Parabola Volume 45, Issue 2 (2009)

2009 University of New South Wales School Mathematics

Competition
Junior Division – Problems and Solutions

Problem 1
Let ABC be a right triangle with right angled at A. Let D, E be points on BC with
BD = DE = EC. Prove that AD 2 + AE 2 = 95 BC 2 .
Solution 1(a)
Let points R, S be on AB so that ER and DS are perpendicular to AB.
Let AB = c, AC = b and BC = a.

E
P a
b
D
Q

A R S B
c

So P E k QD k AB and DS k ER k AC.
Hence since EC = ED = DB, then by the Equal Intercepts between Parallel Lines
Theorem, we have
b c
CP = P Q = QA = and AR = RS = SB = .
3 3
Alternatively, △P CE ||| △QCD ||| △ACB and △SDB ||| △REB ||| △ACB, using
equal right angles and corresponding angles in parallel lines, which gives the same
result.

1
 Now by Pythagoras’ Theorem,
AD 2 = AS 2 + DS 2 = AS 2 + AQ2 as DS = AQ
 2  2
2c b 4c2 b2
= + = +
3 3 9 9
2 2 2 2 2
AE = AR + RE = AR + AP as RE = AP
2
c2 4b2
 c 2  
2b
= + = +
3 3 9 9
2 2 2 2
4c b c 4b 5c2 5b2 5
∴ AD 2 + AE 2 = + + + = + = (c2 + b2 )
9 9 9 9 9 9 9
2
5a 5
= = BC 2 .
9 9
Solution 1(b)
Consider the triangle shown.

B D C

Apollonius’ Theorem states that for any triangle ABC with a point D on BC such
that
BD n
= (n and m need only be positive reals!)
DC m
then m AB 2 + n AC 2 = m BD 2 + n DC 2 + (m + n) AD 2 .

In particular if m = n = 1, so D is the midpoint of BC.

Then AB 2 + AC 2 = BD2 + DC 2 + 2AD 2 = 2BD2 + 2AD 2

This theorem is easily proved by just dropping an altitude from A to BC.
Hence in the original diagram with again CE = ED = DB = x, using this midpoint
case twice,
AD 2 + AC 2 = 2AE 2 + 2CE 2 = 2AE 2 + 2x2
and AB 2 + AE 2 = 2AD 2 + 2DE 2 = 2AE 2 + 2x2 .

So adding these equations

(AD 2 + AE 2 ) + (AB 2 + AC 2 ) = 2(AD 2 + AE 2 ) + 4x2

2
but AB 2 + AC 2 = BC 2 = 9x2 , so

∴ (AD 2 + AE 2 ) + 9x2 = 2(AD 2 + AE 2 ) + 4x2

5
so AD 2 + AE 2 = 5x2 = BC 2 .
9
Problem 2
A box contains 100 red marbles, 80 green marbles, 60 blue marbles and 40 yellow
marbles.
1. What is the smallest number of marbles, which if selected at random, will be
guaranteed to contain at least ten pairs of marbles?
2. What is the smallest number of marbles, which if selected at random, will be
guaranteed to contain at least ten triplets of marbles?
In the above a pair of marbles is defined as two marbles of the same colour and a triplet is defined
as three of the same colour. No marble may be counted in more than one pair (or triplet).
Solution 2
If n marbles are chosen then n = r + g + b + y with r red marbles, g green marbles,
b blue marbles and y yellow marbles.
a) Write r = 2r1 + r2 etc. with r1 , r2 integers and r2 = 0 or 1 (Quotient Remainder or
Division Algorithm Theorem).
Hence the number of pairs is
 
n r2 + g2 + b2 + y2
r1 + g1 + b1 + y1 = − .
2 2

For n even, this has minimum value

 
n 1+1+1+1 n
− = −2
2 2 2
which is ≥ 10 iff n ≥ 24.
For n odd, this has minimum value
 
n 1+1+1+0 n 3
− = −
2 2 2 2
which is ≥ 10 iff n ≥ 23.
Hence the minimum number of marbles required to guarantee 10 pairs is 23.
b) This time write r = 3r1 + r2 etc. where r1 , r2 are integers and r2 = 0, 1 or 2, then the
number of triplets is
 
n r2 + g + 2 + b2 + y2
r1 + g1 + b1 + y1 = −
3 3

3
 For n ≡ 0 (mod 3), this has minimum value
 
n 2+2+2+0 n
− = −2
3 3 3
which is ≥ 10 iff n ≥ 36.
For n ≡ 1 (mod 3), the number of triplets has minimum value
 
n 2+2+2+1 n 7
− = −
3 3 3 3
which is ≥ 10 iff n ≥ 37.
For n ≡ 2 (mod 3), the number of triplets has minimum value
 
n 2+2+2+2 n 8
− = −
3 3 3 3
which is ≥ 10 iff n ≥ 38.
Hence the minimum no. of marbles required to guarantee 10 triplets is 36.
Problem 3
A bowl in the shape of a conical frustum is placed out in the rain at the start of a
downpour. At the end of the downpour the water level r is equal to (i) half the height
of the bowl, (ii) the radius of the base of the bowl, and (iii) half the radius of the top of
the bowl.

111111111111
000000000000
000000000000
111111111111
000000000000
111111111111
r 000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111

The volume of a cone is one-third the area of the circular base times the height. What
was the reported rainfall over the catchment area?
Solution 3
The figure shows the cross-section of the conical frustum perpendicular to its base
containing the axis of the cone,

C R1 = 2r D
r
B R2
E
r
O r F
p

4
 Let R1 be the radius of the top of the bowl, R2 be the radius of the top of the water
level and r be the radius of the bottom of the bowl, and let p = OA.
Let V be the volume of water in the bowl.
We are given that OB = BC = r and R1 = 2r.
Since △ AOF ||| △ ABE ||| △ ACD,
p p + 2r p + 2r p
= = so p = + r
r R1 2r 2
∴ p = 2r
R2 r R2 1
∴ = so =
r+p p r + 2r 2
3r
∴ R2 = .
2
1 1
Hence V = π R22 (p + r) − π r 2 p
3 3
 2
π 3r π
= · 3r − · r 2 · 2r
3 2 3
3
πr 3 19
 
πr 27
= −2 = ·
3 4 3 4
3
19πr
= .
12
Rainfall is measured as the volume of water per unit area on which the water fell
(assumed perpendicularly) over some time period. The water in the bowl fell on the
circular cross-section at the top of the bowl, which has area πR12 = π(2r)2 = 4πr 2 , hence
the rainfall is
19πr 3 1 19r
· 2
= .
12 4πr 48
Problem 4
There are many ways to substitute digits for letters in the expression

CAT + EMU = LION,

with different letters representing different digits, to obtain a valid equation. Prove
that in every solution, LION is a multiple of 9. Is there any higher number than 9 for
which this is still true?
Solution 4
We wish to show that
LION ≡ 0 (mod 9).
Without loss of generality we write

CAT ≡ a mod 9
EMU ≡ b mod 9
LION ≡ c (mod 9),

5
and using the digit sum rule

C + A + T ≡ a (mod 9)
E + M + U ≡ b (mod 9)
L + I + O + N ≡ c (mod 9).

(The digit sum rule can be seen by expanding

CAT = 100C + 10A + T = 99C + C + 9A + A + T

so that CAT (mod 9) ≡ C + A + T (mod 9) etc.)

Using the given equation

CAT + EMU = LION

we have
a + b ≡ c (mod 9).
Now note that there are ten different letters in total so that
C +A+T +E+M +U +L+I +O+N = 0+1+2+3+4+5+6+7+8+9
= 45 = 9 · 5,

and hence
a + b + c ≡ 0 (mod 9).
Combining the results, a + b ≡ c (mod 9) and a + b + c ≡ 0 (mod 9), we deduce that
2c ≡ 0 (mod 9), and hence c ≡ 0 (mod 9) as required.
For the second part note that 1089 = 324+765 and 1098 = 346+752 are both LIONs
and their greatest common divisor is 9.
Problem 5
Let n be a positive integer and let Sn = 1n + 2n + 3n + 4n .

1. Find S1 , S2 , S3 , S4 .

2. Show that n5 has the same last digit (units digit) as n does.

3. Prove that 10 is a factor of Sn unless n is a multiple of 4.

Solution 5

1. S1 = 10, S2 = 30, S3 = 100, S4 = 354

2. One can check n5 ≡ n (mod 10) for n = 0, 1, 2, 3, · · · , 9 In fact we need only check
this for n = 0, 1, 2, 3, 4, 5 as 6 ≡ −2 etc. and we have odd powers here.
Hence this is true for all integers as any integer n is ≡ 0, 1, 2, 3 · · · , 9 (mod 10)
and if n ≡ m (mod 10) then np ≡ mp (mod 10) for any positive integer p, OR

6
 write n = 10k + r with k a non-negative integer and r = 0, 1, 2, 3, · · · 8 or 9 (Di-
vision Algorithm or Quotient-Remainder Theorem), then by the Binomial Theo-
rem,
n5 = (10k + r)5
= (10k)5 + 5 · (10k)4 r + 10 · (10k)3 r 2 + 10 · (10k)2 r 3 + 5 · (10k)r 4 + r 5
≡ r5 ≡ r ≡ n (mod 10).

3. For all integers n, and all non-negative integers k and positive integers r (and for
r = 0 if k > 0), we have
n4k+r ≡ nr (mod 10)
by induction on k. Since it is trivial for k = 0 and if true for some k, then

n4(k+1)+r ≡ n4k+4+r ≡ n5+4k+r−1 ≡ n5 · n4k+r−1

≡ n · n4k+r−1 ≡ n4k+r ≡ nr (mod 10).

Hence for r = 1, 2, 3, 4 and k a positive integer, we have mod 10,

S4k+r ≡ 14k+r + 24k+r + 34k+r + 44k+r

(
0 if r = 1, 2, 3
≡ 1 r + 2 r + 3 r + 4 r ≡ Sr ≡
4 if r = 4.

Hence 10 divides Sn if and only if 4 does not divide n.

Problem 6
A quadrophage is an insect which eats checkerboards (square boards tiled with
squares of alternating light and dark colours). A 5 × 5 checkerboard is shown below.
00
11 00
11
11
0000
1100
1100
11
00
1100
1100
110011
1100
00
11
1100
11
0011 00
11
0011
11 00
11
00
110000
110000
11
00
1100
11 00
11
0011
0011
11 00
00
110011
1100
0011
00 11
11 00
00 11
11 00
The quadrophage starts off in a given direction and always eats one square before
proceeding to the next. As long as there is a square to be eaten it will always proceed
in a straight line along a row or column of the board. If it encounters the edge of the
board, or a square it has already eaten, it turns through a right angle (either left or
right) and keeps going. If it reaches a point with no neighbouring uneaten squares it
must stop.

1. If the quadrophage starts on the middle square of a 7 × 7 checkerboard, what is

the maximum number of squares it can eat? Draw a diagram of the path taken.

7
2. The quadrophage jumps on to a random square of a large checkerboard and even-
tually finds that it has eaten the whole board. On which squares might it have
started?

Solution 6

1. The case in which the quadrophage first moves down and then to the right is
shown in the figure on page 38. Note that the quadrophage has no choice until it
gets to the square labelled A in this figure.

11 11
00 00 11
00
00
11
0011
11 00
1100
1100
110011
11
00
00 A
11 00
11
00 11
11 0000
001100
11
0011
110000
1100
1100
1100
1100
11
0011
110011
0011
0011
0011
0011
00
00
11 00
11 00
11 00
11
0011
1100
00
1100
1100
11
00
1100
1100
11
00
1100
11
00
110011
11000011
11000011
1100
0011
110000
1100
1100
1100
1100
11
00 11
11 00 11
00

Thus except for rotations and reflections, there are only 2 possible paths starting
at the middle square, one path misses 12 squares so 37 are eaten, the other path
misses 5 squares so 44 are eaten. The maximal number of squares that can be
eaten is thus 44.

2. It is sufficient to consider a 7 × 7 checkerboard for the general result. We begin

by identifying starting squares, that result in all squares being eaten, under the
restriction that the quadrophage first moves down and then moves to the right
at the bottom of the board. The complete set of starting squares that result in all
squares being eaten can then be arrived at by considering reflections and rota-
tions of the board.
All possibilities in which all squares are eaten with the quadrophage first moving
downwards and then to the right are shown in the figure on page 39.

After considering reflections and rotations we then arrive at five possible starting
points within each 3 × 3 corner region. The complete set of starting points is then
as shown by filled circles in the 13 × 13 checkerboard below.

8
 11 11
00 00 11
00 11 11
00 00 11
00
00 11
11 00 11
00 00 11
11 00 11
00
0011
110011
0011
0011
0011
0011
00 0011
110011
0011
0011
0011
0011
00
0011
110000
1100
1100
1100
1100
11 0011
110000
1100
1100
1100
1100
11
0011
11 0011
11 0011
11 00
11 0011
11 0011
11 0011
11 00
11
0000
11 0000
0011 0000
00110011
00 0000
11 0000
0011 0000
00110011
00
0011
11 00
1111
00 11
11 00
1111
0011
00 11 00 0011
11 00
1111
00 11
11 00
1111
0011
00 11 00
0011
110011
0011
0011
0011
0011
00 0011
110011
0011
0011
0011
0011
00
0011
110011
0011
0011
0011
0011
00 0011
110011
0011
0011
0011
0011
00
00 11
11 00 11
00 00 11
11 00 11
00

00 11
11 00 11 00 00 11
11 00 11
00
1111
000011
000011
11 00 0011
11
0000 1111
00 0011
000011
11000011
11
0000
0011
110011
00
11000011
11
00
11 00 11
00
11 00
11 0011
11 0011
00
11000011
11
00
110011
00
1100
11
0011
11 0011
11 0011
11 00
11 0011
11 0011
11 0011
11 00
11
000011
11 000011
00001100 000011
11 000011
000011
00
0011
11 00
11
00 11 00
11
00 11 0011
00 0011
11 00
11
00 11 00
110011
00 11 00
0011
110011
0011
0011
0011
0011
00 0011
11 0011
0011
0011
0011
0011
00
0011
11 00
11
00 11 00
11
00 11 0011
00 0011
11 00
11
00 11 00
110011
00 11 00
00 11
11 00 0011
00 11
11 0011
00 11
00 0011
00 11
11 00 11
00 11
00
11
000011
11000011
11000011
11000011
11000011
11000011
1100
0011
1100
00
110011
1100
00
110011
1100
00
110011
1100
00
110011
1100
00
110011
1100
00
1100
11
00
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
11
0011
1100
00
110011
1100
00
11001100
1100
00
110011
1100
00
110011
1100
00
11001100
1100
00
1111
00
110011
1100001100
11001111
000011
1100001100
11001111
00
00
11 00
11 00
11 00
11 00
11 00
11 00
11
0011
11000011
11000011
11000011
11000011
11000011
110000
11
00
1100
1100
1100
1100
1100
00
1100
1100
1100
1100
1100
1100
00
1100
11
00
11 00
11 00
11 00
11 00
11 00
11 00
11
0011
1100
00
110011
1100
00
110011
1100
00
110011
1100
00
110011
1100
00
110011
1100
00
1100
11
00
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
11
0011
1100
00
110011
1100
00
11001100
1100
00
110011
1100
00
110011
1100
00
11001100
1100
00
1111
00
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
11
00
11 00
11 00
11 00
11 00
11 00
11 00
11
0011
110000
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
11
00 11
11 00 11
00 11
00 11
00 11
00

Senior Division – Problems and Solutions

Problem 1
See Problem 6 in the Junior Competition.
Solution 1
See Problem 6 solution in the Junior Competition.
Problem 2
A triangle has one side that is the space diagonal (or long diagonal) of a cube and
the other two sides lie on the surface of the cube whose side has length a.

1. What is the smallest perimeter that such a triangle can have?

2. What is the smallest area that such a triangle can have?

Solution 2
The triangle shown in the figure has both the smallest perimeter and the smallest
area given the constraints.

9
 111111
000000
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
000000
111111
000
111
a 000000
111111
000
111
000000
111111
000
111
2 000000
111111
000
111
000000
111111
000
111
a

1. It is easy to see that the given triangle has the smallest perimiter. Flatten out
the faces containing the edges of the triangle, the resultant straight line is the
shortest distance between the two points connecting the space diagonal. Some
straightforward applications of Pythagoras’s Theorem then yield the result
r  a 2 √ √ √
ℓ = 2 a2 + + 3a = ( 3 + 5)a.
2

2. Consider the geometry shown with the vertices of the triangle labelled A, B, C.
B
111111
000000
000
111
000
111
000000
111111
000
111
000
111
000000
111111
000
111
000
111
a−h 000000
111111
000
111
000
111
000000
111111
000
111
000
111
A 000000
111111
000
111
000
111
000000
111111
000
111
000
111
000000
111111
h
000
111
000
111
C
a

Define the semi-perimeter s = AB+BC+CA

2
and the area from Heron’s formula
p
Area = s(s − AB)(s − BC)(s − CA).

Use Pythagoras’s Theorem to write

p
AB = a2 + (a − h)2 ,
√
BC = 3a,
√
CA = a2 + h2 ,

and then after a little algebra

a √
Area = √ a2 + h2 − ah.
2

10
 d(Area) a d2 (Area) a
Note that = 0 when h = and 2
> 0 when h = so that the
dh √ 2 dh 2
minimum area is Area = 46 a2 .
An alternate method is to consider the minimum distance from the vertex on the
edge of the cube to the space diagonal. This minimum distance is s = √a2 from
√ √
which the mimimum area Area = 21 √a2 3a = 46 a2 follows.

Problem 3
Suppose that f (0) = 1 and a > 0 is a fixed real number. Show that f (x) = 1 is the
only continuous function f : R → R that satisfies

2f (x − y) = f (a − y)f (x) + f (a − x)f (y)

for all x, y ∈ R.
Solution 3
Put x = y then
1 1
f (0) = f (a − x)f (x) + f (a − x)f (x)
2 2
so that
1
f (x) = .
f (a − x)
Note that f (a) = 1/f (0) = 1 follows from this.
Now consider y = 0 then
1 1
f (x) = f (a)f (x) + f (a − x)f (0)
2 2
and hence
1 1
f (x) = f (x) + f (a − x)
2 2
so that
f (x) = f (a − x).
Combining the above results we now obtain

1
f (a − x) =
f (a − x)

from which it follows that f (a − x) = ±1 for all x ∈ R and thus f (x) = ±1 for all x ∈ R.
We now reject the possibility that f (x) = −1 since f (0) = +1 so that there would be
a jump discontinuity at x corresponding to the smallest value of |x| where f (x) = −1.
Note that it is obvious by inspection that f (x) = 1 is a solution. Here we have proved
it is the only continuous solution.
Problem 4
A flea starts on the brick in the centre of a portion of brickwall (as shown) and hops
from one brick to any one of the neighbouring six bricks with equal probability. If it

11
hops onto one of the shaded bricks it is promptly squashed. Otherwise it hops again
to one of the neighbouring bricks with equal probability. The flea continues hopping
in this fashion and is allowed to visit the same white brick more than once.
What is the probability pA that the flea is squashed on the shaded brick labelled A?
What is the probability pB that the flea is squashed on the shaded brick labelled B?
What is the probability pC that the flea is squashed on the shaded brick labelled C?

0000000000
11111111110000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111
B 0000000000
1111111111
0000000000
1111111111
1111111111
0000000000 0000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
1111111111 0000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
1111111111
0000000000
1111111111
A 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
11111111110000000000
1111111111
C
00000000001111111111
111111111100000000001111111111
00000000001111111111
0000000000

Solution 4
It is convenient to number the white bricks 1, 2, 3 from left to right, and let En denote
the expectation that the flea arrives on the nth brick (after any number of hops) before
it hops onto one of the shaded bricks.

1111111111
00000000000000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111
B 0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111
00000000001111111111
0000000000
1111111111 0000000000
0000000000
1111111111
00000000001111111111
1111111111
0000000000
1111111111 000000000011111111110000000000
1111111111
0000000000
1111111111
0000000000
1111111111
A 1 2 3 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
11111111110000000000
1111111111
C
00000000001111111111
111111111100000000001111111111
00000000001111111111
0000000000

From each white brick the flea can hop to any one of the six neighbouring bricks
from which we deduce
1
pA = E1
6
1 1
pB = E2 + E3
6 6
1
pC = E3
6
and E1 = 16 E2 , E3 = 61 E2 . It is clear from the symmetry of the problem that E1 = E3
1
so that we can express each of the probabilities in terms of E2 , as follows: pA = 36 E2 ,
7 1
pB = 36 E2 , and pC = 36 E2 . Given that the probabilities must sum up to unity we also
have 2pA + 4pC + 4pB = 1 so that 36 + 36 + 36 E2 = 1. We now solve for E2 = 36
2 4 28


34
so
6 1 7 1
that E1 = 34 and then pA = 34 , pB = 34 and pC = 34 .
An alternate method of solution is to consider the probabilities for the location of
the flea after the first two hops as shown:

12
 0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
1111111111
0000000000 0000000000
1111111111
1/6 0000000000
1111111111
1/6 0000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
1111111111 0000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
1111111111 1/6 0000000000
1111111111
0000000000
1111111111 1/6 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
00000000001111111111
0000000000
00000000001111111111
0000000000
00000000001111111111
1/6 1/6 0000000000
111111111111111111111111111111 0000000000
1111111111
1111111111
00000000000000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
1111111111
1/36 1111111111
0000000000
7/36 1111111111
0000000000
7/36 1111111111
0000000000
0000000000
1111111111
0000000000
1111111111 0000000000
11111111110000000000
11111111110000000000
1111111111
1/36
0000000000
1111111111
0000000000
1111111111
0000000000
1111111111 0000000000
11111111110000000000
11111111110000000000
1111111111
0000000000
1111111111
0000000000
1111111111 0000000000
1111111111
0000000000
1111111111
1/36 1/18 0000000000
1111111111
1/36
0000000000
1111111111
0000000000
11111111110000000000
11111111110000000000
1111111111 0000000000
1111111111
0000000000
1111111111
0000000000
11111111110000000000
1111111111
1/36 0000000000 0000000000
11111111110000000000
1111111111
0000000000
1111111111
00000000001111111111
00000000001111111111
0000000000
00000000001111111111
7/36 7/36 0000000000
1/36
1111111111111111111111111111110000000000
1111111111

We can now deduce that after many hops

   2  3
1 1 1 1 1 1 1
pC = + + + + ...
36 18 36 18 36 18 36
and by summing the geometric series
 
1 1 1
pC = 1 = .
36 1 − 18 34
1 7
Similarly we find pA = 34 and pB = 34 .
Problem 5
Consider a sequence of integers a1 , a2 , a3 , . . . with an = an−1 −an−2 if n ≥ 3. Find the
sum of the first 2009 terms in the sequence given that the sum of the first 2007 terms is
2008 and the sum of the first 2008 terms is 2007.
Solution 5
Starting with a1 , a2 we find that a3 = a2 − a1 , a4 = −a1 , a5 = −a2 , a6 = a1 − a2 ,
a7 = a1 , a8 = a2 so that


 a1 if n = 1 + 6k
a if n = 2 + 6k

2



a2 − a1 if n = 3 + 6k

an =

 −a1 if n = 4 + 6k
−a if n = 5 + 6k

2



a1 − a2 if n = 6 + 6k,


where k = 1, 2, 3, . . .. From the above it is straightforward to compute the sum of the

first n terms which we denote by sn . This yields


 a1 if n = 1 + 6k,
a1 + a2 if n = 2 + 6k,




2a2 if n = 3 + 6k,

sn =

 2a2 − a1 if n = 4 + 6k,
a 2 − a1 if n = 5 + 6k,




0 if n = 6 + 6k.


13
 Now note that 2007 = 3 + 6(334), 2008 = 4 + 6(334), 2009 = 5 + 6(334) hence s2007 =
2a2 = 2008 and s2008 = 2a2 − a1 = 2007, from which we deduce a2 = 1004, a1 = 1 and
then s2009 = a2 − a1 = 1003.
Problem 6
Let ABC be a triangle with acute angles at B and C. For any X on BC, let M and
N be the feet of the perpendiculars from X to AB and AC. Show how to find X so that
MN is parallel to BC.
Solution 6
Let P, Q be the midpoints of AB, AC and let R be the circumcentre of the triangle;
let X be the intersection of AR and BC.

P Q
N
M
R

C
B X

We note that:
AP AB
△AP Q ||| △ABC so that = ,
AQ AC
AQ AR
△AQR ||| △ANX so that = ,
AN AX
AP AR
△AP R ||| △AMX so that = .
AM AX
From the above we now deduce that
AM AN
=
AB AC
so that △AMN ||| △ABC and MN is parallel to BC.
An alternate construction that yields the same result is shown below.

Let Y be the unique intersection of the perpendicular to AB through B and the perpen-
dicular to AC through C. The X is the intersection of AY and BC. The points ABY C
are concyclic so that ∠BY A = γ and then ∠BAY = γ ∗ . The points AMXN are also
concyclic so that now ∠MNX = γ ∗ . therefore MN k BC (alternate angles are equal).

14
 A

γ∗

N
M
γ∗
γ∗ γ
C
B X

15