2021 ARML Local Problems and Solutions

Team Round (45 minutes)

T-1 The number 999992 has a prime factor p greater than 500. Compute p.

Answer: 2551

Solution: Notice that 999992 = 1000000 − 8 = 1003 − 23 . This factors as

(100 − 2)(1002 + 100 · 2 + 22 ).

Note that 100 − 2 = 98 = 2 · 7 · 7. Note also that 1002 + 100 · 2 + 22 = 10204 = 22 · 2551. Look
for prime factors of 2551. Because the problem states that 999992 has a prime factor greater
than 500, check whether 2, 3, or 5 are factors; none of these primes is a factor of 2551. (One
can independently confirm that 2551 is a prime.) Therefore the prime factor of 999992 greater
than 500 is 2551.

10
 2 1 1 1 A
T-2 The sum = + + ··· + is equal to for some integer A. Compute A.
n(n2 − 1) 12 30 495 165
n=3

Answer: 26

2 1 1
Solution: Notice that = − . The given sum is thus equal to
n(n2 − 1) n(n − 1) n(n + 1)
1 1 1 1 1 1
− + − + ··· + −
3(3 − 1) 3(3 + 1) 4(4 − 1) 4(4 + 1) 10(10 − 1) 10(10 + 1)
1 1
= − ,
3 · 2 10 · 11
26
after telescoping, which is , so A = 26.
165
1 2 1
− +
Similarly, the full partial fraction decomposition of the summand is . After
n−1 n n+1
1 1 1 1 26
expanding and cancelling, the sum reduces to − − + = .
2 3 10 11 165

T-3 Kandyce recently moved, and all of her belongings are packed in 285 boxes. Kandyce unpacks
boxes at a constant rate of 3 boxes in 2 hours. Kandyce’s friend Keenan also unpacks boxes
at a constant rate, and when they work together they unpack 75% more boxes than Kandyce
can unpack alone over the same time. Kandyce unpacks boxes for 4 hours every day (until all
of the boxes are unpacked). Given that Kandyce wants to finish unpacking in exactly 4 weeks,
compute the total number of hours she should ask Keenan to help unpack boxes.

Answer: 104

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Solution: Kandyce will unpack boxes for 4 · 7 · 4 = 112 hours over the four weeks, and in that
time she will unpack 112 · 32 = 168 boxes. The remaining 285 − 168 = 117 boxes are unpacked
by Keenan. Keenan’s rate of unpacking is 75% of Kandyce’s rate, so Keenan unpacks boxes at
a constant rate of 34 · 32 = 98 boxes per hour. To unpack 117 boxes, Keenan should unpack for
117 ÷ 98 = 104 hours.

T-4 A total of 2021 teams participated in an ARML Team Round. The table below shows the
percentage of teams that answered each of the ten questions correctly, rounded to the nearest
percent. A team’s score was five times the number of correct answers.
T-1 T-2 T-3 T-4 T-5 T-6 T-7 T-8 T-9 T-10
93% 86% 85% 82% 75% 72% 66% 43% 16% 3%
Compute the integer closest to the mean score on that ARML Team Round.

Answer: 31

Solution: Suppose there were N participating teams. Let pn where n ∈ {1, . . . , 10} denote the
+n
nth entry of the above table. Then there were N · pn100 correct answers on the nth question,
for some rounding error −0.5 ≤ n < 0.5. The mean score equals the total number of points
awarded across all teams divided by the number of teams, which equals
10
1  pn +  n p1 + · · · + p10 1 + · · · + 10
5·N · = +
N 100 20 20
n=1
621 1 + · · · + 10
=
+
20 20
1 + · · · + 10
= 31.05 + .
20
 
 10  | |+···+| |
By the Triangle Inequality, it follows that  1 +···+
20  ≤ 1 20 10 ≤ 10·0.5
20 = 14 . Thus the mean
score is between 30.80 and 31.30, and so the answer is 31.

T-5 Compute the number of subsets S of {1, 2, . . . , 15} such that S has at least two elements and
no two elements of S are relatively prime.

Answer: 153

Solution: Note that S cannot contain more than one prime, and it cannot contain any primes
greater than 7 because given any prime p > 7, no other positive multiple of p is in {1, 2, . . . , 15}.
Furthermore, S cannot contain 1, so S is a subset of {2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 15}. Doing
casework on which prime is in S can give an organized way to compute the number of desired
subsets S. It is sufficient to consider only the primes because if a subset S has a common com-
posite factor p ≥ 4, then p has a prime factor q ≥ 2, and therefore q is a common prime factor
of S. Note that that prime must be less than or equal to 7.

If 2 ∈ S, then every element must be even, and there are 6 other even elements of {1, 2, . . . , 15}.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Because S must have at least 2 elements, it follows that there must be at least one other even
element in S, and each of these 6 other even elements can either be in S or not. Subtracting out
the possibility that none of the 6 other even elements is in S, it follows that there are 26 − 1 = 63
such subsets S. Note that combining 2 with one or more of the elements of {4, 6, 8, 10, 12, 14}
satisfies the desired conditions, so each one of the 63 such subsets S satisfies the desired condi-
tions.

If 3 ∈ S, then every element must be a multiple of 3, and there are 4 other multiples of 3
in {1, 2, . . . , 15}. By a similar calculation to the one in the first case, there are 24 − 1 = 15 such
subsets S.

If 5 ∈ S, then every element must be a multiple of 5, and there are 2 other multiples of 5
in {1, 2, . . . , 15}. By a similar calculation to the one in the first case, there are 22 − 1 = 3 such
subsets S.

If 7 ∈ S, then every element must be a multiple of 7, and because 14 is the only other multiple
of 7 in {1, 2, . . . , 15}, there is only one such subset S, namely {7, 14}.

If S has no primes, then S is a subset of {4, 6, 8, 9, 10, 12, 14, 15}. If 9 ∈  15 ∈
/ S and / S, then any
such subset with at least two elements would work, so there would be 26 − 61 − 60 = 57 total
subsets in this case.

Otherwise, if 9 ∈ S but 15 ∈ / S, then only 6 or 12 can be in S, which yields 3 such subsets

in this case ({6, 9}, {9, 12}, and {6, 9, 12}). If 15 ∈ S but 9 ∈
/ S, then only 6, 10, 12 can be in S
3
for a total of 2 − 1 = 7 subsets in this case (because each of 6 or 10 or 12 could be in or not,
but subtract out the case where none of them appears because the subset must have at least
two elements). Finally, if 9 ∈ S and 15 ∈ S, then only 6 or 12 can be in S, which yields 22 = 4
such subsets in this case.

Hence, in total, there are 63 + 15 + 3 + 1 + 57 + 3 + 7 + 4 = 153 such subsets S.

Sketch of an Alternate Solution: The total number of subsets with 2 as one of the common
factors is 27 − 7 − 1 = 120 (there are 7 even elements, but the empty set and subsets of one
element are not allowed). Similarly there are 25 − 6 = 26 subsets with 3 as one of the common
factors, 23 − 4 = 4 subsets with 5 as one of the common factors, and 22 − 3 = 1 subset with 7
as one of the common factors.

Note that this solution only considers the primes up to 7 as one of the common factors. This is
because if p is one of the common factors in a subset, then the smallest element in the subset
is greater than or equal to p ≥ 2, and the next smallest element (which surely exists) is greater
than or equal to 2p, so it follows that 2p ≤ 15, which implies p ≤ 7.5. The concern is only
with prime factors because if there is a common factor between two elements of a subset, that
common factor must have a prime factor.

There is one subset with 6 as one of the common factors that was double-counted. The reader
may verify that no other subset was double-counted.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 It is also possible to have such a subset without a common factor greater than 1 by select-
ing {6, 10, 15}, {6, 12, 15}, or {10, 12, 15} (the reader may verify this), for a grand total of
120 + 26 + 4 + 1 − 1 + 3 = 153 subsets.

T-6 Let D1 and D2 be standard, fair, six-sided dice. Let D3 be a fair six-sided die with the numbers
1, 2, 2, 3, 3, 4 on its faces. Let D4 be a fair six-sided die such that for all integers k between 2
and 12 inclusive, the probability that the sum of a roll of D3 and D4 equals k is equal to the
probability that the sum of a roll of D1 and D2 equals k. Two of the four dice are chosen at
random. The two dice are rolled independently of one another. Compute the probability that
the positive difference of the numbers rolled is 2.

25
Answer: 108

1
Solution: There is a 36 probability of getting a sum of 2 when rolling dice D1 and D2 , and
the sum of the least numbers on D3 and D4 must also be 2; further, it must be that the least
number on D4 appears only once. Thus D4 has exactly one face that is 2 − 1 = 1. There is also
1
a 36 chance of getting a sum of 12, and the sum of the greatest number on D3 and D4 must
also be 12; further, it must be that the greatest number on D4 appears only once. Thus D4 has
exactly one face that is 8.
2
There is a 36 probability of getting a sum of 11 when rolling dice D1 and D2 , so there must
be two ways to obtain a sum of 11 when rolling D3 and D4 . These are obtained by the two
occurrences of 3 + 8.
3
There is a 36 probability of getting a sum of 10 when rolling dice D1 and D2 , so there must be
three ways to obtain a sum of 10 when rolling D3 and D4 . Two of these possibilities for a sum
of 10 using the D3 and D4 dice come from 2 + 8 and 2 + 8. If D4 had a 7, there would be a
total of four possible 10’s because of the two 3’s on D3 , and this is a contradiction. Thus D4
has exactly one face that is 10 − 4 = 6.
4
There is a 36 probability of getting a sum of 9 when rolling dice D1 and D2 , so there must
be four ways to obtain a sum of 9 when rolling D3 and D4 . Three out of the four possibilities
for a sum of 9 using the D3 and D4 dice come from 1 + 8, 3 + 6, and 3 + 6. One more possibility
must be created, and this can only occur without contradiction if D4 has exactly one face that
is 9 − 4 = 5.
2
There is a 36 probability of getting a sum of 3 when rolling dice D1 and D2 , so there must
be two ways to obtain a sum of 3 when rolling D3 and D4 . This is already confirmed because
D3 has two 2’s and D4 has one 1.
3
There is a 36 probability of getting a sum of 4 when rolling dice D1 and D2 , so there must
be three ways to obtain a sum of 4 when rolling D3 and D4 . Two of the three possibilities for a
sum of 4 using the D3 and D4 dice come from 3 + 1 and 3 + 1. This means that D4 has exactly
one face that is 4 − 1 = 3.

Finally, this means that D4 has exactly one face that is 4 so D4 has the numbers 1, 3, 4, 5, 6, 8

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 on its faces. (The reader can independently confirm that for all integers k between 2 and 12
inclusive, the probability that the sum of a roll of D3 and D4 equals k is equal to the probability
that the sum of a roll of D1 and D2 equals k.)

If two of the standard dice are rolled, the positive difference of the numbers rolled is 2 if the
numbers rolled are 1 & 3, 2 & 4, 3 & 5, 4 & 6, or the reverses of those cases. Thus this occurs
8
with probability 36 .

If one of the standard dice and D3 are rolled, the positive difference of the numbers rolled
is 2 if the numbers rolled are 1 & 3 (twice), 2 & 4, 3 & 1, 4 & 2 (twice), 5 & 3 (twice), or 6 &
9
4. There are 9 favorable outcomes, so this occurs with probability 36 .

If one of the standard dice and D4 are rolled, the positive difference of the numbers rolled
is 2 if the numbers rolled are 1 & 3, 2 & 4, 3 & 1, 3 & 5, 4 & 6, 5 & 3, 6 & 4, or 6 & 8. There
8
are 8 favorable outcomes, so this occurs with probability 36 .

Finally, if D3 and D4 are rolled, the positive difference of the numbers rolled is 2 if the num-
bers rolled are 1 & 3, 2 & 4 (twice), 3 & 1 (twice), 3 & 5 (twice), or 4 & 6. There are 8
8
favorable outcomes, so this occurs with probability 36 . Accounting for the probability of each
case happening, the probability that the positive difference of the numbers rolled is 2 equals
8 1 9 2 8 2 8 1 50 25
36 · 6 + 36 · 6 + 36 · 6 + 36 · 6 = 216 = 108 .

Note: The two nonstandard dice from the problem are known as Sicherman dice. More infor-
mation about Sicherman dice can be found at https://www.cut-the-knot.org/arithmetic/
combinatorics/Sicherman.shtml.

T-7 In tetrahedron ARM L, AR = RM = M L = AM = RL = 1 and AL = 85 . Compute the volume

of ARM L.
√
2 11
Answer: 75

Solution: Consider M AL as the base of the tetrahedron.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Let B be the midpoint of AL; then M B is an altitude of M AL. Then AM = 1 and AB = 45 ,
so M B = 35 , and [M AL] = 12
25 . Now let C be the circumcenter of M AL. The circumradius of
(AM )(M L)(AL)
M AL is 4·[M AL] = 56 . Because AR = M R = LR, it follows that RC is perpendicular
to the plane of M AL, and thus M CR has a right angle at C, with M C = 56 and M R = 1.
√ √ √
11 1 1 12 11 2 11
Therefore RC = 6 , and the volume of ARM L is 3 · [M AL] · (RC) = 3 · 25 · 6 = 75 .

T-8 Compute the least positive integer n such that each of 20n and 21n has exactly 60 positive
divisors.

Answer: 540

Solution: Let {pi } be the increasing sequence of all primes (where i ∈ N). Consider the prime
factorization of n = pai i where the ai are nonnegative integers. Then the prime factorizations
i≥1
 ai  ai
of 20n and 21n will be 20n = p1a1 +2 pa22 p3a3 +1 pa44 pi and 21n = pa11 p2a2 +1 pa33 pa44 +1 pi . So
i≥5 i≥5
20n has exactly (a1 + 3)(a2 + 1)(a3 + 2)(a4 + 1) (ai + 1) positive divisors and 21n has exactly
 i≥5
(a1 + 1)(a2 + 2)(a3 + 1)(a4 + 2) (ai + 1) positive divisors.
i≥5

Note that the number 540 = 22 · 33 · 51 satisfies the conditions of the problem. Indeed,
20n = 24 · 33 · 52 has exactly (4 + 1)(3 + 1)(2 + 1) = 60 positive divisors and 21n = 22 · 34 · 51 · 71
has exactly (2 + 1)(4 + 1)(1 + 1)(1 + 1) = 60 positive divisors. The claim to be proved is that
any number n that satisfies all of the conditions of the problem also satisfies n ≥ 540.

Step 1: Because 21n has exactly 60 positive divisors, because 60 is a multiple of 5, and because
5 is prime, it follows that at least one of a1 + 1, a2 + 2, a3 + 1, a4 + 2, and ai + 1 (i ≥ 5) is a
multiple of 5.

Case 1.1: Suppose a1 + 1 is a multiple of 5. If a1 + 1 ≥ 15, then it follows that n ≥ 214 > 540.
If a1 + 1 = 5, then a1 + 3 = 7 and the number of positive divisors of 20n is a multiple of 7 and
therefore cannot be 60, a contradiction. If a1 + 1 = 10, then n is a multiple of 29 . If n = 29 ,
then 21n = 29 · 3 · 7 has exactly (9 + 1)(1 + 1)(1 + 1) = 40 positive divisors. If n > 29 , then
n ≥ 2 · 29 > 540. Therefore if a1 + 1 is a multiple of 5, it follows that n is at least 540.

Case 1.2: Suppose a2 + 2 is a multiple of 5. If a2 + 2 ≥ 10, then it follows that n ≥ 38 > 540.
The case where a2 + 2 = 5 will be considered later.

Case 1.3: Suppose a3 + 1 is a multiple of 5. This implies a3 + 1 ≥ 5 and so n ≥ 54 > 540.

Therefore if a3 + 1 is a multiple of 5, it follows that n is at least 540.

Case 1.4: Suppose a4 + 2 is a multiple of 5. If a4 + 2 ≥ 10, then it follows that n ≥ 78 > 540. If
a4 + 2 = 5, then n is a multiple of 73 . If n = 73 , then 21n = 3 · 74 has exactly (1 + 1)(4 + 1) = 10
positive divisors. If n > 73 , then n ≥ 2 · 73 > 540. Therefore if a4 + 2 is a multiple of 5, it follows
that n is at least 540.

Case 1.5: Suppose ai + 1 is a multiple of 5 where i ≥ 5. This implies that n ≥ p45 ≥ 114 > 540.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Therefore if ai + 1 is a multiple of 5 where i ≥ 5, it follows that n is at least 540.

This establishes that either n ≥ 540 (which would complete the proof) or a2 = 3. Now consider
what follows if a2 = 3.

Step 2: Because 20n has exactly 60 positive divisors, because 60 is a multiple of 5, and because
5 is prime, conclude that at least one of a1 + 3, a2 + 1, a3 + 2, a4 + 1, and ai + 1 (i ≥ 5) is a
multiple of 5.

Case 2.1: Suppose a1 + 3 is a multiple of 5. If a1 + 3 ≥ 15, then it follows that n ≥ 212 > 540.
If a1 + 3 = 10, then a1 + 1 = 8 and the number of positive divisors of 21n is a multiple of 8 and
therefore cannot be 60, a contradiction. The case where a1 + 3 = 5 will be considered later.

Case 2.2: Suppose a2 + 1 is a multiple of 5. This case is impossible because a2 = 3.

Case 2.3: Suppose a3 + 2 is a multiple of 5. If a3 + 2 ≥ 10, then n ≥ 58 > 540. If a3 + 2 = 5,

then a3 = 3, and this implies that n ≥ 33 · 53 > 540. Therefore if a3 + 2 is a multiple of 5, it
follows that n is at least 540.

Cases 2.4 and 2.5: Suppose ai + 1 is a multiple of 5 where i ≥ 4. This implies that
n ≥ p44 ≥ 74 > 540. Therefore if ai + 1 is a multiple of 5 where i ≥ 4, it follows that n is at least
540.

The only case left to be considered is the case where a1 = 2 and a2 = 3. In this case, n = 22 ·33 ·m
where m is relatively prime to 2 and 3 (and therefore cannot be 2 or 3 or 4). If m = 1, then
n = 22 · 33 and 21n = 22 · 34 · 7 has exactly (2 + 1)(4 + 1)(1 + 1) = 30 positive divisors. If m ≥ 5,
then n ≥ 5 · 22 · 33 = 540.

Thus the least positive integer n such that each of 20n and 21n has exactly 60 positive di-
visors is 540.

T-9 The positive integers 1 and 18 each have the property that their square is one tenth of a triangular
number. Compute the least integer greater than 18 with this property.

Answer: 323

Solution: The goal is to find the least integer m > 18 such that there exists a positive integer
n such that 10m2 = n(n+1)2 which is equivalent to 80m2 = 4n2 + 4n + 1 − 1 = (2n + 1)2 − 1.
Letting y = 4m and x = 2n + 1, the problem reduces to finding positive integers x and y such
that x2 − 5y 2 = 1, where y is a multiple of 4 and x is odd. Because m = 1 and n = 4 satisfies
the conditions of the problem (as 10 is the fourth triangular number), this corresponds to the
solution y = 4 and x = 9. Note that there is no lesser solution to x2 − 5y 2 = 1 because y = 4
is the least positive multiple of 4 and if y = 0, then x = 1, which already corresponds to the
number 1 that was given in the problem statement.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Because 5 is not a perfect square, the following well-known property of Pell equations ap-
plies: if (p, q) is the least positive integer solution of the equation x2 − Dy 2 = 1, where D is
not a √ √ all positive integral solutions (pn , qn ) to this equation are given by
perfect square, then
n
(p + q D) = pn + qn D. Notice that
√ √ √
(9 + 4 5)2 = 81 + 72 5 + 80 = 161 + 72 5,
and this corresponds to x = 161 → n = 80 and y = 72 → m = 18.

Also note that

√ √ √ √ √
(9 + 4 5)3 = 93 + 3 · 92 · 4 5 + 3 · 9 · (4 5)2 + (4 5)3 = 2889 + 1292 5,
and this corresponds to x = 2889 → n = 1444 and y = 1292 → m = 323.

T-10 In unit square ARM L, circle Γ is drawn centered at M and passes through R and L, and circle Ω
is inscribed in the square. A circle with radius r is drawn tangent to Γ, Ω, and segment AR
(possibly at the endpoints). Compute the sum of all possible values of r.
√
9−4 2
Answer: 16

Solution: Consider Cartesian coordinates such that R lies at the origin with positive x-axis as
−→ −−→
RA and positive y-axis as RM . Then A = (1, 0), M = (0, 1), L = (1, 1), and O = ( 12 , 12 ), where
O is the center of the square. Consider a circle γ centered at P = (h, k) with radius r, and
suppose that this circle is tangent to the two given circles and segment AR.

Because the circle is tangent to segment AR, point P is a distance r from the x-axis. Also
note that P lies in the upper half plane (y > 0) and between the lines RM and AL (0 ≤ x ≤ 1).
Thus 0 ≤ h ≤ 1 and k = r.
Because circle γ is tangent to the circle Γ centered at M , it follows that the distance M P is
equal to either the sum or difference of the radii of Γ and γ. Thus

h2 + (k − 1)2 = |1 ± r| =⇒ h2 + (k − 1)2 = (1 ± r)2
=⇒ h2 = 4r h2 = 0.
k=r
or
Because circle γ is tangent to the circle Ω centered at O, it follows that the distance OP is equal
to either the sum or difference of the radii of Ω and γ. Thus
  
1 2 1 2  1  1 2 1 2 1 2
h− + k− =  ± r =⇒ h − + k− = ±r
2 2 2 2 2 2
1 2 k=r 1 2
=⇒
= 2r or h−h− = 0.
2 2
Now the problem can be finished by applying casework on the conditions.
2
1
• Case 1: h2 = 4r and h − 2 = 2r. Then
2
1
h2 = 4r = 2 h − ,
2
√ √ √
2 2 h2 3−2 2
which has solutions h = 1± 2 . Because h ≤ 1, this implies h = 1− 2 and r = 4 = 8 .

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 h2
• Case 2: h2 = 4r and (h − 12 )2 = 0. Then r = 4 = 1
16 .
(h− 12 )2
• Case 3: h2 = 0 and (h − 12 )2 = 2r. Then h = 0, so r = 2 = 18 .
• Case 4: h2 = 0 and (h − 12 )2 = 0. Then h = 0 and h = 1
2 which is a contradiction, so there
are no solutions here.
See the diagram below for the three possible circles.
M L

R A
√ √
3−2 2 1 1 9−4 2
The possible radii are 8 , 16 , and 8 for a sum of 16
.

T-11 The Springfield Math Team (Alex, Beth, Carol, David, Ed, and Fran) is taking a break from
ARML Local. Their coach has provided two identical cans of apple juice, two identical cans of
raspberry juice, two identical cans of mango juice, and two identical cans of lemonade. Each
student takes one or two cans, and no cans are left over. Compute the number of ways the cans
can be distributed among the students.

Answer: 12420

Solution: Two
 students will take two cans, and the other four students will take one can each.
There are 62 = 15 ways to choose these two students. Call them X and Y .

If X and Y both take two cans of the same kind of juice, there are 4 choices for 4 the two
cans that X takes and 3 choices for the two cans that Y takes, and then there are 2 ways to
choose which two students
 receive the two cans of one of the other kinds of juice. This results
in a total of 4 · 3 · 42 = 72 possible distributions of this type.

If exactly one of X and Y takes two cans of one kind of juice, then there are 2 ways to choose
which one of X and Y that is, and then there  are 4 ways to choose what kind of juice that
person gets in two cans, and then there are 32 ways to choose which two of the kinds of juice
the other of X and Y gets, and then there are 4 · 3 ways to distribute the3other two “singleton”
cans of juice among the other students. This results in a total of 2 · 4 · 2 · 4 · 3 = 288 possible
distributions of this type.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 The last case is the one in which neither X nor Y takes just one kind of juice. Condition 
on the number of “overlapping” flavors they take. If they take zero, then there are 42 ways
to choose which two kinds of juice X takes (recognizing that Y will take cans of the other two
kinds of juice), and then 4! ways to distribute the other four (all different) cans of juice among
the other four students. This results in a total of 42 · 4! = 144 possible distributions of this type.

If they take exactly one, then there are 4 ways to choose the identical juice can, 3 ways for
X to choose their other juice, 2 ways for Y to choose their other juice, and 4 · 3 ways to choose
how to distribute the two cans of the same type of juice that remain. This results in a total of
4 · 3 · 2 · 4 · 3 = 288 possible distributions of this type.

Finally, if they take exactly two, then there are 42 ways to choose which two types of juice
 4
are chosen by X and Y , and there are 2 ways to distribute one of the pairs of cans of juice
  
among the other four students. This results in a total of 42 42 = 36 possible distributions of
this type.

In total, there are 15(72 + 288 + 144 + 288 + 36) = 12420 ways in which the cans can be
distributed among the students.

T-12 Four spheres of radius 1 are externally tangent to one another, as shown below. Compute the
height of the right circular cone whose base is internally tangent to three of the spheres and
whose lateral surface is internally tangent to all four spheres.

√
2 6
√ √ √
3+2 6+3 3
Answer: 1 + 3 + 3 or 3

Solution: Consider the plane that contains the center O of the base of the cone, the center C1
of one of the spheres tangent to the base, and the center C2 of the sphere not tangent to the
base. Let C1 and C2 be the points in this plane where the circles centered at C1 and C2 are
tangent to the lateral surface of the cone, A be the apex of the cone, and R be the point where
this plane, the base of the cone, and the lateral surface intersect, as shown.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Then C2 C2 = C1 C1 = 1, and m∠C2 C2 C1 = m∠C1 C1 C2 = 90◦ , so it follows that C1 C2 C2 C1 is
a rectangle, and thus C1 C2 is parallel to AR. Now draw a segment through C1 parallel to OR,
intersecting AO at E. The centers of the four spheres form a regular tetrahedron of side length
2, whose
√ base has centroid E. Note that the centroid is on the height of the triangular base (of
length 3) and the centroid divides that height into two segments in a 2 : 1 ratio, so it follows
√ √ 2 √
2 3 2 3 2 6
that EC1 = 3 . The height of this tetrahedron is EC2 = 22 − 3 = 3 . Also note
that EO = 1 because EO has the same length as the height of C1 above the base of the cone.
Finally, AC2 C2 ∼ C2 EC1 , √
and these are right triangles whose side lengths √are in the ratio
√ √ √ 2√ 3 √ √ √
1 : 2 : 3 (note that EC2 = 3 = 2 · 3 = 2 · EC1 and C1 C2 = 2 = 3 · 2 3 3 = 3 · EC1 ).
2 6
√ √ √ √
Thus AC2 = 3 · C2 C2 = 3. The height of the cone is AO = 1 + 2 3 6 + 3.

Alternate Solution: Consider the plane that contains the center of the base of the cone, the
center C1 of one of the spheres tangent to the base, and the center C2 of the sphere not tangent
to the base. Let the center of the base of the cone O be the origin of a coordinate grid and let
C2 lie on the positive y-axis. One can determine the coordinates of C1 and C2 .
Noting that the centers of the spheres form a tetrahedron
√ of side length
√ 2, the length of the
2 3 2 3
segment from the centroid of the base E to C1 is 3 , so C1 = ( 3 , 1). The height of the
√
2 6
tetrahedron is 3 (as determined via the Pythagorean Theorem on the right triangle formed
√
by C1 , C2 , and the centroid of the base of the tetrahedron), so C2 = 0, 1 + 2 3 6 . Further,
←−→
if D = (d, 0) is the intersection of C1 C2 with the x-axis, then using similar triangles as in the
previous solution, it follows that

√ √√
2 3 d 2 2 3
d− = √ →d= + .
3 1+2 6 2 3
3

Let C2 be the point of tangency in the plane of the sphere centered at C2 with the lateral surface
of the cone and A = (0, h) be the apex of the cone. Then ΔAC2 C2 ∼ ΔDOC2 , and it follows

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 C2 A OC2
that C2 C2
=
OD . Thus
√ √ √ √ √ √ √ √
1 + 236 6+4 6 4 3−3 2 24 3 − 18 2 + 16 18 − 12 12 √
C2 A= √ √ = √ √· √ √ = = 2.
2 2 3 4 3+3 2 4 3−3 2 48 − 18
2 + 3
√ √ √
Then AC2 = 3, and the height of the cone is 1 + 2 3 6 + 3.

T-13 For some rational number k, the polynomial x3 + kx + 1 has three real, irrational roots. Given
that one of these roots is 2 less than the square of another of these roots, compute k.

Answer: −3

Solution: First, let f (x) = x3 + kx + 1, and let σ = τ 2 − 2 where τ is one of the roots of
f (x). Consider f (σ) = f (τ 2 − 2), substitute in, and repeatedly simplify, using the fact that
τ 3 = −kτ − 1 to reduce any power of τ greater than 2:
f (τ 2 − 2) = (τ 2 − 2)3 + k(τ 2 − 2) + 1
= τ 6 − 6τ 4 + 12τ 2 + kτ 2 − 2k − 7
= (−kτ − 1)2 − 6τ (−kτ − 1) + 12τ 2 + kτ 2 − 2k − 7
= k 2 τ 2 + 1 + 2kτ + 6kτ 2 + 6τ + 12τ 2 + kτ 2 − 2k − 7
= (k 2 + 7k + 12)τ 2 + (2k + 6)τ − (2k + 6).
The result is an expression of the form aτ 2 + bτ + c, for rational numbers a, b, c. Because σ is
a root of f , this expression must equal zero. If a, b, and c are not all zero, then τ is a root
of a polynomial with rational coefficients of degree less than 3, which must then divide f (x).
Either this polynomial or the quotient is a linear polynomial with rational coefficients, implying
that f has a rational root, a contradiction. It follows that a = b = c = 0. This means that
k 2 + 7k + 12 = 0 and 2k + 6 = 0, which implies k = −3.

T-14 Let x1 , x2 , y1 , y2 be real numbers strictly between − π2 and π

2 such that
cos2 (x1 ) + cos2 (x2 ) ≥ sin2 (x1 + x2 ) and cos2 (y1 ) + cos2 (y2 ) ≥ sin2 (y1 + y2 ).
Compute the minimum possible value of
cos(x1 + y1 ) + cos(x2 + y2 ) + cos(x1 + x2 + y1 + y2 ).

Answer: −1

Solution: Rewrite the given condition on x1 , x2 as

cos2 (x1 ) + cos2 (x2 ) + cos2 (x1 + x2 ) ≥ 1,
then apply the half-angle formula for cosine to deduce that
1 + cos(2x1 ) 1 + cos(2x2 ) 1 + cos(2x1 + 2x2 )
+ + ≥ 1,
2 2 2

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 which can be rewritten as
cos(2x1 ) + cos(2x2 ) + cos(2x1 + 2x2 ) ≥ −1.
Similarly, the given condition on y1 , y2 can be rewritten as
cos(2y1 ) + cos(2y2 ) + cos(2y1 + 2y2 ) ≥ −1.
Now, the key claim is that for −π < a, b < π,
cos(a) + cos(b) + cos(a + b) ≥ −1 ⇐⇒ −π ≤ a + b ≤ π.
Observe that
a+b a−b a+b
cos(a) + cos(b) + cos(a + b) + 1 = 2 cos cos + 2 cos2
2 2 2
 
a+b a+b a−b
= 2 cos · cos + cos
2 2 2
a+b a b
= 4 cos cos cos .
2 2 2
a
From −π < a, b < π, it follows that cos 2 cos b
2 > 0. Thus the inequality is equivalent to
cos a+b
2 ≥ 0, which is equivalent to − π2 ≤ a+b
2 ≤ 2 , as desired.
π

Thus the given conditions imply that −π ≤ 2x1 + 2x2 ≤ π and −π ≤ 2y1 + 2y2 ≤ π. Thus
−π ≤ x1 + y1 + x2 + y2 ≤ π
with −π < x1 + y1 , x2 + y2 < π, so it follows that
cos(x1 + y1 ) + cos(x2 + y2 ) + cos(x1 + x2 + y1 + y2 ) ≥ −1.
One can confirm that x1 = x2 = y1 = y2 = π
4 is valid and produces a value of −1, so the answer
is indeed −1.

T-15 Compute the sum of all positive integers n ≤ 50 such that n4 + n2 + 1 is divisible by a perfect
square greater than 1.

Answer: 143

Solution: Let f (n) = n4 + n2 + 1, and note that

f (n) = (n2 + 1)2 − n2 = (n2 + n + 1)(n2 − n + 1)
for positive integers n. Define g(n) = n2 + n + 1 and h(n) = n2 − n + 1 so that f (n) = g(n)h(n).
If f (n) is not square-free, then it must be divisible by p2 for some prime p > 2 (note that
p = 2 because n4 + n2 + 1 is always odd). If p divides both g(n) and h(n), then p divides
g(n) − h(n) = 2n. Because p is odd, it follows that p divides n, and hence p must divide 1,
which is a contradiction. Therefore p2 divides g(n) or p2 divides h(n).

Begin by examining the former case. First, note that if g(n) ≡ 0 (mod 32 ), then it can be
manually checked that there are no solutions, so assume p > 3.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Because p is odd, p2 divides g(n) if and only if p2 divides 4g(n), or (2n + 1)2 ≡ −3 (mod p2 ).
This means that −3 is a quadratic residue modulo p, and so by the Law of Quadratic Reciprocity,

p 3 3 p
= (−1)(p−1)/2 =⇒ = (−1)(p−1)/2 .
3 p p 3

−3 −1 p −1
However, = (−1)(p−1)/2 . Because = 1 if and only if p ≡ 1 (mod 4),
p p 3 p
and also (−1)(p−1)/2 = 1 if and only if p ≡ 1 (mod 4), it follows that their product is always 1.
−3 p
Therefore = , which is 1 if and only if p ≡ 1 (mod 3).
p 3
Now, note that g(n) = kp2 for some positive integer k. Because n2 < g(n) < (n + 1)2 , it
follows k = 1. Furthermore, g(n) is not even so k = 2, which implies k ≥ 3. Furthermore,
51 √ 17
g(n) < (n + 1)2 so 512 > kp2 ≥ 3p2 . Therefore, p < √ < 30, where the inequality 3 >
3 10
was used. Because p ≡ 1 (mod 6), it follows that p = 7, 13, or 19. Because −3 is a quadratic
residue mod p, there are exactly two solutions to the congruence x2 + 3 ≡ 0 (mod p).

When x2 + 3 ≡ 0 (mod 7), it follows that x = 2 or x = 5. To solve x2 + 3 ≡ 0 (mod 72 ), it

suffices to examine which integers 0 ≤ k < 7 satisfy (7k + 2)2 ≡ −3 (mod 72 ) or (7k + 5)2 ≡ −3
(mod 72 ).

The first congruence is equivalent to 7 being a divisor of 4k + 1, giving k ≡ 5 (mod 7). Taking
k = 5 gives 7k + 2 = 37.

The second congruence is equivalent to 7 being a divisor of 10k + 4, or 7 being a divisor of

3k + 4 which gives k ≡ 1 (mod 7). Taking k = 1 gives 7k + 5 = 12. Therefore, the solutions to
x2 + 3 ≡ 0 (mod 72 ) are x = 12 or x = 37.

When x2 + 3 ≡ 0 (mod 13), it follows that x = 6 or x = 7. Using a similar approach to

the previous case yields that the solutions to x2 + 3 ≡ 0 (mod 132 ) are x = 45 and x = 124.

When x2 + 3 ≡ 0 (mod 19), it follows that x = 4 or x = 15. Using a similar approach to

the first case yields that the solutions to x2 + 3 ≡ 0 (mod 192 ) are x = 137 or x = 224.

Recall that p2 divides g(n) if and only if (2n + 1)2 + 3 ≡ 0 (mod p2 ). When p = 7, it follows
that 2n + 1 ≡ 12 (mod 49) or 2n + 1 ≡ 37 (mod 49). These occur if n ≡ 30 (mod 49) or n ≡ 18
(mod 49) respectively, so the solutions when n ≤ 50 are n = 18 and n = 30. If p = 13, then
2n + 1 ≡ 45 (mod 169) (the second case, and similarly the p = 19 cases, cannot occur because
2n + 1 ≤ 101). This happens when n ≡ 22 (mod 169) so n = 22 is the only appropriate solution.

Therefore g(n) is divisible by a square of a prime when n = 18, 22, 30. (Note that because
n > 0, we consider h(1) = 1 as a special case, but this results in g(0) = 1, in which case g(0)
is not divisible by the square of a prime.) Because g(n) = h(n + 1), this means that h(n) is
divisible by a square of a prime when n = 19, 23, 31. Hence, f (n) will be divisible by a square
larger than 1 when n = 18, 19, 22, 23, 30, 31. The sum of these values of n is 143.

Alternate Solution: As in the previous solution, establish that p2 divides g(n) or p2 divides

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
h(n). Then create a prime factorization table for g(n) = n2 + n + 1 for 0 ≤ n ≤ 50. Such a table
is shown below.

Prime Prime Prime

n g(n) n g(n) n g(n)
Factorization Factorization Factorization

0 1 None 17 307 307 34 1191 3 · 397

1 3 3 18 343 7·7·7 35 1261 13 · 97
2 7 7 19 381 3 · 127 36 1333 31 · 43
3 13 13 20 421 421 37 1407 3 · 7 · 67
4 21 3·7 21 463 463 38 1483 1483
5 31 31 22 507 3 · 13 · 13 39 1561 7 · 223
6 43 43 23 553 7 · 79 40 1641 3 · 547
7 57 3 · 19 24 601 601 41 1723 1723
8 73 73 25 651 3 · 7 · 31 42 1807 13 · 139
9 91 7 · 13 26 703 19 · 37 43 1893 3 · 631
10 111 3 · 37 27 757 757 44 1981 7 · 283
11 133 7 · 19 28 813 3 · 271 45 2071 19 · 109
12 157 157 29 871 13 · 67 46 2163 3 · 7 · 103
13 183 3 · 61 30 931 7 · 7 · 19 47 2257 36 · 61
14 211 211 31 993 3 · 331 48 2353 13 · 181
15 241 241 32 1057 7 · 151 49 2451 3 · 19 · 43
16 273 3 · 7 · 13 33 1123 1123 50 2551 2551

As with the previous solution, the three bolded values in the table (n = 18, 22, 30) lead to the
six values of n that add to 143.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Collaborative Relay Round (4, 5, 6 minutes)
R1-A Consider the system of equations
   
x y
20 + = 103
2 2
   
x y
21 + = 109,
2 2

where x and y are positive integers. Compute the least possible value of x + y.

Answer: 17

Solution: Note that a − a = 0 when a is an integer and a − a = 1 when a is not an
integer. Subtract the first equation of the system from the second to obtain
      
x y y
+ − = 6. (∗)
2 2 2

Now consider the

 following
  two cases.
• Case 1: 2 − y2 = 0. Then y must be an even integer, and substituting into (∗) yields
y
x y y
2 = 6. Thus x = 12 or x = 13, and for both values of x, 2 = 2 = −17. This implies
that y = −34
  < 0, so there are no solutions in this case.
y y
• Case 2: 2 − 2 = 1. Then y must be an odd integer, and substituting into (∗) yields
x y y
2 = 5. Thus x = 10 or x = 11, and for both values of x, 2 = 3 and 2 = 4. The only
integer y that satisfies both of the last two equations is y = 7.
Thus the two positive integral solutions to the given system are (x, y) = (10, 7) or (x, y) = (11, 7),
and the least possible value of x + y is 10 + 7 = 17.

R1-B Let T be the number you will receive. A right triangle has legs of lengths a and b and hypotenuse
c. Given that a and b are integers such that a2 − b2 = T , compute the least possible value of c2 .

Answer: 145

Solution: Because a2 − b2 = (a − b)(a + b) and a and b are integers, it follows that a − b and
a + b have the same parity because they differ by an even number. Let K = 10T . Then either
K is odd and each of a − b and a + b is odd or K is a multiple of 4 and each of a − b and a + b is
even. Positive integer solutions to a2 − b2 = K can be found by solving the system of equations

a−b=d
K
a+b= ,
d
where either d is a positive odd divisor of K if K is odd or d is a positive even divisor of K
if K is even, provided that K/d is also even. Moreover, because a + b > a − b, it follows that

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 √ 1 
d< K. The general solution is (a, b) = K
2( d + d), 12 ( K
d − d) , and
 
1 1 K 2
K2 K2
a 2 + b2 = d2
+ 2K + d2 + d2
− 2K + d2 = + d2 .
4 2 d2
By experimenting with some values of K, note that a2 + b2 is minimized by making d and K
d as
close to each other as possible. This can be seen by noting that
2
K2 K
+ d2 = −d + 2K.
d2 d
17
With T = 10 , K = 17, and because 17 is prime, the only possible system of equations is
a−b=1
a + b = 17,
whose solution is (a, b) = (9, 8). Thus the desired value of c2 is 92 + 82 = 145.

R2-A Suppose that a1 , a2 , . . . , an is an increasing sequence of consecutive positive integers whose sum
is 2021. Compute the least possible value of a1 .

Answer: 20

Solution: The sum of n consecutive integers with first term a1 is n2 (2a1 + (n − 1)). Therefore
n is a factor of 2 · 2021 = 4042 = 2 · 43 · 47. Also, n cannot be greater than 2a1 + (n − 1)
because a1 is positive. To minimize a1 , make n as large as possible, in which case n = 47, and
so 2a1 + (n − 1) = 86 → a1 = 20.

R2-B Let T be the number you will receive. Suppose that ABC is a right triangle with hypotenuse
T and legs of length at least T2 . Compute the number of possible integer values for [ABC].

Answer: 14

Solution: Let x◦ be the measure of the smallest acute angle in ABC. Note that 0◦ < x◦ ≤ 45◦ .
◦ ◦ 2
Then it follows that [ABC] = T sin x 2·T cos x = T4 · sin(2x◦ ). Because sin(2x◦ ) ≤ 1, conclude
2
that [ABC] ≤ T4 . Because T sin x◦ ≥ T2 , it follows that sin x◦ ≥ 0.5 = sin(30◦ ), and so
√
3
x◦ ≥ 30◦ → 2x◦ ≥ 60◦ . Thus sin(2x◦ ) ≥ sin(60◦ ) = 2 because the sine function is increasing
√ 2
√
T2 3 3
on [0◦ , 90◦ ].
Therefore [ABC] ≥ · T
= 8 . With T = 20, the possible areas are between
√ 4 √ 2 √
50 3 and 100, inclusive. Because 1.73 < 3 < 1.74, it follows that 86 < 50 3 < 87, and so
there are 100 − 87 + 1 = 14 possible integral values for the area of ABC. Each of these is made
possible by varying the length of the smaller leg, and so there are indeed 14 possible integral
values for [ABC].

R3-A Let C, R, A, Z, Y be an increasing arithmetic sequence of positive numbers. Given that C, R,

Y
Y is an increasing geometric sequence, compute .
R

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Answer: 3

Solution: Let the common difference of the arithmetic sequence be d. Then Y = C + 4d and
Y R C + 4d C +d
R = C +d. Notice that = → = , and this implies C 2 +4Cd = C 2 +2Cd+d2 ,
R C C +d C
which implies d(d − 2C) = 0. Because the arithmetic sequence is increasing, d = 0 which implies
Y 9C
d = 2C. So the arithmetic sequence is C, 3C, 5C, 7C, 9C, and the value of is = 3.
R 3C

R3-B Let T be the number you will receive. In quadrilateral ABCD, suppose that AB = BC = CD =
1
T and cos ABC = cos BCD = − . Compute AD.
T

Answer: 5

Solution: By the Law of Cosines in ABC or BCD, it follows that AC 2 = BD2 = T 2 +

1
T 2 − 2 · T · T · − = 2T 2 + 2T .
T

Note that ABCD is an isosceles trapezoid because AB = CD and m∠ABC = m∠BCD, so

Ptolemy’s Theorem applies. Also note that AC = BD, so AC · BD = AC 2 = BD2 . Ptolemy’s
Theorem states that if a quadrilateral is cyclic (meaning that it can be inscribed in a circle,
as isosceles trapezoids can because their opposite angles are supplementary), then the product
of the lengths of one pair of opposite sides plus the product of the lengths of the other pair of
opposite sides is equal to the product of the diagonals. In this case, Ptolemy’s Theorem implies
that T 2 + T · AD = 2T 2 + 2T , which implies AD = T + 2. With T = 3, the answer is AD = 5.

RA-1 Clarence attempted to compute 2x2 for some real number x but accidentally computed (2x)2
instead and got an answer that was 77 larger than the correct answer. Compute the correct
value of 2x2 .

Answer: 77

Solution: Because (2x)2 − 2x2 = 2x2 , the correct value is 77.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
RA-2 Let T be the number you will receive. Suppose that

⎧
⎨ n/2 if n is even
f (n) = .
⎩ n + 1 if n is odd

Given that k is an odd number such that f (f (f (k))) = T , compute k.

Answer: 307

Solution: As long as T is odd, it came from f by dividing by 2. The prior step could not have
been adding 1 because then k would be twice that number and hence even, so inverting two
applications of f yields 4T . The odd value of k is therefore 4T − 1. Substituting, it follows that
the answer is 4 · 77 − 1 = 307.

RA-3 Let T be the number you will receive. Keisha thought she would create an interesting sequence
of numbers by letting an = an−12−2T + an−2 for n > 1, but was unfortunate enough to choose
initial values for a0 and a1 such that an ended up being the same for every nonnegative value
of n. Compute a2 .

Answer: 614

an −2T
Solution: Given that the sequence is constant, an = 2 + an , which implies an = 2T for
n ≥ 0. Substituting, it follows that the answer is 614.

RB-1 Compute the number of positive integers N such that when 2021 is divided by N , the remainder
is 8.

Answer: 6

Solution: If 2021 leaves a remainder of 8 when divided by N , then 2013 = 3·11·61 is a multiple
of N . There are 2 · 2 · 2 = 8 factors of 2013. Of these, two (1 and 3) are too small to leave a
remainder of 8. Thus the answer is 8 − 2 = 6.

RB-2 Let T be the number you will receive. Let p and q be the solutions of x2 + T x−2T = 0. Compute
p4 + q 4 .

Answer: 3312

Solution: Notice that p4 + q 4 = (p2 + q 2 )2 − 2(pq)2 = ((p + q)2 − 2(pq))2 − 2(pq)2 . By Vieta’s
Formulas, this is (T 2 − 2(−2T ))2 − 2(−2T )2 . With T = 6, the answer is (36 + 24)2 − 288 = 3312.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
RB-3 Let T be the number you will receive. Let a1 , a2 , a3 , . . . be the sequence of numbers 4, 44, 444,
. . . , such that each term after a1 has one more digit 4 than the term before it. Compute the
remainder when aT is divided by 9.

Answer: 0

Solution: Divide the first few terms by 9 to see that the first few remainders are 4, 8, 3, 7,
2, 6, 1, 5, 0, and so on. (One could also notice that aT is congruent to 4T (mod 9).) With
T = 3312 ≡ 0 (mod 9), the answer is 0.

R3-1 Compute the number of ordered pairs of positive integers (m, n) for which 20m + 21n = 2021.

Answer: 5

Solution: The problem requests the number of positive integer solutions to 20m + 21n =
2021 ↔ 21n − 21 = 2000 − 20m, which is equivalent to 21(n − 1) = 20(100 − m), and this is
100−m
equivalent to n−1
20 = 21 . Because 20 and 21 are relatively prime, it follows that n − 1 is a
multiple of 20 and 100 − m is a multiple of 21. Both fractions in the last equation must have
the same integer value; call that integer k. Then n = 20k + 1 and m = 100 − 21k, which will
solve the original equation.

Because n is a positive integer, it follows that 20k + 1 ≥ 1 and so k ≥ 0. Because m is a

positive integer, it follows that 100 − 21k ≥ 1 → k ≤ 33 5
7 = 4 7 . There are therefore 5 values of k
that satisfy 0 ≤ k ≤ 33 7 , namely 0, 1, 2, 3, and 4. Each of these corresponds to an ordered pair
of positive integers that satisfy 20m + 21n = 2021.

R3-2 Let T be the number you will receive. Given that 2x + 3y = 20 and T x + 4y = 21, compute
16x + 17y.

Answer: 102

Solution: To avoid finding the values of x and y (which involve fractions), instead notice that
with T = 5, 16x + 17y = 3(2x + 3y) + 2(5x + 4y). Therefore the desired value is 3(20) + 2(21) =
102. Note that the values of x and y are x = −17/7 and y = 58/7, and substituting these values
supports the answer.

R3-3 Let T be the number you will receive. The line 3x + 9y + T = 0 intersects the line 5x + 7y = T
at the point (a, b). Compute a − b.

Answer: 102

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Solution: Because the lines cross at (a, b), it follows that 3a + 9b + T = 0 and 5a + 7b − T = 0.
Subtract to obtain −2a + 2b + 2T = 0 → a − b = T . Substituting, a − b = 102.

R3-4 Let T be the number you will receive. The positive integers g1 , g2 , and g3 with g1 < g2 < g3 are
in geometric progression in this order. Given that g1 + g2 = 68 and g3 − g2 = T , compute g2 .

Answer: 51

Solution: Let g2 = rg1 and g3 = r2 g1 . Then the given equations are equivalent to rg1 + g1 =
68 → g1 (r + 1) = 68 and r2 g1 − rg1 = T → rg1 (r − 1) = T . Because the terms of the geometric
sequence are positive and increasing, it follows that r > 0 and g1 = 0, so divide to obtain
r(r − 1) T
= . Cross-multiply to obtain 68r2 − 68r = T r + T . The quadratic formula yields
r+1 √68
2 +408T +4624
r = T +68± T 136 . Substitute T = 102 to obtain r = −1/2 or r = 3. Because r > 0,
reject r = −1/2, so r = 3, which implies g1 = 17 and g2 = 17 · 3 = 51.

R3-5 Let T be the number you will receive. Compute the least possible value of
5x2 − 4xy − 8x + 2y 2 − 16y + T
over all real numbers x and y.

Answer: −29

Solution: Rewrite the given expression as

x2 − 8x + 16 + y 2 − 16y + 64 + 4x2 − 4xy + y 2 + T − 80,
which is equivalent to (x − 4)2 + (y − 8)2 + (2x − y)2 + T − 80. This can be minimized if each
parenthesized term is zero, so the minimum is obtained if x = 4 and y = 8. Substituting, the
minimum value is 51 − 80 = −29.

R3-6 Let T be the number you will receive. The solutions of x4 + 2x3 − 16x2 + T x − 12 = 0 are a, b,
c, and T + 33. Compute ab + ac + bc.

Answer: 8

Solution: Consider that abc + ab(T + 33) + ac(T + 33) + bc(T + 33) = −T by Vieta’s Formulas,
which imply that abc + (ab + ac + bc)(T + 33) = −T . Note that the product of the solutions
is abc(T + 33) = −12, and also note that T + 33 is not zero because T + 33 is a factor in a
nonzero product. Therefore abc = T−12
+33 . This implies that the desired quantity is ab + ac + bc =
−12
−T − T +33 29 − −12
4
. With T = −29, the answer is = 8.
T + 33 4

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Individual Round (10 minutes per pair)
20
I-1 When written as a decimal, is 0.952380. Compute the sum of the first 2021 digits after the
21
20
decimal point in the decimal expansion of .
21

Answer: 9099

Solution: The sum of the 6 digits of the repetend (952380) is 27. Because 2021 = 6 · 336 + 5,
the complete repetend appears 336 times, which gives 2016 digits. The next five digits have the
sum 9 + 5 + 2 + 3 + 8 = 27. The desired sum is 27 · 336 + 27 = 27 · 337 = 9099.

I-2 Regular pentagon SM ILE is fully contained in the interior of regular pentagon F ROWN . The
edges of F ROWN are three times the lengths of the edges of SM ILE. The area of the region
between the two pentagons is 3. Compute [SM ILE].

3
Answer: 8

Solution: The shapes are irrelevant except that they are similar. Because the side lengths
are in a 3 : 1 ratio, the larger shape has 9 times the area of the smaller one. Hence the area
of the region between the two pentagons is 8 times that of the smaller pentagon. Therefore
[SM ILE] = 38 .

I-3 In hexagon SQUARE, all sides have length 10, ∠Q and ∠R are right angles, and ∠S ∼ = ∠U√∼=
∼
∠A = ∠E. Given that the area of SQUARE in simplest form is uniquely expressible as p + q r,
where p and q are integers and r is a prime, compute p + q + r.

Answer: 202

Solution: Because the sum of the measures of the interior angles of a hexagon is (6 − 2) · 180◦ =
720◦ , each of the four congruent angles has measure (720◦ − 180◦ )/4 = 135◦ . Because the
measures of all interior angles of the hexagon are less than 180◦ , the hexagon is convex. Draw
in diagonals SU and EA. Then the area of SQU ARE is the sum of the areas of SQU and
ARE and quadrilateral SUAE.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 S E
10
√
Q 10 2 R

U A
Each of the triangles has area 12 ·10·10 = 50. Note that SQU and ARE are both 45◦ −45◦ −90◦
triangles because SQ = QU = 10 and ER = RA = 10, so m∠U SE = m∠SU A = m∠SEA =
m∠U AE = 135◦ − 45◦ = ◦
√ 90 . This implies√ that SUAE is a rectangle.
√ By √
the Pythagorean
2 2
Theorem, SU = EA =√ 10 + 10 = 10 2, so [SU AE] = 10 · 10 2 = 100 2. The area of
SQU ARE is 100 + 100 2, so the answer is 100 + 100 + 2 = 202.

I-4 A base-b palindrome is a base-b integer whose digits read the same forwards and backwards and
whose leftmost digit is nonzero. Compute the greatest base-10 palindrome less than 500 that is
also a palindrome when converted to base 2.

Answer: 313

Solution: Let N be the desired number. Because the base-2 representation of N must begin and
end with the digit 1, it follows that N is odd. Thus N < 400. Because the base-2 representation
of 399 has nine digits, first consider base-2 palindromes with nine digits. The least of these is
1000000012 = 25710 . Because 257 is not a base-10 palindrome, the candidate values of N are
303, 313, . . . , 393. Note that 303 = 1001011112 , which is not a base-2 palindrome. Adding ten
in base 2 is equivalent to adding 10102 . It follows that 313 = 1001011112 + 10102 = 1001110012 ,
which is a palindrome. Repeatedly adding 10102 in base 2 and keeping track of the carrying, it
can be confirmed that none of 323, 333, . . . , 393 is a base-2 palindrome, so the answer is 313.
Alternate Solution: Again, let N be the desired number. If 300 < N < 400, then N must
equal 25710 (= 1000000012 ) plus some combination of:
0100000102 = 12810 + 210 = 13010 ,
0010001002 = 6410 + 410 = 6810 ,
0001010002 = 3210 + 810 = 4010 ,
0000100002 = 1610 ,
and the base-10 representation of N must end in 3. Thus the only possible candidate values of
N are:
257 + 16 + 40, 257 + 16 + 130, and 257 + 16 + 40 + 130,
but the latter two are greater than 400, so the answer is 257 + 16 + 40 = 313.

I-5 For a positive integer n, let s(n) denote the sum of the digits of n. Compute the sum of all
positive integers n for which n = 37s(n).

Answer: 7992

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Solution: Suppose a positive integer n satisfies n = 37s(n), and let d be the number of digits
in n. If d ≤ 2 with n = A B, then n = 10A + B < 37A + 37B = 37s(n). If d ≥ 5, then
37s(n) ≤ 37 · 9d < 10d−1 ≤ n. (The claim that 37 · 9d < 10d−1 can be proven inductively. If
d = 5, then 37 · 9 · 5 = 1665 < 10000 = 105−1 . Then, assume that 37 · 9 · k < 10k−1 and consider
37 · 9 · (k + 1). It follows that 37 · 9 · (k + 1) = 37 · 9 · k + 37 · 9 < 10k−1 + 333 < 10k−1 + 90000 ≤ 10k ,
as needed.) If d = 4, then n = 37s(n) ≤ 37 · 36 = 1332. Then s(n) < 1 + 3 + 9 + 9, so
37s(n) < 37 · 22 = 814 < n. Thus there is a contradiction in all cases where d = 3, so d = 3.

Thus n can be written as n = A B C with

100A + 10B + C = 37(A + B + C).
This is equivalent to 7A = 3B + 4C, which implies B = 7A−4C 3 and C = 7A−3B
4 . Fix A to be a
constant and imagine moving along the line 7A = 3B + 4C in the (B, C)-coordinate plane from
any lattice point on the line to the next lattice point on the line. Such a movement requires a
translation of 4 units to the right and 3 down or a translation of 4 units to the left and 3 up due
to the slope of the line. In fact, if (b, c) is a solution to 7A = 3B + 4C, then it is the only solu-
tion on the line strictly between the points (b−4, c+3) and (b+4, c−3) (which are both solutions).

Thus it follows that 7A = 3B + 4C has the general solution

(A, B, C) = (A, A + 4t, A − 3t)
for integers t. Because |B − C| ≤ 9, and because both B and C are digits, and because
|B − C| = 7|t|, it follows that 7|t| ≤ 9 ↔ |t| ≤ 97 . Because t is an integer, it follows that
t ∈ {0, 1, −1}.
• If t = 0, there are nine solutions of the form A A A, where A can take on each of the values
1, 2, . . . , 9. These nine solutions sum to
111 + 222 + · · · + 999 = 111(1 + 2 + · · · + 9) = 111 · 45 = 4995.
• If t = 1, there are three solutions of the form A B C, where B = A + 4 and C = A − 3, for
A = 3, 4, 5. These three solutions sum to 370 + 481 + 592 = 1443.
• If t = −1, there are three solutions of the form A B C, where B = A − 4 and C = A + 3,
for A = 6, 5, 4. These solutions sum to 629 + 518 + 407 = 1554.
Thus the sum of all solutions is 4995 + 1443 + 1554 = 7992.

−→
I-6 Let ARM L be a unit square inscribed in a circle. A point T lies on ray AL so that T L = 7,
and line T M intersects the circle at a second point P . Compute P R.

1
Answer: 5

Solution: Refer to the diagram below (not drawn to scale).

T L A

P
M R
P

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 √
By the Pythagorean Theorem, T M = 5 2. √By Power of a Point, T M · T P = T L · T A, so
T P = 57·8
√ . Note that T P = 56
2
√ > 50
5 2
√ = 5 2, so it follows that T P > T M . Consequently,
5 2
56
√ 6
M P = T P − T M = 5√ 2 − 5 2 = 5√ 2 .

Now observe that T LM ∼ M P  P , where P  is the projection of P onto side M R. Because

√ 6
MP 3
= 5√2 = ,
TM 5 2 25
3 
and because T LM ∼ M P  P , it follows that MT PL = M P 
T M = 25 , which implies that M P =
3
7 · 25 = 21  4 1
25 . Thus RP = 25 , so by the Pythagorean Theorem, P R = 5 .

I-7 A parallelogram with a vertex at (0, 0) has its other three vertices as lattice points in the first
quadrant. Given that the parallelogram has area 2021, compute the maximum number of lattice
points in the interior of the parallelogram.

Answer: 2020

Solution: Pick’s Theorem states that the area of the parallelogram is equal to the number of
interior lattice points plus half the boundary lattice points less one. Let I be the number of
interior lattice points and B be the number of boundary lattice points. With the area of the
parallelogram being 2021, Pick’s Theorem implies that I + B2 − 1 = 2021. The four vertices
of the parallelogram are lattice points, so B ≥ 4. Consequently, I = 2022 − B2 ≤ 2020. This
number of interior points is achievable with the parallelogram that has vertices at (0, 0), (1, 1),
(2, 2023), and (1, 2022), as the interior points are (1, n) for n = 2, . . . , 2021. This demonstrates
that 2020 is an achievable value of I.

I-8 Compute tan2 20◦ + tan2 40◦ + tan2 80◦ .

Answer: 33

Solution: Observe that for x = 20◦ , 40◦ , 80◦ ,

 2
3 tan x − tan3 x
= tan2 (3x) = 3.
1 − 3 tan2 x
√ √ 2
3 t−t t
Letting tan2 x = t, this becomes 1−3t = 3, which is equivalent to
√
((3 − t)( t))2 = 3(1 − 3t)2 ↔ t3 − 6t2 + 9t = 27t2 − 18t + 3,

which is equivalent to t3 − 33t2 + 27t − 3 = 0. This implies that the cubic polynomial P (t) =
t3 −33t2 +27t−3 has roots tan2 20◦ , tan2 40◦ , tan2 80◦ . Note that tan2 20◦ , tan2 40◦ , and tan2 80◦
are all distinct and so these are the only roots of the cubic. By Vieta’s Formulas, the sum of
the roots of the polynomial is 33.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Alternate Solution: Observe that for x = 20◦ , 40◦ , 60◦ , 80◦ , the complex number
(cos x + i sin x)9 cos 9x + i sin 9x
(1 + i tan x)9 = =
cos9 x cos9 x
is real, where the last equality follows from DeMoivre’s Formula. Expanding using the Binomial
Theorem and taking the imaginary part yields
tan9 x − 36 tan7 x + 126 tan5 x − 84 tan3 x + 9 tan x = 0.
This implies that the quartic polynomial P (t) = t4 − 36t3 + 126t2 − 84t + 9 has roots tan2 20◦ ,
tan2 40◦ , tan2 60◦ , and tan2 80◦ . Note that each of these values is distinct, as in the previous
solution. By Vieta’s Formulas, the sum of the roots of the polynomial is 36. Because tan2 60◦ =
3, it follows that the desired sum is 36 − 3 = 33.

I-9 Jimmy wants to choose a four-digit PIN that has a digit sum of 6. Two such possible PINs are
0222 and 2301. Compute the number of different four-digit PINs that have a digit sum of 6.

Answer: 84

Solution: This solution uses the “sticks and stones” method of counting. (More informa-
tion is available at, among other places, https://www.rainbowresource.com/pdfs/products/
prod060689_smp02.pdf.) Imagine assigning numbers to the PIN by placing six stones into four
boxes, one representing each digit. For example, the PIN 2301 is equivalent to placing 2 stones
in the thousands box, 3 in the hundreds box, 0 in the tens box, and 1 in the ones box. There
are three dividing sticks between the four boxes, and placing the stones determines uniquely
where the stones must go. Therefore, considering how one might line up the six stones and
three sticks in a line, each possible arrangement is equivalent to a distinct four-digit PIN. This
is the same as the number of ways to place 3 sticks in the 6+3 = 9 slots. The answer is therefore
9 9·8·7
= = 3 · 4 · 7 = 84.
3 3·2·1

I-10 Define a sequence as a1 = x for some real number x and

an = nan−1 + (n − 1)(n!(n − 1)! − 1)
p
for integers n ≥ 2. Given that a2021 = (2021! + 1)2 + 2020!, and given that x = , where p and
q
q are positive integers whose greatest common divisor is 1, compute p + q.

Answer: 10106

Solution: Divide both sides of the given equation for an by n! to yield

an an−1 n−1 an−1 1 1
= + (n − 1)(n − 1)! − = + n! − (n − 1)! − + .
n! (n − 1)! n! (n − 1)! (n − 1)! n!
It follows that
an 1 an−1 1
− n! − = − (n − 1)! − ,
n! n! (n − 1)! (n − 1)!

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
so by induction,
an 1 1
− n! − = a1 − 1! − = a1 − 2.
n! n! 1!
With am = (m! + 1)2 + (m − 1)! when m = 2021, this equation implies that
1 1 1
m! + 2 + + − m! − = a1 − 2,
m! m m!
1
so a1 = 4 + m = 8085
2021 . The answer is 8085 + 2021 = 10106.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
 Tiebreaker (10 minutes)

TB Compute the remainder when 1112021 is divided by 10000.

Answer: 3311

Solution: Consider the trinomial expansion (a + b + c)2021 , where a = 100, b = 10, and c = 1.
Any term containing a2 , b4 , ab2 , or any higher power of a and b is congruent to 0 modulo 10000
and thus can be ignored. This leaves the ac2020 , abc2019 , b3 c2018 , b2 c2019 , bc2020 , and c2021 terms.
In general, the coefficient of ar bs ct in this expansion is equal to
2021!
,
r!s!t!
so the sum of the relevant terms is
2021ac2020 + 2021 · 2020abc2019 + 2021·2020·2019 3 2018
6 b c + 2021·2020 2 2019
2 b c + 2021bc2020 + c2021 .
Substituting in the values of a, b, and c, this is equal to
2021 · 100 + 2021 · 2020 · 1000 + 2021 · 1010 · 673 · 1000 + 1010 · 2021 · 100 + 2021 · 10 + 1.
Finally, taking this modulo 10000 gives
2100 + 0 + 0 + 1000 + 210 + 1 = 3311.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes.
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.