ARML Competition 2024

George Reuter, Head Writer

Chris Jeuell, Lead Editor
Evan Chen
Thinula De Silva
Paul Dreyer
Edward Early
Zuming Feng
Zachary Franco
Silas Johnson
Winston Luo
Jason Mutford
Andy Niedermaier
Graham Rosby
Tristan Shin
April Verser

May 31–June 1, 2024

Sponsored By:

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 Team Round

Problem 1. Compute the minimum number of people that must be gathered to guarantee that either at least two
of them have the same birthday or at least one of them was born in February.

Problem 2. Given that

log2 600 = A · log4 20 + B · log4 24 + C,
where A, B, and C are integers, compute A + B + C.

Problem 3. Let (a1 , a3 , a5 , a7 ) be a permutation of (1, 4, 9, 16), and let (a2 , a4 , a6 , a8 ) be a permutation of (1, 8, 27, 64).
Compute the greatest possible value of

|a1 − a2 | + |a2 − a3 | + |a3 − a4 | + |a4 − a5 | + |a5 − a6 | + |a6 − a7 | + |a7 − a8 | + |a8 − a1 |.

√
Problem 4. The least solution to the equation 8x3 − 6x = 3 can be expressed in the form cos(k ◦ ), where
0 < k < 180. Compute k.

Problem 5. Marcelo and Marta play a game in which a fair coin is flipped repeatedly. The coin either lands heads
(H) or tails (T). They play until three consecutive flips appear in the order HTH or HHH. If HTH appears first,
then Marcelo wins; otherwise, Marta wins. Compute the number of distinct sequences of coin flips that result
in Marcelo winning on the tenth flip.

←
→ ←→
Problem 6. Let IOWACT Y be a regular heptagon. The lines OI and T Y intersect at P . Compute

[IW Y ]2
.
[IP Y ] · [IW CY ]

Problem 7. Triangle ABC has AB = 13, AC = 14, and BC = 15, and point P is selected randomly and uniformly
on AC. Compute
√ the probability that the distance between the circumcenters of triangles ABP and BCP is
at least 5 2.

Problem 8. Compute the number of ordered pairs (a, b) of integers with 1 ≤ a ≤ b ≤ 42 such that ab and ba have
the same remainder when divided by 7.

Problem 9. Let a, b, and c be distinct complex numbers such that a + b + c ̸= 0. Compute the greatest possible
number of unordered pairs of different polynomials in the set
2
ax + bx + c, ax2 + cx + b, bx2 + cx + a, bx2 + ax + c, cx2 + ax + b, cx2 + bx + a

that have at least one common root.

Problem 10. Compute the greatest integer k such that 2k divides

ñÇ åÇ å Ç åÇ åô2
X 2024 2034 2024 2034
− .
i j j i
0≤i<j≤2024

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 1
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
2 Team Round Answers

Answer 1. 338

Answer 2. 2

Answer 3. 202

Answer 4. 130

Answer 5. 24

Answer 6. 1
37
Answer 7.
49
Answer 8. 204

Answer 9. 7

Answer 10. 8

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.
3 Team Round Solutions

Problem 1. Compute the minimum number of people that must be gathered to guarantee that either at least two
of them have the same birthday or at least one of them was born in February.

Solution 1. There are 365 − 28 = 337 possible birthdays not in February (regardless of whether or not the year
is a leap year), so at most 337 people can have distinct non-February birthdays. The answer is therefore 338.

Problem 2. Given that

log2 600 = A · log4 20 + B · log4 24 + C,
where A, B, and C are integers, compute A + B + C.

Solution 2. Note that

log4 600
log2 600 = = 2 log4 600
log4 2
9600
= 2 log4 202 + log4 24 − log4 16

= 2 log4
16
= 4 log4 20 + 2 log4 24 − 4.
The answer is therefore 4 + 2 + (−4) = 2.
2
Alternate Solution: Note that 4log2 600 = 2log2 600 = 6002 . Raise 4 to the power of both the left- and
right-hand sides of the given equation to yield
6002 = 26 · 32 · 54 = 4A·log4 20+B·log4 24+C
= 20A · 24B · 22C
= (22 · 5)A · (23 · 3)B · 22C
= 22A+3B+2C · 3B · 5A .
Matching the exponents of 3 and 5, it follows that B = 2 and A = 4, and hence 6 = 2A + 3B + 2C implies
C = −4. Thus A + B + C = 2, as above.
Remark: In the second solution, the Fundamental Theorem of Arithmetic (stating that prime factorizations
are unique) also implies there is only one triple (A, B, C) of integers that satisfies the given equation.

Problem 3. Let (a1 , a3 , a5 , a7 ) be a permutation of (1, 4, 9, 16), and let (a2 , a4 , a6 , a8 ) be a permutation of (1, 8, 27, 64).
Compute the greatest possible value of

|a1 − a2 | + |a2 − a3 | + |a3 − a4 | + |a4 − a5 | + |a5 − a6 | + |a6 − a7 | + |a7 − a8 | + |a8 − a1 |.

Solution 3. Let S denote the given sum. The optimal sum is S = 202, achieved by
(a1 , a2 , a3 , a4 , a5 , a6 , a7 , a8 ) = (1, 64, 4, 27, 9, 1, 16, 8)
=⇒ S = 63 + 60 + 23 + 18 + 8 + 15 + 8 + 7 = 202.
To show this is optimal, construct a diagram as follows. Draw a number line, then make marks at 1, 4, 9, 16
in blue and 1, 8, 27, 64 in red. Then, draw an arc between ai and ai+1 for i = 1, . . . , 7 as well as between a8
and a1 , joining two points of opposite colors for all eight arcs. By construction, these arcs form a cycle passing
through every mark exactly once. The sum S is equal to the sum of the lengths of the arcs (where the length
of the arc joining u and v is defined as |v − u|). An example showing the preceding optimal sequence is given
below; for legibility, the two marks at 1 are placed slightly apart from one another.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 3
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
I1 I2 I3 I4 I5 I6 I7
1 1 4 89 16 27 64

The eight marked points divide the interval [1, 64] into seven subintervals. Number the intervals I1 , I2 , . . . , I7
from left to right. (For convenience, I1 = [1, 1] is permitted as an interval of length zero.) For 1 ≤ k ≤ 7, let
ℓk be the length of Ik (in particular, ℓ1 = 0, and ℓk does not depend on the choice of the arcs drawn). Let nk
denote the number of arcs that contain the interval Ik . Then S is also equal to

S = n1 ℓ1 + n2 ℓ2 + · · · + n7 ℓ7 .

However, note that n7 ≤ 2, because an arc containing I7 must use 64 as an endpoint, and exactly two arcs
have 64 as an endpoint. Furthermore, n6 ≤ 4 because an arc containing I6 must use 27 or 64 as an endpoint,
and exactly four arcs do. By a similar argument, n1 ≤ 2 and n2 ≤ 4. The same logic implies that n5 ≤ 6
and n3 ≤ 6. The same logic would imply that n4 ≤ 8. However, in fact, n4 ≤ 6. To see why, assume for
contradiction that n4 = 8. This means that the blue 1 and blue 4 must be connected to only the red 27 and
red 64 (and vice-versa). By a similar argument, the red 1 and red 8 must be only connected to the blue 9 and
blue 16. However, this would violate the condition that the eight arcs must form a single contiguous loop, and
there are now (at least) two connected components. And because the eight arcs form a single loop, nk is even
for all k. Therefore it follows that n4 ≤ 6.

To finish, it follows from the above analysis that

S ≤ 2ℓ1 + 4ℓ2 + 6ℓ3 + 6ℓ4 + 6ℓ5 + 4ℓ6 + 2ℓ7

= 2 · 0 + 4 · 3 + 6 · 4 + 6 · 1 + 6 · 7 + 4 · 11 + 2 · 37 = 202.

√
Problem 4. The least solution to the equation 8x3 − 6x = 3 can be expressed in the form cos(k ◦ ), where
0 < k < 180. Compute k.

√
Solution 4. The presence of 3 and the condition that the desired solution is of the form cos(k ◦ ) √suggest that a
trigonometric substitution may be useful. Dividing the given equation by 2 yields 4x3 − 3x = 23 = cos(30◦ ).
√
Letting x = cos(t), note that 4x3 − 3x = 4 cos3 (t) − 3 cos(t) = cos(3t). The solutions to cos(3t) = 23 are
◦ ◦
thus of the form t = ± 303 + 360n 3 = 10◦ + 120n◦ , where n is an integer. The first few positive values of t are
10◦ , 110◦ , 130◦ , 230◦ , 250◦ . Observe that cos(10◦ ), cos(110◦ ), and cos(130◦ ) have distinct values (note that
cos(110◦ ) = cos(250◦ ) and cos(130◦ ) = cos(230◦ )), and because the cosine function is negative and decreases
on the interval (90◦ , 180◦ ], it follows that the desired value of k is 130.

Problem 5. Marcelo and Marta play a game in which a fair coin is flipped repeatedly. The coin either lands heads
(H) or tails (T). They play until three consecutive flips appear in the order HTH or HHH. If HTH appears first,
then Marcelo wins; otherwise, Marta wins. Compute the number of distinct sequences of coin flips that result
in Marcelo winning on the tenth flip.

Solution 5. Let Wn be the number of flip sequences in which Marcelo wins on the nth flip. Observe that

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.
ˆ W1 = W2 = 0 because the game requires at least 3 flips to win,
ˆ W3 = 1 if the sequence is HTH,
ˆ W4 = 2 from HHTH or THTH.
Now suppose n ≥ 5 and consider a “Marcelo-winning” sequence of n flips. Depending on whether this winning
sequence begins with 0, 1, or 2 heads, it must start with T, HTT, or HHTT, respectively, and then be followed
by a winning sequence of length 4, 2, or 1, respectively. This yields the recursion

Wn = Wn−1 + Wn−3 + Wn−4 .

The next few terms are therefore W5 = 2, W6 = 3, W7 = 6, W8 = 10, W9 = 15, and W10 = 24.

←
→ ←→
Problem 6. Let IOWACT Y be a regular heptagon. The lines OI and T Y intersect at P . Compute

[IW Y ]2
.
[IP Y ] · [IW CY ]

Solution 6. Let diagonals W Y and IC meet at Q. First, note that m∠QIY = m∠CIY = 2π 7 and m∠P IY =
π − m∠OIY = π − 5π = 2π
. Hence ∠QIY ∼
= ∠P IY . Similarly, ∠QY I ∼
= ∠P Y I, so △IQY ∼
= △IP Y by
7 7
Angle-Side-Angle, and thus [IQY ] = [IP Y ].

W
O

A b

a I
Q
ϕ
b
C a
P
Y
T

Let a = QI = QY and b = QW = QC, and let φ be the acute angle between lines W Y and IC. Then
1 2 1 1 1 2
[IQY ] = [IP Y ] = a sin φ, [QIW ] = [QY C] = ab sin(π − φ) = ab sin φ, [W QC] = b sin φ.
2 2 2 2
Substitute these into the desired fraction and cancel the recurring factor of ( 21 sin φ)2 to obtain
2
[IW Y ]2 [IQY ] + [QIW ]
=
[IP Y ] · [IW CY ] [IQY ] · [IQY ] + [QIW ] + [W QC] + [QY C]
(a2 + ab)2
= = 1.
a2 · (a2+ ab + b2 + ab)

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 5
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Remark: The hypothesis that IOWACT Y is regular is not used significantly. If IOWACT Y is any convex
heptagon that is symmetric around the perpendicular bisector of IY , then repeating the above argument gives

[IW Y ]2 [IW Y ]2 [IQY ] [IQY ] tan(∠CIY )

= · = = .
[IP Y ] · [IW CY ] [IQY ] · [IW CY ] [IP Y ] [IP Y ] tan(π − ∠OIY )
| {z }
=1

In particular, if m∠CIY = π − m∠OIY (as in the original problem), the resulting answer is 1.

Problem 7. Triangle ABC has AB = 13, AC = 14, and BC = 15, and point P is selected randomly and uniformly
on AC. Compute
√ the probability that the distance between the circumcenters of triangles ABP and BCP is
at least 5 2.

Solution 7. Let X and Y denote the respective circumcenters of triangles ABP and BCP . Observe that tri-
angles AXB and CY B are isosceles. In fact, these triangles are similar: vertex angles ∠AXB and ∠CY B
are each equal to twice the smaller angle formed by lines AP and BP . Hence AB/XB = CB/Y B and
∠ABX ∼ = ∠CBY . Because m∠ABX + m∠XBC = m∠ABC and m∠XBC + m∠CBY = m∠XBY , it follows
that ∠ABC ∼ = ∠XBY , and by Side-Angle-Side, it follows that △ABC ∼ △XBY .
Furthermore, because BP is the common chord of the circumcircles of triangles ABP and BCP , it follows
←→
that XY is the perpendicular bisector of BP . Let N be the intersection of BP and XY ; note that BN is an
altitude of △BXY .

N
X

A D P C

Let D be the foot of the altitude from B to AC. Then BD = 12, and because △ABC ∼ △XBY , it follows
that
XY AC XY · BD 12
= =⇒ BP = 2 · BN = 2 · = · XY.
BN BD AC 7
√ √ √
Thus XY ≥ 5 2 if and only if BP ≥ 607 2 . By the Pythagorean Theorem, DP = BP 2 − BD2 , so the desired
condition becomes
Ç √ å2
Ã
60 2 12
DP ≥ − 122 = .
7 7
12
Hence the desired condition holds if and only if P has distance at least 7 from D along the side AC. This
gives the answer
2 · 12
7 37
1− = .
AC 49

6 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: Let M , Q, and R be the respective midpoints of AC, AP , and P C, and let O be the
circumcenter of △ABC. Then QR = 7, and selecting P randomly on AC is equivalent to fixing segment QR
and choosing a point on QR randomly and uniformly to be M , then setting O to be a fixed distance above
point M (orthogonally to AC). Let QM = x. Then x is selected randomly and uniformly from the interval
[0, 7], and RM = 7 − x. As in the first solution, let X and Y be the respective circumcenters of △ABP and
△BCP . Because X and Y lie on the respective perpendicular bisectors of AB and BC, both of which also
pass through O, it follows that ∠OXQ ∼ = ∠BAC and ∠OY R ∼ = ∠BCA. Then XY is the hypotenuse of a right
triangle with one leg of length 7 and the other leg of length
x 7−x 5 3 7 21
|XQ − Y R| = − = x − (7 − x) = x − .
tan A tan C 12 4 6 4

Y
∠C

X
∠A

A Q M P R C
√ » √
In order for XY ≥ 5 2, it must be the case that 67 x − 21 4 ≥ 5 2 )2 − 72 = 1. This occurs when either
7 25 7 17 75 51
6 x ≥ 4 or 6 x ≤ 4 , which implies x ≥ 14 or x ≤ 14 . The probability of this occurring is
Å ã
75/14 51/14 37
1− + = .
7 7 49

Problem 8. Compute the number of ordered pairs (a, b) of integers with 1 ≤ a ≤ b ≤ 42 such that ab and ba have
the same remainder when divided by 7.

Solution 8. In general, for integers k and e, the value of k e modulo 7 depends only on k mod 7 and e mod 6 (by
Fermat’s Little Theorem). One may calculate all 7 · 6 = 42 values explicitly, and the values are shown in the
following table.
k mod 7, e mod 6 e ≡ 0 e ≡ 1 e ≡ 2 e ≡ 3 e ≡ 4 e ≡ 5
k≡0 0 0 0 0 0 0
k≡1 1 1 1 1 1 1
k≡2 1 2 4 1 2 4
k≡3 1 3 2 6 4 5
k≡4 1 4 2 1 4 2
k≡5 1 5 4 6 2 3
k≡6 1 6 1 6 1 6
For each residue 0 ≤ r < 7, let n(r) be the number of times that r appears in the above table; hence

n(0) = 6, n(1) = 15, n(2) = n(4) = 6, n(3) = n(5) = 2, n(6) = 5.

Consider the given problem, and temporarily ignore the condition a ≤ b, so the only constraint is that a, b ∈
{1, 2, . . . , 42}2 . Because 42 = 6 · 7, by the Chinese Remainder Theorem, over the 422 = 1764 possible pairs

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 7
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
(a, b), the quadruples Q(a, b) defined by

Q(a, b) = (a mod 7, b mod 6, b mod 7, a mod 6)

achieve every possible output exactly once. Define an ordered pair (a, b) of integers with 1 ≤ a, b ≤ 42 to be
good if ab and ba have the same remainder when divided by 7. The number of quadruples Q(a, b) such that
(a, b) is good equals
(n(0))2 + (n(1))2 + (n(2))2 + · · · + (n(6))2 = 366
because for each residue 0 ≤ r < 7, there are n(r) ways to choose the first two entries Q(a, b) so that
ab ≡ r (mod 7), and n(r) ways to choose the last entries Q(a, b) so that ba ≡ r (mod 7). Hence without a ≤ b,
there are 366 good pairs. Because the case where a = b always results in ab and ba having the same remain-
der when divided by 7, there are 366 − 42 = 324 good pairs with a ̸= b. Because (a, b) is good if and only
if (b, a) is good, there are 324
2 = 162 good pairs with a < b. Thus there are 162+42 = 204 good pairs with a ≤ b.

Alternate Solution: Let fc,d denote the number of solutions where a ≡ c (mod 7) and b ≡ d (mod 7). Notice
that

ˆ f0,0 = 36 because any pair (a, b) ≡ (0, 0) (mod 7) works,

ˆ fc,d = fc mod 7,d mod 7 ,
ˆ fc,d = fd,c ,
ˆ fc,0 = 0 for c ̸≡ 0 (mod 7).

It suffices to compute fc,d for 1 ≤ c ≤ d ≤ 6. For an integer x, let Cycle(x) denote the set of all y ∈ {1, . . . , 6}
for which there exists a nonnegative integer k such that xk ≡ y (mod 7), and let ord(x) denote the order of x
modulo 7, equivalently the cardinality of Cycle(x). The cycles and orders are as shown in the following table.

x Cycle(x) ord(x)
1 {1} 1
2 {1, 2, 4} 3
3 {1, 3, 2, 6, 4, 5} 6
4 {1, 4, 2} 3
5 {1, 5, 4, 6, 2, 3} 6
6 {1, 6} 2

The key claim is that

6 6
fc,d = Cycle(c) ∩ Cycle(d) · · (∗)
ord(c) ord(d)
for 1 ≤ c, d ≤ 6. To see why, take an element y ∈ Cycle(c) ∩ Cycle(d). Then ab ≡ ba ≡ y (mod 7) is equivalent
to b being fixed modulo ord(c) and a being fixed modulo ord(d). By the Chinese Remainder Theorem, this is
42 42
equivalent to b being fixed modulo 7ord(c) and a being fixed modulo 7ord(d). Thus there are 7ord(c) · 7ord(d)
solutions to (a, b) ≡ (c, d) (mod 7) and ab ≡ ba ≡ y (mod 7), as desired. This immediately implies that if
36
Cycle(c) ⊇ Cycle(d), then fc,d = ord(c) , which resolves all fc,d for 1 ≤ c ≤ d ≤ 6 except f2,6 and f4,6 . By (∗),
f2,6 = f4,6 = 6. The following table summarizes the computations.

fc,d 0 1 2 3 4 5 6
0 36 0 0 0 0 0 0
1 0 36 12 6 12 6 18
2 0 12 12 6 12 6 6
3 0 6 6 6 6 6 6
4 0 12 12 6 12 6 6
5 0 6 6 6 6 6 6
6 0 18 6 6 6 6 18

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.
Summing up all the entries of this table gives 366 total ordered pairs satisfying 1 ≤ a, b ≤ 42. Then one finishes
as in the first solution.

Problem 9. Let a, b, and c be distinct complex numbers such that a + b + c ̸= 0. Compute the greatest possible
number of unordered pairs of different polynomials in the set
2
ax + bx + c, ax2 + cx + b, bx2 + cx + a, bx2 + ax + c, cx2 + ax + b, cx2 + bx + a
that have at least one common root.

6

Solution 9. First, note that 1 is never a root of any of the polynomials because a + b + c ̸= 0. Classify the 2 = 15
pairs of polynomials into four types, as follows:
Common leading coefficient: e.g., ax2 + bx + c = 0 and ax2 + cx + b = 0 for some x. Then x = 1, which is
impossible, as noted above. Thus the 3 pairs of this type never have common roots.
Common middle coefficient: e.g., bx2 + ax + c = 0 and cx2 + ax + b = 0 for some x. Then x2 = 1, so
x = −1. Thus the 3 pairs of this type arise precisely when b + c = a, c + a = b, or a + b = c.
Common constant coefficient: e.g., bx2 + cx + a = 0 and cx2 + bx + a = 0 for some x. Then x2 = x, so
x = 0. Thus the 3 pairs of this type arise precisely when a = 0, b = 0, or c = 0.
Cyclic shift: e.g., ax2 + bx + c = 0 and bx2 + cx + a = 0 for some x. There are 6 pairs of this type. (The
classification of when these pairs arise is needed, and hence is deferred to the remark at the end.)
The following diagram summarizes the previous list: the six polynomials are represented by the six points, and
two polynomials that could share a common root are joined by an edge. Moreover, in the second and third
cases, the edge is labeled by the condition that must be satisfied in order for those two polynomials to share a
common root.

ax2 + bx + c bx2 + cx + a cx2 + ax + b

c+
a
c=0

b=0
=

a
b

+
a=0

b=

= a
c

b +c

bx2 + ax + c cx2 + bx + a ax2 + cx + b

The key claim is that at most one pair among the second and third types has a common root.
ˆ If two pairs of the second type both have common roots, then c + a = b and a + b = c or some symmetric
variant. However, this would imply that two of the coefficients are equal, which violates the given condition
that a, b, and c are distinct.
ˆ If two pairs of the third type both have common roots, then two of a, b, c are equal to 0, which again
violates the given condition that a, b, and c are distinct.
ˆ If a pair of the second type has a common root and a pair of the third type has a common root, then either
two of a, b, c are equal or two sum to 0 while the other is equal to 0. The former violates the distinctness
condition, while the latter implies that a + b + c = 0, which violates the condition that a + b + c ̸= 0.
Thus the claim is true. Because no pairs of the first type have a common root, there are at most 0 + 1 + 6 = 7
pairs with common roots. This set of pairs can be achieved by taking (a, b, c) = (1, −ω, 0), where ω = e2πi/3 ,
as shown in the following table.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 9
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Polynomial Factorization Root 1 Root 2
ax2 + bx + c (x2 − ωx) x(x − ω) ω 0
ax2 + cx + b (x2 − ω) (x − ω 2 )(x + ω 2 ) ω2 −ω 2
bx2 + cx + a (−ωx2 + 1) −ω · (x − ω)(x + ω) ω −ω
bx2 + ax + c (−ωx2 + x) −ω · x(x − ω 2 ) ω2 0
cx2 + ax + b (x − ω) x−ω ω
cx2 + bx + a (−ωx + 1) −ω(x − ω 2 ) ω2

Three pairs of polynomials have common root ω (which are cyclic shifts as discussed above), three pairs have
common root ω 2 , and one pair has common root 0. Thus the answer is 3 + 3 + 1 = 7.

Remark: If ax2 + bx + c and bx2 + cx + a have a common root x ̸= 1, then that root satisfies

0 = x(ax2 + bx + c) − (bx2 + cx + a) = a(x3 − 1).

Consequently, unless a = 0, it follows that x ∈ {ω, ω 2 }. This sheds light on the appearance of cube roots of
unity in the above table.

Problem 10. Compute the greatest integer k such that 2k divides

ñÇ åÇ å Ç åÇ åô2
X 2024 2034 2024 2034
− .
i j j i
0≤i<j≤2024

Solution 10. Let n = 2024 and m = 2034, and set ai,j = ni m n m 2 2

j − j i (in particular, ai,j = aj,i and ai,i = 0).
Let E denote the expression in the problem statement and note that
n X
n ñÇ åÇ å Ç åÇ åô2
X X X n m n m
2E = a2i,j + a2j,i = a2i,j = − .
i=0 j=0
i j j i
0≤i<j≤n 0≤i,j≤n

Then
n
"Ç å2 Ç å2 Ç å2 Ç å2
n Ç åÇ åÇ åÇ å#
1 XX n m n m n m n m
E= + −2
2 i=0 j=0 i j j i i i j j
n Ç å2 X n Ç å2
" n Ç åÇ å#2
X n m X n m
= −
i=0
i j=0 j i=0
i i
 Ç å2  " n Ç åÇ å#2
n Ç å2 m Ç å2 m
X n X m X m X n m
= · − −
i=0
i j=0
j j=n+1
j i=0
n − i i
Ç å Ç å m−n−1 Ç å2  Ç å2
2n 2m X m  n+m
= ·  − −
n m j=0
j n

by Vandermonde’s Identity. With n = 2024 and m = 2034, this evaluates to

Ç å ñÇ å ô Ç å2
4048 4068 4058
· −S − ,
2024 2034 2024

where Ç å2 Ç å2 Ç å2 Ç å2
2034 2034 2034 2034
S= + + + ··· + .
0 1 2 9

10 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 remains to compute ν2 of the terms, where ν2 (x) is the exponent of
Ä 2 in the prime factorization of x. Let s2 (x)
N
ä
be the sum of the binary digits of x. From Kummer’s Theorem, ν2 K = s2 (K) + s2 (N − K) − s2 (K), which
equals s2 (K) if N = 2K. Because 2024 = 111111010002 , 2034 = 111111100102 , and 4058 = 1111110110102 ,
it follows that ν2 4048 = 7, ν2 4068 = 8, and ν2 4058

2024 2034 2024 = (7 + 8) − 10 = 5. Finally, in the sum S, by
Lucas’s Theorem or manual inspection, the only odd binomial coefficients are 2034 and 2034

0 2 . It follows that
S ≡ 2 (mod 4), implying that ν2 (S) = 1. Therefore
ÇÇ å Ç åå ÇÇ å å Ç å2 !!
4048 4068 4048 4058
ν2 (E) = min ν2 · , ν2 · S , ν2
2024 2034 2024 2024

= min (7 + 8), (7 + 1), (2 · 5) = 8.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 11
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
4 Power Round 2024: The Rainbow Connection
Instructions: The power question is worth 50 points; each part’s point value is given in brackets next to the part.
To receive full credit, the presentation must be legible, orderly, clear, and concise. If a problem says “list” or “com-
pute,” you need not justify your answer. If a problem says “determine,” “find,” or “show,” then you must show
your work or explain your reasoning to receive full credit, although such explanations do not have to be lengthy. If a
problem says “justify” or “prove,” then you must prove your answer rigorously. Even if not proved, earlier numbered
items may be used in solutions to later numbered items, but not vice versa. Pages submitted for credit should be
NUMBERED IN CONSECUTIVE ORDER AT THE TOP OF EACH PAGE in what your team considers to be
proper sequential order. PLEASE WRITE ON ONLY ONE SIDE OF THE ANSWER PAPERS. Put the TEAM
NUMBER (not the team name) on the cover sheet used as the first page of the papers submitted. Do not identify
the team in any other way.

Note: Several problems below request that you “construct an example.” For these problems, you must provide
an example that satisfies the given conditions. You do not need to justify your responses.

In this Power Question, integers will be colored with one of several colors. Given a coloring of a set of integers, a
rainbow arithmetic progression (abbreviated rainbow AP ) is a subset of those integers whose elements are all different
colors and form an arithmetic progression. The length of a rainbow AP is the number of elements it contains. For
example, consider the following coloring of {1, 2, 3, 4, 5, 6, 7} with red (R), green (G), and blue (B ):

1 2 3 4 5 6 7
R G B R G B R

Then {2, 4, 6} is an example of a rainbow AP of length 3, because the colors that appear (green, red, and blue) are
all different. Note that the order does not matter; for example, {2, 4, 6} and {6, 2, 4} represent the same rainbow
AP. Also note that {3, 7} is an example of a rainbow AP of length 2.

1. Refer to the example from the table above.

a. Compute the number of rainbow APs of length 2. [1 pt]

b. Compute the number of rainbow APs of length 3. [1 pt]

For each positive integer n, denote by [n] the set {1, 2, . . . , n}. When [n] is colored with many different colors, there
tend to be many rainbow APs.

2. a. Construct an example of a coloring of [5] with red, green, or blue, each color appearing at least once, such
that there is no rainbow AP of length 3. [2 pts]
b. Compute the number of such colorings. [2 pts]
3. Suppose the elements of [7] are randomly colored with red, green, blue, or yellow (Y ). The colors are chosen
independently, and it is not necessary for all the colors to appear.

a. Compute the expected value of the number of rainbow APs of length 2. [1 pt]
b. Compute the expected value of the number of rainbow APs of length 3. [1 pt]
c. Suppose instead that there are five colors: purple (P ) is added as a fifth possible color. Compute the
expected value of the number of rainbow APs of length 3. [1 pt]

4. Show that any coloring of [5] with four colors, each color appearing at least once, must contain a rainbow AP
of length 3. [3 pts]
5. Construct an example of a coloring of [10] with four colors, red, green, blue, and yellow, each color appearing
at least once, such that there is no rainbow AP of length 3. [2 pts]

12 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.
6. Construct an example of a coloring of [20] with five colors, red, green, blue, yellow, and purple, each color used
exactly four times, such that there is no rainbow AP of length 5. [2 pts]
For a positive integer r, define an r-coloring on a subset S of the integers to be a coloring of the elements of S that
uses exactly r distinct colors, each color being used at least once.

7. a. Find a 3-coloring of the set of positive integers such that there is no rainbow AP of length 3. [2 pts]
b. Is there a positive integer r such that every r-coloring of the set of positive integers must contain a
rainbow AP of length 3? Either find the least such r or prove that no such r exists. [2 pts]
When a coloring of [n] uses many colors (relative to n), there will inevitably be some rainbow APs. For n ≥ k ≥ 3,
let M (n, k) denote the greatest possible number of colors, r, that could appear in an r-coloring of [n] with no rainbow
AP of length k (that is, there exists an r-coloring of [n] such that every AP of length k contains a duplicated color).
For example, M (5, 3) = 3 because
ˆ The result of Problem 2(a) is a 3-coloring of [5] with no rainbow AP of length 3.
ˆ The result of Problem 4 is that any 4-coloring of [5] always contains a rainbow AP of length 3.

8. Each part below consists of an expression of the form M (n, k). For each part, compute M (n, k) and construct
an example of an M (n, k)-coloring of [n] with no rainbow AP of length k.
a. M (5, 4) [2 pts]
b. M (6, 3) [2 pts]
c. M (6, 4) [2 pts]
d. M (8, 3) [3 pts]
9. a. Determine M (8, 4). [4 pts]
b. Determine M (2k, k) for each k ≥ 5. [3 pts]
10. Show that M (n, 3) ≤ log2 n + 1 for all n ≥ 3. [4 pts]

11. For all n ≥ k ≥ 4, prove that M (n, k) ≥ log2 (n − 1) + 2. [5 pts]
12. Is it true that M (n + 1, 3) ≥ M (n, 3) for all n ≥ 3? Justify your answer. [5 pts]

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 13
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
5 Power Round Solutions
7

1. a. The answer is 16. Any two terms form an arithmetic progression of length 2, so there are 2 = 21 such
progressions. Of these, 5 APs ({1, 4}, {1, 7}, {4, 7}, {2, 5}, {3, 6}) are not rainbow APs.
b. The answer is 8. Letting d be the common difference of the arithmetic progression, there are 5 APs of
length 3 with d = 1, 3 with d = 2, and 1 with d = 3. Of these 5 + 3 + 1 = 9 APs, only {1, 4, 7} is not a
rainbow AP.
2. a. One such example is RBRRG.
b. The answer is 12. The example from part (a) is unique up to reflection (i.e., string reversal) and renaming
colors. There are 3 ways to choose the repeated color and 2 ways to choose which of the remaining colors
appears last. Accounting for reflection, this gives a total of 3 · 2 · 2 = 12 colorings.
3. a. The answer is 63
4 (= 15.75). As established in Solution 1(a), there are 21 arithmetic progressions of length
2. The probability that the two randomly colored numbers in such a progression are the same color is 41 ,
so the probability that the progression is a rainbow AP is 34 . By linearity of expectation, the expected
number of rainbow APs of length 2 is 21 · 43 = 63
4 .
b. The answer is 27
8 (= 3.375). As established in Solution 1(a), there are 9 arithmetic progressions of length
3. Given any subset of three elements of [7], there are 43 = 64 different ways to color these elements. Of
these subsets, 4 · 3 · 2 = 24 have elements with three different colors. So the probability that all three
24
numbers of a 3-term arithmetic progression are different colors is 64 = 38 . By linearity of expectation, the
3 27
expected number of rainbow APs of length 3 is 9 · 8 = 8 .
c. The answer is 108
25 (= 4.32). The computation is the same as in the previous part, except that there are
53 = 125 colorings, of which 5 · 4 · 3 = 60 use three different colors. So 38 is replaced with 125
60
= 12
25 in the
12 108
calculation from part (b), and the expected number of rainbow APs of length 3 is 9 · 25 = 25 .

4. Note that some color is used exactly twice, and the other colors are used only once. Therefore at least one of
{1, 2, 3} or {3, 4, 5} must be a rainbow AP.
5. One such example is YRRBRRBRRG.
6. First note that an arithmetic progression of length 5 in [20] must have its common difference, d, be either 1,
2, 3, or 4. One way to avoid rainbow APs is to concentrate two of the colors at opposite ends of the range, so
that any rainbow AP (which must contain both of those colors) must have a large value of d. For instance, if
numbers 1–4 are colored red and 17–20 are colored blue, only a progression with d = 4 could have both a red
and a blue number. Now to ensure there is no rainbow AP of length 5, it suffices to map colors to the numbers
so that in each AP with d = 4, at least two of the numbers are colored the same. A simple example of this
is RRRR | YYPP | YYPP | GGGG | BBBB.
7. a. One such coloring is: RRG | RRG | RRB | RRG | RRG |RRB |. . . . The numbers that are not multiples of
3 are colored red, the multiples of 3 that are not multiples of 9 are colored green, and the multiples of
9 are colored blue. Note that for the 3-term arithmetic progression {a, a + d, a + 2d}, there are three
possibilities:
ˆ d is not divisible by 3, in which case two of the terms in the progression are not divisible by 3, so
both of these terms are colored red.
ˆ d is divisible by 3 but a is not, in which case all three terms of the progression are colored red.
ˆ d and a are both divisible by 3. In this case, there are no red terms in the progression.
b. The answer is no. The following argument establishes that for every r ≥ 3, there exists an r-coloring of
the positive integers with no 3-term rainbow AP. The previous part showed this to be the case for r = 3,
and for r < 3, there are too few colors to have a rainbow AP of length 3. Now, proceed by induction,
and assume there is an r-coloring of the positive integers (call it C) with the colors C1 , C2 , . . . , Cr and
with no 3-term rainbow AP. For a positive integer t, let f (t) = Ci if the color of t in C is Ci (1 ≤ i ≤ r).
To construct an (r + 1)-coloring of the positive integers with no 3-term rainbow AP, color the integers in
the following way: if t is divisible by 3, give it the color f (t/3). Otherwise, color it with a new color, C0 .

14 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 following analysis shows that this coloring has no 3-term rainbow APs. Note that for the arithmetic
progression {a, a + d, a + 2d}:
ˆ If d is not divisible by 3, then at least two of the terms in the progression are not divisible by 3 and
are both colored C0 .
ˆ If d is divisible by 3 but a is not, then all three terms are colored C0 .
ˆ If both d and a are divisible by 3, then there are no terms
in the progression
with color C0 . Then the
colors of a, a + d, and a + 2d are f (a/3), f (a + d)/3 , and f (a + 2d)/3 , respectively, which cannot
be a rainbow AP by the induction hypothesis.
8. a. The answer is M (5, 4) = 4. An example is RYBYG.
b. The answer is M (6, 3) = 3. An example is RRBRRG.
c. The answer is M (6, 4) = 5. An example is RBGGYP.
d. The answer is M (8, 3) = 4. An example is RBBYBYYG.
9. a. The answer is M (8, 4) = 5. A 5-coloring of [8] with no 4-term rainbow AP is RGRGBBYP. To show that
a 6-coloring is impossible, first note that in [8], the two 4-term rainbow APs {1, 2, 3, 4} and {5, 6, 7, 8} are
disjoint and must each contain at most three distinct colors, so in a 6-coloring, three colors must be used
only in the first half, and the other three colors only in the second half. Two colors must be duplicated,
and each of them used exactly twice. Furthermore, the 4-term rainbow APs {1, 3, 5, 7} and {2, 4, 6, 8} must
each contain a duplicate color, so the two pairs of numbers that are the same color cannot be consecutive.
But this implies that {3, 4, 5, 6} must be a rainbow AP. This is a contradiction, so a 6-coloring with no
4-term rainbow AP is impossible, hence M (8, 4) = 5.
b. For k ≥ 5, the answer is M (2k, k) = 2k − 2. To construct this, for k = 5, use • • C1 • C1 C2 • C2 • •, where
C1 and C2 are distinct colors, and each dot represents a different color (and is also different from both
C1 and C2 ). The analogous construction applies for k > 5, in which k − 5 dots (each representing a new
distinct color) are added to each end of the above string (for k = 5). This uses 2k − 2 colors. On the other
hand, {1, . . . , k} and {k + 1, . . . , 2k} are disjoint, and the numbers in each set have distinct colors, so any
(2k − 1)-coloring of [2k] must have a rainbow AP of length k.
10. Consider an r-coloring [n] that has no rainbow APs of length 3. Call an element x ∈ [n] new if there are
no elements in {1, . . . , x − 1} with the same color as x. (In particular, 1 is always new.) Any r-coloring will
have exactly r new elements, one for each color. Let x1 , x2 , . . . , xr be the new elements, indexed so that
x1 < x2 < · · · < xr ≤ n. (In particular, note that x1 = 1.)

Then for each i = 1, . . . , r − 1, it must be true that xi ≥ 2xi−1 . Indeed, assume, by way of contradiction, that
2xi−1 − xi > 0 for some i. Then {2xi−1 − xi , xi−1 , xi } is a rainbow AP because xi−1 and xi are new, which is
a contradiction. Thus it follows that xi ≥ 2xi−1 for all i = 1, . . . , r − 1. In particular,

xr ≥ 2r−1 x1 = 2r−1 .

Because xr ≤ n, it follows that r ≤ log2 n + 1, as desired.

11. In what follows, for each positive integer x, ν2 (x) denotes the exponent of 2 in the prime factorization of x.

Fix n≥ k ≥ 4, as in the problem statement. Let m denote the greatest integer for which 2m ≤ n − 1; that is,
m = log2 (n − 1) . Consider the following (m + 2)-coloring of [n] with the colors C⋆ , C0 , C1 , . . . , Cm :

ˆ The number 1 is assigned color C⋆ .

ˆ For each 2 ≤ x ≤ n, the number x is assigned color Cν2 (x−1) .

Then each of the m + 2 colors is indeed used at least once (for example, the color Ci is used on 2i + 1 for
all 0 ≤ i ≤ m). The problem asks to prove that M (n, k) ≥ m + 2. Thus it suffices to prove that there is no
rainbow AP of length 4 in the above coloring (in which case there will be no rainbow AP of length k > 4 either).

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 15
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Suppose that S = {a, a + d, a + 2d, a + 3d} is an AP of length 4 contained in [n]. Define g = gcd(a − 1, d), and
let b = (a − 1)/g ≥ 0 and r = d/g ≥ 1. Then gcd(b, r) = 1, and S can be rewritten as

1 + gb, 1 + g(b + r), 1 + g(b + 2r), 1 + g(b + 3r) .

To show that this is not a rainbow AP, it suffices to analyze the following two cases:
ˆ If b is odd, then b = a−1
g and b + 2r = a+2d−1
g are both odd numbers. It follows that

ν2 (a − 1) = ν2 (a + 2d − 1) = ν2 (g).

Thus a and a + 2d are the same color, namely Cν2 (g) .

ˆ If b is even, then because gcd(b, r) = 1, it follows that r is odd. Therefore b+r = a+d−1
g and b+3r = a+3d−1
g
are odd numbers. It follows that

ν2 (a + d − 1) = ν2 (a + 3d − 1) = ν2 (g).

Thus a + d and a + 3d are the same color, namely Cν2 (g) . Note that this case works even if a = 1, or
equivalently, b = 0.
12. The answer is no. A counterexample is M (8, 3) = 4, yet M (9, 3) = 3. The following is a 4-coloring of [8] with
no rainbow AP of length 3:
1 2 3 4 5 6 7 8
R G G B G B B Y
This shows that M (8, 3) ≥ 4. The result of Solution 10 implies that M (8, 3) ≤ 4, so M (8, 3) = 4, as desired.

Now consider M (9, 3). If M (9, 3) = 4, then there is a 4-coloring of [9] with no rainbow AP of length 3. Remov-
ing 9 from the set results in either a 4-coloring or 3-coloring of [8] with no rainbow AP of length 3. A 3-coloring
of [8] occurs if 9 is colored differently from every other number. For the original coloring to have no rainbow
APs of length 3, both numbers in each of the pairs {1, 5}, {3, 6}, {5, 7}, and {7, 8} must be the same color. If
the color for {3, 6} is the same color for the four numbers in {1, 5, 7, 8}, then 2 and 4 must be the other two
colors, and {2, 3, 4} is a rainbow AP. Otherwise, either {1, 2, 3} or {3, 4, 5} is a rainbow AP. This means that
removing 9 must result in a 4-coloring of [8]. All such colorings will now be classified.

Using the result and notation of Solution 10, in which xi denotes the least integer that has the ith color, it
follows that x1 = 1, x2 = 2, x3 = 4, and x4 = 8. Using R, G, B, and Y as the four colors in that order, this
forces the pattern RG?B???Y. Now, avoiding rainbow APs in {1, 2, 3} and {2, 3, 4} means that the color of 3
must be G. Then 6 and 7 must be the same color (because they cannot be Y, and {6, 7, 8} cannot be a rainbow
AP), and they must be B to avoid making {4, 6, 8} be a rainbow AP. Finally, 5 must be G to avoid either
{3, 4, 5} or {2, 5, 8} from being a rainbow AP. Thus the only possible 4-coloring of [8] (up to permutation of
the colors) is the one found earlier in this solution. However, none of the four colors can then be assigned to 9:
R or G makes {7, 8, 9} a rainbow AP, and B or Y makes {1, 5, 9} a rainbow AP. This shows that M (9, 3) < 4.
(In fact, the coloring RRGRRGRRB shows that M (9, 3) = 3, though it was not required to demonstrate this.)

Author’s Note: This Power Question was inspired by the paper titled Rainbow arithmetic progressions, which can
be found at https://arxiv.org/abs/1404.7232.

16 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.
6 Individual Round

Problem 1. In rectangle ABCD, point E lies on AB so that BE = 8 and DE = 20. Given that [ABCD] = 288
and [AED] = 96, compute AD + AE.

√ √ √
Problem 2. Given that x + y = 697, y + z = 672, and x + z = 680, compute
√ √ √
x − y + y − z + x − z.

Problem 3. Mabel has a fair six-sided die with faces labeled 1, 2, 3, 4, m, n, where m and n are integers with
1 ≤ m < n ≤ 6. She rolls this die twice. Compute the ordered pair (m, n) for which the probability that the
sum of the two rolls equals 6 is maximized.

Problem 4. Suppose that rectangle ARM L with AR < RM has diagonals that intersect at X. Rectangle AXIS
[AXIS] 23
is constructed so that X, R, and S are collinear, with R between X and S. Given that = ,
[ARM L] 24
Å ã2
AR
compute .
RM

Problem 5. The polynomials P (x) and Q(x) have real coefficients. When P (x) is divided by (x + 1)(x + 10), the
quotient is Q(x) and the remainder is 2x. When P (x) is divided by (x + 4)(x + 7), the quotient is Q(x) and
the remainder is 38x + 18. Compute the remainder when P (x) is divided by (x − 1).

Problem 6. In the left figure below, the cell in the fourth row and third column of a 9 × 9 grid is deleted to obtain
a set S of 80 cells. For each n = 1, 2, 3, . . . , 8, an L-shaped tile Ln consisting of 2n + 1 cells is given, as shown
in the figure at right.

L1 L2 L3 L4 L5

L6 L7 L8

Compute the number of ways to tile S using these eight L-shaped tiles, each exactly once, with rotations of
tiles allowed.

Problem 7. Let a1 , a2 , a3 , . . . be the increasing sequence consisting of all positive integers whose binary repre-
sentations contain an odd number of 1s. For example, 1 = 12 and 19 = 100112 are members of this sequence.
Compute a2024 .

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 17
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Problem 8. For an integer N , the rightmost four digits of 2024 · N are A B C D . Compute the least N > 1 such
that A B + 4 = C D , where A and C may be zero.

Problem 9. Let a be the positive real solution to x4 = 20x2 + 24x + 19. Compute ⌊a⌋.

Problem 10. Circle γ has center O and radius 9. Circle ωA has radius 2 and is internally tangent to γ at point
A. Circle ωB has radius 3 and is internally tangent to γ at point B. The common external tangents of ωA and
ωB meet at T . Given that T O = 2 · AB, compute AB.

18 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.
7 Individual Round Answers

Answer 1. 28

Answer 2. 442

Answer 3. (2, 4)

11
Answer 4.
35
Answer 5. −64

Answer 6. 1568

Answer 7. 4046

Answer 8. 54

Answer 9. 5
√
27 2
Answer 10.
4

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 19
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
8 Individual Round Solutions

Problem 1. In rectangle ABCD, point E lies on AB so that BE = 8 and DE = 20. Given that [ABCD] = 288
and [AED] = 96, compute AD + AE.

Solution 1. Let AD = x and AE = y. Then [AED] = 12 xy = 96 =⇒ xy = 192 and [ABCD] = x(y + 8) = 288.
Thus 192(y+8)
y = 288, so y = 16 and x = 192
16 = 12. Hence AD + AE = 12 + 16 = 28.

Alternate Solution: Note that neither BE norpthe area of ABCD

p is needed to√compute AD + AE.
√ By the
Pythagorean Theorem, x2 + y 2 = 400, so x + y = (x + y)2 = x2 + y 2 + 2xy = 400 + 2 · 192 = 784 = 28.

Alternate Solution: Let F lie on CD so that EF ∥ BC. Then 288 = [ABCD] = [AED]+[DEF ]+[BCF E] =
2·96+[BCF E], from which [BCF E] = 96. Thus AD = EF = 96 2·96
8 = 12 and AE = 12 = 16, so AD +AE = 28.
Note that the length of DE is not needed for this solution.

√ √ √
Problem 2. Given that x + y = 697, y + z = 672, and x + z = 680, compute
√ √ √
x − y + y − z + x − z.

Solution 2. Using the differences of squares,

√ √ √ √
x − y = (x + z) − (y + z) = ( x + z + y + z)( x + z − y + z) = (680 + 672)(680 − 672)
= 1352 · 8 = 676 · 16,
√
so x − y = 26 · 4 = 104. By a similar argument,
√ √ √ √
y − z = (x + y) − (x + z) = ( x + y + x + z)( x + y − x + z) = (697 + 680)(697 − 680)
= 1377 · 17 = 81 · 17 · 17
√
so y − z = 9 · 17 = 153. Lastly, note that
√ √ √ √
x − z = (x + y) − (y + z) = ( x + y + y + z)( x + y − y + z) = (697 + 672)(697 − 672)
= 1369 · 25
√ √ √ √
so x − z = 37 · 5 = 185. Therefore the answer is x−y+ y−z+ x − z = 104 + 153 + 185 = 442.

Problem 3. Mabel has a fair six-sided die with faces labeled 1, 2, 3, 4, m, n, where m and n are integers with
1 ≤ m < n ≤ 6. She rolls this die twice. Compute the ordered pair (m, n) for which the probability that the
sum of the two rolls equals 6 is maximized.

3
Solution 3. The probability that Mabel rolls a sum of 6 by rolling one of the first four faces twice is 36 . The prob-
2
ability that Mabel rolls a sum of 6 by rolling m once and one of the first four faces once is 36 if 2 ≤ m ≤ 5 and
0 otherwise. The same holds for rolling n once and one of the first four faces once. The probability that Mabel
2
rolls a sum of 6 by rolling m once and n once is 36 if m + n = 6 and 0 otherwise. Finally, the probability that
1
Mabel rolls a sum of 6 by rolling either m twice or n twice is 36 if m = 3 or n = 3, respectively, and 0 otherwise.

The probability that Mabel rolls a sum of 6 is maximized when 2 ≤ m < n ≤ 5 and m + n = 6, because if
2 1
m + n ̸= 6, then the 36 probability is lost while at most a 36 probability can be gained as m and n cannot
both be 3. The only ordered pair for which this is true is (2, 4).

20 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.
Problem 4. Suppose that rectangle ARM L with AR < RM has diagonals that intersect at X. Rectangle AXIS
[AXIS] 23
is constructed so that X, R, and S are collinear, with R between X and S. Given that = ,
[ARM L] 24
Å ã2
AR
compute .
RM

Solution 4. Without loss of generality, draw ARM L and AXIS so that L, X, R, and S are collinear in that
order, as shown in the diagram below.

AR
Because ∠LAX and ∠SAR are both complementary to ∠XAR, it follows that △SAR ∼ △SLA. Let RM =r
[SAR] 2 K
and observe that [SLA] = r . Let [SAR] = K so that [SLA] = r2 , and therefore [ALR] = [SLA] − [SAR] =
2
K 1−r 2
Ä ä Ä ä
K 1−r
r 2 . Because X is the midpoint of the diagonals of ARM L, it follows that [RAX] = 2 r 2 and
K 1+r 2 1+r 2
Ä ä [AXIS] 23
[SAX] = 2 r2 . Hence [ARM L] = 2−2r2 . Set this ratio equal to 24 and cross multiply to obtain 24 + 24r2 =
AR 2 11

46 − 46r2 , thus = r2 = .
RM 35

Problem 5. The polynomials P (x) and Q(x) have real coefficients. When P (x) is divided by (x + 1)(x + 10), the
quotient is Q(x) and the remainder is 2x. When P (x) is divided by (x + 4)(x + 7), the quotient is Q(x) and
the remainder is 38x + 18. Compute the remainder when P (x) is divided by (x − 1).

Solution 5. The problem statement implies that

P (x) = (x + 1)(x + 10)Q(x) + 2x and P (x) = (x + 4)(x + 7)Q(x) + 38x + 18.

Equate the two expressions for P (x) and solve for Q(x), as shown below.

(x + 1)(x + 10)Q(x) + 2x = (x + 4)(x + 7)Q(x) + 38x + 18

(x + 1)(x + 10) − (x + 4)(x + 7) Q(x) = 36x + 18
(x2 + 11x + 10 − x2 − 11x − 28)Q(x) = 36x + 18
−18Q(x) = 36x + 18
Q(x) = −2x − 1

It follows that P (x) = (x + 1)(x + 10)(−2x − 1) + 2x, and thus the remainder when P (x) is divided by (x − 1)
is P (1) = 2 · 11 · (−3) + 2 = −64.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 21
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Problem 6. In the left figure below, the cell in the fourth row and third column of a 9 × 9 grid is deleted to obtain
a set S of 80 cells. For each n = 1, 2, 3, . . . , 8, an L-shaped tile Ln consisting of 2n + 1 cells is given, as shown
in the figure at right.

L1 L2 L3 L4 L5

L6 L7 L8

Compute the number of ways to tile S using these eight L-shaped tiles, each exactly once, with rotations of
tiles allowed.

Solution 6. Consider first the task of placing the eight tiles in a 9 × 9 grid, ignoring for now the choice of which
single cell remains uncovered.

The process can be described as follows. The largest tile, L8 , needs to first be placed in one of four orientations:
the “corner” of L8 needs to be placed in the northwest, northeast, southwest, or southeast corner of the grid.
Once it is placed, the remaining cells form an 8 × 8 subgrid, and the second largest L7 tile needs to be placed
in the same way. The pattern continues with L6 , L5 , . . . , down to L1 .

In other words, the possible placements of these eight tiles corresponds exactly to a choice of one of NW, NE,
SW, SE for each n = 8, 7, . . . , 1. (In particular, there are exactly 48 ways to place the tiles.) One example of
a placement, together with its accompanying directional indicators, is shown below.

L8 L8 L8 L8 L8 L8 L8 L8 L8

L6 L4 L4 L4 L4 L4 L5 L7 L8

L6 L3 L3 L3 L3 L4 L5 L7 L8

L6 L2 L1 L3 L4 L5 L7 L8

L6 L2 L1 L1 L3 L4 L5 L7 L8

L6 L2 L2 L2 L3 L4 L5 L7 L8

L6 L5 L5 L5 L5 L5 L5 L7 L8

L6 L6 L6 L6 L6 L6 L6 L7 L8

L7 L7 L7 L7 L7 L7 L7 L7 L8

L8 L7 L6 L5 L4 L3 L2 L1
NE SE SW SE NE NE SW SE

In the given problem, the task is to compute how many of these 48 placements result in the missing square
being located in the fourth row and third column, as in the previous example. The main insight is that this
occurs if and only if

22 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.
ˆ there are exactly three N and five S letters; and
ˆ there are exactly two W and six E letters.

This is because the number of W letters counts the number of columns to the left of the missing square, and
similarly for the other letters. The final answer is therefore 83 · 82 = 1568.

Problem 7. Let a1 , a2 , a3 , . . . be the increasing sequence consisting of all positive integers whose binary repre-
sentations contain an odd number of 1s. For example, 1 = 12 and 19 = 100112 are members of this sequence.
Compute a2024 .

Solution 7. For k ≥ 1, let Sk denote the set of nonnegative integers whose base-2 representations have k or fewer
digits. Note that 0 ∈ Sk for all k. Then |Sk | = 2k , and by induction, one can show that Sk has 2k−1 elements
with an odd number of 1s and 2k−1 elements with an even number of 1s. Noting that 211 = 2048, the great-
est element of S12 is 212 − 1, which is 111,111,111,1112 . It therefore follows that a2048 = 212 − 2 = 4094 =
111,111,111,1102 . Working backwards, to compute a2024 , first compute a2048−32 = a2016 , and note that the 32
numbers a2017 , a2018 , . . . , a2048 differ only in their last six digits, when expressed in base 2 (with the leftmost six
digits being 111,111). Thus a2016 = 111,110,111,1112 = 4031, and counting up from 111,111,000,0012 = a2017 ,
conclude that a2024 = 111,111,001,1102 = 4046.

Alternate Solution: The key observation for this solution is that for any nonnegative integer m, exactly one
of 2m and 2m + 1 has a binary representation with an odd number of 1s (because 2m + 1 has exactly one more
1 in its binary representation than 2m). Therefore one of 2 · 2023 or 2 · 2023 + 1 is the 2024th number in the
sequence. Because 404610 = 1111110011102 has nine 1s, it follows that a2024 = 4046.

Note: Positive integers whose binary representations contain an odd number of 1s are called odious numbers.

Problem 8. For an integer N , the rightmost four digits of 2024 · N are A B C D . Compute the least N > 1 such
that A B + 4 = C D , where A and C may be zero.

Solution 8. Let RN be the residue of 2024N modulo 10000. Note that if RN satisfies the conditions given in
the problem, then RN −1 will be a multiple of 101. Accordingly, let M = N − 1. The goal is to find the
least positive integer M such that the residue of 2024M modulo 10000 is a multiple of 101. Noting that 8 is
a common factor of 2024 and 10000, and 8 and 101 are coprime, the problem simplifies to finding the least
positive integer M such that the residue of 253M modulo 1250 is a multiple of 101. That means the residue
of 253M modulo 1250 is of the form 101K, 0 ≤ K ≤ 12. The multiplicative inverse of 253 modulo 1250 is
667, and 101 · 667 ≡ 1117 (mod 1250). Subsequent residues can be computed by subtracting 133 modulo 1250,
giving the residue sequence 0, 1117, 984, 851, 718, 585, 452, 319, 186, 53, 1170, 1037, 904. Accordingly, the
residue of 253 · 53 modulo 1250 is a multiple of 101, so N = 54 is the least such N greater than 1. To check,
2024 · 54 = 109296.

Problem 9. Let a be the positive real solution to x4 = 20x2 + 24x + 19. Compute ⌊a⌋.

Solution 9. Because both x4 and 20x2 + 24x + 19 are increasing functions for x ≥ 0, the positive real solution
to the given equation occurs between integer values of x where there is a change in which function is greater.
Note that 54 = 625 < 639 = 20 · 52 + 24 · 5 + 19, but 64 = 1296 > 883 = 20 · 62 + 24 · 6 + 19, so there is a
solution to x4 = 20x2 + 24x + 19 between 5 and 6. Thus it follows that ⌊a⌋ = 5.

Problem 10. Circle γ has center O and radius 9. Circle ωA has radius 2 and is internally tangent to γ at point
A. Circle ωB has radius 3 and is internally tangent to γ at point B. The common external tangents of ωA and
ωB meet at T . Given that T O = 2 · AB, compute AB.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 23
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Solution 10. Note that T lies on line P Q. Moreover, T is the center of positive homothety from ωA to ωB , while
A is the center of positive homothety from γ to ωA , and B is the center of positive homothety from γ to ωB .
Hence, by Monge’s Theorem, T lies on line AB.

A x B
T
2 3
P
Q
2x

Let x = AB. Then by Menelaus’ Theorem,

AP OQ BT 2 6 BT
−1 = · · = · · ,
P O QB T A 7 3 TA
BT
from which it follows that the directed ratio TA equals − 74 . It follows that T A = 43 x. Hence the power of the
point T with respect to γ equals
4 7 28 2
TA · TB = x· x= x .
3 3 9
However, the power of T is also equal to

T O2 − 92 = (2x)2 − 92 .

Set these equal and solve to obtain

√
2 28 2 8 2 27 2
4x − 81 = x =⇒ x = 81 =⇒ x = .
9 9 4
Alternate Solution: Let x = AB. First compute the distance between the centers P , Q of ωA , ωB using the
Law of Cosines:
92 + 92 − x2 72 + 62 − P Q2 14
= =⇒ P Q2 = 1 + x2 .
2·9·9 2·7·6 27
Next compute the distance of T along line P Q using similar triangles:
TP TQ
= =⇒ T P = 2P Q and T Q = 3P Q.
2 3
Now apply Stewart’s Theorem to △OT Q with cevian OP :

2P Q · P Q · 3P Q + 72 · 3P Q = 62 · 2P Q + T O2 · P Q
=⇒ 6P Q2 + 75 = T O2 .

Substitute P Q2 from above and T O2 = 4x2 to obtain

√
28 2 2 27 2
6 + x + 75 = 4x =⇒ x = .
9 4

24 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.
9 Relay Round

Problem 1-1. Compute the number of positive integers less than or equal to 100 that are not multiples of 2 or 3.

Problem 1-2. Let T be the number you will receive. Let ARM L be a square of side length T . Let S be the set
of all points P such that a circle of radius 1 centered at P lies completely within ARM L. Compute the area
of S.

Problem 1-3. Let T be the number you will receive. For some positive integer K, the number of subsets of
{1, 2, 3, . . . , K} that contain at least one even number and at least one odd number is T . Compute K.

Problem 2-1. Compute the greatest perfect square that divides both 1020 − 1 and 1024 − 1.

Problem 2-2. Let T be the number you will receive. Compute the greatest prime number that divides

(T − 1)! + T ! .

Problem 2-3. Let T be the number you will receive. Compute the greatest perfect square that divides

(T − 1)! + T ! + (T + 1)! .

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 25
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
10 Relay Round Answers

Answer 1-1. 33

Answer 1-2. 961

Answer 1-3. 10

Answer 2-1. 9

Answer 2-2. 7

Answer 2-3. 9216

26 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.
11 Relay Round Solutions

Problem 1-1. Compute the number of positive integers less than or equal to 100 that are not multiples of 2 or 3.

Solution 1-1. There are 100

100
= 33 multiples of 3, and 100

2 = 50 multiples of 2, 3 6 = 16 multiples of both 2 and
3. Thus the desired number of integers is 100 − (50 + 33 − 16) = 33.

Problem 1-2. Let T be the number you will receive. Let ARM L be a square of side length T . Let S be the set
of all points P such that a circle of radius 1 centered at P lies completely within ARM L. Compute the area
of S.

Solution 1-2. The center of the circle must be at least 1 unit away from each side of the square, so the desired
region is a square of side length T − 2, which has area (T − 2)2 . With T = 33, the answer is 312 = 961.

Problem 1-3. Let T be the number you will receive. For some positive integer K, the number of subsets of
{1, 2, 3, . . . , K} that contain at least one even number and at least one odd number is T . Compute K.

Solution 1-3. Let S be a subset of {1, 2, . . . , K} that contains at least one even number and at least one odd number.
Then S = E ∪ O, where E is a nonempty set of even numbers, and O is a nonempty set of odd numbers. If
K is even, then K = 2n for some positive integer n. There are 2n − 1 nonempty subsets of {2, 4, . . . , 2n}, and
likewise, there are 2n − 1 nonempty subsets of {1, 3, . . . , 2n − 1}. Thus T = (2n − 1)2 . A similar analysis reveals
that if K is odd, then K = 2n − 1 for some positive integer n, and T = (2n−1 − 1)(2n − 1). With T = 961,
a perfect square, it follows that there must be an integer n for which (2n − 1)2 = 961 = 312 =⇒ 2n = 32, so
n = 5, and K = 10.

Problem 2-1. Compute the greatest perfect square that divides both 1020 − 1 and 1024 − 1.

Solution 2-1. If an integer n divides both 1020 − 1 and 1024 − 1, then n must divide their difference: 1024 − 1020 .
But n cannot have any factors of 2 or 5, so n must divide 104 − 1 = 32 · 11 · 101. Thus the greatest candidate
is 9, which indeed divides both of the given numbers (which consist of 20 9s and 24 9s, respectively).

Problem 2-2. Let T be the number you will receive. Compute the greatest prime number that divides

(T − 1)! + T ! .

Solution 2-2. Factor the given expression as (T − 1)! · (1 + T ). With T = 9, this becomes 8! · 10. The greatest prime
number less than 8 is 7, while no greater prime divides 10, so the answer is 7.

Problem 2-3. Let T be the number you will receive. Compute the greatest perfect square that divides

(T − 1)! + T ! + (T + 1)! .

Solution 2-3. Factor the given expression as (T − 1)! · 1 + T + T · (T + 1) = (T − 1)! · (T + 1)2 . With T = 7, this

becomes 6! · 82 = (2 · 3) · 4 · 5 · 6 · 82 = (6 · 2 · 8)2 · 5. The greatest perfect square that divides this number is thus
(6 · 2 · 8)2 = 9216.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 27
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
12 Super Relay
1. Let x, y, and z be positive integers such that x + y = 20 and x + yz = 24. Compute the number of possible
ordered triples (x, y, z).

2. Let T = TNYWR. The figure at right (not necessarily drawn to scale)

shows two concentric circles. The radius of the larger circle is 2, and the
area of the shaded region outside the smaller circle and inside the larger
circle is T π. Compute the radius of the smaller circle.

3. Let T = TNYWR. Three standard, fair six-sided dice are rolled. Compute the probability that the sum of the
numbers rolled is 16 + ⌊T ⌋.
4. Let T = TNYWR. Compute the value of x that satisfies the system of simultaneous equations
1 1 1
√ +√ = √
2 x y 3 2
1 1 √
√ − √ = T.
2 x y

5. Let T = TNYWR. Suppose that

log2 (20T ) = log2 (24 − x) + log4 (100).

Compute log2 (x) .
6. Let T = TNYWR. Sphere S has radius R. A plane intersects S in circle C, centered at P and with radius
T + 8. The shortest distance from P to sphere S is 1. Compute R.
7. Let T = TNYWR. Working individually, each of Abby, Robert, and Morgan takes T − 1 hours to complete
a job. Working individually, Leo takes 20 hours to complete the same job. Working together, compute the
number of hours it takes the four of Abby, Robert, Morgan, and Leo to complete the job.

15. Albert makes a list of all unordered triples of three of the six letters in his name. Rich, on the other hand,
makes a list of all ways to order the letters R, I, C, and H. Albert’s list has M triples, and Rich’s list has L
orderings of the four letters in his name. Compute M + L.
14. Let T = TNYWR. Compute the value of x that satisfies the equation

(T + 1)2 = 2024 + x.
√
13. Let T = TNYWR, and let i = −1. Compute the number of integers n such that 1 ≤ n ≤ 2024 + T and
iT = in .
12. Let T = TNYWR, and let K = T − 1. Let p be the least prime factor of K, and let q be the greatest prime
factor of K. Compute the sum of the positive divisors of p3 q.
11. Let T = TNYWR. Compute the number of values of x with 0 ≤ x ≤ T such that sin(x◦ ) = sin(2x◦ ).
10. Let T = TNYWR. Each interior angle of a regular (T + 4)-gon measures a degrees, and each interior angle of
a regular 20-gon measures b degrees. Compute the integer closest to |a − b|.
9. Let T = TNYWR. In the arithmetic sequence a1 , a2 , a3 , . . . , let a20 = 3 and a24 = 47. Compute aT .

8. Let x0 be the positive number you will receive from position 7, and let y0 be the positive number you will
receive from position 9. Let A = (0, 0), R = (x0 − 2, 0), and L = (0, y0 − 1). Line ℓ1 has slope −5 and passes
through R. Line ℓ2 has slope −1 and passes through L. Point M is the intersection of lines ℓ1 and ℓ2 . Compute
the area of quadrilateral ARM L.

28 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.
13 Super Relay Answers

1. 3
2. 1
1
3. 72

4. 8
5. 3
6. 61
7. 10
15. 44
14. 1
13. 507
12. 360
11. 5
10. 22
9. 25
8. 128

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 29
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
14 Super Relay Solutions

Problem 1. Let x, y, and z be positive integers such that x + y = 20 and x + yz = 24. Compute the number of
possible ordered triples (x, y, z).

Solution 1. Subtract the first equation from the second and factor to obtain y(z − 1) = 4. The 3 positive integer
solutions to this equation are (y, z) = (1, 5), (2, 3), and (4, 2). Each of these determines a unique positive
integer x, so the answer is 3.

Problem 2. Let T = TNYWR. The figure at right (not necessarily drawn to scale)
shows two concentric circles. The radius of the larger circle is 2, and the
area of the shaded region outside the smaller circle and inside the larger
circle is T π. Compute the radius of the smaller circle.

Solution 2. Let r be the radius of the smaller circle. Then the shaded region has area 4π − r2 π, so T = 4 − r2 .
With T = 3, r = 1.

Problem 3. Let T = TNYWR. Three standard, fair six-sided dice are rolled. Compute the probability that the
sum of the numbers rolled is 16 + ⌊T ⌋.

Solution 3. Let n = 16 + ⌊T ⌋. Note that if either n < 3 or n > 18, then the answer is trivially 0. Otherwise, let
f (n) be the number of ordered triples (x1 , x2 , x3 ) such that x1 + x2 + x3 = n and 1 ≤ x1 , x2 , x3 ≤ 6. Then
the answer is f6(n)
3 . With T = 1, n = 17, and f (n) = 3 (the possible triples are (5, 5, 6), (5, 6, 5), and (6, 5, 5)).
3 1
Thus the answer is 216 = 72 .

Problem 4. Let T = TNYWR. Compute the value of x that satisfies the system of simultaneous equations

1 1 1
√ +√ = √
2 x y 3 2
1 1 √
√ − √ = T.
2 x y

√
Solution 4. Adding the two equations yields √1x = 3√1
2
+ T . This is equivalent to x = √ 18 . With
1
√ 1+6 2T +18T
T = 72 , 6 2T = 66 = 1 and 18T = 14 . Thus the answer is x = 1+1+1/4
18
= 8.

Problem 5. Let T = TNYWR. Suppose that

log2 (20T ) = log2 (24 − x) + log4 (100).

Compute log2 (x) .

Solution 5. Note that log2 (20T ) = log2 (2T ) + log2 (10) and that log4 (100) = log 2 (100) 2 log2 (10)
log2 (4) = 2 log2 (2) = log2 (10).
Thus
the given equation is equivalent to log2 (2T ) = log2 (24 − x). Thus x = 24 − 2T . With T = 8, x = 8, hence
log2 (x) = 3.

30 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.
Problem 6. Let T = TNYWR. Sphere S has radius R. A plane intersects S in circle C, centered at P and with
radius T + 8. The shortest distance from P to sphere S is 1. Compute R.

−−→
Solution 6. Let O be the center of S, and let OP intersect S at Q. Let P ′ be any point on C, and note that
′
△OP P is a right triangle with legs OQ − P Q = R − 1 and T + 8, and hypotenuse R. Thus (T + 8)2 =
R2 − (R − 1)2 = 2R − 1, so R = 21 (T + 8)2 + 1 . With T = 3, the answer is thus 12 (121 + 1) = 61.

Problem 7. Let T = TNYWR. Working individually, each of Abby, Robert, and Morgan takes T − 1 hours
to complete a job. Working individually, Leo takes 20 hours to complete the same job. Working together,
compute the number of hours it takes the four of Abby, Robert, Morgan, and Leo to complete the job.

1
Solution 7. In one hour, the fraction of the job that each of Abby, Robert, and Morgan completes is T −1 . In one
1
hour, the fraction of the job that Leo completes is 20 . If it takes h hours for the four of them to complete the
3 1
job, then T −1 + 20 = h1 . With T = 61, this implies that h1 = 60 3 1
+ 20 1
= 10 , hence h = 10.

Problem 15. Albert makes a list of all unordered triples of three of the six letters in his name. Rich, on the other
hand, makes a list of all ways to order the letters R, I, C, and H. Albert’s list has M triples, and Rich’s list has
L orderings of the four letters in his name. Compute M + L.

6

Solution 15. Note that M = 3 = 20 and L = 4! = 24. Thus M + L = 20 + 24 = 44.

Problem 14. Let T = TNYWR. Compute the value of x that satisfies the equation

(T + 1)2 = 2024 + x.

Solution 14. Rearranging the given equation yields x = (T + 1)2 − 2024. With T = 44, x = 452 − 2024 =
2025 − 2024 = 1.

√
Problem 13. Let T = TNYWR, and let i = −1. Compute the number of integers n such that 1 ≤ n ≤ 2024 + T
and iT = in .

Solution 13. If T ≤ −2024, then the given inequality is never satisfied for any n, hence the answer is trivially 0.
Also, if T is not an integer, then iT ∈
/ {±i, ±1}, and again, the answer is trivially 0. So assume that T is an
integer such that T ≥ −2023. Because integer powers of i cycle in groups of 4, the number of integers n in
[1, 2024 + T ] such that in = iT is 2024+T

4 . With T = 1, the answer is 507.

Problem 12. Let T = TNYWR, and let K = T − 1. Let p be the least prime factor of K, and let q be the greatest
prime factor of K. Compute the sum of the positive divisors of p3 q.

Solution 12. A positive divisor of p3 q has the form pa q b , where a ∈ {0, 1, 2, 3} and b ∈ {0, 1}. Note that the terms
in the expansion of (1 + p + p2 + p3 )(1 + q) precisely correspond to the positive divisors of p3 q. With T = 507,
K = 506 = 2 · 11 · 23, p = 2, and q = 23. Thus the answer is (1 + 2 + 4 + 8)(1 + 23) = 15 · 24 = 360.

Problem 11. Let T = TNYWR. Compute the number of values of x with 0 ≤ x ≤ T such that sin(x◦ ) = sin(2x◦ ).

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 31
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
Solution 11. Using the identity sin(2t) = 2 sin(t) cos(t), the given equation implies that either sin(x◦ ) = 0 or
cos(x◦ ) = 21 . The nonnegative values of x that satisfy sin(x◦ ) = 0 are 180k, where k is a nonnegative integer.
The nonnegative values of x that satisfy cos(x◦ ) = 12 are of the form |60 + 360k ′ |, where k ′ is an integer. With
T = 360, sin(x◦ ) = 0 has solutions x = 0, 180, or 360, and cos(x◦ ) = 12 has solutions x = 60 or 300. This is a
total of 5 possible values of x.

Problem 10. Let T = TNYWR. Each interior angle of a regular (T +4)-gon measures a degrees, and each interior
angle of a regular 20-gon measures b degrees. Compute the integer closest to |a − b|.

Solution 10. The sum of the interior angles of an n-gon is 180◦ (n − 2), so each interior angle of a regular n-gon
measures 180(n−2)
n degrees. Thus a = 180(T +2)
T +4 and b = 180·18
20 = 162. With T = 5, it follows that a = 140.
Thus the closest integer to |a − b| is 22.

Problem 9. Let T = TNYWR. In the arithmetic sequence a1 , a2 , a3 , . . . , let a20 = 3 and a24 = 47. Compute aT .

Solution 9. Let d be the common difference. Then a24 −a20 = 47−3 = 44 = 4d, so d = 11. Then a1 = a20 −19·11,
and the general term of the sequence is an = −217+11n. With T = 22, a22 = −217+11·22 = 25. (Alternatively,
note that a22 = a20 +a
2
24
= 3+47
2 = 25.)

Problem 8. Let x0 be the positive number you will receive from position 7, and let y0 be the positive number
you will receive from position 9. Let A = (0, 0), R = (x0 − 2, 0), and L = (0, y0 − 1). Line ℓ1 has slope −5 and
passes through R. Line ℓ2 has slope −1 and passes through L. Point M is the intersection of lines ℓ1 and ℓ2 .
Compute the area of quadrilateral ARM L.

Solution 8. Note that [ARM L] = [ARM ] + [AM L]. The equation of ℓ1 is y = −5x + (5x0 − 10). The equation
of ℓ2 is y = −x + (y0 − 1). The point M thus satisfies the system of equations

y = −5x + (5x0 − 10)

y = −x + (y0 − 1).

The solution to this system is (x, y) = 5x0 −y 0 −9 −5x0 +5y0 +5

. Using “ 12 base times height” for the area of a

4 , 4
triangle, it follows that

−5x0 + 5y0 + 5 5x0 − y0 − 9

Å ã Å ã
1 1
[ARM ] = (x0 − 2) and [AM L] = (y0 − 1) .
2 4 2 4

With x0 = 10 and y0 = 25, R = (8, 0), L = (0, 24), and M = (4, 20). (Note the intentional appearance of
“20” and “24” in those coordinates!) Plugging the values of x0 and y0 into the above equations for [ARM ] and
[AM L] yields [ARM L] = 21 · 8 · 20 + 12 · 24 · 4 = 128.

32 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.
15 Tiebreaker Round

Problem 1. Let S be the set of all positive integers n such that the numbers of divisors of 8n, 9n, and 10n form an
arithmetic sequence (in some order) and such that n is not a multiple of 3. Compute the second least element
of S.

Problem 2. The domain of the function f (x) = log6 (log5 (log4 (log3 (log2 (x))))) is the set of x such that x > A,
where A is an integer. Compute the number of digits in A.

Problem 3. Eddy and Sarah are at an ARML coaches dinner together, and the menu has 7 different dishes.
Compute the number of ways that Eddy and Sarah can order dinner so that each of them orders at least one
dish, no dish can be ordered twice, and no dish is ordered by both Eddy and Sarah.

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 33
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
16 Tiebreaker Round Answers

Answer 1. 350

Answer 2. 25

Answer 3. 1932

34 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.
17 Tiebreaker Round Solutions

Problem 1. Let S be the set of all positive integers n such that the numbers of divisors of 8n, 9n, and 10n form an
arithmetic sequence (in some order) and such that n is not a multiple of 3. Compute the second least element
of S.

Solution 1. Let n = 2a · 5b · K, where gcd(K, 30) = 1, and let d(m) denote the number of positive divisors of m.
Then d(8n) = (a + 4)(b + 1)d(K), d(9n) = 3(a + 1)(b + 1)d(K), and d(10n) = (a + 2)(b + 2)d(K). This means
that (a + 4)(b + 1), 3(a + 1)(b + 1), and (a + 2)(b + 2) are in an arithmetic sequence (in some order). First, note
that 3(a + 1)(b + 1) ≥ (a + 4)(b + 1) unless a = 0. Furthermore, 3(a + 1)(b + 1) − (a + 2)(b + 2) = 2ab + a + b − 1
by expanding, so 3(a + 1)(b + 1) ≥ (a + 2)(b + 2) unless a = b = 0; however, this possibility can be discarded,
because n = 1 does not satisfy the desired conditions.

If a = 0, then 3(b + 1), 2(b + 2), and 4(b + 1) must be in an arithmetic sequence (in some order) for some integer
b ≥ 1. Because 2(b + 2) ≤ 3(b + 1) ≤ 4(b + 1), the common difference would have to both be b − 1 (from the
first two terms) and b (from the last two terms), so this is not possible.

Otherwise, if a ̸= 0, then 3(a + 1)(b + 1) is the last term in the arithmetic sequence. If (a + 4)(b + 1) is the
least term in the sequence, then

3(a + 1)(b + 1) + (a + 4)(b + 1) = 2(a + 2)(b + 2) =⇒ (4a + 7)(b + 1) = (2a + 4)(b + 2),

so 4ab + 7b + 4a + 7 = 2ab + 4b + 4a + 8 =⇒ 2ab + 3b = 1, which is impossible because b is a nonnegative integer.

(Indeed, 2ab + 3b = 0 when b = 0, and 2ab + 3b > 1 when b > 0.) Thus a > 0 and b > 0, so 2ab + 3b > 1. If
(a + 2)(b + 2) is the least term in the sequence, then

3(a + 1)(b + 1) + (a + 2)(b + 2) = 2(a + 4)(b + 1) =⇒ (a + 2)(b + 2) = (5 − a)(b + 1),

so a ≤ 5. The only nonnegative solution is (a, b) = (1, 2), and because gcd(K, 30) = 1, the least element of S
is n = 50 · 1, and the second least element of S is n = 50 · 7 = 350.

Problem 2. The domain of the function f (x) = log6 (log5 (log4 (log3 (log2 (x))))) is the set of x such that x > A,
where A is an integer. Compute the number of digits in A.

Solution 2. The domain of the function log6 (t) is t > 0, so it follows that log5 (log4 (log3 (log2 (x)))) > 0. Thus

log4 (log3 (log2 (x))) > 1

=⇒ log3 (log2 (x)) > 4
=⇒ log2 (x) > 34
4
=⇒ x > 23 .

Hence A = 281 . To count the number of digits in A, use common logarithms. Note 81
that log(2 ) = 81 log(2),
and log(10) − log(5) = log(2), so the problem is to estimate 81 log(10) − log(5) ≈ 81(1 − 0.7) = 24.3. Thus
281 has 25 digits.

Problem 3. Eddy and Sarah are at an ARML coaches dinner together, and the menu has 7 different dishes.
Compute the number of ways that Eddy and Sarah can order dinner so that each of them orders at least one
dish, no dish can be ordered twice, and no dish is ordered by both Eddy and Sarah.

Solution 3. If the two coaches did not have to order at least one dish, then there would be 37 = 2187 possibilities.
It suffices to subtract the number of cases in which at least one of the two coaches orders no dish. If Eddy
orders no dishes, then there are 27 = 128 possibilities. Similarly, there are 27 ways for Sarah to order no

ARML encourages the reproduction of our contest problems for non-commercial, educational purposes. 35
Commercial usage of ARML problems without permission and posting entire contests or contest books are prohibited.
dishes, and only 1 way for both coaches to order no dishes. By the Inclusion-Exclusion Principle, there are
2187 − 2 · 128 + 1 = 1932 ways for Eddy and Sarah to order dinner subject to the given constraints.

Alternate Solution: Let S(n, k) be the number of ways to partition a set of n elements into k nonempty
subsets (the numbers S(n, k) are known as the Stirling numbers of the second kind). If every dish is assigned
to the two coaches, the 7 dishes can be partitioned into 2 nonempty groups, and there are 2! ways to order
the so there are 2! · S(7, 2) possibilities. If there are some dishes not assigned to either coach, the 7 dishes
can be partitioned into 3 nonempty groups, and there are 3! ways to order the groups, so there are 3! · S(7, 3)
possibilities. The values of S(7, 2) and S(7, 3) can be computed to be 63 and 301, respectively, so the overall
count is 2 · 63 + 6 · 301 = 1932.

36 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.