IMO 2010 Solution Notes

Evan Chen《陳誼廷》
15 December 2024

This is a compilation of solutions for the 2010 IMO. 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 IMO 2010/1, proposed by Pierre Bornsztein (FRA) . . . . . . . . . . . . . 3
1.2 IMO 2010/2, proposed by Tai Wai Ming and Wang Chongli (HKG) . . . . 4
1.3 IMO 2010/3, proposed by Gabriel Carroll (USA) . . . . . . . . . . . . . . 5

2 Solutions to Day 2 6
2.1 IMO 2010/4, proposed by Marcin Kuczma (POL) . . . . . . . . . . . . . . 6
2.2 IMO 2010/5, proposed by Netherlands . . . . . . . . . . . . . . . . . . . . 8
2.3 IMO 2010/6, proposed by Morteza Saghafiyan (IRN) . . . . . . . . . . . . 9

1
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§0 Problems
1. Find all functions f : R → R such that for all x, y ∈ R,

f (bxc y) = f (x) bf (y)c .

2. Let I be the incenter of a triangle ABC and let Γ be its circumcircle. Let line AI
intersect Γ again at D. Let E be a point on arc BDC
’ and F a point on side BC
such that
∠BAF = ∠CAE < 12 ∠BAC.
Finally, let G be the midpoint of IF . Prove that DG and EI intersect on Γ.

3. Find all functions g : Z>0 → Z>0 such that

(g(m) + n) (g(n) + m)

is always a perfect square.

4. Let P be a point interior to triangle ABC (with CA 6= CB). The lines AP , BP

and CP meet again its circumcircle Γ at K, L, M , respectively. The tangent line
at C to Γ meets the line AB at S. Show that from SC = SP follows M K = M L.

5. Each of the six boxes B1 , B2 , B3 , B4 , B5 , B6 initially contains one coin. The

following two types of operations are allowed:
a) Choose a non-empty box Bj , 1 ≤ j ≤ 5, remove one coin from Bj and add
two coins to Bj+1 ;
b) Choose a non-empty box Bk , 1 ≤ k ≤ 4, remove one coin from Bk and swap
the contents (possibly empty) of the boxes Bk+1 and Bk+2 .
Determine if there exists a finite sequence of operations of the allowed types, such
that the five boxes B1 , B2 , B3 , B4 , B5 become empty, while box B6 contains exactly
coins.
2010
20102010

6. Let a1 , a2 , a3 , . . . be a sequence of positive real numbers, and s be a positive integer,

such that
an = max{ak + an−k | 1 ≤ k ≤ n − 1} for all n > s.
Prove there exist positive integers ` ≤ s and N , such that

an = a` + an−` for all n ≥ N .

2
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§1 Solutions to Day 1
§1.1 IMO 2010/1, proposed by Pierre Bornsztein (FRA)
Available online at https://aops.com/community/p1935849.

Problem statement

Find all functions f : R → R such that for all x, y ∈ R,

f (bxc y) = f (x) bf (y)c .

The only solutions are f (x) ≡ c, where c = 0 or 1 ≤ c < 2. It’s easy to see these work.
Plug in x = 0 to get f (0) = f (0) bf (y)c, so either

1 ≤ f (y) < 2 ∀y or f (0) = 0

In the first situation, plug in y = 0 to get f (x) bf (0)c = f (0), thus f is constant. Thus
assume henceforth f (0) = 0.
Now set x = y = 1 to get
f (1) = f (1) bf (1)c
so either f (1) = 0 or 1 ≤ f (1) < 2. We split into cases:

• If f (1) = 0, pick x = 1 to get f (y) ≡ 0.

• If 1 ≤ f (1) < 2, then y = 1 gives

f (bxc) = f (x)

from y = 1,in particular f (x) = 0 for 0 ≤ x < 1. Choose (x, y) = 2, 12 to get

f (1) = f (2) f 12 = 0.



3
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§1.2 IMO 2010/2, proposed by Tai Wai Ming and Wang Chongli (HKG)
Available online at https://aops.com/community/p1935927.

Problem statement

Let I be the incenter of a triangle ABC and let Γ be its circumcircle. Let line AI
intersect Γ again at D. Let E be a point on arc BDC
’ and F a point on side BC
such that
∠BAF = ∠CAE < 12 ∠BAC.
Finally, let G be the midpoint of IF . Prove that DG and EI intersect on Γ.

Let EI meet Γ again at K. Then it suffices to show that KD bisects IF . Let AF meet
Γ again at H, so HE k BC. By Pascal theorem on

AHEKDD

we then obtain that P = AH ∩ KD lies on a line through I parallel to BC.

Let IA be the A-excenter, and set Q = IA F ∩ IP , and T = AIDIA ∩ BF C. Then
F
−1 = (AI; T IA ) = (IQ; ∞P )

where ∞ is the point at infinity along IP Q. Thus P is the midpoint of IQ. Since D is
the midpoint of IIA by “Fact 5”, it follows that DP bisects IF .

Q P I

B F T C

H E
D

IA

4
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§1.3 IMO 2010/3, proposed by Gabriel Carroll (USA)

Available online at https://aops.com/community/p1935854.

Problem statement

Find all functions g : Z>0 → Z>0 such that

(g(m) + n) (g(n) + m)

is always a perfect square.

For c ≥ 0, the function g(n) = n + c works; we prove this is the only possibility.
First, the main point of the problem is that

Claim — We have g(n) ≡ g(n0 ) (mod p) =⇒ n ≡ n0 (mod p).

Proof. Pick a large integer M such that

νp (M + g(n)), νp (M + g(n0 )) are both odd.

(It’s not hard to see this is always possible.) Now, since each of

(M + g(n)) (n + g(M ))
M + g(n0 ) n0 + g(M )

is a square, we get g(n) ≡ g(n0 ) ≡ −M (mod p).

This claim implies that

• The numbers g(n) and g(n + 1) differ by ±1 for any n, and

• The function g is injective.

It follows g is a linear function with slope ±1, hence done.

5
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§2 Solutions to Day 2
§2.1 IMO 2010/4, proposed by Marcin Kuczma (POL)
Available online at https://aops.com/community/p1936916.

Problem statement

Let P be a point interior to triangle ABC (with CA 6= CB). The lines AP , BP

and CP meet again its circumcircle Γ at K, L, M , respectively. The tangent line at
C to Γ meets the line AB at S. Show that from SC = SP follows M K = M L.

We present two solutions using harmonic bundles.

¶ First solution (Evan Chen). Let N be the antipode of M , and let N P meet Γ again
at D. Focus only on CDM N for now (ignoring the condition). Then C and D are feet
of altitudes in 4M N P ; it is well-known that the circumcircle of 4CDP is orthogonal
to Γ (passing through the orthocenter of 4M P N ).

N
C
K

L
P

S
A
B

D
M

Now, we are given that point S is such that SC is tangent to Γ, and SC = SP . It follows
that S is the circumcenter of 4CDP , and hence SC and SD are tangents to Γ.
Then −1 = (AB; CD) = (KL; M N ). Since M N is a diameter, this implies M K =
P

M L.
Remark. I think it’s more natural to come up with this solution in reverse. Namely, suppose
we define the points the other way: let SD be the other tangent, so (AB; CD) = −1. Then
project through P to get (KL; M N ) = −1, where N is the second intersection of DP .
However, if M L = M K then KM LN must be a kite. Thus one can recover the solution in
reverse.

¶ Second solution (Sebastian Jeon). We have

SP 2 = SC 2 = SA · SB =⇒ ]SP A = ]P BA = ]LBA = ]LKA = ]LKP

(the latter half is Reim’s theorem). Therefore SP and LK are parallel.

6
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

Now, let SP meet Γ again at X and Y , and let Q be the antipode of P on (S). Then

SP 2 = SQ2 = SX · SY =⇒ (P Q; XY ) = −1 =⇒ ∠QCP = 90◦

that CP bisects ∠XCY . Since XY k KL, it follows CP bisects to ∠LCK too.

7
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§2.2 IMO 2010/5, proposed by Netherlands

Available online at https://aops.com/community/p1936917.

Problem statement

Each of the six boxes B1 , B2 , B3 , B4 , B5 , B6 initially contains one coin. The

following two types of operations are allowed:

1. Choose a non-empty box Bj , 1 ≤ j ≤ 5, remove one coin from Bj and add two
coins to Bj+1 ;

2. Choose a non-empty box Bk , 1 ≤ k ≤ 4, remove one coin from Bk and swap

the contents (possibly empty) of the boxes Bk+1 and Bk+2 .

Determine if there exists a finite sequence of operations of the allowed types, such
that the five boxes B1 , B2 , B3 , B4 , B5 become empty, while box B6 contains exactly
coins.
2010
20102010

First,

(1, 1, 1, 1, 1, 1) → (0, 3, 1, 0, 3, 1) → (0, 0, 7, 0, 0, 7)

→ (0, 0, 6, 2, 0, 7) → (0, 0, 6, 1, 2, 7) → (0, 0, 6, 1, 0, 11)
→ (0, 0, 6, 0, 11, 0) → (0, 0, 5, 11, 0, 0).

and henceforth we ignore boxes B1 and B2 , looking at just the last four boxes; so we
write the current position as (5, 11, 0, 0).
We prove a lemma:

Claim — Let k ≥ 0 and n > 0. From (k, n, 0, 0) we may reach (k − 1, 2n , 0, 0).

Proof. Working with only the last three boxes for now,

(n, 0, 0) → (n − 1, 2, 0) → (n − 1, 0, 4)
→ (n − 2, 4, 0) → (n − 2, 0, 8)
→ (n − 3, 8, 0) → (n − 3, 0, 16)
→ · · · → (1, 2n−1 , 0) → (1, 0, 2n ) → (0, 2n , 0).

Finally we have (k, n, 0, 0) → (k, 0, 2n , 0) → (k − 1, 2n , 0, 0).

Now from (5, 11, 0, 0) we go as follows:

 
  11
11 211 22
(5, 11, 0, 0) → (4, 2 , 0, 0) → 3, 2 , 0, 0 → 2, 2 , 0, 0
11
!
11 2
 
22 22
→ 1, 22 , 0, 0 → 0, 22 , 0, 0 .

211
22
Let A = 22 = B. Then by using move 2 repeatedly on the fourth
2010
> 20102010
box (i.e., throwing away several coins by swapping the empty B5 and B6 ), we go from
(0, A, 0, 0) to (0, B/4, 0, 0). From there we reach (0, 0, 0, B).

8
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

§2.3 IMO 2010/6, proposed by Morteza Saghafiyan (IRN)

Available online at https://aops.com/community/p1936918.

Problem statement

Let a1 , a2 , a3 , . . . be a sequence of positive real numbers, and s be a positive integer,

such that
an = max{ak + an−k | 1 ≤ k ≤ n − 1} for all n > s.
Prove there exist positive integers ` ≤ s and N , such that

an = a` + an−` for all n ≥ N .

Let
a1 a2 as
w1 = , w 2 = , . . . , ws = .
1 2 s
(The choice of the letter w is for “weight”.) We claim the right choice of ` is the one
maximizing w` .
Our plan is to view each an as a linear combination of the weights w1 , . . . , ws and track
their coefficients.
To this end, let’s define an n-type to be a vector T = ht1 , . . . , ts i of nonnegative integers
such that

• n = t1 + · · · + ts ; and

• ti is divisible by i for every i.

We then define its valuation as v(T ) = si=1 wi ti .

P
Now we define a n-type to be valid according to the following recursive rule. For
1 ≤ n ≤ s the only valid n-types are

T1 = h1, 0, 0, . . . , 0i
T2 = h0, 2, 0, . . . , 0i
T3 = h0, 0, 3, . . . , 0i
..
.
Ts = h0, 0, 0, . . . , si

for n = 1, . . . , s, respectively. Then for any n > s, an n-type is valid if it can be written
as the sum of a valid k-type and a valid (n − k)-type, componentwise. These represent
the linear combinations possible in the recursion; in other words the recursion in the
problem is phrased as
an = max v(T ).
T is a valid n-type

In fact, we have the following description of valid n-types:

Claim — Assume n > s. Then an n-type ht1 , . . . , ts i is valid if and only if either

• there exist indices i < j with i + j > s, ti ≥ i and tj ≥ j; or

• there exists an index i > s/2 with ti ≥ 2i.

9
IMO 2010 Solution Notes web.evanchen.cc, updated 15 December 2024

Proof. Immediate by forwards induction on n > s that all n-types have P this property.
The reverse direction is by downwards induction on n. Indeed if i tii > 2,Pthen we
may subtract off on of {T1 , . . . , Ts } while preserving the condition; and the case i tii = 2
is essentially by definition.

Remark. The claim is a bit confusingly stated in its two cases; really the latter case should
be thought of as the situation i = j but requiring that ti /i is counted with multiplicity.

Now, for each n > s we pick a valid n-type Tn with an = v(Tn ); if there are ties, we
pick one for which the `th entry is as large as possible.

Claim — For any n > s and index i 6= `, the ith entry of Tn is at most 2s + `i.

Proof. If not, we can go back i` steps to get a valid (n − i`)-type T achieved by decreasing
the ith entry of Tn by i`. But then we can add ` to the `th entry i times to get another
n-type T 0 which obviously has valuation at least as large, but with larger `th entry.

Now since all other entries in Tn are bounded, eventually the sequence (Tn )n>s just
consists of repeatedly adding 1 to the `th entry, as required.

Remark. One big step is to consider wk = ak /k. You can get this using wishful thinking or
by examining small cases. (In addition this normalization makes it easier to see why the
largest w plays an important role, since then in the definition of type, the n-types all have a
sum of n. Unfortunately, it makes the characterization of valid n-types somewhat clumsier
too.)

10