The 2nd US Ersatz Math Olympiad

Solutions and Results

Evan Chen

24 January 2024
Contents
1 Summary 3
1.1 Overview of the 2nd USEMO . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Thanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Proposers of problems . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Reviewers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.3 Graders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Results 5
2.1 Top Scores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Special awards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Honorable mentions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.4 Distinction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3 Solutions to the Problems 7

3.1 Solution to USEMO1, proposed by David Altizio . . . . . . . . . . . . . . 7
3.2 Solution to USEMO2, proposed by Pitchayut Saengrungkongka . . . . . . 9
3.3 Solution to USEMO3, proposed by Anant Mudgal . . . . . . . . . . . . . 11
3.4 Solution to USEMO4, proposed by Borislav Kirilov and Galin Totev . . . 17
3.5 Solution to USEMO5, proposed by Ankan Bhattacharya . . . . . . . . . . 19
3.6 Solution to USEMO6, proposed by Pitchayut Saengrungkongka . . . . . . 21

4 Marking schemes 24
4.1 Rubric for USEMO1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2 Rubric for USEMO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.3 Rubric for USEMO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.4 Rubric for USEMO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.5 Rubric for USEMO5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.6 Rubric for USEMO6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

5 Statistics 28
5.1 Summary of scores for USEMO 2020 . . . . . . . . . . . . . . . . . . . . . 28
5.2 Problem statistics for USEMO 2020 . . . . . . . . . . . . . . . . . . . . . 28
5.3 Rankings for USEMO 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.4 Histogram for USEMO 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.5 Full stats for USEMO 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2
1 Summary
§1.1 Overview of the 2nd USEMO
The second USEMO was held on October 24 – 25, 2020. A total of 135 students took
part.
Overall, I am happy with the quality of all six of the chosen problems. The main
surprise on the paper was an unusually difficult problem 4; I think one can make the
case that problem 5 was similar in difficulty, so the order of these two problems on the
second day was rather arbitrary. So this is the usual lesson in why it is valuable to try
all problems on a given day.
Given the difficulty of the competition, solving any single problem is a fine achievement;
like last year I did not “water down” the exam despite the fact that it is a public event,
and instead probably made it too difficult.

I continue to get requests to open the USEMO to a broader audience, either by having
a multi-division contest or allowing international students to compete. As you all suspect,
the concern is a grading bottleneck; even this year, we had difficulties completing the
grading and resolving disagreements in the scores within a timely manner. Therefore, I
think it is unlikely that the 2021 USEMO (which will likely take place in fall 2021, by
which point the pandemic has hopefully ended) would be enlarged, but the possibility is
still on the table for future contests as the volunteer base gradually grows larger over
time. We will have to wait and see!
We hope everyone stays safe as this year draws to a close and look forward to the next
event.

§1.2 Thanks
I am once again grateful to many individuals who helped make this competition possible.
I would like to thank the Art of Problem Solving for offering the software and platform
for us to run the competition. Special thanks to Corinne who was my main contact this
time.

3
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§1.2.1 Proposers of problems

I gratefully acknowledge the receipt of 23 proposals from Anas Chentouf, Anant Mudgal,
Ankan Bhattacharya, Arnav Pati, Borislav Kirilov, David Altizio, Galin Totev, Hu Man
Keat, Jaedon Whyte, Jeffery Li, Konstantin Garov, Luke Robitaille, Nikolai Beluhov,
Pitchayut Saengrungkongka, Pulkit Agarwal, Valentio Iverson.

§1.2.2 Reviewers
I am indebted to the reviewers of the packet, namely Anant Mudgal, Andrew Gu, Ankan
Bhattacharya, Ashwin Sah, Krit Boonsiriseth, Mihir Singhal, Nikolai Beluhov, Sasha
Rudenko, Tristan Shin, Vincent Huang.

§1.2.3 Graders
Thanks to everyone who signed up to help grade the competition (even if you ended
up not being able to contribute during these difficult times): Aayam Mathur, Adam
Kelly, Ahmed Shaaban, Anant Mudgal, Anas Chentouf, Ankan Bhattacharya, Anubhab
Ghosal, Arifa Alam, Aritra Barua, Arman Raayatsanati, Aron Thomas, Bobby Shen,
Brandon Wang, Brian Reinhart, Carl Schildkraut, Daniel Naylor, Daniel Sheremeta,
David Schmitz, Dylan Dalida, Ejaife Ogheneobukome, Hadyn Tang, Hector Osuna,
Hu Man Keat, Ivan Borsenco, Jeffery Li, Jeffrey Kwan, Jit Wu Yap, Justin Hua, Kai
Wang, Kazi Aryan Amin, Krit Boonsiriseth, Le Duc Minh, Lim Jeck, Luke Robitaille,
Matija Delic, Michael Greenberg, Mihir Singhal, Milica Vugdelić, Minjae Kwon, RedPig,
Risto Atanasov, Rohan Goyal, Sasha Rudenko, Shashwat Kasliwal, Srijon Sarkar, Taes
Padhihary, Tahmid Hameem Chowdhury, Tamim Iqbal, Thomas Luo, Ting-Wei Chao,
Tristan Shin, Valentio Iverson, Xinke Guo-Xue, Yannick Yao, Zawadul Hoque.

4
2 Results
If you won one of the seven awards, please reach out to usemo@evanchen.cc to claim
your prize!

§2.1 Top Scores

Congratulations to the top three scorers, who win the right to propose problems to future
instances of USEMO.
1st place Noah Walsh (42 points)
2nd place Ankit Bisain (41 points)
3rd place Gopal Goel (33 points)

§2.2 Special awards

See the Rules for a description of how these are awarded. For the purposes of awarding
monetary prizes, ties are broken more or less arbitrarily by considering the presentation
of elegance of solutions (which is obviously subjective). When this occurs, the names of
tied students are noted as well.
Top female Sanjana Das (28 points)
Youth prize Ram Goel (28 points); tied with. . .
• Kevin Wu
Top day 1 Brandon Chen (21 points on Day 1)
Top day 2 Edward Yu (14 points on Day 2); tied with. . .
• Kevin Wu
• Rishabh Das
• Samuel Wang
• Sogand Kamani
• William Yue

§2.3 Honorable mentions

This year we award Honorable Mention to anyone scoring at least 26 points. The HM’s
are listed below in alphabetical order.
Brandon Chen
Kevin Wu
Ram Goel
Rishabh Das
Sanjana Das

5
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§2.4 Distinction
We award Distinction to anyone scoring at least 14 points (two fully solved problems).
The Distinction awards are listed below in alphabetical order.
Alex Hu
Amol Rama
Andrew Yuan
Benjamin Jeter
David Dong
Dennis Chen
Derek Liu
Edward Xiong
Edward Yu
Ethan Liu
Ethan Zhou
Jason Cheah
Jeffrey Chen
Justin Lee
Kevin Min
Matthew Chen
Maximus Lu
Maxwell Sun
Paul Hamrick
Ramyro Corrêa Aquines
Raymond Feng
Rich Wang
Robert Yang
Rowechen
Ryan Li
Samuel Wang
Serena An
Sogand Kamani
Warren Bei
William Yue

6
3 Solutions to the Problems
§3.1 Solution to USEMO1, proposed by David Altizio

Problem statement

Which positive integers can be written in the form

lcm(x, y) + lcm(y, z)
lcm(x, z)

for positive integers x, y, z?

Let k be the desired value, meaning

−k lcm(x, z) + lcm(x, y) + lcm(y, z) = 0.

Our claim is that the possible values are even integers.

Indeed, if k is even, it is enough to take (x, y, z) = (1, k/2, 1).
For the converse direction we present a few approaches.

¶ First approach using ν2 only. We are going to use the following fact:

Lemma
If u, v, w are nonzero integers with u + v + w = 0, then either

ν2 (u) > ν2 (v) = ν2 (w);

ν2 (v) > ν2 (w) = ν2 (u); or
ν2 (w) > ν2 (u) = ν2 (v).

Proof. Let’s assume WLOG that e = ν2 (w) is minimal. If both ν2 (u) and ν2 (v) are
strictly greater than e, then ν2 (u + v + w) = e which is impossible. So assume WLOG
again that ν2 (v) = ν2 (w) = e. Then

u = −(2e · odd + 2e · odd) = −2e · even

so ν2 (u) ≥ e + 1.

However, if we assume for contradiction that k is odd, then

ν2 (−k lcm(x, z)) = max(ν2 (x), ν2 (z))

ν2 (lcm(x, y)) = max(ν2 (x), ν2 (y))
ν2 (lcm(y, z)) = max(ν2 (y), ν2 (z)).

In particular, the largest two numbers among the three right-hand sides must be equal.
So by the lemma, there is no way the three numbers (−k lcm(x, z), lcm(x, y), lcm(y, z))
could have sum zero.

7
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

¶ Second approach using νp for general p. We’ll prove the following much stronger
claim (which will obviously imply k is even).

Claim — We must have lcm(x, z) | lcm(x, y) = lcm(y, z).

Proof. Take any prime p and look at three numbers νp (x), νp (y), νp (z). We’ll show that

max(νp (x), νp (z)) ≤ max(νp (x), νp (y)) = max(νp (y), νp (z)).

If νp (y) is the (non-strict) maximum, then the claim is obviously true.

If not, by symmetry assume WLOG that νp (x) is largest, so that νp (x) > νp (y) and
νp (x) ≥ νp (z). However, from the given equation, we now have νp (lcm(y, z)) ≥ νp (x).
This can only occur if νp (z) = νp (x). So the claim is true in this case too.

¶ Third approach without taking primes (by circlethm). By scaling, we may as well
assume gcd(x, y, z) = 1.
Let dxy = gcd(x, y), etc. Now note that gcd(dxy , dxz ) = 1, and cyclically! This allows
us to write the following decomposition:

x = dxy dxz a
y = dxy dyz b
z = dxz dyz c.

We also have gcd(a, b) = gcd(b, c) = gcd(c, a) = 1 now.

Now, we have

lcm(x, y) = dxy dxz dyz ab

lcm(y, z) = dxy dxz dyz bc
lcm(x, z) = dxy dxz dyz ac

and so substituting this in to the equation gives

 
1 1
k =b· + .
a c

For a, b, c coprime this can only be an integer if a = c, so k = 2b.

Remark. From a = c = 1, the third approach also gets the nice result that lcm(x, y) =
lcm(y, z) in the original equation.

8
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§3.2 Solution to USEMO2, proposed by Pitchayut

Saengrungkongka

Problem statement

Calvin and Hobbes play a game. First, Hobbes picks a family F of subsets of
{1, 2, . . . , 2020}, known to both players. Then, Calvin and Hobbes take turns
choosing a number from {1, 2, . . . , 2020} which is not already chosen, with Calvin
going first, until all numbers are taken (i.e., each player has 1010 numbers). Calvin
wins if he has chosen all the elements of some member of F, otherwise Hobbes wins.
What is the largest possible size of a family F that Hobbes could pick while still
having a winning strategy?

The answer is 41010 − 31010 . In general, if 2020 is replaced by 2n, the answer is 4n − 3n .

¶ Construction. The construction is obtained as follows: pair up the numbers as {1, 2},
{3, 4}, . . . , {2019, 2020}. Whenever Calvin picks a numbers from one pair, Hobbes elects
to pick the other number. Then Calvin can never obtain a subset which has both numbers
from one pair. There are indeed 22n − 3n subsets with this property, so this maximum is
achieved.

¶ Bound. The main claim is the following.

Claim —
Fix a strategy for Hobbes and an integer 0 ≤ k ≤ n. Then there are at
least nk 2k sets with k numbers that Calvin can obtain after his kth turn.

Proof, due to Andrew Gu. The number of ways that Calvin can choose his first k moves
is
2n · (2n − 2) · (2n − 4) · . . . (2n − 2(k − 1)).
But each k-element set can be obtained in this way in at most k! ways (based on what
order its numbers were taken). So we get a lower bound of
 
2n · (2n − 2) · (2n − 4) · . . . (2n − 2(k − 1)) k n
=2 .
k! k

Thus by summing k = 0, . . . , n the family S is missing at least nk=0 2k nk = (1+2)n = 3n

P

subsets, as desired.

¶ Alternate proof of bound. Fix a strategy for Hobbes, as before. We proceed by

induction on n to show there are at least 3n missing sets (where a “missing set”, like in
the previous proof, is a set that Calvin can necessarily reach). Suppose that if Calvin
picks 1 then Hobbes picks 2. Then the induction hypothesis on the remaining game gives
that:

• there are 3n−1 missing sets that contain 1 but not 2;

• there are also 3n−1 missing sets that contain neither 1 nor 2.

9
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

• But imagining Calvin picking 2 first instead, applying the induction hypothesis
again we find that there are 3n−1 missing sets which contain 2.

These categories are mutually exclusive, so we find there are at least 3n missing sets, as
needed.

10
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§3.3 Solution to USEMO3, proposed by Anant Mudgal

Problem statement

Let ABC be an acute triangle with circumcenter O and orthocenter H. Let Γ denote
the circumcircle of triangle ABC, and N the midpoint of OH. The tangents to Γ at
B and C, and the line through H perpendicular to line AN , determine a triangle
whose circumcircle we denote by ωA . Define ωB and ωC similarly.
Prove that the common chords of ωA , ωB , and ωC are concurrent on line OH.

We begin by introducing several notations. The orthic triangle is denoted DEF and the
tangential triangle is denoted Ta Tb Tc . The reflections of H across the sides are denoted
Ha , Hb , Hc . We also define the crucial points P and Q as the poles of Hc B and Hb C
with respect to Γ.
The solution, based on the independent solutions found by Anant Mudgal and Nikolai
Beluhov, hinges on two central claims: that ωA is the circumcircle of 4Ta P Q, and that
EF is the radical axis of Γ and ωA . We prove these two claims in turn.
Tc

Hc A

F Tb
Hb
P E
H
Q
N
O

B D C

Ha

Ta

11
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Claim (Characterization of ωA ) — Line P Q passes through H and is perpendicular

to AN .

Proof. The fact that H lies on line P Q is immediate by Brokard’s theorem.

Showing the perpendicularity is the main part. Denote by B 0 and C 0 the antipodes of
B and C on Γ. Also, define L = Hc C 0 ∩ Hb B 0 and K = BHc ∩ CHb , as shown.
K

L
Tc

Hc A

F Tb
Hb
P E
C′ H B′
Q
N
O

B D C

Ha

Ta

We observe that:

• We have OK ⊥ P Q since K is the pole of line P Q (again by Brokard).

• The points O, K, L are collinear by Pascal’s theorem on BHc C 0 CHb B 0 .

• The point L is seen to be the reflection of H across A, so it follows AN k OL by a

2 -factor homothety at H.
1

Putting these three observations together completes the first claim.

12
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Remark (First claim is faster with complex numbers). It is also straightforward to prove
the first claim by using complex numbers. Indeed, in the usual setup, we have that the
intersection of the tangents at B and Hc is given explicitly by

2b · − ab


c 2ab
p= =
b − ab
c
a −c

and one explicitly checks p − (a + b + c) ⊥ (b + c − a), as needed.

Claim (Radical axis of ωA and Γ) — Line EF coincides with the radical axis of ωA
and Γ.

Proof. Let lines EF and Ta Tc meet at Z. It suffices to show Z lies on the radical axis,
and then repeat the argument on the other side.
Tc

Hc A
Z

F Tb
P E
H
Y
Q
O

B X D C

Ta

Since ]F BZ = ]ABZ = ]BCA = ]EF A = ]ZF B, it follows ZB = ZF . We

introduce two other points X and Y on the perpendicular bisector of BF : they are the
midpoints of BC and BHc .
Since OX · OTa = OB 2 = OY · OP , it follows that XY P Ta is cyclic. Then

ZP · ZTa = ZX · ZY = ZB 2

with the last equality since the circumcircle of 4BXY is tangent to Γ (by a 12 -homothety
at B). So the proof of the claim is complete.

13
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Finally, we are ready to finish the problem.

Claim — Line DTa coincides with the radical axis of ωB and ωC .

Proof. The point D already coincides with the radical axis because it is the radical center
of Γ, ωB and ωC . As for the point Ta , we let the tangent to Γ at Ha meet Ta Tc at U and
V ; by the first claim, these lie on ωC and ωB respectively.

Tc

F Tb
E
H

B D C

V
Ha
U

Ta

We need to show Ta U · Ta Tc = Ta V · Ta Tb .
But U V Tb Tc is apparently cyclic: the sides Tb Tc and U V are reflections across a
line perpendicular to AHa , while the sides U Tc and V Tb are reflections across a line
perpendicular to BC. So this is true.

Now since 4DEF and 4Ta Tb Tc are homothetic (their opposite sides are parallel), and
their incenters are respectively H and O, the problem is solved.

Remark (Barycentric approaches with respect to 4Ta Tb Tc ). Because the first claim is so
explicit, it is possible to calculate the length of the segment P B. This opens the possibility
of using barycentric coordinates with respect to the reference triangle Ta Tb Tc , and in fact
some contestants were able to complete this approach. Writing a = Tb Tc , b = Tc Ta , c = Ta Tb
one can show that the radical center is the point
 
a b c
: :
s−a s−b s−c

which is checked to be collinear with the circumcenter and incenter of 4Ta Tb Tc .

14
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

¶ Alternate inversion approach replacing the last two claims, by Serena An. After
finding P and Q, it’s also possible to solve the problem by using inversion. This eliminates
the need to identify line EF as the radical axis of ωA and Γ.
Inverted points are denoted with •∗ as usual, but we will only need two points: P ∗ , the
midpoint of BHc , and Q∗ , the midpoint of CHb . Now, let Ma = Ta∗ denote the midpoint
BC and let K be a point on ray HA with
3
KH = AH.
2
K

Hc A

F Hb
P
E
P∗ H
Q∗ Q

B Ma D C

Ha

Claim — The points D, P ∗ , Q∗ lie on the circle with diameter KMa .

Proof. Consider the homothety at H with scale factor 32 . It maps F to the midpoint of
F Hc and A to K, so we find KP ∗ is the perpendicular bisector of F Hc . As Ma P ∗ k CH,
we conclude ∠KP ∗ Ma = 90◦ .
Similarly ∠KQ∗ Ma = 90◦ . And ∠KDMa = 90◦ is given.

Claim — Line CH coincides with the radical axis of ωA

∗ and ω ∗ . In particular, the
B
circles ωA , ωB , (COHc ) are coaxial.

Proof. Letting Γ and Γ9 denote circumcircle and nine-point circle,

Pow(C, ωA
∗
) = CD · CMa = Pow(C, Γ9 )
Pow(H, ωA
∗
) = HM · HD = 32 HA · 12 HHa = 3
4 Pow(H, Γ).

The same calculation holds with ωB

∗ . Now line CH inverts to (COH ), as needed.
c

Since the circles (AOHa ), (BOHb ), (COHc ) have common radical axis equal to line OH,
the problem is solved.

15
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Remark (Nikolai Beluhov — generalization with variable ABC and fixed H). Take a fixed
circle Γ and a fixed point H in its interior. Then there exist infinitely many triangles ABC
with orthocenter H and circumcircle Γ. In fact, for every point A on Γ we get a unique pair
of B and C, determined as follows: Let line AH meet Γ again at SA , and take B and C to
be the intersection points of the perpendicular bisector of segment HSA with Γ.
With this framework, the following generalization is true: The radical center W of ωA , ωB ,
and ωC is the same point for all such triangles. Indeed, the Euler circle e of triangle ABC
is constant because it depends only on H and Γ. Let inversion relative to Γ map e onto
Ω. Then all three of Ta , Tb , and Tc lie on Ω, and so, by the solution, W is the homothety
center of e and Ω. (We take the homothety center with positive ratio when triangle ABC is
acute, and with negative ratio when it is obtuse. When triangle ABC is right-angled, Ω
degenerates into a straight line.)
Explicitly, let Γ be the unit circle and put H on the real axis at h. Then W is also on
the real axis, at 4h/(h2 + 3).
Furthermore, it turns out the power of W with respect to ωA , ωB , and ωC is constant as
well. This, however, is much tougher to prove; we are not aware of a purely geometric proof
at this time. Explicitly, in the setting above where Γ is the unit circle and H is on the real
axis at h, the power of W with respect to ωA , ωB , and ωC equals 12(h2 − 1)/(h2 + 3)2 .

16
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§3.4 Solution to USEMO4, proposed by Borislav Kirilov and

Galin Totev
Problem statement

A function f from the set of positive real numbers to itself satisfies

f (x + f (y) + xy) = xf (y) + f (x + y)

for all positive real numbers x and y. Prove that f (x) = x for all positive real
numbers x.

We present two solutions.

¶ First solution (Nikolai Beluhov). We first begin with the following observation.

Claim — We must have f (y) ≥ y for all y > 0.

Proof. Otherwise, choose 0 < x < 1 satisfying that f (y) = (1 − x) · y. Then plugging
this P (x, y) gives xf (y) = 0, contradiction.

Now, we make the substitution f (x) = x + g(x), so that g is a function R>0 → R≥0 .
The given function equation reads g(x + xy + (y + g(y))) + x + (y + g(y)) = (xy + xg(y)) +
(x + y + g(x + y)), or

g(x + y + xy + g(y)) = (x − 1)g(y) + g(x + y). (†)

We have to show that g is the zero function from (†).

Claim (Injectivity for nonzero outputs) — If g(a) = g(b) for a 6= b, then we must
actually have g(a) = g(b) = 0.

Proof. Setting (a, b) and (b, a) in (†) gives (a − 1)g(b) = (b − 1)g(a) which, since a − 1 6=
b − 1, forces g(a) = g(b) = 0.

Claim (g vanishes on (1, ∞)) — We have g(t) = 0 for t > 1.

Proof. If we set x = 1 in (†) we obtain that g(g(y) + 2y + 1) = g(1 + y). As the inputs
are obviously unequal, the previous claim gives g(1 + y) = 0 for all y > 0.

Now x = 2 in (†) to get g(y) = 0, as needed.

¶ Second solution (from authors). We start with the same opening of showing f (y) ≥ y,
defining f (x) = x + g(x), so g satisfies (†). Here is another proof that g ≡ 0 from (†).

Claim — If g is not the zero function, then for any constant C, we have g(t) > C
for sufficiently large t.

17
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Proof. In (†) fix y to be any input for which g(y) > 0. Then

g ((1 + y)x + (y + g(y))) ≥ (x − 1)g(y)

so for large x, we get the conclusion.

Remark. You could phrase the lemma succinctly as “limx→∞ g(x) = +∞”. But I personally
think it’s a bit confusing to do so because in practice we usually talk about limits of continuous
(or well-behaved) functions, so a statement like this would have the wrong connotations,
even if technically correct and shorter.

On the other hand, by choosing x = 1 and y = t − 1 for t > 1 in (†), we get

g(2t + g(z) − 1) = g(t)

and hence one can generate an infinite sequence of fixed points: start from t0 = 100, and
define tn = 2tn−1 + g(tn−1 ) − 2 > tn−1 + 98 for n ≥ 1 to get

g(t0 ) = g(t1 ) = g(t2 ) = · · ·

and since the ti are arbitrarily large, this produces a contradiction if g 6≡ 0.

18
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§3.5 Solution to USEMO5, proposed by Ankan Bhattacharya

Problem statement

The sides of a convex 200-gon A1 A2 . . . A200 are colored red and blue in an alternating
fashion. Suppose the extensions of the red sides determine a regular 100-gon, as do
the extensions of the blue sides.
Prove that the 50 diagonals A1 A101 , A3 A103 , . . . , A99 A199 are concurrent.

We present a diagram (with 100 replaced by 6, for simplicity).

B5 R4
X4
R5
B4

X5 X3

B6 R3
O
R6
B3
X6
X2

B1 R2
R1 X1
B2

Let B1 . . . B100 and R1 . . . R100 be the regular 100-gons (oriented counterclockwise), and
define Xi = A2i+1 = Bi Bi+1 ∩ Ri Ri+1 for all i, where all indices are taken modulo 100.
We wish to show that X1 X51 , . . . , X50 X100 are concurrent.
We now present two approaches.

¶ First approach (by spiral similarity). Let O be the spiral center taking B1 . . . B100 →
R1 . . . R100 (it exists since the 100-gons are not homothetic). We claim that O is the
desired concurrency point.

Claim — ]Xi OXi+1 = π

50 for all i.

Proof. Since 4OBi Bi+1 ∼ 4ORi Ri+1 , we have 4OBi Ri ∼ 4OBi+1 Ri+1 , so O, Xi ,
+ +

Bi+1 , Ri+1 are concyclic. Similarly O, Xi+1 , Bi+1 , Ri+1 are concyclic, so
π
]Xi OXi+1 = ]Xi Bi+1 Xi+1 =
50
as wanted.

It immediately follows that O lies on all 50 diagonals Xi Xi+50 , as desired.

19
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

¶ Second approach (by complex numbers). Let ω be a primitive 100th root of unity.
We will impose complex coordinates so that Rk = ω k , while Bk = pω k + q, where m and
b are given constant complex numbers.
In general for |z| = 1, we will define f (z) as the intersection of the line through z and
ωz, and the line through pz + q and p · ωz + q.
ωz
pz + q

f (z)

p · ωz + q

In particular, Xk is f (ω k ).

Claim — There exist complex numbers a, b, c such that f (z) = a + bz + cz 2 , for

every |z| = 1.

Proof. Since f (z) and f (z)−q

p both lie on the chord joining z to ωz we have

z + ωz = f (z) + ωz 2 · f (z)
f (z) − q f (z) − q
z + ωz = + ωz 2 · .
p p

Subtracting the first equation from the p times the second to eliminate f (z), we get that
f (z) should be a degree-two polynomial in z (where p and q are fixed constants).

Claim — Let f (z) = a + bz + cz 2 as before. Then the locus of lines through f (z)
and f (−z), as |z| = 1 varies, passes through a fixed point.

Proof. By shifting we may assume a = 0, and by scaling we may assume b is real (i.e.
b = b). Then the point −c works, since

f (z) + c c + bz + cz 2
=
f (−z) + c c − bz + cz 2

is real — it obviously equals its own conjugate. (Alternatively, without the assumptions
a = 0 and b ∈ R, the fixed point is a − bc
b
.)

Remark (We know a priori the exact coefficients shouldn’t matter). In fact, the exact value
is
−ωqz 2 + (1 − p)(1 + ω)z − pp q
f (z) = .
1 − pp
Since p and q could be any complex numbers, the quantity c/b (which is all that matters for
concurrence) could be made to be equal to any value. For this reason, we know a priori the
exact coefficients should be irrelevant.

20
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§3.6 Solution to USEMO6, proposed by Pitchayut

Saengrungkongka

Problem statement

Prove that for every odd integer n > 1, there exist integers a, b > 0 such that, if we
let Q(x) = (x + a)2 + b, then the following conditions hold:

• we have gcd(a, n) = gcd(b, n) = 1;

• the number Q(0) is divisible by n; and

• the numbers Q(1), Q(2), Q(3), . . . each have a prime factor not dividing n.

Let p1 < p2 < · · · < pm denote the odd primes dividing n and call these primes small.
The construction is based on the following idea:

Claim — For each i = 1, . . . , m we can choose a prime qi ≡ 1 (mod 4) such that

−1 if j = i
  (
pj
=
qi +1 otherwise.

Proof. Fix i. By quadratic reciprocity, it suffices that qi ≡ 1 (mod 4) and that qi is a

certain nonzero quadratic residue (or not) modulo pj for j 6= i.
By Chinese remainder theorem, this is a single modular condition, so Dirichlet theorem
implies such primes exist.

We commit now to the choice

b = kq1 q2 . . . qm

where k ≥ 1 is an integer (its value does not affect the following claim).

Claim (Main argument) — For this b, there are only finitely many integers X
satisfying the equation
X 2 + b = pe11 . . . pemm (♠)
where ei are some nonnegative integers (i.e. X 2 + b has only small prime factors).

Proof. In (♠) the RHS is a quadratic residue modulo b. For any i > 0, modulo qi we find
Y  pei 
+1 = i
= (−1)ei
qi
j

so ei must be even. This holds for every i though! In other words all ei are even.
Hence (♠) gives solutions to X 2 + b = Y 2 , which obviously has only finitely many
solutions.

We now commit to choosing any k ≥ 1 such that

1
k≡− (mod n)
q1 q2 . . . q m

21
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

which in particular means gcd(k, n) = 1. Now as long as a ≡ 1 (mod n), we have

Q(0) ≡ 0 (mod n), as needed. All that remains is to take a satisfying the second claim
larger than any of the finitely many bad integers in the fist claim.

Remark (Motivational comments from Nikolai Beluhov). The solution I ended up with is
not particularly long or complicated, but it was absurdly difficult to find. The main issue I
think is that there is nothing in the problem to latch onto; no obvious place from which you
can start unspooling the yarn. So what I did was throw an awful lot of different strategies
at it until one stuck.
Eventually, what led me to the solution was something like this. I decided to focus on
the simplest nontrivial case, when n contains just two primes. I spent some time thinking
about this, and then at some point I remembered that in similar Diophantine equations
I’ve seen before, like 2x + 3y = z 2 or 3x + 4y = 5z , the main trick is first of all to prove
that the exponents are even. After that, we can rearrange and factor a difference of squares.
This idea turned out to be fairly straightforward to implement, and this is how I found the
solution above.

Remark (The problem is OK with n even). The problem works equally well for n even, but
the modifications are both straightforward and annoying, so we imposed n odd to reduce
the time taken in solving this problem under exam conditions.
On the other hand, for odd n, one finds that a simplified approach is possible where
one proves the main argument by choosing b ≡ 2 (mod 4) and then using the quadratic
reciprocity argument to force the right-hand side of (♠) to be 1 (mod 4). In this case, there
are no integers X at all satisfying (♠), and the “sufficiently large” leverage provided by the
choice of a is not needed.

Remark (On the choice of conditions). The equation (♠), and the goal to show it has finitely
many solutions (or no solutions), is the heart of the problem. But the other conditions have
been carefully chosen to prevent two “trivial” constructions to this.
If the condition that gcd(a, n) = gcd(b, n) = 1 or n | Q(0) is dropped, the problem
becomes much easier because one can simply ensure that νp (Q(x)) is bounded for all p | n,
by taking b = n (or b = rad n, etc.). However, these two conditions jointly together ensure
that νp (Q(x)) may be unbounded, by a Hensel-type argument.
If b < 0 is permitted, an easier approach to make sure that Q(x) factors as the product
of two polynomials by requiring b to be the negative of a perfect square. Several easier
approaches become possible in this situation. For example, one can try to use Kobayashi’s
theorem to generate the value of a after ensuring the first two conditions are true.

Remark (Author remarks on generalization). In general, any b satisfying gcd(b, n) = 1

should still have finitely many solutions to (♠). The author comments that this would
√ be a
statement of Kobayashi’s theorem in the ring of integers of the quadratic field Q( −b).
A confirmation this (and much more) is indeed true is given by user Loppukilpailija
at https://aops.com/community/p18545077. An excerpt from this post goes:
If there are infinitely many x which do not have this property, by the pigeonhole
principle there is some integer c with rad(c) | n and νp (c) ≤ 2 for all primes
p such that the equation Q(x) = cy 3 has infinitely many solutions. This
rearranges to the polynomial cy 3 − b attaining infinitely many square values.
Since b 6= 0, the roots of cy 3 − b are simple. This contradicts Theorem 3 of the
paper starting at page 381 of https://www.mathunion.org/fileadmin/ICM/
Proceedings/ICM1978.1/ICM1978.1.ocr.pdf.
In general, the results presented in the linked article are, together with elemen-

22
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

tary arguments, enough to characterize all polynomials which attain infinitely

many perfect powers as their values. (Exercise!)

23
4 Marking schemes
§4.1 Rubric for USEMO1
ν2 solutions
• 0 points for stating that even integers are the only solutions

• 1 point for the construction for even integers

• 1 points for considering ν2 of x, y and z and resolving at least one substantial case,
such as ν2 (y) being maximal (additive). This point can also be awarded if done
with a general prime p instead of 2.

• 7 points for a complete solution

Factoring solutions
• 0 points for stating that even integers are the only solutions

• 0 points for stating WLOG gcd(x, y, z) = 1

• 1 point for the construction for even integers

• 1 points for factoring pairwise greatest common divisors (additive)

• 7 points for a complete solution

General νp solutions
• 0 points for stating that even integers are the only solutions

• 1 point for the construction for even integers

• 1 point for stating lcm(x, z) | lcm(x, y) = lcm(y, z) (additive)

• 7 points for a complete solution

§4.2 Rubric for USEMO2

• 1 point for a correct construction, with the correct strategy for Hobbes which
results in the correct answer (but they need not correctly calculate the answer, as
long as the strategy is right).

• 2 points for claiming a statement (no proof needed) of the form “for each integer
k, Calvin can ensure a victory if the number of sets of size k is at most —” (or “the
number of missing sets is at least —”) and the bound listed is the correct (tight)
bound.

• 3 points for both of the above items.

• 5 points for a correct proof of the upper bound

24
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

• 7 points for a fully correct solution

• -2 points for solutions which would satisfy rubric items (3) or (4), except that the
expressions given in terms of n and k are missing or not bounded correctly.
For example: a solution might have a claim that Hobbes needs to be missing some
number of sets of size k, and show how to use induction to show a bound on n, k
based on n–1, k–1, but not use the correct quantities to do the bounding

• -1 point for a wrong answer, only for solutions that would otherwise receive a 7.
Deduct no points if an answer is unsimplified but correct.

Deductions for other minor/major errors are as usual. None of the items above are
additive (though see the third rubric item).
In general, for solutions that do not fit a rubric item but which you feel deserves partial
credit, use your judgment to assign partial credit.

§4.3 Rubric for USEMO3

As usual, incomplete computational approaches earn partial credits only based on the
amount of synthetic progress which is made.
No points are awarded for just drawing a diagram or simple observations.
Follow the notation in the typeset official solution. The following rubric items are
totally additive:

(a) 1 point for proving that P and Q are the poles of lines BHc and CHb .

(b) 1 point for proving that Ta is on the radical axis of ωB and ωC . This point can
be awarded if the proof is conditional on some reasonable description of P and Q,
such as (a).

(c) 2 points for proving that D is on the radical axis of ωB and ωC . This point can
be awarded if the proof is conditional on some reasonable description of P and Q,
such as (a).

(d) 0 points for commenting that the homothety center of Ta Tb Tc and DEF lies on
the line OH.

The four rubric items above, when combined, give a perfect solution worth 7.
If none of the items above are earned: the following rubric item (not additive) is
possible:

• 1 point for both claiming that P and Q are the poles of BHc and CHb , and that
the radical axis of ωB and ωC is exactly DTa .

§4.4 Rubric for USEMO4

General remarks
• Unlike most functional equations, this one doesn’t really have that many steps of
“partial progress.” As such, the steps below are mainly intermediate claims, not
specific equations.

25
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

• This problem does not ask the contestant to find all functions f that satisfy the
given property. As a result, the fact that f (x) = x satisfies the property does not
need to be stated nor proven.

• The results below may come in many different forms, and may not be stated
explicitly. As such, care must be taken to determine whether a contestant has
found any of the claims below.

• Most solutions follow the following general path:

(a) Show f (x) ≥ x, and define g(x) = f (x) − x.
(b) Find some property that shows g(x) must be small or fixed in some places.
(c) Find some property that shows g(x) must be large in some places.
(d) Combine the two to finish.
The first claim is worth 1 point, and points (b) and (c) are meant to each be worth
3 points (non-additively), with a full solution (including finish) worth 7, of course.
Solutions that make enough progress that a “standard finish” is applicable should
be worth 5 points (in essence, points for both (b)-like and (c)-like things should be
given). The rubric below attempts to codify most (b)-like and (c)-like progress we
expect to see, but it is certainly possible that other progress exists.

Rubric items
None of these items are additive.

• 0 points for solving the equation over some domain that is not the positive reals
(e.g. reals, nonnegative reals)

• 1 point for showing f (x) ≥ x for all x.

• 1 point for showing that, if f (x) = x for some x, then f (x) = x for all x.

• 3 points for showing that, if f (x) + y = f (y) + x, then f (x) = x and f (y) = y.

• 3 points for finding distinct values of x and y for which f (x) + y = f (y) + x.

• 3 points for showing that f (x) − x is eventually greater than any specified real.

• 5 points for finding (or showing existence of) any value of x for which f (x) = x.

• 6 points for a complete solution with a minor error that does not affect the solution.

• 7 points for a complete solution.

§4.5 Rubric for USEMO5

Most solutions are worth 0 or 7.

• 0 points for no progress, special cases, etc.

• 5-6 points for any tiny slip which the contestant could have easily repaired

• 7 points for a correct solution

For solutions which are not complete, the following items are additive:

26
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

• 1 point for considering the spiral similarity taking P1 ...P100 to Q1 ...Q100 AND
claiming that the center of the spiral similarity is the point of concurrency.

• 1 point for claiming that ∠Ri ORi+1 = π

50

• 1 point for proving that O, Ri , Pi+1 , Qi+1 is concyclic

• 1 point for further extending the above to proving that O, Ri , Ri+1 , Pi+1 , Qi+1
concyclic

There is no deduction for small configuration issues (such as not using directed angles)
or small typos (such as labelling points).
Usually, computational approaches which are not essentially completed are judged by
any geometric content and do not earn other marks. However, the following marks (not
additive with anything) are possible:

• Following the notation of the complex solution, 1 point for showing that the
intersection point is quadratic AND making the general claim that a + bz + cz 2 is
sufficient regardless of what the numbers a, b, c are.

§4.6 Rubric for USEMO6

In general, not much partial credit is expected for this problem.
The heart of the problem can be thought of as studying the equation

X 2 + b = pe11 . . . pekk

where p1 , . . . , pk are a fixed set of primes, and showing that the equation cannot hold
for all sufficiently large X.

• No points are given for steps related to the first two conditions, e.g. for the
Hensel’s-lemma type observation that ei may be unbounded. This is equivalent to
reducing to the case where n is squarefree.

• No points are given for taking specific moduli, e.g. taking the equation modulo 4.

• No points are given for special cases of k, such as k = 1.

 
• 1 point is awarded for the idea to select a prime q for which pi
q is known in
order to control the parity of the exponents ei .

• There is no deduction for quoting the theorems of Dirichlet or quadratic reciprocity,

or in general the quoting of any named theorems which the grader can indeed verify
exists.

• 1 point is deducted if the student fails to verify the Hensel argument, but their
construction holds anyways.
 
• 1 point is deducted if the student asserts control over pqi with no justification
whatsoever. We require the student to at least mention that quadratic reciprocity
is used to get a modular condition and then use Dirichlet. (This justification may
be very terse: even “by QR and Dirichlet” is accepted).

27
5 Statistics
This year the scoring data I received did not distinguish between blank papers and
those with a score of 0, but the statistics only include students with at least one submitted
file.

§5.1 Summary of scores for USEMO 2020

N 135 1st Q 6 Max 42
µ 9.77 Median 8 Top 3 33
σ 7.99 3rd Q 14 Top 12 22

§5.2 Problem statistics for USEMO 2020

P1 P2 P3 P4 P5 P6
0 19 69 127 77 106 131
1 8 44 1 30 3 1
2 9 0 2 1 1 0
3 0 3 0 4 1 0
4 0 0 0 0 0 0
5 4 0 1 1 4 0
6 4 1 0 2 1 1
7 91 18 4 20 19 2
Avg 5.24 1.37 0.28 1.49 1.24 0.16
QM 5.93 2.71 1.31 2.91 2.83 1.00
# 5+ 99 19 5 23 24 3
% 5+ % 73.3 % 14.1 % 3.7 % 17.0 % 17.8 % 2.2

28
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

§5.3 Rankings for USEMO 2020

Sc Num Cu Per Sc Num Cu Per Sc Num Cu Per
42 1 1 0.74% 28 4 7 5.19% 14 9 38 28.15%
41 1 2 1.48% 27 0 7 5.19% 13 1 39 28.89%
40 0 2 1.48% 26 1 8 5.93% 12 1 40 29.63%
39 0 2 1.48% 25 0 8 5.93% 11 6 46 34.07%
38 0 2 1.48% 24 0 8 5.93% 10 4 50 37.04%
37 0 2 1.48% 23 1 9 6.67% 9 8 58 42.96%
36 0 2 1.48% 22 5 14 10.37% 8 23 81 60.00%
35 0 2 1.48% 21 1 15 11.11% 7 18 99 73.33%
34 0 2 1.48% 20 0 15 11.11% 6 3 102 75.56%
33 1 3 2.22% 19 1 16 11.85% 5 1 103 76.30%
32 0 3 2.22% 18 0 16 11.85% 4 1 104 77.04%
31 0 3 2.22% 17 1 17 12.59% 3 3 107 79.26%
30 0 3 2.22% 16 0 17 12.59% 2 6 113 83.70%
29 0 3 2.22% 15 12 29 21.48% 1 8 121 89.63%
0 14 135 100.00%

§5.4 Histogram for USEMO 2020

23

18

14
12
9
8 8
6 6
5
4 4
3 3
1 1 1 1 1 1 1 1 1 1 1 1
1 2 3 4 5 6 8 9 0 1 2 3 5 6 7 8 9 0 2 3 4 5 6 7 9 0 1 2 3 4 6 7 8 9 0 1
0 7 14 21 28 35 42

§5.5 Full stats for USEMO 2020

Rank P1 P2 P3 P4 P5 P6 Σ
1. 7 7 7 7 7 7 42
2. 7 7 7 7 7 6 41
3. 7 7 5 7 7 0 33
4. 7 7 0 7 7 0 28
4. 7 7 0 7 7 0 28
4. 7 7 0 7 7 0 28
4. 7 7 0 0 7 7 28
8. 7 7 7 5 0 0 26
9. 7 7 2 7 0 0 23
10. 7 7 0 1 7 0 22
10. 7 7 0 1 7 0 22
10. 7 1 7 7 0 0 22
10. 7 1 0 7 7 0 22
10. 7 1 0 7 7 0 22

29
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Rank P1 P2 P3 P4 P5 P6 Σ
15. 7 7 0 0 7 0 21
16. 7 6 0 1 5 0 19
17. 7 3 0 7 0 0 17
18. 7 7 0 1 0 0 15
18. 7 7 0 1 0 0 15
18. 7 7 0 1 0 0 15
18. 7 1 0 7 0 0 15
18. 7 1 0 7 0 0 15
18. 7 1 0 7 0 0 15
18. 7 1 0 7 0 0 15
18. 7 1 0 7 0 0 15
18. 7 1 0 7 0 0 15
18. 7 1 0 7 0 0 15
18. 7 0 0 1 7 0 15
18. 6 3 0 0 6 0 15
30. 7 7 0 0 0 0 14
30. 7 7 0 0 0 0 14
30. 7 7 0 0 0 0 14
30. 7 1 0 6 0 0 14
30. 7 1 0 1 5 0 14
30. 7 0 0 0 7 0 14
30. 7 0 0 0 7 0 14
30. 0 0 0 7 7 0 14
30. 0 0 0 7 7 0 14
39. 7 0 0 1 5 0 13
40. 5 0 0 0 7 0 12
41. 7 3 0 1 0 0 11
41. 7 1 0 3 0 0 11
41. 7 1 0 3 0 0 11
41. 7 1 0 0 3 0 11
41. 7 0 0 3 1 0 11
41. 6 0 0 0 5 0 11
47. 7 1 2 0 0 0 10
47. 7 1 0 1 1 0 10
47. 7 0 1 0 2 0 10
47. 7 0 0 3 0 0 10
51. 7 1 0 1 0 0 9
51. 7 1 0 1 0 0 9
51. 7 1 0 1 0 0 9
51. 7 1 0 1 0 0 9
51. 7 1 0 1 0 0 9
51. 7 1 0 1 0 0 9
51. 7 1 0 0 1 0 9
51. 7 0 0 2 0 0 9
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8

30
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Rank P1 P2 P3 P4 P5 P6 Σ
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 1 0 0 0 0 8
59. 7 0 0 1 0 0 8
59. 7 0 0 1 0 0 8
59. 7 0 0 1 0 0 8
59. 7 0 0 1 0 0 8
59. 7 0 0 1 0 0 8
59. 7 0 0 1 0 0 8
59. 7 0 0 0 0 1 8
59. 0 0 0 1 7 0 8
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 7 0 0 0 0 0 7
82. 5 1 0 1 0 0 7
82. 1 0 0 6 0 0 7
100. 6 0 0 0 0 0 6
100. 6 0 0 0 0 0 6
100. 5 0 0 1 0 0 6
103. 5 0 0 0 0 0 5
104. 2 1 0 1 0 0 4
105. 2 1 0 0 0 0 3
105. 2 0 0 1 0 0 3
105. 2 0 0 1 0 0 3
108. 2 0 0 0 0 0 2
108. 2 0 0 0 0 0 2
108. 2 0 0 0 0 0 2
108. 2 0 0 0 0 0 2
108. 2 0 0 0 0 0 2

31
Evan Chen — 24 January 2024 The 2nd US Ersatz Math Olympiad

Rank P1 P2 P3 P4 P5 P6 Σ
108. 1 0 0 1 0 0 2
114. 1 0 0 0 0 0 1
114. 1 0 0 0 0 0 1
114. 1 0 0 0 0 0 1
114. 1 0 0 0 0 0 1
114. 1 0 0 0 0 0 1
114. 1 0 0 0 0 0 1
114. 0 1 0 0 0 0 1
114. 0 1 0 0 0 0 1
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0
122. 0 0 0 0 0 0 0

32