USAMO 2009 Solution Notes

Evan Chen《陳誼廷》
29 March 2023

This is a compilation of solutions for the 2009 USAMO. Some of the

solutions are my own work, but many are from the official solutions provided
by the organizers (for which they hold any copyrights), and others were found
by users on the Art of Problem Solving forums.
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 USAMO 2009/1, proposed by Ian Le . . . . . . . . . . . . . . . . . . . . . 3
1.2 USAMO 2009/2, proposed by Kiran Kedlaya, Tewordos Amdeberhan . . . 4
1.3 USAMO 2009/3, proposed by Sam Vandervelde . . . . . . . . . . . . . . . 5

2 Solutions to Day 2 7
2.1 USAMO 2009/4, proposed by Titu Andreescu . . . . . . . . . . . . . . . . 7
2.2 USAMO 2009/5, proposed by Zuming Feng . . . . . . . . . . . . . . . . . 8
2.3 USAMO 2009/6, proposed by Gabriel Carroll . . . . . . . . . . . . . . . . 9

1
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§0 Problems
1. Given circles ω1 and ω2 intersecting at points X and Y , let `1 be a line through
the center of ω1 intersecting ω2 at points P and Q and let `2 be a line through the
center of ω2 intersecting ω1 at points R and S. Prove that if P , Q, R, and S lie on
a circle then the center of this circle lies on line XY .

2. Let n be a positive integer. Determine the size of the largest subset of {−n, −n +
1, . . . , n − 1, n} which does not contain three elements a, b, c (not necessarily
distinct) satisfying a + b + c = 0.

3. We define a chessboard polygon to be a simple polygon whose sides are situated

along lines of the form x = a or y = b, where a and b are integers. These lines divide
the interior into unit squares, which are shaded alternately grey and white so that
adjacent squares have different colors. To tile a chessboard polygon by dominoes is
to exactly cover the polygon by non-overlapping 1 × 2 rectangles. Finally, a tasteful
tiling is one which avoids the two configurations of dominoes and colors shown on
the left below. Two tilings of a 3 × 4 rectangle are shown; the first one is tasteful,
while the second is not, due to the vertical dominoes in the upper right corner.

Distasteful tilings

Prove that (a) if a chessboard polygon can be tiled by dominoes, then it can be
done so tastefully, and (b) such a tasteful tiling is unique.

4. For n ≥ 2, let a1 , a2 , . . . , an be positive real numbers such that

   2
1 1 1 1
(a1 + a2 + · · · + an ) + + ··· + ≤ n+ .
a1 a2 an 2

Prove that max (a1 , . . . , an ) ≤ 4 min (a1 , . . . , an ).

5. Trapezoid ABCD, with AB k CD, is inscribed in circle ω and point G lies inside
triangle BCD. Rays AG and BG meet ω again at points P and Q, respectively.
Let the line through G parallel to AB intersect BD and BC at points R and S,
respectively. Prove that quadrilateral P QRS is cyclic if and only if BG bisects
∠CBD.

6. Let s1 , s2 , s3 , . . . be an infinite, nonconstant sequence of rational numbers, meaning

it is not the case that s1 = s2 = s3 = . . . . Suppose that t1 , t2 , t3 , . . . is also
an infinite, nonconstant sequence of rational numbers with the property that
(si − sj )(ti − tj ) is an integer for all i and j. Prove that there exists a rational
number r such that (si − sj )r and (ti − tj )/r are integers for all i and j.

2
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§1 Solutions to Day 1
§1.1 USAMO 2009/1, proposed by Ian Le
Available online at https://aops.com/community/p1485133.

Problem statement

Given circles ω1 and ω2 intersecting at points X and Y , let `1 be a line through

the center of ω1 intersecting ω2 at points P and Q and let `2 be a line through the
center of ω2 intersecting ω1 at points R and S. Prove that if P , Q, R, and S lie on
a circle then the center of this circle lies on line XY .

Let r1 , r2 , r3 denote the circumradii of ω1 , ω2 , and ω3 , respectively.

O
Q

X
S

P R
O1 O2

We wish to show that O3 lies on the radical axis of ω1 and ω2 . Let us encode the
conditions using power of a point. Because O1 is on the radical axis of ω2 and ω3 ,

Powω2 (O1 ) = Powω3 (O1 )

=⇒ O1 O22 − r22 = O1 O32 − r32 .

Similarly, because O2 is on the radical axis of ω1 and ω3 , we have

Powω1 (O2 ) = Powω3 (O2 )

=⇒ O1 O22 − r12 = O2 O32 − r32 .

Subtracting the two gives

(O1 O22 − r22 ) − (O1 O22 − r12 ) = (O1 O32 − r32 ) − (O2 O32 − r32 )
=⇒ r12 − r22 = O1 O32 − O2 O32
=⇒ O2 O32 − r22 = O1 O32 − r12
=⇒ Powω2 (O3 ) = Powω1 (O3 )

as desired.

3
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§1.2 USAMO 2009/2, proposed by Kiran Kedlaya, Tewordos Amdeberhan

Available online at https://aops.com/community/p1485139.

Problem statement

Let n be a positive integer. Determine the size of the largest subset of {−n, −n +
1, . . . , n − 1, n} which does not contain three elements a, b, c (not necessarily distinct)
satisfying a + b + c = 0.

The answer is n with n even and n + 1 with n odd; the construction is to take all odd
numbers.
To prove this is maximal, it suffices to show it for n even; we do so by induction on
even n ≥ 2 with the base case being trivial. Letting A be the subset, we consider three
cases:

(i) If |A ∩ {−n, −n + 1, n − 1, n}| ≤ 2, then by the hypothesis for n − 2 we are done.

(ii) If both n ∈ A and −n ∈ A, then there can be at most n − 2 elements in A \ {±n},

one from each of the pairs (1, n − 1), (2, n − 2), . . . and their negations.

(iii) If n, n − 1, −n + 1 ∈ A and −n ∈ / A, and at most n − 3 more can be added, one

from each of (1, n − 2), (2, n − 3), . . . and (−2, −n + 2), (−3, −n + 3), . . . . (In
particular −1 ∈/ A. Analogous case for −A if n ∈ / A.)

Thus in all cases, |A| ≤ n as needed.

Remark. A few examples of equality cases:

• All odd numbers
• All numbers with absolute value at least n/2
• For n even, the set {1, 2, . . . , n}
This list isn’t exhaustive e.g. for n = 4, the set {−3, 2, 3, 4} achieves the optimum.

4
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§1.3 USAMO 2009/3, proposed by Sam Vandervelde

Available online at https://aops.com/community/p1485321.

Problem statement

We define a chessboard polygon to be a simple polygon whose sides are situated

along lines of the form x = a or y = b, where a and b are integers. These lines divide
the interior into unit squares, which are shaded alternately grey and white so that
adjacent squares have different colors. To tile a chessboard polygon by dominoes is
to exactly cover the polygon by non-overlapping 1 × 2 rectangles. Finally, a tasteful
tiling is one which avoids the two configurations of dominoes and colors shown on
the left below. Two tilings of a 3 × 4 rectangle are shown; the first one is tasteful,
while the second is not, due to the vertical dominoes in the upper right corner.

Distasteful tilings

Prove that (a) if a chessboard polygon can be tiled by dominoes, then it can be
done so tastefully, and (b) such a tasteful tiling is unique.

Proof of (a): This is easier, and by induction. Let P denote the chessboard polygon
which can be tiled by dominoes.
Consider a lower-left square s of the polygon, and WLOG is it white (other case
similar). Then we have two cases:

• If there exists a domino tiling of P where s is covered by a vertical domino, then

delete this domino and apply induction on the rest of P. This additional domino
will not cause any distasteful tilings.

• Otherwise, assume s is covered by a horizontal domino in every tiling. Again delete

this domino and apply induction on the rest of P. The resulting tasteful tiling
should not have another horizontal domino adjacent to the one covering s, because
otherwise we could have replaced that 2 × 2 square with two vertical dominoes to
arrive in the first case. So this additional domino will not cause any distasteful
tilings.

Remark. The second case can actually arise, for example in the following picture.

Thus one cannot just try to cover s with a vertical domino and claim the rest of P is tile-able.
So the induction is not as easy as one might hope.
One can phrase the solution algorithmically too, in the following way: any time we see a
distasteful tiling, we rotate it to avoid the bad pattern. The bottom-left corner eventually

5
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

becomes stable, and an induction shows the termination of the algorithm.

Proof of (b): We now turn to proving uniqueness. Suppose for contradiction there
are two distinct tasteful tilings. Overlaying the two tilings on top of each other induces
several cycles of overlapping dominoes at positions where the tilings differ.
Henceforth, it will be convenient to work with the lattice Z2 , treating the squares as
black/white points, and we do so. Let γ be any such cycle and let s denote a lower
left point, and again WLOG it is black. Orient γ counterclockwise henceforth. Restrict
attention to the lattice polygon Q enclosed by γ (we consider points of γ as part of Q).
In one of the two tilings of (lattice points of) Q, the point s will be covered by a
horizontal domino; in the other tiling s will be covered by a vertical domino. From now
on we will focus only on the latter one. Observe that we now have a set of dominoes
along γ, such that γ points from the white point to the black point within each domino.
Now impose coordinates so that s = (0, 0). Consider the stair-case sequence of points
p0 = s = (0, 0), p1 = (1, 0), p2 = (1, 1), p3 = (2, 1), and so on. By hypothesis, p0 is
covered by a vertical domino. Then p1 must be covered by a horizontal domino, to avoid
a distasteful tiling. Then if p2 is in Q, then it must be covered by a vertical domino to
avoid a distasteful tiling, and so on. We may repeat this argument as long the points pi
lie inside Q. (See figure below; the staircase sequence is highlighted by red halos.)

s b

The curve γ by definition should cross y = x − 1 at the point b = (1, 0). Let a denote
the first point of this sequence after p1 for which γ crosses y = x − 1 again.
Now a is tiled by a vertical domino whose black point is to the right of `. But the line
segment ` cuts Q into two parts, and the orientation of γ has this path also entering
from the right. This contradicts the fact that the orientation of γ points only from white
to black within dominoes. This contradiction completes the proof.

Remark. Note the problem is false if you allow holes (consider a 3 × 3 with the middle
square deleted).

6
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§2 Solutions to Day 2
§2.1 USAMO 2009/4, proposed by Titu Andreescu
Available online at https://aops.com/community/p1485147.

Problem statement

For n ≥ 2, let a1 , a2 , . . . , an be positive real numbers such that

1 2
   
1 1 1
(a1 + a2 + · · · + an ) + + ··· + ≤ n+ .
a1 a2 an 2

Prove that max (a1 , . . . , an ) ≤ 4 min (a1 , . . . , an ).

Assume a1 is the largest and a2 is the smallest. Let M = a1 /a2 . We wish to show M ≤ 4.
In left-hand side of given, write as a2 + a1 + · · · + an . By Cauchy Schwarz, one obtains

1 2
   
1 1 1 1
n+ ≥ (a2 + a1 + a3 + · · · + an ) + + + ··· +
2 a1 a2 a3 an
r r 2
a2 a1
≥ + + 1 + ··· + 1
a1 a2
p √ 2
≥ 1/M + M + (n − 2) .

Expanding and solving for M gives 1/4 ≤ M ≤ 4 as needed.

7
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§2.2 USAMO 2009/5, proposed by Zuming Feng

Available online at https://aops.com/community/p1485200.

Problem statement

Trapezoid ABCD, with AB k CD, is inscribed in circle ω and point G lies inside
triangle BCD. Rays AG and BG meet ω again at points P and Q, respectively.
Let the line through G parallel to AB intersect BD and BC at points R and S,
respectively. Prove that quadrilateral P QRS is cyclic if and only if BG bisects
∠CBD.

Perform an inversion around B with arbitrary radius, and denote the inverse of a point
Z with Z ∗ .

A B B

R S
G Q∗
A∗ ∗
P D∗ C∗ P
D C
R∗ S∗

Q G∗

After inversion, we obtain a cyclic quadrilateral BS ∗ G∗ R∗ and points C ∗ , D∗ on BS ∗ ,

BR∗ , such that (BC ∗ D∗ ) is tangent to (BS ∗ G∗ R∗ ) — in other words, so that C ∗ D∗
is parallel to S ∗ R∗ . Point A∗ lies on line C ∗ D∗ so that A∗ B is tangent to (BS ∗ G∗ R∗ ).
Points P ∗ and Q∗ are the intersections of (A∗ BG∗ ) and BG∗ with line C ∗ D∗ .
Observe that P ∗ Q∗ R∗ S ∗ is a trapezoid, so it is cyclic if and only if it isosceles.
Let X be the second intersection of line G∗ P ∗ with (BS ∗ R∗ ). Because

]Q∗ P ∗ G∗ = ]A∗ BG∗ = ]BXG∗

we find that BXS ∗ R∗ is an isosceles trapezoid.

If G∗ is indeed the midpoint of the arc then everything is clear by symmetry now.
Conversely, if P ∗ R∗ = Q∗ S ∗ then that means P ∗ Q∗ R∗ S ∗ is a cyclic trapezoid, and hence
that the perpendicular bisectors of P ∗ Q∗ and R∗ S ∗ are the same. Hence B, X, P ∗ , Q∗
are symmetric around this line. This forces G∗ to be the midpoint of arc R∗ S ∗ as desired.

8
USAMO 2009 Solution Notes web.evanchen.cc, updated 29 March 2023

§2.3 USAMO 2009/6, proposed by Gabriel Carroll

Available online at https://aops.com/community/p1485204.

Problem statement

Let s1 , s2 , s3 , . . . be an infinite, nonconstant sequence of rational numbers, meaning it

is not the case that s1 = s2 = s3 = . . . . Suppose that t1 , t2 , t3 , . . . is also an infinite,
nonconstant sequence of rational numbers with the property that (si − sj )(ti − tj )
is an integer for all i and j. Prove that there exists a rational number r such that
(si − sj )r and (ti − tj )/r are integers for all i and j.

First we eliminate the silly edge case:

Claim — For some i and j, we have (si − sj )(ti − tj ) 6= 0.

Proof. Assume not. WLOG s1 6= s2 , then t1 = t2 . Now select i such that ti 6= t1 = t2 .

Then either ti − s1 6= 0 or ti − s2 6= 0, contradiction.

So, WLOG (by permutation) that n = (s1 − s2 )(t1 − t2 ) 6= 0. By shifting and scaling
appropriately, we may assume

s1 = t1 = 0, s2 = 1, t2 = n.

Thus we deduce
si ti ∈ Z, si tj + sj ti ∈ Z ∀i, j.

Claim — For any index i, ti ∈ Z, si ∈ n1 Z.

Proof. We have si ti ∈ Z and ti + nsi ∈ Z by problem condition. By looking at νp of this,

we conclude nsi , ti ∈ Z (since if either as negative p-adic value, so does the other, and
then si ti ∈
/ Z).

Last claim:

Claim — If d = gcd t• , then dsi ∈ Z for all i.

Proof. Consider a prime p | n, and let e = νp (tj ). We will show νp (si ) ≥ −e for any i.
This is already true for i = j, so assume i 6= j. Assume for contradiction νp (si ) < −e.
Then νp (ti ) > e = νp (tk ). Since νp (sk ) ≥ −e we deduce νp (si tk ) < νp (sk ti ); so νp (si tk ) ≥
0 and νp (si ) ≥ −e as desired.