PMO 2019 Qualifying Stage

Carl Joshua Quines

October 20, 2018

The qualifying stage this year has become more difficult than the previous years. I’ll present the
questions and some solutions to all the problems. Are any explanations unclear? If so, contact me
at cj@cjquines.com. More material is available on my website: https://cjquines.com.

PART I. Choose the best answer. Each correct answer is worth three points.

1. The measures of the angles of a pentagon form an arithmetic sequence with common difference
15◦ . Find the measure of the largest angle.

(a) 78◦ (b) 103◦ (c) 138◦ (d) 153◦

Answer: (c) 138◦ .

Solution: Suppose that the angles had measures (a − 30)◦ , (a − 15)◦ , a◦ , (a + 15)◦ , and
(a + 30)◦ . Their sum is 5a◦ , which must be equal to 540◦ , hence a = 108. The largest angle is
108◦ + 30◦ = 138◦ .

2. If x − y = 4 and x2 + y 2 = 5, find the value of x3 − y 3 .

(a) −24 (b) −2 (c) 2 (d) 8

Answer: (b) − 2 .

Solution: We know that x3 − y 3 = (x − y)(x2 + xy + y 2 ), hence we need the value of xy.

Squaring the first equation gives x2 − 2xy + y 2 = 16. Combined with the second equation, we
11
get xy = − . Thus
2
 
3 3 2 2 11
x − y = (x − y)(x + y + xy) = (4) 5 − = −2.
2

3. Five numbers are inserted between 4 and 2916 so that the resulting seven numbers form a
geometric sequence. What is the fifth term of this geometric sequence?

(a) 324 (b) 416 (c) 584 (d) 972

Answer: (a) 324 .

Solution: The first term of the sequence is 4; let its common ratio be r. Note that 2916 would
be the seventh term of this geometric sequence, so by the formula for the nth term, we get
2916 = 4r6 . Thus, r = ±3. The fifth term of the sequence is 4r4 = 324.

1 6
 
2
4. The constant term in the expansion of 3x − is
x
2 Carl Joshua Quines

(a) 189 (b) 135 (c) 90 (d) 54

Answer: (b) 135 .

 4
1
Solution: We use the binomial theorem. The constant term is produced by (x2 )2 . This
x
gives the term as
1 4
   
6 2 2
(3x ) − = 15(3)2 = 135.
2 x

5. Juan has 4 distinct jars and a certain number of identical balls. The number of ways that he can
distribute the balls into the jars such that each jar has at least one ball is 56. How many balls
does he have?

(a) 9 (b) 8 (c) 7 (d) 6

Answer: (a) 9 .

Solution: Consider lining up the n balls in a row, and placing 3 dividers in the n − 1 spaces
between them. These divide the n balls into four groups, and counting the number of ways to

Since there are n − 1 spaces
do this is equivalent to the original problem.
in between and we
choose 3 of them as dividers, there are n−1
3 ways to do so. Setting n−1
3 = 56 and solving, we
get n = 9.

Remark: This is known as the balls-and-urns problem. We are distributing identical objects
into distinct containers such that each container has at least one object.

6. A regular octagon of area 48 is inscribed in a circle. If a regular hexagon is inscribed in the same
circle, what would its area be?
√ √ √ √
(a) 12 10 (b) 18 6 (c) 24 3 (d) 30 2
√
Answer: (b) 18 6 .

R 45◦ R R
60◦
R

Solution: Let the radius of the circle be r. Draw the center of the circle, and connect it to each
of the octagon’s vertices. This creates eight congruent isosceles triangles with vertex angle 45◦ .
1
We know the area of a triangle with side lengths a and b and included angle C is ab sin C. Thus
2
3 Carl Joshua Quines

the area of the octagon is

1 2
8· · r sin 45◦ = 48,
2
√
and we get r2 = 12 2. By similar reasoning, the area of the hexagon would be
1 2 √
6· · r sin 60◦ = 18 6.
2

7. What is the smallest positive integer which when multiplied to 244 + 64 makes the product a
perfect square?

(a) 1037 (b) 2074 (c) 5185 (d) 10370

Answer: (c) 5185 .

Solution: We use the Sophie-Germain factorization a4 +4b4 = a2 + 2b2 + 2ab a2 + 2b2 − 2ab .

It can be derived by completing the square like so:

a4 + 4b4 = a4 + 4a2 b2 + 4b4 − 4a2 b2

= (a2 + 2b2 )2 − (2ab)2
= (a2 + 2b2 + 2ab)(a2 + 2b2 − 2ab).

Here, we get

244 + 4 · 24 = (242 + 2 · 22 + 2 · 24 · 2)(242 + 2 · 22 − 2 · 24 · 2)

= (680)(488)
= (23 · 5 · 17)(23 · 61)
= 26 · 5 · 17 · 61.

The prime factorization of a perfect square must have even exponents for each prime. To make
the product a perfect square, we need to multiply by at least 5 · 17 · 61 = 5185.

8. A bowl of negligible thickness is in the shape of a truncated circular cone, with height 4 in and
upper and lower radii of 9 in and 6 in, respectively. What is the volume of the bowl?

(a) 276π in3 (b) 248π in3 (c) 234π in3 (d) 228π in3

Answer: (d) 228π in3 .

Solution 1: Extend the sides of the truncated cone to meet at the tip of the original cone.
The volume is the difference of the volumes of two cones: the whole cone, and the cone that is
truncated.
4 Carl Joshua Quines

Consider the right triangle with legs h, the height of the truncated cone, and the radius 6. This
is similar to right triangle with legs h + 4, the height of the whole cone, and radius 9. Since they
are similar triangles, we get
h h+4
= ,
6 9
and solving yields h = 8.
1
The volume of a cone with height h and radius r is πr2 h. Thus the volume of the truncated
3
cone is
1 1
π · 92 · (8 + 4) − π · 62 · 8 = 228π in3 .
3 3

πh 2
Solution 2: We use the formula (R + Rr + r2 ), where h is the height, and R and r are the
3
upper and lower radii, respectively. This gives 228π in3 .

Remark: The formula in Solution 2 is derived through a similar solution in Solution 1.

9. A circle is tangent to the line 2x − y + 1 = 0 at the point (2, 5) and the center is on the line
x + y − 9 = 0. Find the radius of the circle.
√ √ √
(a) 14 (b) 4 (c) 3 2 (d) 2 5
√
Answer: (d) 2 5 .

Solution: We know that the line joining the center of a circle to a point of tangency is
perpendicular to the tangent line. Hence, the line perpendicular to 2x − y + 1 = 0 and passing
through (2, 5) passes through the center of the circle. This is the line x + 2y − 12 = 0.

We are given the center also lies on x + y − 9 = 0. Solving the system of equations, we get the
center as (6, 3). Since (2, 5) lies on the circumference
p of the circle, its radius√must be the distance
from the center (6, 3) to the point (2, 5). This is (6 − 2)2 + (3 − 5)2 = 2 5.

10. Suppose that 16 points are drawn on a plane such that exactly 7 of those points are collinear.
Any set of three points which do not all belong to the 7 are noncollinear. If 3 random points
are selected from the 16 points, what is the probability that a triangle can be formed by joining
those points?
5 Carl Joshua Quines

15 17 19 63
(a) (b) (c) (d)
16 20 20 80

15
Answer: (a) .
16

Solution: We will use complementary counting, and consider the probability a triangle cannot
be formed. A triangle
cannot be formed only if the 3 points
chosen are among the 7 collinear
points, so there are 73 ways to do so. In total, there are 16
3 ways to choose the points. Thus
the probability we cannot form a triangle is
7


3 7·6·5 1
16 = 16 · 15 · 14 = 16 .


3
1 15
So the probability we can form a triangle is 1 − = .
16 16
11. The points (0, −1), (1, 1), and (a, b) are distinct collinear points on the graph of y 2 = x3 − x + 1.
Find a + b.

(a) −6 (b) −2 (c) 1 (d) 8

Answer: (d) 8 .

Solution: The line passing through the two points is y = 2x − 1. Substituting into the equation
and simplifying, we get x3 − 4x2 + 3x = 0. We know that x = 0 and x = 1 are solutions of this
(because they lie on both the line and the curve), hence we know x(x − 1) must be a factor.
Indeed, the polynomial is x(x − 1)(x − 3) = 0, and the remaining point must have x = 3. Since
it lies on y = 2x − 1, we get the point as (3, 5), and 3 + 5 = 8.

Remark: Curves of the form y 2 = x3 + ax + b for real numbers a and b are known as elliptic
curves. The problem describes collinear points on elliptic curves, which is how “point addition”
is defined on them. A theorem about rational points on elliptic curves, called the modularity
theorem, is famously used to prove Fermat’s Last Theorem.

12. What is the probability that a positive divisor of 220 317 also divides 28 36 ?
12 2 1 1
(a) (b) (c) (d)
85 9 9 6

1
Answer: (d) .
6

Solution: All divisors are of the form 2a 3b , where 0 ≤ a ≤ 20 and 0 ≤ b ≤ 17 are integers. For
this to also be a divisor of 28 36 , we must have a ≤ 8 and b ≤ 6. Thus the possible choices for a
9
are 0, 1, . . . , 8 out of 0, 1, . . . , 20, for a probability of . Similarly, the probability that b satisfies
21
7
the condition is . Multiplying, since we want both a and b to satisfy the condition, we get
18
9 7 1
· = .
21 18 6
13. Let ABC be a right triangle where AB = 7, BC = 24, and with hypotenuse AC. Point D is on
6 Carl Joshua Quines

AC such that AD : DC = 2 : 3. Let m and n be the relative prime positive integers such that
BD2 = mn . What is m + n?

(a) 554 (b) 550 (c) 544 (d) 540

Answer: (a) 554 .

Solution 1: Let the foot of the perpendiculars from D to AB and BC and be E and F ,
2 48
respectively. Then triangles AED and ABC are similar with the ratio 2 : 5, so ED = BC = .
5 5
3 21
Also, triangles DF C and ABC are similar with the ratio 3 : 5, so DF = AB = .
5 5

D
E

B F C

48
Note that EDF B is a rectangle, so ED = F B = . By the Pythagorean theorem on right
5
triangle DF B,
 2  2
2 2 2 21 48 2745 549
BD = DF + F B = + = = ,
5 5 25 5
so the answer is 549 + 5 = 554.

Solution 2: We use coordinates. Set A(0, 7), B(0, 0), C(24, 0).

A(0, 7)

D= 35 A + 25 C

B(0, 0) C(24, 0)

3 2
Since AD : DC = 2 : 3, we must have D = A + C, in terms of coordinates. To see this, we
5 5
can use the well-known formula or similar triangles, as above. Here, we will use vectors.

~ 2 ~ ~ ~ ~ 2 ~ ~ 2 ~ ~
 3
~ + 2 C.
~
Note that D is of the way from A on AC. Hence D = A + AC = A + C −A = A
5  5 5  5 5
3 2 3 2 48 21
Using this, we can solve the coordinates of D as · 0 + · 24, · 7 + · 0 = , .
5 5 5 5 5 5
549
Using the Pythagorean Theorem to find the distance we get , so the answer is 549 + 5 = 554.
5
Solution 3: By the Pythagorean theorem, AC = 25, hence AD = 10 and DC = 15. By
7 Carl Joshua Quines

Stewart’s Theorem,

AD · AC · DC + BD2 · AC = BC 2 · AD + AB 2 · DC
(10)(25)(15) + BD2 (25) = (242 )(10) + (72 )(15)
25BD2 = 2745,

549
so BD2 = and the answer is 549 + 5 = 554.
5
14. In chess, a knight moves by initially taking two steps in any horizontal or vertical direction
and then taking one more step in any direction that is perpendicular to its initial movement.
Suppose Renzo places a knight on a random tile on an 8 × 8 chessboard. Find the probability
that he can land on a corner tile in exactly two moves.
3 1 9 5
(a) (b) (c) (d)
16 4 16 8

5
Answer: (d) .
8

Solution: It is easier to solve the problem backwards. Beginning from the corners, we count
the number of squares that can be reached in exactly two moves. We can do this by drawing a
diagram:

From left to right are the number of possible squares that can be reached from the corners with
zero, one, and two moves. There are 40 possible squares that can be reached from the corners in
two moves, so there are 40 possible squares that can reach the corners in two moves. This gives
40 5
the probability as = .
64 8
15. In rectangle ABCD, point Q lies on side AB such that AQ : QB = 1 : 2. Ray CQ is extended
past Q to R so that AR is parallel to BD. If the area of triangle ARQ is 4, what is the area of
rectangle ABCD?

(a) 108 (b) 120 (c) 132 (d) 144

Answer: (b) 120 .

8 Carl Joshua Quines

Solution 1: Let BD and CQ intersect at point S. Triangles ARQ and BSQ are similar in the
ratio 1 : 2, hence their areas are in the ratio 1 : 4, and the area of BSQ is 16. We now try to
relate the area of triangle BSQ to the area of triangle ADB. To do this, we find the ratios of
their heights and their bases.

R
A Q B

D C

Triangle BSQ is similar to triangle DSC in the ratio 2 : 3. Hence their heights must be in the
ratio 2 : 3. Hence the height from S to QB and the height from S to DC are in the ratio 2 : 3.
But, in total, they form the length AD. Thus the height from S to QB and the side AD are in
the ratio 2 : 5.

But QB : AB = 2 : 3. So the areas of BSQ and ADB are in the ratio 2 · 2 : 3 · 5, or 4 : 15. Hence
the area of ADB is 60, making the area of the rectangle 120.

Solution 2: Knowing the answer is numerical, and that we only care about the ratios of areas,
we can choose ABCD to be a square of side length 3. We also use coordinates. Set A(0, 3),
B(3, 3), C(3, 0), and D(0, 0). This gives Q(1, 3).

The slope of BD is 1, so the line through A parallel to BD  is y− x − 3 = 0. The line CQ can be
3 18
solved as 3x + 2y − 9 = 0. Intersecting them gives R , . We can see the distance from
5 5
18 3 3
R to AQ is − 3 = , and since AQ = 1, its area is . Thus the ratio of areas of ARQ to
5 5 10
3
ABCD is : 9. Since its actual area is 4, the area of ABCD must be 120.
10
 
3 18
R ,
5 5
Q(1, 3)
A(0, 3) B(3, 3)

D(0, 0)
C(3, 0)
9 Carl Joshua Quines

Remark: Solution 2 can be justified as follows. Imagine compressing the plane horizontally
until ABCD becomes a square. The ratios of the areas do not change, the ratios of points on a
line do not change, and neither do parallel lines. Also, you can scale ABCD such that its side
length is 3, and again, the ratios of the areas do not change again.

Scaling and compressing are both examples of affine transformations, which preserve parallel
lines, ratios of areas, and ratios of distances between points on a line. Affine transformations
make solving problems like these much easier.

PART II. Choose the best answer. Each correct answer is worth three points.

1. How many two-digit numbers are there such that the product of their digits is equal to a prime
raised to a positive integer exponent?

(a) 27 (b) 28 (c) 29 (d) 30

Answer: (c) 29 .

Solution: Since the digits are single-digit numbers, the prime must be a single-digit number as
well. Thus it is either 2, 3, 5, or 7.

In the case of 2, the digits could be any of 1, 2, 4, or 8. This gives 42 = 16 possibilities. We must
exclude the case 1 · 1, so there are 15 possibilities.

Similarly, in the case of 3, the digits could be any of 1, 3, or 9. This gives 32 = 9 possibilities.
Excluding the case 1 · 1, we get 8 possibilities.

Similarly, for 5 and 7, we get 22 − 1 = 3 possibilities each. This gives a total of 15 + 8 + 3 + 3 = 29

possibilities in all.

2. ABCDEF is a six-digit perfect square in base 10 such that DEF = 8 × ABC. What is
A + B + C + D + E + F ? (Note that ABCDEF , ABC, and DEF should be interpreted as
numerals in base 10 and not as products of digits.)

(a) 18 (b) 27 (c) 36 (d) 45

Answer: (b) 27 .

Solution: Note that the number is 1000 · ABC + DEF , or 1000 · ABC + 8 · ABC, or 1008 · ABC.
Observe that 1008 = 24 · 32 · 7.

Since the exponents in the prime factorization of a perfect square must all be even, ABC must be
7 times a perfect square. Since ABC is a three-digit number, it must be at least 100. Since DEF
1000
is a three-digit number, ABC must at most = 125. The only possibility is 7 · 16 = 112,
8
making the perfect square 112896, with digit sum 27.

3. Quadrilateral ABCD has AB = 25, BC = 60, CD = 39, DA = 52, and AC = 65. What is the
inradius of 4BCD?

(a) 14 (b) 15 (c) 16 (d) 18

10 Carl Joshua Quines

Answer: (a) 14 .

Solution 1: We note that triangle ABC has side lengths 25, 60, and 65. Thus, it is similar to
the triangle with side lengths 5, 12, and 13, which is a right triangle. So triangle ABC is right
with ∠ABC = 90◦ . Similarly, triangle CDA is similar to a 3-4-5 right triangle, so ∠CDA = 90◦ .

F
A C
E

Drop perpendiculars E and F to AC from B and D, respectively. Since 4AEB ∼ 4ABC,

AB 125 BC 300
AE = AB · = and BE = AB · = . Similarly, because 4DF A ∼ 4CDA,
AC 13 AC 13
DA 208 CD 156
AF = DA · = and DF = AD · = .
AC 5 AC 5
By the Pythagorean Theorem, BD2 = (BE + F D)2 + EF 2 = (BE + F D)2 + (AF − AE)2 = 3969,
so BD = 63. By Heron’s theorem, the area of BCD is
p √
81(81 − 63)(81 − 60)(81 − 39) = 81 · 18 · 21 · 42 = 1134.
1134
But since the area is also the semiperimeter times the inradius, the inradius must be = 14.
81
Solution 2: By the previous solution, we know that ∠ABC = ∠CDA = 90◦ . Thus, the
circumcircle of triangle ABC has diameter AC, and so does the circumcircle of triangle CDA.
So they must share the same circumcircle; in other words, quadrilateral ABCD is cyclic. We can
also get this from the fact ∠ABC + ∠CDA = 180◦ .

Since ABCD is cyclic, by Ptolemy’s theorem,

AB · CD + BC · DA = AC · BD
(25)(39) + (60)(52) = (65)BD
BD = 63.

Then a similar solution to the above can be used.

4. Find the sum of the first 20 positive integers that are multiples of either 3 or 7 but not both.

(a) 336 (b) 399 (c) 529 (d) 592

Answer: (c) 529 .

11 Carl Joshua Quines

Solution: We first estimate the largest integer among them. Suppose the largest integer is n.
Then the number of integers less than or equal to n that are multiples of 3 or 7 but not both, by
PIE, is jnk jnk jnk n n n 8n
20 = + −2 ≈ + −2 ≈ ,
3 7 21 3 7 21 21
so n is about 52.5. Checking, we see n = 52 satisfies the first equation, so 52 is an upper bound.

Summing the multiples of 3 less than 52 gives

(17)(18)
3 + 6 + · · · + 51 = 3(1 + 2 + · · · + 17) = 3 · = 459.
2
Summing the multiples of 7 less than 52 gives

7 + 14 + · · · + 49 = 7(1 + 2 + · · · + 7) = 196.

The multiples of 21 less than 52 are 21 and 42, with sum 63. Hence the total sum is 459 + 196 −
2(63) = 529.

5. Q is a rational function with xQ(x + 2018) = (x − 2018)Q(x) for all x 6∈ {0, 2018}. If Q(1) = 1,
what is Q(2017)?

(a) 2020 (b) 2019 (c) 2018 (d) 2017

Answer: (d) 2017 .

Solution 1: By repeatedly substituting x + 2018 for x, we get

(x − 2018)Q(x) = xQ(x + 2018) = (x + 2018)Q(x + 4036) = (x + 4036)Q(x + 6054) = · · · .

Let R(x) = (x − 2018)Q(x) + 2017. By the above, we see it is equal to 0 for all of 1, 1 + 2018,
2017
1 + 4036, and so on. Hence, it must be the zero function. Solving, we get Q(x) = − ,
x − 2018
and thus Q(2017) = 2017.

Solution 2: Since the answer is numerical, any rational function Q that satisfies the conditions
should give the right answer. We see the x − 2018 multiplied on the right-hand side, and wanting
to cancel it out, try that as the denominator.
1
We try Q(x) = . This satisfies the equation, but does not satisfy Q(1) = 1, so we
x − 2018
2017
multiply it by −2017. We get Q(x) = − , which satisfies all the conditions. Then
x − 2018
Q(2017) = 2017.

6. How many ordered pairs (x, y) of positive integers are there such that 1 ≤ x ≤ y ≤ 20 and both
y y+2
and are integers?
x x+2

(a) 38 (b) 36 (c) 34 (d) 32

Answer: (d) 32 .
12 Carl Joshua Quines

y y+2 y y+2
Solution: Since and are integers, then − 1 and − 1 are both integers as well.
x x+2 x x+2
y−x y−x
But these are and , so y − x is divisible by both x and x + 2.
x x+2
We first note that y = x is always a solution, so we only deal with the case y > x:

• If x = 1, then y − x is divisible by 1 and 3, or by 3. Then y − x can be 3, 6, . . . 18: 6 pairs.

• If x = 2, then y − x is divisible by 2 and 4, or by 4. Then y − x can be 4, 8, 12, 16: 4 pairs.

• If x = 3, then y − x is divisible by 3 and 5, or by 15. Then y − x can only be 15: 1 pair.

• If x = 4, then y − x is divisible by 4 and 6, or by 12. Then y − x can only be 12: 1 pair.

• No larger choice of x is possible, because y − x would have to be too large.

This gives a total of 20 + 6 + 4 + 1 + 1 = 32 pairs.

Remark: Note that

y y+2 2(y − x)
− =
x x+2 x(x + 2)
must be an integer, so 2(y − x) is divisible by both x and x + 2. After this, a similar solution to
the above works. It is also possible to write a solution using a bounding argument.

7. An entertainment agency has seven trainees. Each of the trainees does at least one of dancing,
singing, and rapping, and no two trainees have the same skill set. How many ways can the agency
choose three trainees to form a group, provided that the group must have at least one dancer,
one singer, and one rapper (who are not necessarily distinct)?

(a) 26 (b) 29 (c) 32 (d) 35

Answer: (c) 32 .

Solution: We use complementary counting and count the number of ways their group does not
have a dancer, singer, or rapper. Suppose that no one in a group can dance. Then the only
possible skill sets are {singer}, {rapper}, and {singer, rapper}. By the conditions in the problem,
the group must consist of these three people, and there is only 1 group.

Similarly, there is only 1 possible group where no one is a singer, and only 1 possible group where
no one is a rapper. It can be seen that these groups are all
different. This makes 3 ways for a
group to not have a dancer, singer, or rapper. There are 73 = 35 ways to choose in total, and
subtracting 3 ways that do not work give 32 remaining ways.

8. Let 4ABC be a right triangle such that its hypotenuse AC has length 10 and ∠BAC = 15◦ .
Let O be the center of the circumcircle of 4ABC. Let E be the point of intersection of the lines
tangent to the circumcircle at points A and B, and F be the point of intersection of the lines
tangent to the circumcircle at points B and C. The area of 4OEF is
25 225
(a) (b) 50 (c) 100 (d)
8 8
13 Carl Joshua Quines

Answer: (b) 50 .

E O

B C
F

Solution: It is well-known that O is the midpoint of AC. Observe that triangles EBO, ABC,
and OBF are similar: they share right angles ∠EBO and ∠OBF due to tangent lines, and it can
be shown that the other angles are congruent. But ∠EOF = ∠EOB + ∠BOF = 15◦ + 75◦ = 90◦ ,
so the area of triangle OEF is
1 1 AC · BO AC · BO AC 2 · BO2 BO2 52
· OE · OF = · · = = = 1 = 50.
2 2 BC AB 2 · AC sin 15◦ · AC cos 15◦ sin 30◦ 2

From left-to-right, we apply similar triangles, use the fact that ∠ACB = 15◦ , and then use the
double-angle formula.

Remark: The figure provided is not to scale. There are many other ways to solve this problem
that use the same observations. It is possible to use the values of sin 15◦ and cos 15◦ to solve this
problem: one can memorize them, or derive them from the half-angle formula. Here is a hint for
a synthetic way to derive their values: let D be the point on side BC such that ∠BAD = 60◦ ,
then triangle BAD is 30-60-90, and triangle ADC is isosceles.

9. A real number x is chosen randomly from the interval (0, 1). What is the probability that
blog5 (3x)c = blog5 xc? (Here, bxc denotes the greatest integer less than or equal to x.)
1 5 1 1
(a) (b) (c) (d)
6 36 9 12

1
Answer: (a) .
6

Solution: Fix n = blog5 xc; we find the probability that n is also blog5 (3x)c. For n = blog5 xc,
we need n ≤ log5 x < n + 1, or x ∈ [5n , 5n+1 ).

For n = blog5 (3x)c to be true, we need 5n ≤ 3x < 5n+1 to be true as well. It is clear that 3x ≥ 5n
5n+1 5n+1

is already true for all x ∈ [5n , 5n+1 ). We need to be true x < , or x ∈ 5n , .
3 3

To find the probability, we find the lengths of the intervals and divide them. This gives
5n+1
3 − 5n 5n 53 − 1 1
= · = .
5n+1 −5 n 5 n 5−1 6
14 Carl Joshua Quines

1
Since it does not depend on n, the total probability is also .
6
Remark: Compare to AMC 12B 2006/20: Let x be chosen at random from the interval (0, 1).
What is the probability that blog10 4xc − blog10 xc = 0?

10. What is the remainder when 12018 + 22018 + · · · + 20172018 is divided by 2018?

(a) 0 (b) 2 (c) 1009 (d) 2017

Answer: (c) 1009 .

Solution 1: We note that 2018 factorizes as 2 · 1009. We will evaluate the sum modulo 2 and
modulo 1009, and combine the results to get the sum modulo 2018.

Modulo 2, the sum is

12018 + 02018 + · · · + 12018 ≡ 1 + 0 + · · · + 1 ≡ 1.

Modulo 1009, the sum is

12018 + 22018 + · · · + 10082018 + 02018 + 12018 + · · · + 10082018 ≡ 2 12018 + 22018 + · · · + 10082018 .

2
By Fermat’s little theorem, a2018 ≡ a1008 · a2 ≡ 12 · a2 ≡ a2 for any a 6≡ 0. So by the formula
for the sum of squares, the sum is
(1008)(1009)(2017) (1008)(2017)
2 12 + 22 + . . . + 10082 ≡ 2 ·


≡ 1009 · ≡ 0.
6 3
Since the sum is 1 modulo 2 and 0 modulo 1009, the whole sum must be 1009 modulo 2018.

Solution 2: Here is an alternative way to evaluate the sum modulo 1009. Let g be a primitive
root. Then the set {1, 2, . . . , 1008} is the same set as 1, g, g , . . . , g 1007 , by definition of a
2

primitive root. The sum is

  g (1008)(2018) − 1
2 1 + g 2018 + g (2)(2018) + · · · + g (1007)(2018) ≡ 2 · ≡0
g 2018 − 1
by Fermat’s little theorem.

Remark: The same method in Solution 2 shows the following well-known lemma: let p be a
prime and m an integer not divisible by p − 1. Then 1m + 2m + · · · + (p − 1)m is divisible by p.
It works even when m is negative.

PART III. All answers should be in simplest form. Each correct answer is worth six points.

1. Suppose two numbers are randomly selected in order, and without replacement, from the set
{1, 2, 3, . . . , 888}. Find the probability that the difference of their squares is not divisible by 8.

555
Answer: .
887

Solution: The square of a number modulo 8 can be either 0, if it is divisible by 4; 1, if it is odd;

or 4, if it 2 modulo 4.
15 Carl Joshua Quines

We will use complementary counting and determine the probability it is divisible by 8. Note
that the difference can only be 0 modulo 8 if their squares are the same modulo 8. Thus we are
looking for the probability that the two numbers chosen are both:
222 221 221
• divisible by 4, with probability · = ,
888 887 3548
444 443 886
• odd, with probability · = ,
888 887 3548
222 221 221
• or 2 modulo 4, with probability · = .
888 887 3548
1328 332 332 555
The total probability is = , so the probability it is not divisible by 8 is 1 − = .
3548 887 887 887
2. Let P (x) be a polynomial with degree 2018 whose leading coefficient is 1. If P (n) = 3n for
n = 1, 2, . . . , 2018, find P (−1).

Answer: 2019! − 3 .

Solution: Note that P (n) − 3n is a degree 2018 polynomial, with leading coefficient 1, whose
roots are 1, 2, . . . , 2018. Hence

P (n) − 3n = (n − 1)(n − 2) · · · (n − 2018),

and substituting −1 gives

P (−1) − 3(−1) = (−1 − 1)(−1 − 2) · · · (−1 − 2018)

P (−1) + 3 = (−2)(−3) · · · (−2019)
P (−1) = (−1)2018 (2)(3) · · · (2019) − 3
P (−1) = 2019! − 3.

3. A sequence {an }n≥1 of positive integers is defined by a1 = 2 and for integers n > 1,
1 1 1 n
+ + ··· + + = 1.
a1 a2 an−1 an
∞
X 3k (k + 3)
Determine the value of .
4k ak
k=1

21
Answer: .
8

Solution: Applying the condition for both n and n + 1, and subtracting them, we get
   
1 1 1 n 1 1 1 1 n+1 n−1 n+1
0= + + ··· + + − + + ··· + + + = − .
a1 a2 an−1 an a1 a2 an−1 an an+1 an an+1
From this, we get (n − 1)an+1 = (n + 1)an . We write the first few terms of the sequence:
2, 4, 12, 24, 40, 60, . . . . Ignoring a1 , it seems that the terms are 2n(n − 1), and it is easy to prove
this by induction.
16 Carl Joshua Quines

Since we have a polynomial denominator, we decompose into partial fractions. This gives us
k+3 4 3
= − ,
k(k − 1) k−1 k

and now the sum telescopes as

∞ ∞
3k (k + 3) 31 (1 + 3) X 1 3k
 
X 4 3
= + · · −
4k ak 41 a1 2 4k k−1 k
k=1 k=2
∞ 
3k 3k+1 1

3 1 X 1
= + · · − k · .
2 2 4k−1 k − 1 4 k
k=2

3 1 9 1 21
The remaining terms are + · · = .
2 2 4 1 8
Remark: Compare to Stanford Mathematics Tournament Advanced Topics 2011/2: Compute
∞
X (7n + 32) · 3n
.
n · (n + 2) · 4n
n=1

4. In triangle ABC, D and E are points on sides AB and AC respectively, such that BE is
perpendicular to CD. Let X be a point inside the triangle such that ∠XBC = ∠EBA and
∠XCB = ∠DCA. If ∠A = 54◦ , what is the measure of ∠EXD?

Answer: 36◦ .

Solution 1: Let Y be the intersection of BE and CD. Let Z be the foot of the altitude from
X to BC.

D
Y

B Z C

Observe that as ∠XBZ = ∠DBY and ∠DY B = ∠XZB = 90◦ , triangles DBY and XBZ are
DB YB DB XB
similar by AA. Then = , or = . Combined with ∠DBY = ∠XBZ again,
XB ZB YB ZB
we get triangles DBX and Y BZ are similar by SAS. Similarly, triangles ECX and Y CZ are
similar.
17 Carl Joshua Quines

Using quadrilateral ABY C, and then triangle XBC, we get

360◦ = ∠BAC + ∠ACY + ∠CY B + ∠Y BA

360◦ = 54◦ + ∠ACY + 270◦ + ∠Y BA
36◦ = ∠ACY + ∠Y BA
36◦ = ∠XCB + ∠XBC.
144◦ = ∠BXC.

Also, by the two similarities, ∠DXB + ∠EXC = ∠Y ZB + ∠Y ZC = 180◦ . Thus, we get

∠DXE = 360◦ − ∠DXB + ∠EXC − ∠BXC = 360◦ − 180◦ − 144◦ = 36◦ .

Solution 2: We cite the following well-known lemma without proof: the isogonal conjugate of a
point P with respect to quadrilateral ABCD exists if and only if ∠AP B + ∠CP D = 180◦ .

D
Y

B C

Let Y be the intersection of BE and CD. Apply the lemma to point Y in quadrilateral BCDE,
as ∠BY C + ∠DY E = 180◦ . It can be easily shown its isogonal conjugate must be X: hence X
and Y are isogonal conjugates with respect to BCDE. Then

∠EXD = 180◦ − ∠XDE − ∠XED

= 180◦ − ∠Y DB − ∠Y EC
= 180◦ − (180◦ − ∠Y DA) − (180◦ − ∠Y EA)
= ∠Y DA + ∠Y EA − 180◦
= (360◦ − ∠DAE − ∠DY E) − 180◦
= 360◦ − 54◦ − 90◦ − 180◦
= 36◦ .

Solution 3: Knowing the answer is numerical, we take the limit as D approaches A. Then E is
the foot from B to AC, and X coincides with B. Then ∠EXD = ∠EBA. But triangle BEA is
right, hence ∠EBA = 90◦ − ∠EAB = 36◦ .
18 Carl Joshua Quines

D
Y

B C

Solution 4: Knowing the answer is numerical, we carefully draw a figure and measure ∠EXD
with a protractor, seeing it is about 36◦ . But 36◦ = 90◦ − 54◦ , which is relatively nice, so that
must be the answer.

Remark: Solution 1 reveals a spiral similarity. A spiral similarity is a rotation followed by a

dilation, with the same center for both transformations. Here, note that 4XBZ ∼ 4DBY is a
spiral similarity with center B. It carries D to X and Y to Z.

In the solution, we proved a special case of the spiral similarity lemma: the spiral similarity
that carries A to B and C to D is the same spiral similarity that carries A to C and B to D.
Here, the spiral similarity must also carry D to Y and X to Z. Since the center is B, that means
4DBX ∼ 4Y BZ.

The lemma in Solution 2 is proved as follows, with the notation of the problem. Let the
perpendiculars from Y to BC, CA, AB, and DE be P , Q, R, and S respectively. Let X 0 be the
isogonal conjugate of Y with respect to ADE. Note that X 0 is the circumcenter of QRS by the
well-known isogonal pedal triangle lemma, and similarly X is the circumcenter of P QR from the
same lemma. But we can show P QRS is cyclic, hence X = X 0 .

Don’t use Solution 4 as a model. In an answers-only contest, Solution 3 is both easier and safer.
√
5. Define g : R \ {1} → R and h : R \ 3 → R as follows:
√
1+x 3 + 3x
g(x) = and h(x) = √ .
1−x 3 − 3x
How many ways are there to choose f1 , f2 , f3 , f4 , f5 ∈ {g, h}, not necessarily distinct, such that
(f1 ◦ f2 ◦ f3 ◦ f4 ◦ f5 )(0) is well-defined and equal to 0?

Answer: 8 .

Solution: Let y = arctan x. Then x = tan y. Observe that

1+x tan 45◦ + tan y
g(tan y) = g(x) = = = tan(y + 45◦ ).
1−x 1 − tan 45◦ tan y
19 Carl Joshua Quines

Similarly,
√ √
1 3
3 + 3x 3 + x tan 30◦ + tan y
h(tan y) = h(x) = √ · 3
= √ = = tan(y + 30◦ ).
3 − 3x 1
3 1− 3 x3 1 − tan 30◦ tan y

The value of y begins at 0◦ , and g adds 45◦ to it, while h adds 30◦ to it. The largest possible y
we get at the end is 45◦ · 5 = 225◦ . Thus, for the tangent to be equal to 0, the total must be 180◦ .

The only possible way to make 180◦ is with two 45◦ s and three 30◦ s, or two gs and three hs.
5!
There are = 10 different ways to arrange these.
2!3!
However, we cannot make an intermediate total 90◦ , which would make the tangent undefined.
This happens only in the cases g ◦ g ◦ h ◦ h ◦ h and h ◦ h ◦ h ◦ g ◦ g. Subtracting from 10, we get a
total of 8 ways.

With thanks to Nathanael Joshua Balete for figures; and Paco Adajar, Kurt Ang, and Sean Ty for
solutions.