Answer key:

1. 42
2. 779 (p = 754, q = 25)
3. 749 (a = 191, b = 367)

4. 20 (a = 14, b = 3, c = 3)
5. 714
6. 876 (a100 = 248 · 317 · 517 )
7. 498 ({a, b, c} = {3, 3, 61}, d = 17)

8. 973 (p = 973, q = 213 )

9. 231 (a = 225, b = 6)
10. 300
11. 510

12. 523
13. 528
14. 911 (N = 5911)

15. 68

1
Solutions:

1. Two circles ω1 and ω2 , each of radius 112, are drawn such that their centers
lie on each others’ circumferences. Then, two smaller circles with equal
radii are drawn such that they are both internally tangent to ω1 and ω2 ,
and are externally tangent to each other. What is the length of the radius
of one of the smaller circles?
Answer (42): Let the centers of ω1 and ω2 be A and B, and let the
center of a smaller circle be O. Using symmetry, AB is tangent to both of
the smaller circles at its midpoint, which we will call M . Let the radius
of the smaller circles be r, and let P be the point where the smaller circle
with center O is internally tangent to the circle with center A. Then
AM = 112 2 = 56, OM = r, and OA = P A − P O = 112 − r. Since AM O
6
2 2 2
is right, by the Pythagorean theorem AM + OM = OA , so:

1122 − 562
562 + r2 = (112 − r)2 ⇒ 562 = 1122 − 2 · 112r ⇒ r = = 42
2 · 112
√ √
2. √ √ functions f (x) and g(x) is either equal to x − 1 + x − 9 or
Each of the
x − 1 − x − 9. The equation f (g(x)) = 2 has exactly one real solution,
which can be expressed as pq , where p and q are relatively prime positive
integers. Compute p + q.
Answer p (779): First,
p let’s solve for the
p value of g(x).
p Since f (g(x)) = 2,
either g(x) − 1 + g(x) − 9 = 2 or g(x) − p 1 − g(x) −p 9 = 2. How-
ever, in both cases it must
√ be that g(x) ≥ 3, so g(x) − 1 + g(x) − 9 ≥
√ √
9 −p 1 + 9 − 9 =p 2 2 > 2, rendering the first case impossible. There-
fore, g(x) − √ 1 − √ − 9 = 2, and by inspection and the fact that
g(x)
the function x − 1 − x − 9 is strictly decreasing, g(x) = 10 is the only
solution.
√ √ √ √
Now, either x − 1 √ + x − 9√ = 10 or x√− 1 − x − 9 = 10. Since the
maximum value of x − 1 −√ x − 9 is√2 2, which is less than 10, the
second case is impossible, so x − 1 + x − 9 = 10. We now solve for x,
starting by squaring both sides:

√ √ p
( x − 1 + x − 9)2 = (x − 1) + (x − 9) + 2 (x − 1)(x − 9) = 100
p
⇒ x2 − 10x + 9 = 55 − x
⇒ x2 − 10x + 9 = (110 − 2x)2 = 552 − 110x + x2
58 · 13 754
⇒ 100x = 552 − 9 = 58 · 52 ⇒ x = =
25 25

Thus p + q = 754 + 25 = 779 .

2
3. Noelle writes down the factors of 24 · 179 from least to greatest. Starting
with the first factor, she circles every other factor. Noelle then records
the sum of all of the circled factors, as well as sum of all of the uncircled
factors. The ratio of the smaller recorded sum to the larger recorded sum
is ab , where a and b are relatively prime positive integers. Compute 2a + b.
Answer (749): Since 24 < 17, the factors of 24 · 179 when written in
order will be the factors of 24 , then the factors of 24 multiplied by 17k for
each successive positive integer k up to 10. Written in a chart-like array,
with every other factor circled, the factors look like:

20 21 22 23 24
20 · 17 21 · 17 22 · 17 23 · 17 24 · 17
20 · 172 21 · 172 22 · 172 23 · 172 24 · 172
20 · 173 21 · 173 22 · 173 23 · 173 24 · 173
.. .. .. .. ..
. . . . .

Note that each group of every two rows repeats, except that all terms
are multiplied by 172 each time. So the ratio of the sum of circled terms
to uncircled terms is the same as the ratio computed just for the first
0 2
+24 +21 ·17+23 ·17
two rows, which is 212+2+2 191
3 +20 ·17+22 ·17+24 ·17 = 367 . Therefore, 2a + b =
2 · 191 + 367 = 749 .

4. Point P is located in the interior of regular hexagon ABCDEF so that the

distance from P to the sides√AB, CD, and EF are 8, 4, and 9 respectively.
Then the length of AB is a c b , where a and c are relatively prime positive
integers and b is not divisible by the square of any prime. Find a + b + c.
Answer (20): Let AB = s, and let the lines AB and EF intersect at X,
lines AB and CD intersect at Y , and lines CD and EF intersect at Z.
Then 4XY Z is equilateral,√
and has
√
side length XY = XZ = Y Z = 3s.
(3s)2 3 9s2 3
Note that [P QR] = 4 = 4 . Since we know the lengths of the
altitudes from P to the side lengths of 4XY Z, we can write the area of
4XY Z in another way: [XY Z] = [XP Y ]+[Y P Z]+[ZP X] = 21 (XY )(8)+
1 1 21
2 (Y Z)(4) + 2 (ZX)(9) = 2 (3s). Setting the two expressions equal, we
get: √ √
9s2 3 21 14 3
= (3s) ⇒ s =
4 2 3
So, a + b + c = 14 + 3 + 3 = 20 .

3
5. A token is placed in the top left square of a 5 × 4 checkerboard. A move
consists of moving the token one square up, down, to the left, or to the
right, so long as the token does not move off of the checkerboard. How
many nine-move sequences move the token to the bottom right square?
Answer (714): Let U , D, L, and R denote a move upwards, downwards,
leftwards, and rightwards. A seven-move path from the top left corner to
the bottom right corner would be a permutation of DDDRRRR; an nine-
move sequence would have an additional U D or a LR mixed in, with some
constraints. For the first case, where U D is added in to DDDRRRR, the
U must not be before or after all of the Ds, as that would mean the token
moves off of the grid at some point. So we first permute DDDDDRRRR,
and then change one of the middle Ds to a U ; there are 94 · 3 = 378

ways to do this. Similarly, for the second case where LR is added into
DDDRRRR, the L cannot go before or after all of the Rs, so we first
permute DDDRRRRRR, and then change one of the middle Rs to an L;
there are 93 · 4 = 336 ways in this case. The total number of sequences

of moves is 378 + 336 = 714 .

6. Define the sequence an so that a0 = 1, a1 = 2, a2 = 3, and for all n > 2,

an = lcm(an−1 , an−2 + an−3 ).

Compute the remainder when the number of divisors of a100 is divided by

1000.
Answer (876): We compute the first few terms of the sequence:

n 0 1 2 3 4 5 6
an 1 2 3 3 3·5 2·3·5 2 · 32 · 5
n 7 8 9 10
an 2 · 32 · 5 23 · 32 · 5 23 · 32 · 5 23 · 32 · 52

Note that the numbers in bold are three consecutive terms that are re-
peated with a gap of 6, except all of them are multiplied by 23 · 3 · 5. Using
the fact that lcm(km, kn) = klcm(m, n) for all integer k, m, n, this pattern
will continue forever, so an+6 = 23 · 3 · 5 · an for all n ≥ 2. Therefore:

a100 = a4+16·6 = (23 · 3 · 5)16 · a4 = 248 · 317 · 517

The number of divisors of a100 is then (48 + 1)(17 + 1)(17 + 1) = 15 876 .

4
7. Suppose a, b, c and d are (not necessarily distinct) prime numbers such
that a + b + c − 3d = 42 and ab + bc + ca − 3d = 182 . Compute abc − 3d.
Answer (498): Since both equations have a −3d that otherwise seems
hard to use, we consider taking the equations modulo 3 to get rid of it.
We have:

a + b + c ≡ 42 ≡ 1 mod 3
ab + bc + ca ≡ 182 ≡ 0 mod 3

Looking at the second equation, the possibilities for the moduli of ab,
bc, and ca are limited: (ab, bc, ca) ≡ (0, 0, 0), (1, 1, 1), (2, 2, 2), (0, 1, 2) or
permutations thereof. The case (0, 1, 2) is impossible, as a, b, or c would
be 0 mod 3, but that would mean two of ab, bc, and ca would be 0 mod 3.
For (2, 2, 2), each of the pairs (a, b), (b, c), and (c, a) must have one 1 mod
3 element and one 2 mod 3 element, which is an impossible task. For
(1, 1, 1), it must be that a ≡ b ≡ c mod 3 (and of nonzero modulus), but
that means a+b+c ≡ 0 6≡ 1 mod 3. We are left with (ab, bc, ca) = (0, 0, 0),
in which case without loss of generality, a ≡ b ≡ 0 mod 3 and c ≡ 1 mod 3.
Since a ≡ b ≡ 0 mod 3 and a and b are prime, a = b = 3. Substituting
into the given equations yields a system of equations in two variables:
6 + c − 3d = 16 and 9 + 6c − 3d = 324. Solving, we get c = 61 and d = 17
(check: c and d are both prime!). To finish, abc−3d = 3·3·61−3·17 = 498 .
8. Seven teams play a round-robin tournament, where each pair of teams
plays each other exactly once. In any match, the two teams are equally
likely to win, with no ties, and all match results are independent. The
probability that at most five teams win at least two matches is pq , where
p and q are relatively prime positive integers. Find p.
Answer (973): More simply put, we wish to find the probability that at
least two teams win at most one match each. There are two main cases:
Case 1: Team A and Team B lose all games except their head-to-head
matchup (where one
of them wins their one game, and the other loses all
games). There are 72 ways to choose two teams, and the probability that
they each lose all of their games against the other five teams is 215 · 215 = 2110 ,
so this case has a probability of 72 · 2110 = 221


10 of occurring.

Case 2: Team A,B,C beat each other in a cyclical fashion (A → B → C

→ A or A → C → B
→ A), and at least two of these teams lose all other
games. There are 73 ways to choose three teams, and the probability that
they beat each other in a cyclical fashion is 14 . There are 3 ways to choose
which two teams lose all other games; given the choice, the possibility
then happens with probability 214 · 214 = 218 . But we count the case where
all three teams lose all of their other games three times, overcounting this
possibility twice, so we subtract off 2 · 214 · 214 · 214 = 2111 to get that the
probability that at least two of the three teams lose all of their other

5
 games is 3 · 218 − 2111 = 23
211 . In all, the probability that this case occurs is
7

1 23 805
3 · 4 · 211 = 213 .
21 805 973
The total probability is 210 + 213 = 213 , and the numerator is 973 .
9. A spotlight is 30 meters away from a very large wall, and can cast a beam
of light in the shape of a right circular cone, with its apex at the spotlight’s
location. The region of the wall that the spotlight illuminates is a circular
disk with center A and an area of 300π square meters. The spotlight is
then swiveled so that point A lies on the perimeter of √ the illuminated
area. The new illuminated area, in square meters, is aπ b, where b is not
divisible by the square of any prime. Compute a + b.
Answer (231): The initial
√ circular disk, which we’ll call γ for future
reference, has radius 10 3, and since the height, slant height, and radius
of the disk create a 30-60-90 triangle, the angle that the height and slant
height make in the cone is 30◦ . When the spotlight is swiveled, the new
illuminated region is an ellipse, since the intersection of a cone is always
a conic. We now proceed to find the major and minor axes of the ellipse.
Let P be the vertex of the cone, let AB be the major axis of the ellipse,
let CD be the minor axis of the ellipse, let Q be the intersection of γ with
AB, and let R be on γ such that QR is a diameter of γ.
If we take the cross-section of the cone using the plane perpendicular to the
wall and containing the line AB, we can see that 6 P AB =√30◦ +30√ ◦
= 60◦ ,
so 4BAP is a 30-60-90√triangle. √ Therefore, AB = AP 3 = 30 3, and
the major axis is 12 · 30 3 = 15 3.
To find the minor axis, draw 4CP D. Since CP = P D and 6 CP D =
30◦ +30◦ = 60◦ , 4CP D ∼ 4QP R. So, the ratio of the heights from point
P is the equal to CD
QR . If M is on CD so that P M is an altitude, then
√
M is the center of the ellipse, so M A = 15 2, and by the Pythagorean
theorem:
p q √ √
P M = M A + P A = (15 2)2 + 302 = 15 6
2 2

√ 15√6 √ √
So, CD = QR· PPM 1
A = 20 3· 30 = 30 2, and the minor axis is 2 ·30 2 =
√
15 2.
√ √ √
We conclude that the area of the ellipse is (15 3)(15 2)π = 225π 6, so
a + b = 225 + 6 = 231 .
10. Delthea brings two copies each of four different books to a book club and
gives them to five friends so that no one gets two copies of the same book.
She then observes that the only non-empty subset of friends for which the
books she gave them can be paired up with their identical copies is the
entire group of five friends. If N is the number of ways that Delthea could
have given the books, find N/10.

6
 Answer (300): Since these books come in identical pairs, we may visu-
alize the books as edges of different colors, and the five friends as vertices.
Then the given condition “the only non-empty subset of friends for which
the books she gave them can be paired up with their identical copies is
the entire group of five friends” simply becomes “the graph is connected”.
However, there are only four edges, so the graph must be a tree. It is a
known fact that there are nn−2 trees on n vertices, so there are 55−2 = 125
trees in this case (but if one does not know this formula, the number of
trees on 5 vertices can be found with a small amount of casework). Lastly,
there are 4! = 24 ways to color the edges to ascertain which book copies
were given to which friends, so the total number of ways Delthea could
have given the books is 125·24 = 3000. To finish, N/10 = 3000/10 = 300 .
11. Let P (x) = ak xk + ak−1 xk−1 + . . . + a1 x1 + a0 be the polynomial that
k
X
satisfies (x9 + x − 1)P (x) = (x9 − 1)9 − x + 1 for all x. Compute |ai |.
i=0
9 9 9
Answer (509): To find P (x), we divide (x −1) −x+1 by x +x−1. We
have (x9 −1)9 = (x9 +x−1)(x9 −1)8 −x(x9 −1)8 , and to repeat the process
in general, x9−k (x9 −1)k = (x9 +x−1)(x9−k )(x9 −1)k−1 −x9−k (x9 −1)k−1 .
Applying this rule repeatedly to the remainders, the signs alternate while
the power of x increases by one each time, resulting in:

(x9 − 1)9 − x + 1 = (x9 + x − 1)((x9 − 1)8 − x(x9 − 1)7 + x2 (x9 − 1)6

− . . . − x7 (x9 − 1) + x8 ) − x9 − x + 1
= (x9 + x − 1)((x9 − 1)8 − x(x9 − 1)7 + x2 (x9 − 1)6
− . . . − x7 (x9 − 1) + x8 − 1)
⇒ P (x) = (x9 − 1)8 − x(x9 − 1)7 + . . . − x7 (x9 − 1) + x8 − 1

To find the sum of the absolute values of the coefficients of P (x), we

can observe that none of the expansions of the binomial expressions will
overlap, due to each one taking a different modulus modulo 9 for the
degrees of their terms, so we can take the sums of the absolute values of the
coefficients of the binomial expressions separately. The only expressions
that overlap are (x9 − 1)8 and −1, where the constant coefficients cancel,
so we will subtract 1 from the total of the binomial coefficient sum. Note
that the sum of the absolute values of (x9 − 1)k is equal to the sum of the
coefficients of (x9 + 1)k , which upon plugging in x = 1 is just 2k . So the
sum of the absolute values of the coefficients of P (x) is 28 + 27 + 26 + . . . +
21 + 20 − 1 = (29 − 1) − 1 = 510 .

7
12. Points A, B, C, D, E, and F lie on a circle, in that order. The region
enclosed by the chords AD, BE, and CF is an equilateral√ triangle. If
AC 2 + CE 2 + EA2 = 475 and [4ACE] − [4BDF ] = 4 3, find the value
of BD2 + DF 2 + F B 2 . (Here [K] denotes the area of the region K.)
Answer (523): Let P be the intersection of AD and BE, let Q be
the intersection of AD and CF , and let R be the intersection of BE
and CF . Then 4P QR is an equilateral as given in the problem; let
P Q = QR = RP = s. Let AP = a, BP = b, CQ = c, DQ = d, ER = e,
and F R = f , and assume that none of these segments contain a side
of 4P QR. (If this isn’t the case, then we can just reassign the points
B ↔ E and C ↔ F so that it works out while all given equations remain
the same.)
Now, by Power of a Point on each pair of chords:

a(s + d) = b(s + e)
c(s + f ) = d(s + a)
e(s + b) = f (s + c)

Adding all three equations together yields s(a + c + e) + ad + cf + eb =

s(b + d + f ) + be + da + f c, and simplifying, a + c + e = b + d + f .
Turning our attention to the second equation, we split the areas into mul-
tiple triangles and use the sine area formula with the 120◦ angles:
√
4 3 = [ACE] − [BDF ] = ([AQC] + [CRE] + [EP A] + [P QR])
− ([BP D] + [DQF ] + [F RB] + [P QR])
√
3
= · ((ac + ce + ea) − (bd − df − f b))
4
⇒ ac + ce + ea = bd + df + f b + 16

For the first equation, we use the law of cosines with the 120◦ angles:

475 = AC 2 + CE 2 + EA2
= (a + s)2 + (a + s)c + c2 + (c + s)2 + (c + s)e + e2
+ (e + s)2 + (e + s)a + a2
= 2(a2 + c2 + e2 ) + (ac + ce + ea) + s(a + c + e) + 3s2
= 2(a + c + e)2 − 3(ac + ce + ea) + s(a + c + e) + 3s2
= 2(b + d + f )2 − 3(bd + df + f b + 16) + s(b + d + f ) + 3s2
= 2(b + d + f )2 − 3(bd + df + f b) + s(b + d + f ) + 3s2 − 48
= BD2 + DF 2 + F B 2 − 48

We conclude that BD2 + DF 2 + F B 2 = 475 + 48 = 523 .

8
13. Find the least positive integer n such that 126n − 5n + 11n2 is divisible
by 553 .
Answer (528): We can solve the modular equations 126n −5n +11n2 ≡ 0
mod 113 and 126n − 5n + 11n2 ≡ 0 mod 53 separately, and combine them
afterwards by the Chinese Remainder Theorem.
First, we solve 126n − 5n + 11n2 ≡ 0 mod 113 . Since 126 − 5 = 121 = 112 ,
we can use the LTE lemma, which yields υ11 (126n − 5n ) = υ11 (126 − 5) +
υ11 (n) = 2 + υ11 (n). This means that if 11|n, then 113 |126n − 5n , but
we also have 113 |11n2 at the same time, so n ≡ 0 mod 11 are solutions to
126n − 5n + 11n2 ≡ 0 mod 113 . Since we always have 112 |126n − 5n , and
112 |11n2 if and only if 11|n, 112 |126n − 5n + 11n2 if and only if 11|n, and
therefore these are the only solutions.
Next, we solve 126n − 5n + 11n2 ≡ 1 + 11n2 ≡ 0 mod 53 . Solving for
n2 gives n2 ≡ (−11)−1 ≡ 34 mod 53 . Taking the equation modulo 5 gives
n2 ≡ 4 mod 5, or n ≡ 2, 3 mod 5, so either n = 5a + 2 or n = 5a + 3 for
some integer a.
For the case n = 5a + 2, plugging into n2 ≡ 34 mod 125 gives (5a + 2)2 ≡
25a2 + 20a + 4 ≡ 34 mod 53 , or 5a2 + 4a ≡ 6 mod 52 ; taking this equation
modulo 5 gives 4a ≡ 1 mod 5, so a ≡ 4 mod 5. Write a = 5b + 4 for some
integer b. Substituting into 5a2 + 4a ≡ 6 mod 52 gives 5(5b + 4)2 + 4(5b +
4) ≡ 20b + 21 ≡ 6 mod 52 , or 4b ≡ 2 mod 5, so b ≡ 3 mod 5. Upon writing
b = 5c + 3 for some integer c, we get n = 125c + 97, so n ≡ 97 mod 125.
For the case n = 5a + 3, plugging into n2 ≡ 34 mod 125 gives (5a + 3)2 ≡
25a2 + 30a + 9 ≡ 34 mod 53 , or 5a2 + 6a ≡ 5 mod 52 ; taking this equation
modulo 5 gives a ≡ 0 mod 5. Write a = 5b for some integer b. Substituting
into 5a2 + 6a ≡ 5 mod 52 gives 5(5b)2 + 6(5b) ≡ 5b ≡ 5 mod 52 , or b ≡
1 mod 5. Upon writing b = 5c + 1 for some integer c, we get n = 125c + 28,
so n ≡ 28 mod 125.
Finally, we want to find the least value of n such that 11|n and ei-
ther n ≡ 28 mod 125 or n ≡ 97 mod 125. The first few positive in-
tegers that are 28 mod 125 are 28, 153, 278, 403, 528, . . ., where 528 is a
multiple of 11. The first few positive integers that are 97 mod 125 are
97, 222, 347, 472, 597, . . ., none of which are a multiple of 11. Therefore,
the least possible value of n is 528 .
14. Start with the number 1, and perform a sequence of moves to turn it into
the number 100. A move can be either of these operations:

• Add 1 to the number.

• If the number has two digits, swap the digits. This may only be done
if doing so would increase the value of the number.

Let N be the sum of the number of digit swaps that happen, over all
different sequences of moves. Compute the remainder when N is divided
by 1000.

9
Answer (911): Some first initial observations: since the only way to go
from 1 to 11 is adding 1, and likewise for 99 to 100, we can start at 11
and end at 99. Also, we really only need to focus on swap move, as the
add move fills out the space between. So, drawing a grid where each row
is m0, m1, . . . , m9 and each column is 0n, 1n, . . . , 9n, we can see that the
sequences of moves going from aa to (a + k)(a + k) are in bijection with
the sequences of moves going from bb to (b + k)(b + k), just by shifting the
moves so that the starting point moves from aa to bb. Inspired by this,
we find a recursion to work out the number of sequences going from 11 to
nn for n = 1, 2, 3, . . . , 8 which we’ll call An , so that we can compute the
number of moves that use each possible swap move.
The base case is, of course, A1 = 1. To find an expression for An , we
will consider the use of the swap moves an → na for 1 ≤ a < n. If
none of them are used, then the number of sequences is An−1 : we go
from 11 to (n − 1)(n − 1) without having any of the new swap moves,
and then repeat adding 1 for the rest of the way. At most one of the
swap moves can be used, so let an → na be used for some a; then the
number of sequences using this swap move is Aa : we go from 11 to aa,
then add 1 repeatedly until we get to an, then use the swap move na,
and are forced to keep adding 1 the rest of the way. So our recursion is
An = An−1 + (An−1 + An−2 + . . . + A1 ) = 2An−1 + An−2 + An−3 + . . . + A1 .
Writing out the first few terms of the recursion, A2 = 2, A3 = 5, and
A4 = 13, at which point we can guess that An = F2n−1 , where Fn is the
Fibonacci sequence (F1 = F2 = 1, Fn = Fn−1 +Fn−2 ). This can be proven
by induction and the identity F2k = F2k−1 + F2k−3 + . . . + F1 .
Now let’s count the number of times each swap move is used. The swap
move ab → ba, for 1 ≤ a < b ≤ 9, is used Aa A10−b times: we go from 11
to aa, then repeatedly add 1 until ab, use the swap move ab → ba, then
repeatedly add 1 until bb, and finally go from bb to 99. We sum all of these
up to get our answer; using the identity F2k = F2k−1 + F2k−3 + . . . + F1
again, the total number of times that swap moves are used over all possible
sequences of moves is:

8
X 9
X 8 9−a
X X 8
X 9−a
X
Aa A10−b = Aa Ab = F2a−1 F2b−1
a=1 b=a+1 a=1 b=1 a=1 b=1
8
X 8
X
= F2a−1 F2(9−a) = F2a−1 F18−2a
a=1 a=1
= F1 F16 + F3 F14 + F5 F12 + F7 F10 + F9 F8 + F11 F6 + F13 F4 + F15 F2
= 1 · 987 + 2 · 377 + 5 · 144 + 13 · 55 + 34 · 21 + 89 · 8 + 233 · 3 + 610 · 1
= 5 911

10
15. Let S be the set of all complex numbers w such that |w| = 53. Let a1
aj − (7 + 13i)
and a2 be two distinct complex numbers in S such that is
aj − (55 − 7i)
imaginary for j = 1, 2, and let b1 and b2 be two distinct complex numbers
bk − (41 + 27i)
in S such that is imaginary for k = 1, 2. Let z be the
bk − (81 + 39i)
z − a1 z − b1
complex number that satisfies the property that both and
z − a2 z − b2
are real. Then z = m + ni for real numbers m and n. Find m + n.
Answer (68): Denote the circle traced out by |w| = 53 by β. The
condition on a1 and a2 is that 7 + 13i, aj , and 55 − 7i form a right
triangle with the right angle at aj , which is equivalent to the condition
that ai lies on the circle with the diameter as the hypotenuse. The circle,
which we will denote by ξ, has center (7+13i)+(55−7i) = 31 + 3i and radius
√ 2
|(7 + 13i) − (31 + 3i)| = 242 + 102 = 26. Similarly, the condition on
b1 and b2 is that 41 + 27i, bk , and 81 + 39i form a right triangle with
the right angle at bk , which is equivalent to the condition that bk lies
on the circle with the diameter as the hypotenuse. This circle, which
we will denote by δ, has center (41+27i)+(81+39i) = 61 + 33i and radius
√ 2 √
2 2
|(41 + 27i) − (61 + 33i)| = 14 + 12 = 2 109.
The conditions on z mean that a1 , a2 , and z are collinear, and b1 , b2 , and
z are collinear. Since a1 and a2 are the intersections of circles β and ξ,
the line through a1 and a2 is the radical axis of β and ξ, and since b1 and
b2 are the intersections of circles β and δ, the line through b1 and b2 is the
radical axis of β and δ. So z is the intersection of these two radical axes,
and by the Radical Axis Theorem, the radical axis of ξ and δ also passes
through z. We then notice that the slope of the line passing through the
centers of ξ and δ is 1, so the slope of the radical axis of ξ and δ, which
is perpendicular, has slope −1, which means that all points on the radical
axis will have the same sum of real and imaginary parts. This means that
if we find a complex number on the radical axis, we will be done!
One such complex number on the radical axis we can find is the intersection
of the radical axis with the line passing through the centers of ξ and δ,
which we will denote P and Q respectively. Let A be the intersection of
P Q with the radical axis, and let B be√ one of the intersections
√ of circles
ξ and δ. Then BP = 26, BQ = 2 109, and P Q = 30√ 2. By the
Pythagorean theorem, AP 2 + AB 2 = BP 2 = 676 and (30 2 − AP )2 +
AB 2 √= AQ2 + AB 2 2
√ =2 BQ = 436. Subtracting the two equations
√ gives
2 · 30 2AP − (30 2) = 240, and so AP = 240+1800 √
60 2
= 17 2. Therefore,
A as a complex number is (31 + 3i) + (17 + 17i) = 48 + 20i, and the sum
of the real and imaginary parts of z is that of A, which is 48 + 20 = 68 .

11