Official Solutions: 2016 Sun Life Financial CMO Qualifying Repêchage Competition
Official Solutions: 2016 Sun Life Financial CMO Qualifying Repêchage Competition
4n + 7n + 20n
0≡ n
(mod 31)
7n n
20 100
≡ +1+ (mod 31)
35 35
n n
20 100
≡1+ + (mod 31)
4 4
n n
≡ 1 + 5 + 25 (mod 31)
≡ 1 + 5n + 52n (mod 31)
Since 53 ≡ 1 (mod 31), this sum only depends on n (mod 3). We calculate:
(a) If S is a point on the line y = x and T is a point on the horizontal x-axis so that P is on the
line segment ST, determine the minimum possible area of triangle OST.
Solution: Let (a, a) be the coordinates of S and (b, 0) the coordinates of T. The line segments
a−1 1 6a
SP, P T, and P T all have the same slope, so a−7 = 7−b . Solving for b we get b = a−1 .
6a 2
3a 2
The triangle has a base of OT = b and a height of a, so the area is A = 2(a−1) = a−1 =
q
3 3 3
3a + 3 + a−1 = 3(a − 1) + 6 + a−1 . By the AM-GM inequality, A ≥ 2 3(a − 1) a−1 + 6 = 12,
3
with equality when 3(a − 1) = a−1
. Equality holds when a = 2, and we have a = 2, b = 12 and
an area of 12.
1
Official Solutions: 2016 Sun Life Financial CMO Qualifying Repêchage Competition
(b) If U is a point on the line y = x and V is a point on the horizonal x-axis so that P is on the
line segment U V, determine the minimum possible perimeter of triangle OU V.
Solution: Consider a circle that is tangent to the lines OU, OV, and U V at the points A, B,
and C respectively. We notice that AU = CU and BV = CV, so the perimeter of OU V equals
OA + OB = 2OA.
In order to minimize this sum, we seek to minimize the size of the circle. Notice that if the
circle is tangent to U V at a point that is not P, then there exists a line through P that does
not touch the circle. Taking this line as U V would allow for the circle to be smaller. Thus,
the perimeter will be minimized when the circle is tangent to the point P.
p
Let U = (a, a), V = (b, 0), and OA = x. Then V B = V C, so (b − 7)2 + 1 = x − b. Solving
x2 −50 2 −50
for b yields b = 2x−14 . Similarly U P = U A and we find that a = 2√x 2x−16 .
6a
As in part (a), we have b = a−1
. Substituting in a and b to this equation gives:
x2 −50
x2 −50 6 √
2 2x−16
2x−14
= x2 −50
√ −1
2 2x−16
x2 −50
x2 −50 6 √
2 √2x−16
2x−14
= x2 −2
√ 2x−34
2 2x−16
x −50 2 6(x2 −50)
= x2 −2√2x−34
√ 2x−14
(x2 − 50)(x√2 − 2 2x − 34) = (x2 − 50)(12x − 84)
(x − 50)(x2 − (2 2 + 12)x + 50) = 0
2
√ √ q √
Solving this for x yields: x = ± 50, x = 6 + 2 ± 2 3( 2 − 1)
√
When x = ± 50, a and b are both 0, which is not a solution. Of the two remaining solutions,
we take the larger one, as the other occurs from q the circle that is contained inside triangle
√ √
OU V. thus the minimal perimeter is 2(6 + 2 + 2 3( 2 − 1)).
3. Given an n × n × n grid of unit cubes, a cube is good if it is a sub-cube of the grid and has side
length at least two. If a good cube contains another good cube and their faces do not intersect, the
first good cube is said to properly contain the second. What is the size of the largest possible set
of good cubes such that no cube in the set properly contains another cube in the set?
Solution: Let S be the set of good cubes with side length 2 or 3. For any s ∈ S, we define Cs to
be the set of good cubes that have s as their centre. Then we can see that for any s and any pair of
distinct cubes in Cs , that one cube properly contains the other. Also, for any good cube c1 in the
grid, there exists a cube c2 ∈ S such that c2 is the centre of c1 . Thus, the size of largest possible
set of good cubes with no cube properly containing each other is at most |S|. Finally, we observe
that no cube in S can properly contain another cube in S. Thus |S| = (n − 1)3 + (n − 2)3 is the
size of the largest possible set.
2
Official Solutions: 2016 Sun Life Financial CMO Qualifying Repêchage Competition
f (x + f (y)) + f (x − f (y)) = x.
Solution: Let a = f (0) and let y = 0, x = a. The given equation becomes f (2a) + f (0) = a which
is equivalent to f (2a) = 0. Now, let y = 2a and we get 2f (x) = x or f (x) = x/2. If we substitute
this into the given relation, we see that the equation holds with equality. Thus, f (x) = x/2 is the
only such function.
5. Consider a convex polygon P with n sides and perimeter P0 . Let the polygon Q, whose vertices are
the midpoints of the sides of P, have perimeter P1 . Prove that P1 ≥ P20 .
Solution: When P0 is a triangle, the sides of P1 are parallel to the sides of P0 and exactly half
of the length of the side they are parallel to, so the result holds. When P0 has at least 4 sides,
label the verties v0 , v1 , . . . vn−1 clockwise. Label the midpoints m0 , m1 , . . . , mn−1 clockwise, where
m0 is the midpoint of v0 v1 . We observe that 2mi mi+1 = vi vi+2 , so it is equivalent to show that
v0 v2 + v1 v3 + · · · + vn−1 v1 ≥ v0 v1 + v1 v2 + · · · vn−1 v0 .
The segment vi vi+2 intersects vi vi+1 at a point w and intersects vi+1 vi+2 at the point x. Observe that
vi w + wvi+1 ≥ vi vi+1 by the triangle inequality. We can continue this all around the the polygon.
We observe that over all segments vi vi+1 , the segments vi w and wvi+1 are disjoint subsegments of
the set of segments vi vi+2 . Thus, we have v0 v2 + v1 v3 + · · · + vn−1 v1 ≥ v0 v1 + v1 v2 + · · · vn−1 v0 and
P1 ≥ P20 .
6. Determine all ordered triples of positive integers (x, y, z) such that gcd(x + y, y + z, z + x) >
gcd(x, y, z).
Solution: Let g = gcd(x+y, y+z, z+x). We observe that gmust divide (x+y)+(y+z)−(x+z) = 2y.
Similarly, g must also divide 2x and 2z. Thus g divides 2 gcd(x, y, z). In order to have g > gcd(x, y, z)
we must have g = 2 gcd(x, y, z).
Assume that gcd(x, y, z) = 1. Then 2 must divide x + y, y + z, and z + x. This can only happen
when all of x, y, z have the same parity. Since their gcd is 1, they must therefore all be odd. Thus,
we have that x, y, z are odd numbers multiplied by a common factor. This is equivalent to having
the largest power of 2 that divides each number being the same.
7. Starting at (0, 0), Richard takes 2n+1 steps, with each step being one unit either East, North, West,
or South. For each step, the direction is chosen uniformly at random from the four possibilities.
Determine the probability that Richard ends at (1, 0).
Solution: In order for Richard to end up at (1, 0) he must take the same number of steps north as
he does south, and the number of steps he takes east must be one more than the number he takes
west. Thus, we see that the number of steps taken north or west is n as is the number
taken south
2n+1
or west. The number of ways to choose the n steps that are north or west is n , which is the
3
Official Solutions: 2016 Sun Life Financial CMO Qualifying Repêchage Competition
same as the number of ways to choose the steps that are south or west. A step is west if and only
if it is chosen in both sets. Since there are a total of 2n + 1 steps, the probability of ending at (1, 0)
2
(2n+1
n )
will be 42n+1 .
8. Let n ≥ 3 be a positive integer. A chipped n-board is a 2 × n checkerboard with the bottom left
square removed. Lino wants to tile a chipped n-board and is allowed to use the following types of
tiles:
Two tilings T1 and T2 are considered the same if there is a set of consecutive Type 1 tiles in both
rows of T1 that can be vertically swapped to obtain the tiling T2 . For example, the following three
tilings of a chipped 7-board are the same:
For any positive integer n and any positive integer 1 ≤ m ≤ 2n − 1, let cm,n be the number of
distinct tilings of a chipped n-board using exactly m tiles (any combination of tile types may be
used), and define the polynomial
2n−1
X
Pn (x) = cm,n xm .
m=1
for all n ≥ 3.
Solution: Given a chipped (n − 1)-board, we can extend this to a chipped n-board in 4 different
ways. For each of the two new spots in the last column, we can either have the tile be attached
to the previous column, or we can have it be a new tile. This gives a recurrence of Pn (x) =
4
Official Solutions: 2016 Sun Life Financial CMO Qualifying Repêchage Competition
(1+2x+x2 )Pn−1 (x). However, note that we have overcounted, as some of these tilings are equivalent
to each other and some are not valid.
We get equivalent tilings when the k rightmost columns of the (n − 1)-board contain two 1 × k
boards, in which case the tilings that consist of attaching one from the last column and not attaching
the other are equivalent. Note that the number of such overcountings is given by the polynomial
x3 Pn−1−k , as this tells us how mamy ways we can tile the left-most n − k − 1 columns.
Also, note that if the (n − 1)-board is a single piece, then neither of the two constructions which
attach a single piece is valid. This mean that there are two invalid tilings of size 2, which has
polynomial 2x2 . Thus, our recurrence is actually
Pn−1 (x) = (1 + 2x + x2 )Pn−2 (x) − x3 (Pn−3 (x) + Pn−4 (x) + · · · P1 (x)) − 2x2 .