JMO 2025 Solution Notes

Evan Chen《陳誼廷》
27 March 2025

This is a compilation of solutions for the 2025 JMO. The ideas of the
solution are a mix of my own work, the solutions provided by the competition
organizers, and solutions found by the community. However, all the writing
is maintained by me.
These notes will tend to be a bit more advanced and terse than the “official”
solutions from the organizers. In particular, if a theorem or technique is not
known to beginners but is still considered “standard”, then I often prefer to
use this theory anyways, rather than try to work around or conceal it. For
example, in geometry problems I typically use directed angles without further
comment, rather than awkwardly work around configuration issues. Similarly,
sentences like “let R denote the set of real numbers” are typically omitted
entirely.
Corrections and comments are welcome!

Contents
0 Problems 2

1 Solutions to Day 1 3
1.1 JMO 2025/1, proposed by John Berman . . . . . . . . . . . . . . . . . . . 3
1.2 JMO 2025/2, proposed by John Berman . . . . . . . . . . . . . . . . . . . 4
1.3 JMO 2025/3, proposed by Wilbert Chu . . . . . . . . . . . . . . . . . . . 5

2 Solutions to Day 2 8
2.1 JMO 2025/4, proposed by Hung-Hsun Yu . . . . . . . . . . . . . . . . . . 8
2.2 JMO 2025/5, proposed by Carl Schildkraut . . . . . . . . . . . . . . . . . 9
2.3 JMO 2025/6, proposed by Enrique Treviño . . . . . . . . . . . . . . . . . 10

1
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§0 Problems
1. Prove that if f : Z → Z is any function, then there are infinitely many integers c
such that the function g(x) = f (x) + cx is not a bijection.

2. Fix positive integers k and d. Prove that for all sufficiently large odd positive
integers n, the digits of the base-2n representation of nk are all greater than d.

3. Let m and n be positive integers, and let R be a 2m × 2n grid of unit squares. A

domino is a 1 × 2 or 2 × 1 rectangle. An up-right path is a path from the lower-left
corner of R to the upper-right corner of R formed by exactly 2m + 2n edges of the
grid squares.
In terms of m and n, find the number of up-right paths that divide R into two
subsets (possibly empty), each of which can be tiled with dominoes.

4. Let n be a positive integer, and let a0 ≥ a1 ≥ · · · ≥ an ≥ 0 be integers. Prove that

n    
X ai 1 a0 + a1 + · · · + an
i ≤ .
2 2 2
i=0

5. Let H be the orthocenter of an acute triangle ABC, let F be the foot of the altitude
from C to AB, and let P be the reflection of H across BC. Suppose that the
circumcircle of triangle AF P intersects line BC at two distinct points X and Y .
Prove that CX = CY .

6. Let S be a set of integers with the following properties:

• {1, 2, . . . , 2025} ⊆ S.
• If a, b ∈ S and gcd(a, b) = 1, then ab ∈ S.
• If for some s ∈ S, s + 1 is composite, then all positive divisors of s + 1 are in
S.
Prove that S contains all positive integers.

2
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§1 Solutions to Day 1
§1.1 JMO 2025/1, proposed by John Berman
Available online at https://aops.com/community/p34326771.

Problem statement

Prove that if f : Z → Z is any function, then there are infinitely many integers c
such that the function g(x) = f (x) + cx is not a bijection.

Assume for contradiction that there exists a finite “bad” set S such that f (x) + cx is
bijective for all c ∈
/ S.
The first observation is basically that given just f (0) and f (1), or any two consecutive
f -values, we can already find bad values of c.

Claim — The numbers f (0) − f (1), f (1) − f (2), f (2) − f (3), . . . are all in S. That
is, consecutive differences of f only take finitely many values.

Proof. To see that f (0) − f (1) ∈ S, just note that

x 7→ f (x) + (f (0) − f (1)) · x

is obviously not a bijective function, since it takes the same values at x = 0 and x = 1,
namely f (0). The same is true for other elements.

In particular, suppose M > maxs∈S |s|. Then the function

g(x) = f (x) + (M + 100)x

is supposed to be a bijection (because M + 100 ∈

/ S), and yet g(x + 1) − g(x) > 100 for
all x, which is a contradiction.

3
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§1.2 JMO 2025/2, proposed by John Berman

Available online at https://aops.com/community/p34326777.

Problem statement

Fix positive integers k and d. Prove that for all sufficiently large odd positive integers
n, the digits of the base-2n representation of nk are all greater than d.

The problem actually doesn’t have much to do with digits: the idea is to pick any length
` ≤ k, and look at the rightmost ` digits of nk ; that is, the remainder upon division by
(2n)` . We compute it exactly:

Claim — Let n ≥ 1 be an odd integer, and k ≥ ` ≥ 1 integers. Then

nk mod (2n)i = c(k, `) · ni

for some odd integer 1 ≤ c(k, `) ≤ 2` − 1.

Proof. This follows directly by the Chinese remainder theorem, with c(k, `) being the
residue class of nk−i (mod 2` ) (which makes sense because n was odd).

In particular, for the `th digit from the right to be greater than d, it would be enough
that
c(k, `) · n` ≥ (d + 1) · (2n)`−1 .
But this inequality holds whenever n ≥ (d + 1) · 2`−1 .
Putting this together by varying `, we find that for all odd

n ≥ (d + 1) · 2k−1

we have that

• nk has k digits in base-2n; and

• for each ` = 1, . . . , k, the `th digit from the right is at least d + 1

so the problem is solved.

Remark. Note it doesn’t really matter that c(k, i) is odd per se; we only need that c(k, i) ≥ 1.

4
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§1.3 JMO 2025/3, proposed by Wilbert Chu

Available online at https://aops.com/community/p34326818.

Problem statement

Let m and n be positive integers, and let R be a 2m × 2n grid of unit squares. A

domino is a 1 × 2 or 2 × 1 rectangle. An up-right path is a path from the lower-left
corner of R to the upper-right corner of R formed by exactly 2m + 2n edges of the
grid squares.
In terms of m and n, find the number of up-right paths that divide R into two
subsets (possibly empty), each of which can be tiled with dominoes.

Apply the usual black/white checkerboard coloring. We define a staircase to be the below
above an up-right path (equivalently, a Young diagram rotated 180◦ ).
The proof is composed of two steps. One is rewriting the “domino-tileable” condition
to one that is actually usable; the other is the counting part.

¶ Determining when a staircase can be domino-tileable. It turns out the obvious

guess is right.

Lemma
A staircase can be tiled with dominoes if and only if the number of black and white
cells in it is equal.

Proof. It’s obvious equal black and white cells is necessary; we prove it’s sufficient.
Assume the number of black and white cells is equal. The proof is by induction with
empty staircases being vacuous. We consider the following four cases (which could overlap,
pick one arbitrarily if so). Note that the latter two cases are just mirrors of the first two.

Case 1 If the rightmost two columns of the staircase have the same height, tile those
two columns.

Case 2 If some two consecutive columns of the staircase differ by more than two cells in
height, say the (i − 1)st and the ith , then tile the top two cells in the ith column
and rightwards.

Case 3 If the bottom two rows of the staircase have the same length, tile those two rows.

Case 4 If some two consecutive rows of the staircase differ by more than two cells in
length, say the (j − 1)st and the j th , then tile the leftmost two cells in the j th
column and below.

Needless to say, each induction step preserves that the number of black squares equals the
number of white squares. Hence the remaining cells can then be tiled by the induction
hypothesis. See the image below for an illustration of all four cases.

5
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

Case 1 Case 2

Case 2 Case 4

I claim that at least one of these four cases must apply unless the staircase is empty.
This could only occur if the heights of the staircase are 1, 2, . . . , k in that order. However,
in that case the number of black cells and white cells are obviously not equal (unless
k = 0), so this can never occur. This completes the induction.

¶ Counting. Back to the main problem. The point is that we need to find the number
of up-right paths for which the two resulting regions have equal numbers of black and
white squares; call such paths balanced. Since the overall 2m × 2n grid has 2mn black
and 2mn white squares, it’s sufficient for just the bottom-right staircase to have equal
black and white cells.
Then in general, an up-right path is characterized by integer sequences

0 ≤ h1 ≤ h2 ≤ · · · ≤ h2n ≤ 2m

corresponding to the heights of the staircase.

Claim — The (hi ) correspond to a balanced up-right path if and only if {i : hi ≡ 1

(mod 2)} has an equal number of even and odd indices.

Proof. WLOG let’s fix our coloring so that the bottom-left square is black. Then in the
ith column, for i = 1, . . . , 2n, has (i) one more black square than white if hi is odd and
i ≡ 1 (mod 2); (ii) one more white square than black if hi is odd and i ≡ 0 (mod 2); (iii)
equal black and white squares if hi is even. The conclusion follows immediately.

The trick is to instead define

ti := hi + i.
That is, we will instead describe the up-right paths by integer sequences ti with

1 ≤ t1 < t2 < · · · < t2n ≤ 2n + 2m.

Claim — The (ti ) correspond to a balanced up-right path if and only if the set
{t1 , . . . , t2n } has an equal number of even and odd elements.

6
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

Proof. Translating the four cases between the two notations gives the following table:

i even i odd i even i odd

hi ≡ 1 (mod 2) a b ⇐⇒ ti ≡ 1 (mod 2) a 2n − b
hi ≡ 0 (mod 2) 2n − a 2n − b ti ≡ 0 (mod 2) 2n − a b.

Then we have

balanced ⇐⇒ a = b ⇐⇒ a + (2n − b) = (2n − a) + b ⇐⇒ #{odd ti } = #{even ti }

as desired.

Hence we must count the number of 2m-element subsets of {1, 2, . . . , 2n + 2m} with
m even and m odd terms. Since the even and odd terms can be chosen separately, this
gives an answer of n+m .

2
m

7
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§2 Solutions to Day 2
§2.1 JMO 2025/4, proposed by Hung-Hsun Yu
Available online at https://aops.com/community/p34335897.

Problem statement

Let n be a positive integer, and let a0 ≥ a1 ≥ · · · ≥ an ≥ 0 be integers. Prove that

n    
X ai 1 a0 + a1 + · · · + an
i ≤ .
2 2 2
i=0

For n = 0 (which we permit) there is nothing to prove. Hence to prove by induction on

n, it would be sufficient to verify
     
an a0 + a1 + · · · + an a0 + a1 + · · · + an−1
2n ≤ − .
2 2 2

Rearranging the terms around, that’s equivalent to proving

⇐⇒ 2n(a2n − an ) ≤ a2n + an · (2(a0 + · · · + an−1 ) − 1)

⇐⇒ 0 ≤ 2an (a0 + · · · + an−1 − nan ) + an (an + 2n − 1).

However, the last line is obvious because min(a0 , . . . , an−1 ) ≥ an , and an ≥ 0.

Remark. The only equality case is when a0 ∈ {0, 1} and ai = 0 for i ≥ 1.

The bound in the problem is extremely loose and pretty much anything will work.

8
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§2.2 JMO 2025/5, proposed by Carl Schildkraut

Problem statement

Let H be the orthocenter of an acute triangle ABC, let F be the foot of the altitude
from C to AB, and let P be the reflection of H across BC. Suppose that the
circumcircle of triangle AF P intersects line BC at two distinct points X and Y .
Prove that CX = CY .

Let Q be the antipode of B.

Claim — AHQC is a parallelogram, and AP CQ is an isosceles trapezoid.

Proof. As AH ⊥ BC ⊥ CQ and CF ⊥ AB ⊥ AQ.

N Q

F
M
H O

B C

Let M be the midpoint of QC.

Claim — Point M is the circumcenter of 4AF P .

Proof. It’s clear that M A = M P from the isosceles trapezoid. As for M A = M F ,

let N denote the midpoint of AF ; then M N is a midline of the parallelogram, so
M N ⊥ AF .

Since CM ⊥ BC and M is the center of (AF P ), it follows CX = CY .

9
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

§2.3 JMO 2025/6, proposed by Enrique Treviño

Available online at https://aops.com/community/p34335894.

Problem statement

Let S be a set of integers with the following properties:

• {1, 2, . . . , 2025} ⊆ S.

• If a, b ∈ S and gcd(a, b) = 1, then ab ∈ S.

• If for some s ∈ S, s + 1 is composite, then all positive divisors of s + 1 are in S.

Prove that S contains all positive integers.

We prove by induction on N that S contains {1, . . . , N } with the base cases being
N = 1, . . . , 2025 already given.
For the inductive step, to show N + 1 ∈ S:

• If N + 1 is composite we’re already done from the third bullet.

• Otherwise, assume N + 1 = p ≥ 2025 is an (odd) prime number. We say a number

is good if the prime powers in its prime factorization are all less than p. Hence by
the second bullet (repeatedly), good numbers are in S. Now our proof is split into
three cases:
Case 1. Suppose neither p − 1 nor p + 1 is a power of 2 (but both are still even).
We claim that the number

s := p2 − 1 = (p − 1)(p + 1)

is good. Indeed, one of the numbers has only a single factor of 2, and the
other by hypothesis is not a power of 2 (but still even). So the largest power
of 2 dividing p2 − 1 is certainly less than p. And every other prime power
divides at most one of p − 1 and p + 1.
Hence s := p2 − 1 is good. As s + 1 = p2 , Case 1 is done.
Case 2. Suppose p + 1 is a power of 2; that is p = 2q − 1. Since p > 2025, we
assume q ≥ 11 is odd. First we contend that the number
  
s0 := 2q+1 − 1 = 2(q+1)/2 − 1 2(q+1)/2 + 1

is good. Indeed, this follows from the two factors being coprime and both less
than p. Hence s0 + 1 = 2q+1 is in S.
Thus, we again have

s := p2 − 1 = (p − 1)(p + 1) ∈ S

as we did in the previous case, because the largest power of 2 dividing p2 − 1

will be exactly 2q+1 which is known to be in S. And since s + 1 = p2 , Case 2
is done.

10
JMO 2025 Solution Notes web.evanchen.cc, updated 27 March 2025

Case 3. Finally suppose p − 1 is a power of 2; that is p = 22 + 1 is a Fermat prime.

e

Then in particular, p ≡ 2 (mod 3). Now observe that

s := 2p − 1 ≡ 0 (mod 3)

and moreover 2p − 1 is not a power of 3 (it would imply 22 +1 + 1 = 3k , which

e

is impossible for k ≥ 3 by Zsigmondy/Mihailescu/etc.). So s is good, and

since s + 1 = 2p, Case 3 is done.
Having finished all the cases, we conclude p ∈ S and the induction is done.

Remark. In fact just 2025 ∈ S is sufficient as a base case; however this requires a bit more
work to check. Here is how:
• From 2025 ∈ S we get 2026 = 2 · 1013, so 2, 1013 ∈ S.
• From 1013 + 1 = 1014 = 2 · 3 · 132 we get 3, 13 ∈ S.
• From 3 + 1 = 4 we get 4 ∈ S.
• 3 · 13 + 1 = 40 = 23 · 5 we get 5 ∈ S.
• Once {1, 2, . . . , 5} ⊆ S, the induction above actually works fine; that is, N ≤ 5 are
sufficient as base cases for the earlier cases to finish the rest of the problem. (Case 2
works once q ≥ 3, and Case 3 works once e ≥ 2.)
However, {1, 2, 3, 4} ⊆ S is not sufficient; for example S = {1, 2, 3, 4, 6, 12} satisfies all the
problem conditions.

11