1

International Mathematical Olympiad

Preliminary Selection Contest 2010 Hong Kong

Outline of Solutions

Answers:
1. 1000 2. 311550 3. 3888 4. 88
5. 501 6. 721 7. 2024 8. 32768
9. 3100
10. 3 3 t 11. 2 3 + 12. 2 3 3 3 t +
13. 203 14. 10 15. 213 16. 132
17. 8 3
18.
329
100

19.
15
9

20. 5

Solutions:

1. Note that
3 3
( ) ( 1) f n n n = . It follows that
3 3 3 3 3 3 3 3
(1) (2) (2010) (1 0 ) (2 1 ) (2010 2009 ) 2010 201 1000 f f f + + + = + + + = = .
Since
3
201 has unit digit 1, it is clear that the answer is 1000.
Remark. Even without realising
3 3
( ) ( 1) f n n n = , one could still easily observe that
3
(1) (2) ( ) f f f n n + + + = by computing this sum for some small n, which is a natural
approach given that the question asks for something related to (1) (2) (2010) f f f + + + .

2. The digit which is even must be the unit digit 0. Hence it remains to find the smallest integer
positive k for which all digits of 201k are odd.
Clearly we must have 150 k > , for if 100 k < then the hundreds digit of 201k is even (same as
the unit digit of 2k); if 100 150 k s < the ten thousands digit of 201k is 2 which is even.
Furthermore k must be odd as the unit digit of 201k has to be odd. Hence we try
201 151 30351 = , 201 153 30753 = and 201 155 31155 = to get the answer is 311550.
2
3. We consider the following cases.
If n is of the form 2
a
, then we have 2 1200 2400 n > = , and the smallest n of this form is
12
2 4096 = .
If n is of the form 3
b
, then we have 3 1200 3600 n > = , and the smallest n of this form is
8
3 6561 = .
If n is of the form 2 3
a b
, then we have 3 1200 3600 n > = , and after some trial (for
example, by first fixing a) we find that the smallest n of this form is
4 5
2 3 3888 = .
If n contains a prime factor greater than or equal to 5, then we have 5 1200 6000 n > = .
Combining the above cases, the answer is 3888.

4. First, note that 87 N > . This is because if there are 75 green, 12 red, 12 white and 12 blue balls
in the bag, we do not necessarily get balls of at least three different colours when at most 87
balls are drawn (for instance we may only get green and red balls), and this combination of
balls satisfies the condition if 100 balls are drawn, we can ensure getting balls of all four
colours because, when 100 balls are drawn, only 11 balls are missing and hence no colour can
be missing from the balls drawn as there are at least 12 balls of each colour.
Next we show that 88 N = is enough. The given condition ensures at least 12 balls of each
colour (for otherwise if there are, say, 11 or fewer red balls, it is possible for all of them to be
missing when 100 out of the 111 balls are drawn, contradicting the given condition), so when
88 balls are drawn, 23 are left over and it is impossible to have 2 colours missing (which
requires at least 24 balls to be left over). It follows that the answer is 88.

5. We have 1 a b > and 1 c d > . As
2 2 2 2
2010 ( )( ) ( )( ) ( ) ( ) 2010 a b c d a b a b c d c d a b c d = + = + + + > + + + = ,
equality holds throughout and hence we must have 1 a b c d = = . This gives
( 1) ( 1) 2010 b b d d + + + + + = , or 1004 b d + = , where b and d are positive integers with
2 b d > + . Hence we get 501 choices for (b, d), namely, (1003, 1), (1002, 2), , (503, 501),
and hence 501 choices for (a, b, c, d), namely, (1004, 1003, 2, 1), (1003, 1002, 3, 2), , (504,
503, 502, 501).

6. Note that
2 2
( 1) 1 ( ) 2( 1) 1 x P x x + s s + for any real number x. Hence we must have
2
( ) ( 1) 1 P x a x = + for some 1 2 a s s . Since (11) 181 P = , we have
2
181 (11 1) 1 a = + , or
9
5
a = . It follows that
2
9
(21) (21 1) 1 721
5
P
| |
= + =
|
\ .
.
3
7. Let n be the integer. Note that there must be no integer k for which
144 135
n n
k s s , for if such
integer exists, we must be able to write n as the sum of k numbers, each between 135 and 144.
For instance, if 2500 n = , then 17.4
144
n
~ and 18.5
135
n
~ . We may then take 18 k = . Since
135 18 2430 = and 144 18 2592 = , to form 2500 we may start by using 135 only (and we are
short by 70), and then change 14 of the 135 to 140, thus getting 2500 135 4 140 14 = + .
If 1
135 144
n n
> , then the above-mentioned integer k must exist. Hence 1
135 144
n n
< and so
9 135 144 n < , or 135 16 2160 n < = . When n is slightly less than 2160, we can still take
15 k = until n is less than 135 15 2025 = .
Indeed, the answer is 2024. This is because with 14 summands, the value of n is at most
144 14 2016 = ; with 15 summands, the value of n is at least 135 15 2025 = . It follows that
2024 cannot be expressed in the form described in the question.

8. Let 1000 n a b = + , where 0 999 b s s . By removing the last three digits of n, we get a. Hence
we have
3
1000 a a b = + , or
2
( 1000) b a a = . Since 0 b > , we have
2
1000 a > and hence
32 a > . If 33 a > , then
2
33(33 1000) 1000 b > > which is impossible. Hence we must have
32 a = , which gives
2
32(32 1000) 768 b = = and hence 32768 n = .

9. Let o and | be the roots of the equation
2
10 0 x px q + + = . Then we have
10
p
o | + = and
10
q
o| = , and so 2010 ( 10)( ) 10 p q o | o| = + = + + , which gives 201 o| o | = or
( 1)( 1) 202 o | = . As o and | are positive integers, the two factors 1 o and 1 | may be
1 and 202, or 2 and 101. Hence o and | may be 2 and 203, or 3 and 102. These correspond
to 2050 p = and 1050 p = respectively. It follows that the answer is 3100.

10. Note that the hexagon must have as its vertices the midpoints of six edges of the cube. Hence
the side length of the hexagon is 2 and its area is
2
1
6 ( 2) sin60 3 3
2
= .
On the other hand, P passes through the centre of the inscribed sphere by symmetry, so it cuts
out a cross section of radius 1 (which is the radius of the inscribed sphere), whose area (which
is contained entirely inside the hexagon) is t .
Therefore the answer is 3 3 t .
4
11. As Q is the circumcentre of ABPC, it lies on the
perpendicular bisector of BC, and we have
PQ CQ = . As D is the circumcentre of APQA,
we have 1 DQ DA = = and hence Q lies on the
circle with centre D and radius 1. These two
loci intersect at two points, giving two possible
positions of Q. We choose the one outside
AABC in order to maximise PQ (= CQ).
Let E and F be the midpoints of AD and BC respectively. Since AAPQ is equilateral with side
length 1, we have
3
2
EQ = . Appling Pythagoras Theorem in ACFQ , we have
2
2
2 2 2 2
1 3
1 2 3
2 2
PQ CQ CF FQ
| |
| |
= = + = + + = +
|
|
|
\ .
\ .
.

12. As shown in the left hand figure below, the set of points P for which 60 APB Z > is the union
of two segments (the major segment formed by AB in the circumcircle of AABP, where P may
take one of the two positions for which AABP is equilateral). Hence the red region R is the
intersection of the four such congruent shapes (denoted by XYWZ in the centre figure below).
Let O be the centre of major arc AXWB. By symmetry, ZXBA = 45, and so ZXOA = 90.
Similarly, ZWOB = 90. Also, 120 AOB Z = as angle at centre is twice angle at
circumference. Thus ZXOW = 360 90 120 90 = 60.
As shown in the right hand figure below, R consists of four identical minor segments (like
XPW) and a square XYZW. Since 3 AB = , each segment has radius 3 . Thus the square
XYZW has area 3 and each segment has area
( ) ( )
2 2
1 3 3 3
3 3
6 4 2 4
t
t = . It follows that
the answer is
3 3
3 4 2 3 3 3
2 4
t
t
| |
+ = +
|
|
\ .
.


D
P
C
A
B
A
B
C D
Q
E F
A B
R
D
C
X W
Y
Z
A
B
P
D
C
X W
O
5
13. We only have to consider which points are to be connected by red lines, and the rest of the
lines must be blue. Note that the restriction means that the 6 points must be divided into some
number of groups, such that two points are joined by a red line if and only if they are in the
same group. So the question becomes counting the number of such groupings, and we count
using the pattern of grouping, as follows:
Pattern Number
6 1
5-1
6
1
6 C =
4-2
6
4
15 C =
4-1-1
6
4
15 C =
3-3
6
3
2 10 C =
3-2-1
6 3
3 2
60 C C =
3-1-1-1
6
3
20 C =
2-2-2
6 4
2 2
3! 15 C C =
2-2-1-1
6 4
2 2
2 45 C C =
2-1-1-1-1
6
2
15 C =
1-1-1-1-1-1 1
Hence the answer is 1 6 15 15 10 60 20 15 45 15 1 203 + + + + + + + + + + = .
Remark. This question essentially asks for the number of partitions of the set {1, 2, 3, 4, 5,
6}. The answer is in fact the 6th Bell number.

14. If the binary representation of n is
1 2 2 1 k k k
a a a a a

, where each
i
a is 0 or 1, then we have
1 2 2
...
2
k k k
n
a a a a

(
=
(

and
1
2
2
n
n a
(
=
(

. It follows that
1 2 2 1
1 2 2 1
1 2 3 2 1
1 2 2 1
1 2 2 1
( ) ( ... )
( ... )
( ... )
( )
k k k
k k k
k k k
k k k
k k k
f n f a a a a a
f a a a a a
f a a a a a a
f a a a a a
a a a a a





=
= +
= + +
=
= + + + + +
= + + + + +

which is number of 1s in the binary representation of n. For 0 2010 n s s , the binary
representation of n has at most 11 digits since
10 11
2 2010 2 < < , and consists of at most ten 1s
(as
2
1111111111 2047 2010 = > ). Hence the maximum value of ( ) f n is 10, as we can check,
6
for instance, that
10
2
(1023) (2 1) (1111111111 ) 10 f f f = = = .

15. We first observe that by extending three mutually non-
adjacent sides of an equiangular hexagon, we get an
equilateral triangle. From this we see that every
equiangular hexagon is formed by removing three small
equilateral triangles from a large equilateral triangle.
Conversely, any such removal gives an equiangular
hexagon.
It is thus easy to see that if the lengths of the consecutive sides of an equiangular hexagon are
a, b, c, d, e, f, then we must have a d b e c f = + = (note that opposite sides form a pair
and the lengths alternate in sign in the above equality; this also says that the sum of the lengths
of three consecutive sides, which is also the side length of the equilateral triangle formed, may
only take two different possible values, namely, either a b c c d e e f a + + = + + = + + or
b c d d e f f a b + + = + + = + + ). Furthermore, since an equilateral triangle of unit length has
area
3
4
, we see that the equiangular hexagon formed from removing equilateral triangles of
sides lengths a, c, e from a large equilateral triangle of side length n has area
2 2 2 2
3
( )
4
n a c e , where n a b c = + + .
Now we must divide the lengths 6, 7, 8, 9, 10, 11 into three pairs with equal difference. Since
the perimeter is 51, which is odd, the possible equal differences include 1 (the pairing is {6,7},
{8,9}, {10,11}) and 3 (the pairing is {6,9}, {7,10}, {8,11}).
In the first case, the two sides adjacent to 6 must be 9 and 11 (i.e. the two larger elements of
the other two pairs, since 6 is the smaller in its pair), and the side lengths in order must be (6,
9, 10, 7, 8, 11). In this case the area of the hexagon is
2 2 2 2
3 425
(25 6 10 8 ) 3
4 4
= .
(Another expression would be
2 2 2 2
3
(26 9 7 11 )
4
which is the same.)
Similarly, in the second case, the two sides adjacent to 6 must be 10 and 11, and the side
lengths in order must be (6, 10, 8, 9, 7, 11), and the area of the hexagon in this case is
2 2 2 2
3 427
(24 6 8 7 ) 3
4 4
= .
Combining these two cases, we see that the answer is
425 427
213
4 4
+ = .

7
16. First of all, note that a 18-24-30 triangle is a 6-
time magnification of a 3-4-5 triangle, and so is
right-angled. Thus the midpoint of the
hypotenuse of each triangle is the centre of
their common circumcircle, and the in-radius is
(18 24 30) 2 6 + = . Let one of the triangles
be AABC where 90 A B C Z < Z < Z = , with
circumcentre O.
Let be the diameter of the circumcircle
passing through the mid-point of AC. Since O is
the mid-point of AB, is parallel to BC, and
since 24 AC = , it is 12 units from BC. As the
diameter of the inscribed circle is also 12 units,
is tangent to the inscribed circle.
Therefore must be the hypotenuse of the other triangle. It is then easy to see that the regions
which belong to exactly one of the two triangles (i.e. the six shaded triangular regions in the
figure) are all similar to AABC.
The area of AABC is 18 24 2 216 = . We focus on the three shaded regions at A, B and C. At
A, the area of the shaded region is clearly 216 4 54 = (in fact, it is a 9-12-15 triangle). The
region at B is a 6-8-10 triangle with area 6 8 2 24 = . (To see this, we note that the line
segment 15 BO = is split into two parts, with the second part being of length 9 as can be seen
by comparing with the 9-12-15 shaded triangle at A.) Similarly, the region at C is a 3-4-5
triangle with area 3 4 2 6 = . It follows that the answer is 216 54 24 6 132 = .

17. Let S be the mid-point of AC. Then R lies on BS with
: 2:1 BR RS = since R is the centroid of AABC. We
have 3 BS = (as S is the centre and AB is a diameter
of the circumcircle of AABC) and so 1 RS = . Let
AR x = and CR y = . Since the area of AARC is
1 1
sin150
2 4
xy xy = , the area of AABC is
3
4
xy and
hence the area of ABCD is
3
2
xy .
To find xy, we first apply the cosine law in AARC, AARS and ACRS to get
2 2
2 2 2 2
36 2 cos150
(3 1 2 3 1 cos ) (3 1 2 3 1 cos ) 3
20 3
x y xy
RSA RSC xy
xy
= +
= + Z + + Z +
= +
.
A B
C
D
S
R
P
Q
3
3
150
x
y
C
B
A
O
8
which gives (note that cos cos RSA RSC Z = Z ). It follows that
16
3
xy = and hence the
answer is
3 16
8 3
2 3
| |
=
|
\ .
.

18. Upon expansion, we get
2 2
2(100 260 169)(5 13 8) 1 x x x x + + + + = . If we let
2
5 13 8 u x x = + + ,
the equation becomes 2(20 9) 1 u u + = , which gives
1
20
u = or
1
2
u = , i.e.
2
159
5 13 0
20
x x + + =
or
2
17
5 13 0
2
x x + + = . The former gives two real roots with product
159
100
, and the latter gives
two complex roots with product
17
10
. The two roots in the same equation must be paired up in
order for pq rs + to be real. It follows that the answer is
159 17 329
100 10 100
+ = .

19. Since
2 2 2 2
3 2 2 4 ( 1) ( 1) 3 2 0 a ab b a b a b ab + + + = + + + > for all a, b, we see that a b
is always well-defined and positive when a, b are positive. When a is positive, we have
2 2
( 2) 6 4 2 4 4 1
2
2 4 2( 2) 2
a a a a
a
a a
+ + + +
= = =
+ +
.
Hence
1 15
(( ((2010 2009) 2008) 2) 1) 1
2 9
= = .
Remark. Clearly, the crucial observation of this question is that 2 a is a constant. What
leads us to consider this? One way is to write the expression in the square root as a function of
a, i.e.
2 2
(3 2) ( 2 4) a b a b b + + + , with discriminant
2 2 2
(3 2) 4( 2 4) 5 4 12 b b b b b + = .
By some trial and error we see that this is a perfect square when 2 b = .

20. Let
5 3 3 5
20 20 z x x x x = + = . Then we have
5 3
20 z x x = + and
3 5
20 . z x x = Adding, we
have
5 3 5 3
z z x x + = + , or
5 5 3 3
0 z x z x + = . Hence we have
4 3 2 2 3 4 2 2
0 ( )( ) z x z z x z x zx x z zx x = + + + + + + + .
As x is non-zero, z is also non-zero. Thus we have
2
2 2 2
3
0
2 4
x
z zx x z x
| |
+ + = + >
|
\ .
and
2 2
4 3 2 2 3 4 2 2 2 2
1
0
2 2 2
x z
z z x z x zx x z z x x z x
| | | |
+ + + + = + + + + >
| |
\ . \ .
. Therefore the second factor
on the right hand side of the above equation is positive. This forces z x = and hence we get
5 3
20 x x x = + , or
5 3 2 2
0 20 ( 4)( 5) x x x x x x = = + . The possible non-zero real values of x
are thus 5 and so the answer is 5.