JMO 2016 Solution Notes

Evan Chen《陳誼廷》
15 April 2024

This is a compilation of solutions for the 2016 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 2016/1, proposed by Ivan Borsenco, Zuming Feng . . . . . . . . . . 3
1.2 JMO 2016/2, proposed by Evan Chen . . . . . . . . . . . . . . . . . . . . 5
1.3 JMO 2016/3, proposed by Iurie Boreico . . . . . . . . . . . . . . . . . . . 6

2 Solutions to Day 2 8
2.1 JMO 2016/4, proposed by Gregory Galperin . . . . . . . . . . . . . . . . . 8
2.2 JMO 2016/5, proposed by Zuming Feng, Jacek Fabrykowski . . . . . . . . 9
2.3 JMO 2016/6, proposed by Titu Andreescu . . . . . . . . . . . . . . . . . . 11

1
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§0 Problems
1. The isosceles triangle 4ABC, with AB = AC, is inscribed in the circle ω. Let
P be a variable point on the arc BC that does not contain A, and let IB and IC
denote the incenters of triangles 4ABP and 4ACP , respectively. Prove that as
P varies, the circumcircle of triangle 4P IB IC passes through a fixed point.

2. Prove that there exists a positive integer n < 106 such that 5n has six consecutive
zeros in its decimal representation.

3. Let X1 , X2 , . . . , X100 be a sequence of mutually distinct nonempty subsets of a set

S. Any two sets Xi and Xi+1 are disjoint and their union is not the whole set S,
that is, Xi ∩ Xi+1 = ∅ and Xi ∪ Xi+1 6= S, for all i ∈ {1, . . . , 99}. Find the smallest
possible number of elements in S.

4. Find, with proof, the least integer N such that if any 2016 elements are removed
from the set {1, 2, . . . , N }, one can still find 2016 distinct numbers among the
remaining elements with sum N .

5. Let 4ABC be an acute triangle, with O as its circumcenter. Point H is the foot
of the perpendicular from A to line BC, and points P and Q are the feet of the
perpendiculars from H to the lines AB and AC, respectively.
Given that
AH 2 = 2AO2 ,
prove that the points O, P , and Q are collinear.

6. Find all functions f : R → R such that for all real numbers x and y,

(f (x) + xy) · f (x − 3y) + (f (y) + xy) · f (3x − y) = (f (x + y))2 .

2
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§1 Solutions to Day 1
§1.1 JMO 2016/1, proposed by Ivan Borsenco, Zuming Feng
Available online at https://aops.com/community/p6213607.

Problem statement

The isosceles triangle 4ABC, with AB = AC, is inscribed in the circle ω. Let P be
a variable point on the arc BC that does not contain A, and let IB and IC denote
the incenters of triangles 4ABP and 4ACP , respectively. Prove that as P varies,
the circumcircle of triangle 4P IB IC passes through a fixed point.

Let M be the midpoint of arc BC not containing A. We claim M is the desired fixed
point.

MB MC

IC
IB
B C

P
M

Let MB , MC be the second intersections of P IB and P IC with circumcircle.

Claim — We have 4IB MB M ∼

= 4IC MC M .

Proof. Note that

MB IB = MB B = MC C = MC IC
M MB = M MC
1‘
∠IB MB M = P M = ∠IC MC M.
2
This implies the desired congruence.

Since ∠M P A = 90◦ and ray P A bisects ∠IB P IC , the conclusion M IB = M IC finishes

the problem.

3
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

Remark 1.1. Complex in the obvious way DOES NOT WORK, because the usual claim
(“the fixed point is arc midpoint”) is FALSE if the hypothesis that P lies in the interior of
the arc is dropped. See figure below.

P
IB IC

B C

Fun story, I pointed this out to Zuming during grading; I was the only one that realized the
subtlety.

4
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§1.2 JMO 2016/2, proposed by Evan Chen

Available online at https://aops.com/community/p6213569.

Problem statement

Prove that there exists a positive integer n < 106 such that 5n has six consecutive
zeros in its decimal representation.

We will prove that n = 20 + 219 = 524308 fits the bill.

First, we claim that

5n ≡ 520 (mod 520 ) and 5n ≡ 520 (mod 220 ).

Indeed, the first equality holds since both sides are 0 (mod 520 ), and the second by
ϕ(220 ) = 219 and Euler’s theorem. Hence

5n ≡ 520 (mod 1020 ).

In other words, the last 20 digits of 5n will match the decimal representation of 520 ,
with leading zeros. However, we have
1 1
520 = 20
· 1020 < · 1020 = 10−6 · 1020
2 10002
and hence those first six of those 20 digits will all be zero. This completes the proof! (To
be concrete, it turns out that 520 = 95367431640625 and so the last 20 digits of 5n will
be 00000095367431640625.)

Remark. Many of the first posts in the JMO 2016 discussion thread (see https://aops.
com/community/c5h1230514) claimed that the problem was “super easy”. In fact, the
problem was solved by only about 10% of contestants.

¶ Authorship comments. This problem was inspired by the observation 58 ≡ 54

(mod 104 ), i.e. that 58 ended with 0625.
I noticed this one day back in November, when I was lying on my bed after a long
afternoon and was mindlessly computing powers of 5 in my head because I was too tired
to do much else. When I reached 58 I noticed for the first time that the ending 0625 was
actually induced by 54 . (Given how much MathCounts I did, I really should have known
this earlier!)
Thinking about this for a few more seconds, I realized one could obtain arbitrarily
long strings of 0’s by using a similar trick modulo larger powers of 10. This surprised me,
because I would have thought that if this was true, then I would have learned about it
back in my contest days. However, I could not find any references, and I thought the
result was quite nice, so I submitted it as a proposal for the JMO, where I thought it
might be appreciated.
The joke about six consecutive zeros is due to Zuming Feng.

5
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§1.3 JMO 2016/3, proposed by Iurie Boreico

Available online at https://aops.com/community/p6213589.

Problem statement

Let X1 , X2 , . . . , X100 be a sequence of mutually distinct nonempty subsets of a set

S. Any two sets Xi and Xi+1 are disjoint and their union is not the whole set S,
that is, Xi ∩ Xi+1 = ∅ and Xi ∪ Xi+1 6= S, for all i ∈ {1, . . . , 99}. Find the smallest
possible number of elements in S.

Solution with Danielle Wang: the answer is that |S| ≥ 8.

¶ Proof that |S| ≥ 8 is necessary. Since we must have 2|S| ≥ 100, we must have
|S| ≥ 7.
To see that |S| = 8 is the minimum possible size, consider a chain on the set S =
{1, 2, . . . , 7} satisfying Xi ∩ Xi+1 = ∅ and Xi ∪ Xi+1 6= S. Because of these requirements
any subset of
size
4 or more can only be neighbored by sets of size 2 or less, of which
there are 1 + 2 = 28 available. Thus, the chain can contain no more than 29 sets of
7 7

size

4 or more and no more than 28 sets of size 2 or less. Finally, since there are only
3 = 35 sets of size 3 available, the total number of sets in such a chain can be at most
7

29 + 28 + 35 = 92 < 100, contradiction.

¶ Construction. We will provide an inductive construction for a chain of subsets

X1 , X2 , . . . , X2n−1 +1 of S = {1, . . . , n} satisfying Xi ∩ Xi+1 = ∅ and Xi ∪ Xi+1 6= S for
each n ≥ 4.
For S = {1, 2, 3, 4}, the following chain of length 23 + 1 = 9 will work:

34 1 23 4 12 3 14 2 13 .

Now, given a chain of subsets of {1, 2, . . . , n} the following procedure produces a chain
of subsets of {1, 2, . . . , n + 1}:

1. take the original chain, delete any element, and make two copies of this chain,
which now has even length;

2. glue the two copies together, joined by ∅ in between; and then

3. insert the element n + 1 into the sets in alternating positions of the chain starting
with the first.

For example, the first iteration of this construction gives:

345 1 235 4 125 3 145 2 5

34 15 23 45 12 35 14 25

It can be easily checked that if the original chain satisfies the requirements, then so does
the new chain, and if the original chain has length 2n−1 + 1, then the new chain has length
2n + 1, as desired. This construction yields a chain of length 129 when S = {1, 2, . . . , 8}.

6
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

Remark. Here is the construction for n = 8 in its full glory.

345678 1 235678 4 125678 3 145678 2 5678

34 15678 23 45678 12 35678 14 678
345 1678 235 4678 125 3678 145 2678 5
34678 15 23678 45 12678 35 78
3456 178 2356 478 1256 378 1456 278 56
3478 156 2378 456 1278 356 1478 6
34578 16 23578 46 12578 36 14578 26 578
346 1578 236 4578 126 8
34567 18 23567 48 12567 38 14567 28 567
348 1567 238 4567 128 3567 148 67
3458 167 2358 467 1258 367 1458 267 58
3467 158 2367 458 1267 358 7
34568 17 23568 47 12568 37 14568 27 568
347 1568 237 4568 127 3568 147 68
3457 168 2357 468 1257 368 1457 268 57
3468 157 2368 457 1268

7
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§2 Solutions to Day 2
§2.1 JMO 2016/4, proposed by Gregory Galperin
Available online at https://aops.com/community/p6220314.

Problem statement

Find, with proof, the least integer N such that if any 2016 elements are removed from
the set {1, 2, . . . , N }, one can still find 2016 distinct numbers among the remaining
elements with sum N .

The answer is

N = 2017 + 2018 + · · · + 4032 = 1008 · 6049 = 6097392.

¶ Proof that N ≥ 6097392 is necessary. To see that N must be at least this large,
consider the situation when 1, 2, . . . , 2016 are removed. Among the remaining elements,
any sum of 2016 elements is certainly at least 2017 + 2018 + · · · + 4032.

¶ Proof that N = 6097392 does in fact work. Consider the 3024 pairs of numbers
(1, 6048), (2, 6047), . . . , (3024, 3025). Regardless of which 2016 elements of {1, 2, . . . , N }
are deleted, at least 3024 − 2016 = 1008 of these pairs have both elements remaining.
Since each pair has sum 6049, we can take these pairs to be the desired numbers.

8
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§2.2 JMO 2016/5, proposed by Zuming Feng, Jacek Fabrykowski

Available online at https://aops.com/community/p6220305.

Problem statement

Let 4ABC be an acute triangle, with O as its circumcenter. Point H is the foot
of the perpendicular from A to line BC, and points P and Q are the feet of the
perpendiculars from H to the lines AB and AC, respectively.
Given that
AH 2 = 2AO2 ,
prove that the points O, P , and Q are collinear.

We present two approaches.

¶ First approach (synthetic). First, since AP · AB = AH 2 = AQ · AC, it follows that

P QCB is cyclic. Consequently, we have AO ⊥ P Q.

Q
K

P
B H C

Let K be the foot of A onto P Q, and let D be the point diametrically opposite A.
Thus A, K, O, D are collinear.
Since quadrilateral KQCD is cyclic (∠QKD = ∠QCD = 90◦ ), we have

AH 2 AH 2
AK · AD = AQ · AC = AH 2 =⇒ AK = = = AO
AD 2AO
so K = O.

¶ Second approach (coordinates), with Joshua Hsieh. We impose coordinates with

H at the origin and A = (0, a), B = (−b, 0), C = (c, 0), for a, b, c > 0.

Claim — The circumcenter has coordinates ( c−b a

2 ,2 −
bc
2a ).

9
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

Proof. This is a known lemma but but we reproduce its proof for completeness. It uses
the following steps:

• By power of a point, the second intersection of line AH with the circumcircle is

(0, − bc
a ).

• Since the orthocenter is the reflection of this point across line BC, the orthocenter
is given exactly by (0, bc
a ).

• The centroid is is
~ B+
A+ ~ C
~
3 = ( c−b a
3 , 3 ).

• Since H~ −O~ = 3(G ~ according to the Euler line, we have O

~ − O) ~ = 3G
2
~ This
~ − 1 H.
2
gives the desired formula.

Note that HQ = HA·HC

AC = √ ac
a2 +c2
. If we let T be the foot from Q to BC, then
∼
4HQT + 4AHC and so the x-coordinate of Q is given by HQ · AH AC =
a2 c
a2 +c2
. Repeating
the analogous calculation for Q and P gives
 2
ac2

a c
Q= ,
a2 + c2 a2 + c2
a2 b ab2
 
P = − 2 , .
a + b2 a2 + b2

Then, O, P , Q are collinear if and only if the following shoelace determinant vanishes
(with denominators cleared out):

−a2 b ab2 a2 + b2 ab2 a2 + b2

   
−ab
0 = det  a2 c ac2 a2 + c2  = a det  ac ac2 a2 + c2 
2
a(c − b) a − bc 2a 2
c − b a − bc 2a
2 2 2 2
   
−a(b + c) a(b − c ) b − c −a a(b − c) b − c
= a det  ac ac2 a2 + c2  = a(b + c) det  ac ac2 a2 + c2 
c−b 2
a − bc 2a 2
c − b a − bc 2a
= a(b + c) · − a(a2 c2 − a4 + bc(a2 + c2 )) + ac(b − c) −a2 − bc − (b − c)2 · a3



= a2 (b + c)(a4 − a2 b2 − b2 c2 − c2 a2 ).

On the other hand,

AH 2 = a2
b2 c2
" 2 2 #
a2 + b2 + c2 +

2 a c−bbc a2
2AO = 2 + − − =
2 2a2 2
b2 c2
 
1
=⇒ AH − 2AO =
2 2 2 2 2
a −b −c − 2 .
2 a

So the conditions are equivalent.

10
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

§2.3 JMO 2016/6, proposed by Titu Andreescu

Available online at https://aops.com/community/p6220308.

Problem statement

Find all functions f : R → R such that for all real numbers x and y,

(f (x) + xy) · f (x − 3y) + (f (y) + xy) · f (3x − y) = (f (x + y))2 .

We claim that the only two functions satisfying the requirements are f (x) ≡ 0 and
f (x) ≡ x2 . These work.
First, taking x = y = 0 in the given yields f (0) = 0, and then taking x = 0 gives
f (y)f (−y) = f (y)2 . So also f (−y)2 = f (y)f (−y), from which we conclude f is even.
Then taking x = −y gives
∀x ∈ R : f (x) = x2 or f (4x) = 0 (F)
for all x.
Remark. Note that an example of a function satisfying (F) is

if |x| < 1

2
x

f (x) = 1 − cos π2 · x1337 if 1 ≤ |x| < 4

if |x| ≥ 4.

0


So, yes, we are currently in a world of trouble, still. (This function is even continuous; I
bring this up to emphasize that “continuity” is completely unrelated to the issue at hand.)

Now we claim

Claim — f (z) = 0 ⇐⇒ f (2z) = 0 (♠).

Proof. Let (x, y) = (3t, t) in the given to get

f (t) + 3t2 f (8t) = f (4t)2 .

Now if f (4t) 6= 0 (in particular, t 6= 0), then f (8t) 6= 0. Thus we have (♠) in the reverse
direction.
(F) (♠)
Then f (4t) 6= 0 =⇒ f (t) = t2 6= 0 =⇒ f (2t) 6= 0 implies the forwards direction, the
last step being the reverse direction (♠).

By putting together (F) and (♠) we finally get

∀x ∈ R : f (x) = x2 or f (x) = 0 (♥)
We are now ready to approach the main problem. Assume there’s an a 6= 0 for which
f (a) = 0; we show that f ≡ 0.
Let b ∈ R be given. Since f is even, we can assume without loss of generality that
a, b > 0. Also, note that f (x) ≥ 0 for all x by (♥). By using (♠) we can generate c > b
such that f (c) = 0 by taking c = 2n a for a large enough integer n. Now, select x, y > 0
such that x − 3y = b and x + y = c. That is,
 
3c + b c − b
(x, y) = , .
4 4

11
JMO 2016 Solution Notes web.evanchen.cc, updated 15 April 2024

Substitution into the original equation gives

0 = (f (x) + xy) f (b) + (f (y) + xy) f (3x − y) ≥ (f (x) + xy) f (b).

But since f (b) ≥ 0, it follows f (b) = 0, as desired.

12