For any positive real numbers a1 6 a2 6 a3 6 a4 6 a5 , consider the five fractions


a1 a3 a1 a2 a4
, , , , .
a2 a4 a5 a3 a5

Each of them lies in the interval [0,1]. Therefore, by the Pigeonhole Principle, at
least three of them must lie in 0, 21 or lie in 12 , 1 simultaneously. In particular,

there must be two consecutive terms which belong to an interval of length 21 . In other
words, the difference of these two fractions is less than 12 . As the indices involved
in these fractions are distinct, we conclude C 6 21 . This shows that C = 12 is the
smallest possible choice.

C1. Let cn denote the colour of positive integers x such that 2n 6 x < 2n+1 . We
will determine colours cn inductively. First, let us choose c0 , c1 and c2 to be three
different colours. Next, for each n > 3 let cn be the colour different from cbn/2c and
cbn/2c+1 . Note that this is well-defined since bn/2c + 1 < n for all n > 3.
By construction condition i) holds. Let us prove that condition ii) also holds. Let x,
y and z be positive integers of the same colour such that x + y = z 2 . Without loss of
generality we may assume that x 6 y. Let n be an integers such that 2n 6 y < 2n+1 .
Then we obviously have
2n < x + y < 2n+2 ,
and hence 2n/2 < z < 2(n+2)/2 . It follows that 2bn/2c < z < 2bn/2c+2 . Since y and z
have the same colour, by construction it follows that n 6 2. From here we conclude
x 6 y < 8 and z = 2 or 3. Direct verification shows that the only possibility is
x = y = z = 2.

G1. Let N 0 be the second intersection of the circumcircle of triangle ABC and line
BI. It is the midpoint of the arc CA
¯ which does not contain B, it is diametrically
opposite to N and M ∈ N N . By Thales’ theorem, we have ^N BI = ^N BN 0 = 90◦ ,
0

so it suffices to show that


^IM N 0 = ^N IN 0 .

N
B

C M I′ A

N′
Croatian Mathematical Olympiad 27

To prove the equality of angles, we will prove that the triangles IM N 0 and N IN 0 are
similar. Since these triangles have the common angle at N 0 , it is enough to prove
|IN 0 | |N N 0 |
0
= ⇐⇒ |IN 0 |2 = |M N 0 | · |N N 0 |.
|M N | |IN 0 |

Consider the triangle N AN 0 . It has a right angle at vertex A. By Euclid’s theorem


it follows that
|AN 0 |2 = |M N 0 | · |N N 0 |.
^CAB ^ABC
Since ^AIN 0 = ^IAN 0 = + , triangle AIN 0 is isosceles and we have
2 2
|AN 0 | = |IN 0 |. From there it follows that |IN 0 |2 = |AN 0 |2 = |M N 0 | · |N N 0 |, which
proves the desired statement.

N1. Let us write the equation in the form


5m3 = n(27n3 − 2n + 1)
and note that n and 27n3 − 2n + 1 are relatively prime numbers. We consider two
cases:
a) n = 5a3 and 27n3 − 2n + 1 = b3 for some positive integers a and b.
Inequality (3n − 1)3 = 27n3 − 27n2 + 9n − 1 < 27n3 − 2n + 1 is equivalent to the
inequality 27n2 − 11n + 2 > 0, which holds for all positive integers n. Also, we
have 27n3 − 2n + 1 < 27n3 − 2 + 1 < 27n3 = (3n)3 .
Hence, (3n − 1)3 < 27n3 − 2n + 1 < (3n)3 , which is impossible.
b) n = a3 and 27n3 − 2n + 1 = 5b3 for some positive integers a and b.
From the second equation we conclude that n cannot be divisible by 5. If we have
n ≡ ±1 (mod 5), then 27n3 − 2n + 1 ≡ 1 (mod 5). If we have n ≡ 2 (mod 5),
then 27n3 − 2n + 1 ≡ 3 (mod 5), and if n ≡ 3 (mod 5), then 27n3 − 2n + 1 ≡ 4
(mod 5). In all cases the left-hand side of the equation cannot be divisible by 5.
Hence the equation has no solution.

∗∗∗

Day 2

A2. Let us denote by P (x, y) the equality xf (x) − yf (y) = (x − y)(f (x + y) − xy).
Equality P (x, 1) is xf (x) − f (1) = (x − 1)(f (x + 1) − x), so we conclude
xf (x) − (x − 1)f (x + 1) = f (1) − x(x − 1), ∀x ∈ R. (1)

On the other hand, P (x + 1, −1) is (x + 1)f (x + 1) + f (−1) = (x + 2)(f (x) + x + 1),


so we have
(x + 2)f (x) − (x + 1)f (x + 1) = f (−1) − (x + 1)(x + 2), ∀x ∈ R. (2)
28 Croatian Mathematical Olympiad

Equalities (1) and (2) form a linear system of two equations with indeterminates f (x)
and f (x + 1). By elimination of f (x + 1) we get
f (1) − f (−1)
f (x) = x2 + x + (f (1) + f (−1) − 1), ∀x ∈ R.
2

f (1) − f (−1)
If we denote b = and c = f (1) + f (−1) − 1 we have f (x) = x2 + bx +
2
c, ∀x ∈ R, where b, c ∈ R. We check directly that each such function is indeed a
solution.

C2. Let A, B, C be committees and let them have exactly a, b, c persons, respectively.
To each person assign a point in the plane so that no three points are collinear. We
connect the points corresponding to persons that know each other by a blue segment,
and for those that do not know each other by a red segment. Points corresponding
to persons in the same committee are not connected.
First, let us count the monochromatic triangles (all three sides are of the same colour).
For each pair of persons x ∈ A, y ∈ B there are exactly 10 monochromatic triangles
(if xy is a blue segment, there are 10 points z ∈ C such that segments xz and yz
are blue, and analogously for red segments). Hence, the number of monochromatic
triangles is 10ab. In the same way we conclude that the number of monochromatic
triangles is 10bc and 10ca, so we conclude a = b = c.
The number of all triangles is a3 , while the number of monochromatic triangles is
10a2 . Let us count bichromatic triangles. For each pair x ∈ A, y ∈ B we have exactly
10 bichromatic triangles such that the segments xz and yz have the same colour, i.e.
there are 10a2 such triangles, while there are 10a2 triangles such that yx and zx are
of the same colour, and 10a2 triangles such that xy and zy are of the same colour.
Hence there are 30a2 bichromatic triangles.
Finally, we have a3 = 10a2 + 30a2 , so a = 40. This means that there are in total 120
members in all three committees.

G2. Since
|AD| |BE| |CF | |AM | + |M D| |BM | + |M E| |CM | + |M F |
+ + = + +
|M D| |M E| |M F | |M D| |M E| |M F |
|AM | |BM | |CM |
=3+ + + ,
|M D| |M E| |M F |
it suffices to show that
|AM | |BM | |CM | 3
+ + > .
|M D| |M E| |M F | 2
Quadrilaterals M BDC and M CEA are cyclic, so

^BCD = ^BM D = ^EM A = ^ECA.

Also, we have
^DBC = ^DM C = 180◦ − ^CM A = ^CEA.
Croatian Mathematical Olympiad 29

Hence we conclude that the triangles BDC and EAC similar. Analogously, we prove
that the triangle BAF is similar to them.

E
D

A B

By Ptolemy’s theorem for quadrilaterals M BDC, M CEA and M AF B, and by using


the ratios from the similarity of BDC, EAC and BAF , we have:
|CD| |DB|
|M D| = |BM | · + |CM | · ,
|BC| |BC|
|AE| |EC| |DB| |BC|
|M E| = |CM | · + |AM | · = |CM | · + |AM | · ,
|CA| |CA| |CD| |CD|
|BF | |F A| |BC| |CD|
|M F | = |AM | · + |BM | · = |AM | · + |BM | · ,
|AB| |AB| |DB| |DB|
from where it follows that
|AM | |BM | |CM | |AM | · |BC|
+ + =
|M D| |M E| |M F | |BM | · |CD| + |CM | · |DB|
|BM | · |CD|
+
|CM | · |DB| + |AM | · |BC|
|CM | · |DB|
+ .
|AM | · |BC| + |BM | · |CD|

Let us denote x = |AM | · |BC|, y = |BM | · |CD|, z = |CM | · |DB|. We need to prove
the inequality
x y z 3
+ + > .
y+z z+x x+y 2
This is Nesbitt’s famous inequality, so the proof is finished.
30 Croatian Mathematical Olympiad

N2. Let n = 3a · b · m, where a is a non-negative integer, b and m positive integers


such that all prime factors of b give remainder 1 when divided by 3, and all prime
factors of m give remainder 2.
Then τ (10n) = (a + 1) · τ (b) · τ (10m).
For τ1 (n), a prime factor 3 cannot be taken, prime factors of b can be taken arbitrarily,
and the only condition is to take an even number of prime factors of m. So, τ1 (10n) =
τ (b) · τ1 (10m).
Hence we have
τ (10n) τ (10m)
= (a + 1) · .
τ1 (10n) τ1 (10m)

Let m = 2c1 · 5c2 · pc33 · · · pckk . We have τ (10m) = (c1 + 2)(c2 + 2)(c3 + 1) · · · (ck + 1).
Let us consider two cases.

• Number τ (10m) is even. One of the brackets (in the expression for τ (10m))
is even. Let it be the bracket corresponding to the factor pj and let it be 2t.
If we choose a prime factor (different from pj ) arbitrarily, then the number of
occurrences of pj can be chosen in exactly t ways. So, in this case τ1 (10m) =
1
2 τ (10m).
In this case, the given fraction can take the value 2(a + 1), i.e. any even positive
integer is a possible value and there are no other.
• Number τ (10m) is odd. All brackets are odd, which means that the exponent
of each prime factor is even. For each combination of exponents giving the even
sum, the first exponent which is not 1 can be reduced by 1, while for the odd
sum the last exponent which is not 0 (or the first if all of them are 1) can be
increased by 1. In that way, we establish a bijection between combinations with
even and odd sum (except for the combination in which all exponents are 0).
1
We conclude τ1 (10m) = (τ (10m) + 1). Let τ (10m) = t. So we have
2

τ (10n) t
= 2(a + 1) · .
τ1 (10n) t+1

Since we are interested in integral values of the fraction, we see that t + 1 divides
2(a + 1), so the fraction above is of the form tt0 , where t0 is a positive integer.
Since all brackets are odd in t, the first two are at least 3, so t is a composite
number. Hence the given fraction cannot be an odd prime number.
It remains to show that the given fraction can achieve the value xy, where x and
y are non-negative integers greater than 1. We can achieve this for
xy−1
n=3 2 · 2x−2 · 5y−2 .

∗∗∗
Croatian Mathematical Olympiad 31

Final test for MEMO team selection

A3. Notice that


1 4a + b 4a + b 4(a + 4b)
= < < = 4,
4 4(4a + b) a + 4b a + 4b
and analogously
1 4b + c
< < 4,
4 b + 4c
1 4c + a
< < 4.
4 c + 4a
Without loss of generality we may assume that a 6 b 6 c. Hence we have
4c + a c + 3c + a c + 4a
= > =1
c + 4a c + 4a c + 4a
and
4a + b a + 3a + b a + 4b
= 6 = 1.
a + 4b a + 4b a + 4b
Finally,
3 1 1 4a + b 4b + c 4c + a
= + +1< + + < 1 + 4 + 4 = 9.
2 4 4 a + 4b b + 4c c + 4a

C3. Imagine the figure of a crazy bishop like a dot from which we draw 3 arrows in
a directions of cells which are attacked by this crazy bishop. By the term diagonal
we consider all the diagonals of the board, not just the main ones. The four corner
squares are also considered diagonals (containing only one square).
On each diagonal there are exactly two directions in which a crazy bishop can move.
Notice that we can have at most one arrow in each direction, otherwise one of the two
arrows which are pointing at the same direction is pointing to the other, i.e. there is
at least one bishop under attack. Therefore, we can have at most 2 arrows on each
diagonal. There are 30 diagonals, which means that there can be at most 60 arrows.
Every crazy bishop contributes with 3 arrows, thus we conclude N 6 20.
An example in which we have 20 crazy bishops (none of them under attack) on an
8 × 8 gaming table is given in the picture.
32 Croatian Mathematical Olympiad

G3. Notice that the circumcentre of triangle BEF is on the segment BC. Therefore,
segment BK is a diameter of the circumcircle of triangle BEF , so ^BM K = 90◦ ,
and analogously ^LN C = 90◦ .

N
E
M

B L D K C

Let lines AD and KM intersect at X. From ^BM X = ^BM K = 90◦ and ^XDB =
^ADB = 90◦ , it follows that the quadrilateral BDXM is cyclic.
Quadrilaterals BM EF and EF CN are cyclic too, so from power of a point theorem
(multiple use) it follows that
|AX| · |AD| = |AM | · |AB| = |AE| · |AF | = |AN | · |AC|,
and because of that, quadrilateral CDXN is also cyclic.
Finally, ^XN C = 180◦ − ^CDX = 90◦ = ^LN C, i.e. point X lies on the line LN ,
which completes the proof.

N3. Notice that n2 + 2 > 3, n2 + 2 > 2n − 1 and also that numbers n2 + 2 and 2n − 1
have the same prime factors.
Let p be a prime factor of n2 + 2 and 2n − 1.
p | n2 + 2 and p | 2n − 1 =⇒ p | (n2 + 2) + (2n − 1) = (n + 1)2 =⇒ p | n + 1.

Therefore, p | 2(n + 1) − (2n − 1) = 3 and we conclude that p = 3.


Hence there are positive integers k and m, k > m, such that n2 + 2 = 3k and
2n − 1 = 3m . We get
4 · 3k = 4n2 + 8 = (3m + 1)2 + 8 = 32m + 2 · 3m + 9,
from which it follows that 3m divides 9, i.e. m 6 2.
Croatian Mathematical Olympiad 33

We have two possibilities:

1) If m = 1, then 2n − 1 = 3, and consequently n = 2 and 3k = 6, which is


impossible.
2) If m = 2, then 2n − 1 = 9, and consequently n = 5 and k = 3.

Therefore, n = 5 is the only positive integer with the given property.

∗∗∗

Final test for IMO team selection

A4. Because of simplicity let A = a2 + b2 + c2 and B = ab + bc + ca. The CSB


inequality gives us

a b c
Å ã
+ + (a(b + c) + b(c + a) + c(a + b)) > (a + b + c)2 ,
b+c c+a a+b

i.e.
a b c A + 2B A
+ + > = + 1.
b+c c+a a+b 2B 2B
Therefore,
  …
a b c ab + bc + ca A B
+ + + > +1+ .
b+c c+a a+b a2 + b2 + c2 2B A

Finally, the inequality of arithmetic and geometric means implies


… … …
A 1 B 1 B 3 AB 3
+ + >3 = ,
2B 2 A 2 A 8BA 2
and the given inequality is proved.

C4. Ship lines maintained by Red dragon we will call red lines, and those maintained
by Blue eye blue lines.
Let an denote the number of good arrangements for an archipelago with n islands.
Consider an archipelago with n + 1 islands and notice island 1. Let A be the set of
all islands which are connected to island 1 with a red line, and let B be the set of all
islands which are connected to island 1 with a blue line.
Note that all lines (if there are any) from some island in A to some island in B must
be blue. Suppose that’s not the case, i.e. that there are islands a ∈ A and b ∈ B such
that a < b and the line a → b is red. We know that line 1 → b is blue, and lines 1 → a
and a → b are red, hence we can go from a to b using only red lines, and also using
only blue lines, which is a contradiction. By the same reasoning we conclude that all
lines (if there are any) from some island in B to some island in A must be red.
34 Croatian Mathematical Olympiad

blue
a b

red

red blue
1

Let set A contain k ∈ {0, 1, . . . , n} islands. That means that set B contains n −
k islands. Note that islands in the set A must form an archipelago with a good
arrangement of ship lines, and the same holds for the set B. On the other hand, for
each selection of archipelagos A and B with good arrangements we get exactly one
archipelago with n + 1 islands and good ship lines arrangement between them.
Now, using mathematical induction, we will prove that an = n!. It is obvious that
a0 = a1 = 1. Suppose that ak = k!, for all k 6 n, for some positive integer n. By
previous reasoning we conclude that
n Ç å n
X n X
an+1 = ak an−k = n! = (n + 1)!.
k
k=0 k=0

Thus, an = n! and finally a2017 = 2017!.

G4. Let Γ be the circle with centre E passing through B and C. Since lines ED and
AC are perpendicular to each other, the point F symmetric to C with respect to D
lies on Γ.
A

Γ
K

D J
E

B C

From ^DCI = ^ICB = ^CBI, the line DC is a tangent to the circumcircle of


triangle IBC. Let K be the symmetric point to I with respect to D. Using power of
Croatian Mathematical Olympiad 35

a point theorem, from

|DC| · |DF | = |DC|2 = |DI| · |DB| = |DK| · |DB|,

we conclude that the point K also lies on Γ. Since segments IK and CF bisect each
other, CKF I is a parallelogram.
From ^F JC = ^CIF = ^F KC, we find that J lies on Γ. This gives |EJ| = |EB|.
Note that line AC is the internal angle bisector of ^BDJ. This shows that DE is
the external angle bisector of ^BDJ as DE ⊥ AC. Together with |EJ| = |EB|, it is
well-known that E lies on the circumcircle of triangle BDJ.

N4. It is given that


f (a) + f (b) − ab | af (a) + bf (b). (3)
Taking a = b = 1 in (3), we have 2f (1) − 1 | 2f (1). Then 2f (1) − 1 | 2f (1) − (2f (1) −
1) = 1 and hence f (1) = 1.
Let p > 7 be a prime. Taking a = p and b = 1 in (3), we have f (p) − p + 1 | pf (p) + 1
and hence
f (p) − p + 1 | pf (p) + 1 − p(f (p) − p + 1) = p2 − p + 1.
If f (p) − p + 1 = p2 − p + 1, then f (p) = p2 . If f (p) − p + 1 6= p2 − p + 1, as p2 − p + 1
is an odd positive integer, we have p2 − p + 1 > 3(f (p) − p + 1), i.e.
1 2
f (p) 6 (p + 2p − 2). (4)
3

Taking a = b = p in (3), we have 2f (p) − p2 | 2pf (p). This implies

2f (p) − p2 | 2pf (p) − p(2f (p) − p2 ) = p3 .

By (4) and f (p) > 1 we get


2 2
−p2 < 2f (p) − p2 6 (p + 2p − 2) − p2 < −p,
3
since p > 7. This contradicts the fact that 2f (p) − p2 is a factor p3 . Thus we have
proved that f (p) = p2 for all primes p > 7.
Let a be a fixed positive integer. Choose a sufficiently large prime p. Consider b = p
in (3). We obtain

f (a) + p2 − pa | af (a) + p3 = a(f (a) + p2 − pa) + p3 − p2 a + pa2 ,

i.e.
f (a) + p2 − pa | p(p2 − pa + a2 ).
As p is sufficiently large and a is fixed, p cannot divide f (a), and so numbers f (a) +
p2 − pa and p are relatively prime. It follows that

f (a) + p2 − pa | p2 − pa + a2 = (f (a) + p2 − pa) + a2 − f (a),


36 Croatian Mathematical Olympiad

i.e.
f (a) + p2 − pa | a2 − f (a).
Note that a2 − f (a) is fixed while f (a) + p2 − pa is chosen to be sufficiently large.
Therefore, we must have a2 − f (a) = 0, so that f (a) = a2 for any positive integer a.
Finally, we check that when f (a) = a2 for any positive integer a, then

f (a) + f (b) − ab = a2 + b2 − ab

and
af (a) + bf (b) = a3 + b3 = (a + b)(a2 + b2 − ab).
The latter expression is divisible by the former for any positive integers a and b. This
shows that f (a) = a2 is the only solution.

c Croatian Mathematical Society, 2017

Editor:
Ivan Kokan

Translation:
Petar Bakić, Matija Bašić, Mea Bombardelli,
Ivan Kokan, Ivan Krijan, Azra Tafro

Figures and typesetting:


Ivan Kokan

Printed in:
Element Ltd.
July 2017, Zagreb, Croatia

Published in 250 copies


Vladimir Varićak – Famous Croatian Mathematician and Physicist

Vladimir Varićak (1865–1942), Croatian mathematician and theoretical physicist was among
the first scientists who discovered deep relationships between two great theories: Lobachevsky’s
(hyperbolic) geometry and Einstein’s relativity theory. Soon after Einstein’s fundamental paper from
1905 on special theory of relativity, Varićak introduced the concept of rapidity (or pseudovelocity) u,
related to the ordinary velocity v by the formula v = c tanh u, where c is the velocity of light.
The first geometric interpretation of the special theory of relativity was given by German
mathematician H. Minkowski in 1908 as a four-dimensional space-time continuum with hyperbolic
metric. A. Einstein and (independently) French mathematician H. Poincaré proved that the relativistic
sum w of velocities v and v' acting in the same direction is given by w = (v + v' ) | (1 + vv' ), where c is
assumed to be normed and equal to 1. A recent simple geometric interpretation of this formula is as
follows. Consider two points on the x-axis, V (v, 0) and V' (v', 0). Now, connect the “south pole”
S (0, −1) with V and let A ≠ S be the intersection of the line SV with the unit circle k ... x2 + y2 = 1,
and similarly, let B ≠ N be the intersection of the line NV' with k, where N (0,1) is the “north pole”
of k. Then the intersection of the line AB with the x-axis is the point W (w, 0). Prove it! (Or see the
paper G. Wanner, From Pappus to Kocik’s diagram for relativistic velocity addition, Elemente der
Mathematik, 72 (2017), 35–38.) So, for instance, the relativistic sum of v = 1/2 and v' = 1/2 is equal
to 4/5, while the sum of 1/2 and 2/3 is equal to 7/8. Varićak’s rapidity shows that the above formula, in
fact, reflects the addition formula for the hyperbolic tangent function: tanh (a + b) = (tanh a + tanh b) |
(1 + tanh a tanh b). He consistently applied hyperbolic trigonometry to prove, for example, Einstein’s
formula for Doppler effect, transformation laws of electromagnetic fields in relativity theory and
many other related results. His main papers on these topics were published in German in journals like
Jahresbericht der Deutschen Mathematiker-Vereinigung, Physikalische Zeitschrift and in journals of
the Croatian (former Yugoslav) Academy of Sciences and Arts. His book written in German, Relativity
in three-dimensional Lobachevsky space was published in Zagreb in 1924, and was translated into
English in 2007.
Academician Vladimir Varićak was a professor of mathematics at the University of Zagreb for
almost 40 years, where he lectured many courses from Real and Complex Analysis, Euclidean and non-
Euclidean Geometry, Set Theory, Topology, Special and General Theory of Relativity, and others. But,
at the beginning of his professional career, 1888–1897, he was a high school teacher of mathematics
and physics, and wrote or translated a couple of textbooks and problem books. He published in total
about 120 papers and books in areas of algebraic analysis, non-Euclidean geometry, theory of relativity,
in history of science about Croatian mathematician, physicist and astronomer Ruđer Josip Bošković
(1711–1787), and books and papers on high school mathematics and other topics. His works are often
quoted in our days, more than 100 years after being published.
Darko Veljan

Вам также может понравиться