43rd International Mathematical Olympiad

Glasgow, United Kingdom, July 2002

1. Let n be a positive integer. let T be the set of points (x, y) in the plane where x and
y are non-negative integers and x + y < n. Each point of T is coloured red or blue. If a
point (x, y) is red, then so are all points (x0 , y 0 ) of T with both x0 ≤ x and y 0 ≤ y. Define
an X-set to be a set of n blue points having distinct x-coordinates, and a Y -set to be a
set of n blue points have distinct y-coordinates. Prove that the number of X-sets is equal
to the number of Y -sets.

Soln. Induct on the number B of blue points in the diagram. The base cases: If there are
no blue points in the bottom row, then there are no X-sets and no Y -sets and so we are
done.
The recursive case: If there is only one blue point at the bottom row, then delete the
botton row to get a new configuration. The number of X-sets in the two configurations are
equal as are the Y -sets. Thus we are done. If there are m(≥ 2) blue points in the bottom
row, then change the leftmost blue point P to red. In the new configuration, the number
of X-sets and Y -sets are equal. Now restore the colour of P to red. The now of X-sets is
changed by a factor of m/(m − 1). The same goes for the number of Y -sets. Thus are are
done again.

Second soln. Assign the value k/(k − 1) to any blue point (x, y) with x + y = n − k,
k = 2, . . . , n. Any point (x, y) with x + y = n − 1 is assigned the value 0 if it is coloured
red and is assigned the value 1 if it is coloured blue. The other red points are assigned
the value 1. Then the number of blue points in each row or column is the product of the
numbers in the row or column. Hence the number of X-sets and the number of Y -sets
are both equal to the product of all the numbers assigned and they are therefore equal in
number.

Third soln. First we observe that when all the points are red, the number of X- and
Y -sets are both zero and so they are equal. Now suppose that in a given configuration,
the number of X-and Y -sets are equal. If (x, y) is a red point with x + y = n − k and
all points (x, y 0 ), (x0 , y) where x0 > x, y 0 > y are blue, then changing the colour of this
point to blue changes the number of blue points in the row and column from k − 1 to k.
Thus the number of X- and Y -sets remain equal in the new configuration. Since we can
construct any configuration starting from the initial configuration with all points red by
changing the red points to blue row by row, we conclude that the number of X- and Y -sets
are equal in any configuration. (Taking this appraoch you must show that you can reach
the configuration you want starting from the initial configuration, i.e., you must indicate
how you can reach the final configuration.)

Fourth soln. This is the reverse of the third solution. If (x, y) is a blue point with
x + y = n − k and all points (x, y 0 ), (x0 , y) where x0 < x, y 0 < y are red, then changing the
colour of this point to red changes the number of blue points in the row and column from k

1
to k−1. The the number of X- and Y -sets are equal in the original configuration if and only
if they are equal in the new configuration. If we continue this process, eventually we reach
the configuration in which all the points are red. In this final configuration the number of
X and Y -sets are equal. Therefore they are also equal in the original configuration. (Note:
The advantage of this apprach over the previous one is that you need not worry about
reaching the final configuration.)

2. Let BC be a diameter of the circle Γ with centre O. Let A be a point on Γ such

that 0◦ < ∠AOB < 120◦ . Let D be the midpoint of the arc AB not containing C. The
line through O parallel to DA meets the line AC at J. The perpendicular bisector of OA
meets Γ at E and at F . Prove that J is the incentre of the triangle CEF .

Soln. Let O be the origin and OA be the y-axis. We may also assume that the radius of
Γ is 2. Let ∠BOA = 2θ, then the coordinates of A, B, C, D, E, F are:

A(0, 2), B(−2 sin 2θ, 2 cos 2θ), C(2 sin 2θ, −2 cos 2θ),
√ √
D(−2 sin θ, 2 cos θ), E( 3, 1), F (− 3, 1).

Then AE = AF = 2. Since arc AF = arc AE, AC bisects ∠F CE. (Note that the
condition ∠AOB < 120◦ implies that A and C are on the opposite sides of EF and so AC
is the internal angle bisector.) Thus J is the incentre of ∆CEF if AJ = AE = AF . Thus
we only need to show that AJ = 2. For some t,

OJ = tDA = t(2 sin θ, 2 − 2 cos θ)

AJ = (2t sin θ, 2t − 2t cos θ − 2)
CA = (2 sin 2θ, −2 cos 2θ − 2) = 2 cos θ(2 sin θ, −2 cos θ)

Since CA is parallel to JA, we have t = 1 or AJ = 2 as required.

A ...............
.............................. .......................
......... .......... ... ........ .......
....... ......... ... ....... ......
....
. .................... .. ..... ......
D ..
..
..........
.. .....
. .
.
.
.
..
.....
.....
.....
....
..... .....
..... .
.
.M .....
.....
.........................................................................................................................................
.. . .
...
F .
...
..
... .....
.....
.....
.
.
....
.....
.....
.....
...... E
......
... ......
....
...............
θ
.....
.....
..... ..
..
..
. .... .
.....
.. . .
....... J
... ...
.. ...
B ..
.. .................. θ
.................. ....... . .
.
.
...................................... .. .
... ....... ..
. ....... .....
..... ..... ....
..... ... ...
.
... .................
.................. ..... .. ...
... .................. ... ...
... .................................
...
...
O ...
.. C
... ...
.
.
... .
.
... .
... ...
... ...
... . ...
..... ..
..... ....
..... .....
...... .....
...... .. .........
........ .
.......... ........
..............................................

Second soln. The fact 0◦ < ∠AOB < 120◦ implies that ∠AOC > ∠OAD and ∠AOJ >
∠OAJ. Thus J is an interior of the segment AC and A, J are on opposite sides of F E
so the diagram is correct. A is the midpoint of arc EAF , so CA bisects ∠ECF . Now

2
since OA = OC, ∠AOD = 21 ∠AOB = ∠OAC so OD is parallel to JA and ODAJ is a
parallelogram. Hence AJ = OD = OE = AF since OEAF (with diagonals bisecting each
other at right angles) is a rhombus. Thus

∠JF E = ∠JF A − ∠EF A = ∠AJF − ∠ECA

= ∠AJF − ∠JCF = ∠JF C.

Therefore, JF bisects ∠EF C and J is the incentre of 4CEF .

3. Find all pairs of integers m, n ≥ 3 such that there exist infinitely many positive integers
a for which
am + a − 1
an + a2 − 1
is an integer.

Soln. The problem is equivalent to the following: Find all pairs of integers m, n ≥ 3 for
which xn + x2 − 1 is a factor of the polynomial xm + x − 1. It’s clear that m > n. Write
m = n + k. Then

xm + x − 1 = xk (xn + x2 − 1) + (1 − x)(xk+1 + xk − 1).

So xn + x2 − 1 divides xk+1 + xk − 1. Now xn + x2 − 1 has a real root α ∈ (0, 1), α is also

a root of xk+1 + xk − 1. Thus αk+1 + αk = 1 and αn + α2 = 1. But k + 1 ≥ n ≥ 3 and so
αn + α2 ≥ αk+1 + αk with equality if and only if k + 1 = n and k = 2. Thus (m, n) = (5, 3)
is the only possible solution. It is easy to check that this is indeed a solution.

4. Let n be an integer greater than 1. The positive divisors of n are d1 , d2 , . . . , dk , where

1 = d1 < d2 < · · · < dk = n.

Define D = d1 d2 + d2 d3 + · · · + dk−1 dk .
(a) Prove that D < n2 .
(b) Determine all n for which D is a divisor of n2 .

Soln. If d is a divisor of n, then so is n/d and n/dk < · · · < n/d2 < n/d1 . Thus

n2
 
2
X 1 X 1 1
D=n ≤ n2 − < = n2 .
di di+1 di di+1 d1

For part (b), note that if n is a prime, then D = n which certainly divides n2 . Now suppose
that n is not prime. As dk−1 is the greatest proper divisor of n, then dk−1 dk = dk−1 n is
the greatest proper divisor of n2 . But n2 > D > dk−1 dk . So D is not a divisor of n2 .
Therefore D is a divisor of n2 if and only if n is prime.

3
5. Find all functions f from the set R of real numbers to itself such that



f (x) + f (z) f (y) + f (t) = f (xy − zt) + f (xt + yz) (∗)

for all x, y, z, t ∈ R.

Soln. It is clear that f (x) ≡ 0, f (x) ≡ 1/2 and f (x) = x2 for all x are solutions. We claim
that there are no other solutions.


Setting x = y = z = 0 gives 2f (0) = 2f (0) f (0) + f (t) . In particular 2f (0) = 4f (0)2
and so f (0) = 0 or 1/2. If f (0) = 1/2 we get f (0) + f (t) = 1 and so f (x) ≡ 1/2.
Suppose f (0) = 0. Then setting z = t = 0 in (∗) gives f (xy) = f (x)f (y). In particular
f (1) = f (1)2 and so f (1) = 0 or 1. If f (1) = 0, then f (x) = f (x)f (1) = 0 for all x.
So we may assume that f (0) = 0 and f (1) = 1. Setting x = 0 and y = t = 1 in (∗),
we have
f (−z) + f (z) = 2f (z) or f (−z) = f (z)
and f is an even function. So it suffices to show that f (x) = x2 for positive x. Note that
f (x2 ) = f (x)2 ≥ 0; as f is even, we have f (y) ≥ 0 for all y. Now put t = x and z = y in
(∗) to get

2
f (x2 + y 2 ) = f (x) + f (y) .
This shows that f (x2 + y 2 ) ≥ f (x)2 = f (x2 ). Hence f is increasing on the positive reals.
Set y = z = t = 1 in (∗) to yield

f (x − 1) + f (x + 1) = 2(f (x) + 1).

By induction on n, it readily follows that f (n) = n2 for all non-negative integers n. As f

is even, f (n) = n2 for all integers n. Since f is multiplicative, f (r) = r2 for all rational
2
numbers r. Supposep f (x) 6= x for some positive x. If f (x) < x2 , take a rational number
a with x > a > f (x). Then f (a) = a2 > f (x), but f (a) ≤ f (x) as f is increasing
on positive reals. This is a contradiction. A similar argument shows that f (x) > x2 is
impossible. Thus f (x) = x2 for all real x.

6. Let Γ1 , Γ2 , . . . , Γn be circles of radius 1 in the plane, where n ≥ 3. Denote their centres

by O1 , O2 , . . . , On , respectively. Suppose that no line meets more than two of the circles.
Prove that
X 1 (n − 1)π
≤ .
Oi Oj 4
1≤i<j≤n

Soln. Consider the circles Γi , Γj and their 4 common tangents. The circle Γi contains 2
minor arcs P Q, RS, each of length

2
θij ≥ sin θij = .
Oi Oj

4
The tangent at any interior point of these minor arcs will intersect only Γj . Thus for each
fixed i, the minor arcs obtained as j varies are disjoint.
.............
..................
... .................................
S β ... ....
......
. ... ..........
... ..... .......................
..................................................................................................................................
β
....................... ..
..
.........
....... ... .............. .....
...
....
..... ... ... ........... R .....
........
. .....
....
...
..
...........
... ......................
....
..
... ... ...
..
.........
. ... ....
............ ...
. ... ....... .. .. ... ...
.... ........ . ..
.... ...
......................................................................................................................
O ...
...
... O i .
... ... .
. ... . . .
. ....... ... . . j ...
.. ...
...
...
...
....
. ...
.
.
.
θ
... .... .......... .
.
.
. ....... .
. ...... ...
..........
. . .
.
.
... ... ..............
... ........ ......
... ........................
.....
..... θ
....... .... ............
. ... ......
Q
......
......
.....................................................................................................................................
.....
... .......
....
...
...
...
....
...
...
..
... . ... ..
. .
. .
......... . ... ... .
...
....
P .. ........ ...... .. .
....
.........................................................................................................

Now enclose the n circles by a convex polygon so that each side is tangent to at least
2 of the circles. So the two sides at vertex Vk is tangent to a circle Γ. The two points
of contact define
P a minor arc. The length of the minor arc is eqaul to the external angle
βk at Vk and k βk = 2π. These minor arcs are disjoint from the minor arcs described
earlier. Thus
XX X XX
2θij + βk ≤ 2nπ and 2θij = 2(n − 1)π.
i j k i j

Therefore XX X X 8
2(n − 1)π ≥ 2θij = 4θij ≥
i j
Oi Oj
1≤i<j≤n 1≤i<j≤n

as required.