Algebra Tiebreaker Solutions

1. Let
1
f (x) = .
1
1−
1−x
Compute f 2016 (2016), where f is composed upon itself 2016 times.

Proposed by Joshua Siktar

Solution. We compute f (2016) = 2015 2015 1 1

2016 , f ( 2016 ) = − 2015 , and f (− 2015 ) = 2016. Therefore, f is periodic
with period 3, and so f 2016 (2016) = f 2016 (mod 3) (2016) = f 3 (2016) = 2016 .

2. Determine the value of the sum

X
ij(−1)i+j .
1≤i<j≤50

Proposed by David Altizio

Solution. Let ai = i(−1)i . Then since

 2
X X X X
 ai  = ai aj = 2 ai aj + a2i ,
1≤i≤50 1≤i,j≤50 1≤i<j≤50 1≤i≤50

we have
X 1
2 ai aj = (−1 + 2 − 3 + 4 + · · · + 50)2 − (12 + · · · + 502 ) = 252 − 50(51)(101) = −42300,
6
1≤i<j≤50

whence the answer is 21150 .

3. Suppose x and y are real numbers which satisfy the system of equations
17 23
x2 − 3y 2 =
and 3x2 − y 2 = .
x y
√
Then x2 + y 2 can be written in the form m n, where m and n are positive integers and m is as small as
possible. Find m + n.

Proposed by David Altizio

Solution. Note that the equations rearrange to x3 − 3xy 2 = 17 and 3x2 y − y 3 = 23. Thus

x3 − 3xy 2 + i(3x2 y − y 3 ) = 17 + 23i =⇒ (x + yi)3 = 17 + 23i.

Taking the magnitude of both sides yields

p √
(x2 + y 2 )3/2 = (17 + 23i)1/2 =⇒ x2 + y 2 =
3 3
172 + 232 = 818.

The requested answer is 818 + 3 = 821 .

 Combinatorics Tiebreaker
√
1. For a set S ⊆ N, define f (S) = {d se | s ∈ S}. Find the number of sets T such that |f (T )| = 2 and
f (f (T )) = {2}.

Proposed by Patrick Lin

Solution. Denote Sn = {k | dke = n}, and an the number of non-empty subsets of Sn . Observe that Sn
contains exactly 2n − 1 elements, and so an contains 22n−1 − 1 elements. Since f 2 (T ) = {2}, it follows that
f (T ) ⊂ S2 = {2, 3, 4}; in particular, f (T ) contains exactly two of those three elements. Thus the number of sets
T that satisfy the problem condition is given by a2 a3 + a2 a4 + a3 a4 = (7)(31) + (7)(127) + (31)(127) = 5043 .
2. Let S = {1, 2, 3, 4, 5, 6, 7}. Compute the number of sets of subsets T = {A, B, C} with A, B, C ∈ S such that
A ∪ B ∪ C = S, (A ∩ C) ∪ (B ∩ C) = ∅, and no subset contains two consecutive integers.

Proposed by Patrick Lin

Solution. The last condition is equivalent to (A ∪ B) ∩ C = ∅, and A ∪ B ∪ C = S tells us each element is

in at least one set. Hence each element has four possible states: in A only, in B only, in both A and B, and
in C only. Letting An be the number of sets T that satisfy the condition for S = {1, 2, . . . , n} such that n is
in only A and similarly for B, AB, and C, we have

An+1 = Bn + Cn
Bn+1 = An + Cn
ABn+1 = Cn
Cn+1 = An + Bn + ABn ,

with A1 = B1 = AB1 = C1 = 1. Solving this up to n = 7 yields A7 + B7 + AB7 + C7 = 394 .

3. Let S be the set containing all positive integers whose decimal representations contain only 3s and 7s, have at
most 1998 digits, and have at least one digit appear exactly 999 times. If N denotes the number of elements
in S, find the remainder when N is divided by 1000.

Proposed by Patrick Lin

Solution. Note that element in S contains at least 999 and at most

1998 digits. For some number
0≤k<
999, the number of integers of length 999+k in S is equal to 2 999+k , and for k = 999 there are 1998
999 elements
P999 999+k
1998

1999

1998

9992000
1998


in S. Hence N = 2 k=0 999 − 999 = 2 999 − 999 = 1000 − 999 . By examining v2 (2000!) and
v2 (1000!), we find that 2000 Wolstenholme’s yields 2000 16




1000 ≡ 0 (mod 8), and using 1000 ≡ 8 ≡ 120 (mod 125).
Hence 2000 1998 500 2000 1998





1000 ≡ 120 (mod 1000). Noting that 999 = 1999 1000 yields 999 ≡ 0 (mod 1000), and so the
answer is 120 − 0 = 120 .
 Computer Science Tiebreaker
1. A planar graph is a connected graph that can be drawn on a sphere without edge crossings. Such a drawing
will divide the sphere into a number of faces. Let G be a planar graph with 11 vertices of degree 2, 5 vertices
of degree 3, and 1 vertex of degree 7. Find the number of faces into which G divides the sphere.

Proposed by Cody Johnson

Solution. By double counting, the total number of edges is 12 (11 · 2 + 5 · 3 + 1 · 7) = 22. Thus, by Euler’s
polyhedron formula, there are F = E − V + 2 = 22 − 17 + 2 = 7 faces. Note: there does exist such a graph,
for example:

2. The Stooge sort is a particularly inefficient recursive sorting algorithm defined as follows: given an array A
of size n, we swap the first and last elements if they are out of order; we then (if n ≥ 3) Stooge sort the first
d 2n 2n 2n α
3 e elements, then the last d 3 e, then the first d 3 e elements again. Given that this runs in O(n ), where
α
α is minimal, find the value of (243/32) .

Proposed by Cody Johnson

Solution. Let T (n) be the number of steps Stooge sort takes on an array of size n. Then
   2 !  3 !
2 2 2
T (n) = 3T n = 32 T n = 33 T n = ....
3 3 3

Let k be the smallest integer such that ( 32 )k n ≤ 1. Then k ≥ log3/2 n, so the algorithm will run in O(3log3/2 n ) =
5
O(nlog3/2 3 ). Finally, (243/32)log3/2 3 = (3/2)log3/2 3 = 35 = 243 .
3. Let ε denote the empty string. Given a pair of strings (A, B) ∈ {0, 1, 2}∗ ×{0, 1}∗ , we are allowed the following
operations:


 (A, 1) → (A0, ε)

(A, 10) → (A00, ε)




(A, 0B) → (A0, B)


 (A, 11B) → (A01, B)
(A, 100B) → (A0012, 1B)





(A, 101B) → (A00122, 10B)


We perform these operations on (A, B) until we can no longer perform any of them. We then iteratively delete
any instance of 20 in A and replace any instance of 21 with 1 until there are no such substrings remaining.
Among all binary strings X of size 9, how many different possible outcomes are there for this process performed
on (ε, X)?

Proposed by Cody Johnson

Solution. Let [·] denote the value when we read · as a binary integer. Now we claim this process performed
on (ε, X) will output b[X]/3c (with enough leading zeroes). It is clear for small enough values of |X|. Now
consider the following algorithm for division when the first digits are 100 or 101:
 Geometry Tiebreaker
1. Point A lies on the circumference of a circle Ω with radius 78. Point B is placed such that AB is tangent to
the circle and AB = 65, while point C is located on Ω such that BC = 25. Compute the length of AC.

Proposed by David Altizio

Solution. Extend P C past C to intersect Ω at D. Then by Power of a Point AB 2 = AC · AD, so

AB 2 652
AD = AC = = 169.
25
Now let D0 be the point on Ω such that AD0 is a diameter of Ω. Then AD0 = 156; combining this with
AB = 65 yields BD0 = 169 as well. From here it’s not hard to see that D ≡ D0 , so 4DAB is a right triangle.
Finally, note that since A and D are antipodal ∠DCA = 90◦ as well. Thus C is the foot of the perpendicular
from A to BD, so
AB · AD 65 · 156
AC = = = 60 .
BC 169
2. Identical spherical tennis balls of radius 1 are placed inside a cylindrical container of radius 2 and height 19.
Compute the maximum number of tennis balls that can fit entirely inside this container.

Proposed by Patrick Lin

Solution. Observe that we can fit two balls into the bottom such that they both touch the bottom; it is
then clear that the optimal way to pack in balls is to place them in layers of two each, then stack them in
such a way that the line formed by connecting the centers of each pair is orthogonal to the pair above and
below.
Since each ball is tangent to the other three, connecting their centers forms a tetrahedron ABCD of side length
2. Let M be the midpoint of AB and N be the midpoint of CD. To find the vertical height between the two
layers, it suffices to compute M N . Project D and N onto the √
plane ABC to points D0 and N 0 , respectively,
then we find that CN 0 = N 0 D0 = DM , and thus M N 0 = 3 . Further, by Pythagorean Theorem we have
2 3
√ √ √ √
DD0 = 2 3 6 and N N 0 = 12 DD0 = 36 . Hence M N = M N 02 + N N 02 = 2.
Let k be the maximum √ number of layers we can fit√inside this container; this is the maximum solution to the
17
inequality 2 + (k − 1) 2 ≤ 19. Noting that √ 2
= 144.5 > 12, we have k = 13, and so the answer is 26 .

3. Triangle ABC satisfies AB = 28, BC = 32, and CA = 36, and M and N are the midpoints of AB and AC
respectively. Let point P be the unique point in the plane ABC such that 4P BM ∼ 4P N C. What is AP ?

Solution. Scale down by a factor of 4, so that AB = 7, BC = 8, and CA = 9.

Note that P is the intersection of the circumcircles of 4AN B and 4AM C. To see this, remark that by the
similarity condition ∠BM P = ∠N CP , so quadrilateral AM QC is cyclic. Similarly, AN P B is also cyclic.
√
Now perform a bc inversion Φ about A. Note that Φ sends B to C and vice versa. Furthermore, Φ sends
−→
M to the point M 0 on ray AC such that

AM · AM 0 = AB · AC =⇒ AM 0 = 2AC.

Similarly, this inversion sends N to the point N 0 such that AN 0 = 2AB. As a result, Φ sends (AN B) to
CN 0 and (AM C) to BM 0 , meaning that the image of Q about Φ is the centroid of the triangle homothetic
to 4ABC with scale factor 2 (i.e. 4AN 0 M 0 ).
The rest is computation. Let ma be the length of the A-median of 4ABC. Then
r
0 4 4 2(AB 2 + AC 2 ) − BC 2 28
AP = ma = = .
3 3 4 3
Therefore, by the definition of inversion,
63 27
AP 0 · AP = AB · AC = 63 =⇒ AP = = .
28/3 4

Scaling back up by a factor of 4 yields the desired answer of 27 .

Proposed by David Altizio

 Number Theory Tiebreaker
√
1. For all integers n ≥ 2, let f (n) denote the largest positive integer m such that m
n is an integer. Evaluate

f (2) + f (3) + · · · + f (100).

Proposed by Cody Johnson

√
Solution. √ Increment once for each perfect power. There are 100 − 1 = 99 first powers, 100 − 1 = 9
squares, b 3 100c − 1 = 3 cubes, etc. for a total of 99 + 9 + 3 + 2 + 1 + 1 = 115 .
2. For each integer n ≥ 1, let Sn be the set of integers k > n such that k divides 30n − 1. How many elements
of the set [
S= Si = S1 ∪ S2 ∪ S3 ∪ . . .
i≥1

are less than 2016?

Proposed by Cody Johnson

Solution. Note that if k | 30n − 1 then gcd(30, k) = 1. Now if gcd(30, k) = 1 then let n = 30−1 (mod k).
We have k | 30n − 1 and k > n, so S = {a | gcd(a, 30) = 1}. Finally, since φ(30) = 8, we can look at just the
first 2016 (mod 30) numbers and add them in. Since 1, 5 are the only numbers in [1, 6] coprime with 30, the
answer is 8 · 2016−6
30 + 2 = 538 .
3. Let {x} denote the fractional part of x. For example, {5.5} = 0.5. Find the smallest prime p such that the
inequality
p2  p 
X n
> 2016
n=1
p2
holds.

Proposed by Andrew Kwon

Solution. For each 1 ≤ n ≤ p2 , let n = kp + `, with 0 ≤ k, ` ≤ p − 1. Note that this fractional part is
equivalent to the sum of the remainders when np is divided by p2 . Then,
   
p p p p−1 p p
n ≡ (kp + `) ≡ (kp)(`) + ` (mod p2 ).
1 0

That is, np ≡ `p (mod p2 ). Then, for each n ≥ 1 not divisible by p,

 p  2
(p − n)p

n
+ = 1,
p2 p2

while there are p2 − p such values of n from 1 to p2 . Thus,

2
p  p
X n p2 − p
= ,
n=1
p2 2

and it follows that p2 − p > 4032. Simple estimates yield p = 67 as the smallest possible prime.