IMO 1997 Solution Notes

Evan Chen《陳誼廷》
30 January 2025

This is a compilation of solutions for the 1997 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 1997/1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 IMO 1997/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 IMO 1997/3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 Solutions to Day 2 7
2.1 IMO 1997/4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 IMO 1997/5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 IMO 1997/6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§0 Problems
1. In the plane there is an infinite chessboard. For any pair of positive integers m
and n, consider a right-angled triangle with vertices at lattice points and whose
legs, of lengths m and n, lie along edges of the squares. Let S1 be the total area
of the black part of the triangle and S2 be the total area of the white part. Let
f (m, n) = |S1 − S2 |.
(a) Calculate f (m, n) for all positive integers m and n which are either both even
or both odd.
(b) Prove that f (m, n) ≤ 1
2 max{m, n} for all m and n.
(c) Show that there is no constant C such that f (m, n) < C for all m and n.

2. Let ABC be a triangle with ∠A < min(∠B, ∠C). The points B and C divide the
circumcircle of the triangle into two arcs. Let U be an interior point of the arc
between B and C which does not contain A. The perpendicular bisectors of AB
and AC meet the line AU at V and W , respectively. The lines BV and CW meet
at T .
Show that AU = T B + T C.

3. Let x1 , x2 , . . . , xn be real numbers satisfying the conditions:

|x1 + x2 + · · · + xn | = 1
n+1
|xi | ≤ for i = 1, 2, . . . , n
2
Show that there exists a permutation y1 , y2 , . . . , yn of x1 , x2 , . . . , xn such that
n+1
|y1 + 2y2 + · · · + nyn | ≤ .
2

4. An n × n matrix whose entries come from the set S = {1, 2, . . . , 2n − 1} is called a

silver matrix if, for each i = 1, 2, . . . , n, the i-th row and the i-th column together
contain all elements of S. Show that:
(a) there is no silver matrix for n = 1997;
(b) silver matrices exist for infinitely many values of n.

5. Find all pairs (a, b) of positive integers satisfying

2
ab = ba .

6. For each positive integer n, let f (n) denote the number of ways of representing n
as a sum of powers of 2 with nonnegative integer exponents. Representations which
differ only in the ordering of their summands are considered to be the same. For
instance, f (4) = 4, because the number 4 can be represented in the following four
ways: 4; 2 + 2; 2 + 1 + 1; 1 + 1 + 1 + 1.
n2 n2
Prove that for any integer n ≥ 3 we have 2 4 < f (2n ) < 2 2 .

2
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§1 Solutions to Day 1
§1.1 IMO 1997/1
Available online at https://aops.com/community/p356696.

Problem statement

In the plane there is an infinite chessboard. For any pair of positive integers m
and n, consider a right-angled triangle with vertices at lattice points and whose
legs, of lengths m and n, lie along edges of the squares. Let S1 be the total area
of the black part of the triangle and S2 be the total area of the white part. Let
f (m, n) = |S1 − S2 |.

(a) Calculate f (m, n) for all positive integers m and n which are either both even
or both odd.

(b) Prove that f (m, n) ≤ 1

2 max{m, n} for all m and n.

(c) Show that there is no constant C such that f (m, n) < C for all m and n.

In general, we say the discrepancy of a region in the plane equals its black area minus
its white area. We allow negative discrepancies, so discrepancy is additive and f (m, n)
equals the absolute value of the discrepancy of a right triangle with legs m and n.
For (a), the answers are 0 and 1/2 respectively. To see this, consider the figure shown
below.
A

M Q
P

C B

Notice that triangles AP M and BQM are congruent, and when m ≡ n (mod 2), their
colorings actually coincide. So, the discrepancy of the triangle is exactly equal to the
discrepancy of CP QB, which is an m × n/2 rectangle and hence equal to 0 or 1/2
according to parity.
For (b), note that a triangle with legs m and n, with m even and n odd, can be
dissected into one right triangle with legs m and n − 1 plus a thin triangle of area 1/2
which has height m and base 1. The former region has discrepancy 0 by (a), and the
latter region obviously has discrepancy at most its area of m/2, hence f (m, n) ≤ m/2 as
needed. (An alternative slower approach, which requires a few cases, is to prove that two
adjacent columns have at most discrepancy 1/2.)
For (c), we prove:

3
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

Claim — For each k ≥ 1, we have

2k − 1
f (2k, 2k + 1) = .
6

Proof. An illustration for k = 2 is shown below, where we use (0, 0), (0, 2k), (2k + 1, 0)
as the three vertices.

WLOG, the upper-left square is black, as above. The 2k small white triangles just below
the diagonal have area sum
1 1 1  2  4k + 1
· · 1 + 22 + · · · + (2k)2 =
2 2k + 1 2k 12
The area of the 2k black polygons sums just below the diagonal to
2k  
X 1 1 1 2 4k + 1 20k − 1
1− · · · i = 2k − = .
2 2k + 1 2k 12 12
i=1

Finally, in the remaining 1 + 2 + · · · + 2k squares, there are k more white squares than
black squares. So, it follows

20k − 1 4k + 1 2k − 1
f (2k, 2k + 1) = −k + − = .
12 12 6

4
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§1.2 IMO 1997/2

Available online at https://aops.com/community/p356701.

Problem statement

Let ABC be a triangle with ∠A < min(∠B, ∠C). The points B and C divide the
circumcircle of the triangle into two arcs. Let U be an interior point of the arc
between B and C which does not contain A. The perpendicular bisectors of AB and
AC meet the line AU at V and W , respectively. The lines BV and CW meet at T .
Show that AU = T B + T C.

Let BT V meet the circle again at U1 , so that AU1 U B is an isosceles trapezoid. Define
U2 similarly.

U1
A

U2

B C
U

Now from the isosceles trapezoids we get

AU = BU1 = BT + T U1 = BT + T C

as desired.

5
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§1.3 IMO 1997/3

Available online at https://aops.com/community/p356706.

Problem statement

Let x1 , x2 , . . . , xn be real numbers satisfying the conditions:

|x1 + x2 + · · · + xn | = 1
n+1
|xi | ≤ for i = 1, 2, . . . , n
2
Show that there exists a permutation y1 , y2 , . . . , yn of x1 , x2 , . . . , xn such that
n+1
|y1 + 2y2 + · · · + nyn | ≤ .
2

WLOG xi = 1 (by negating xi ) and x1 ≤ x2 ≤ · · · ≤ xn . Notice that

P

• The largest possible value of the sum in question is

A = x1 + 2x2 + 3x3 + · · · + nxn .

while the smallest value is

B = nx1 + (n − 1)x2 + · · · + xn .

• Meanwhile, the average value across all permutations is

1 1 1 n+1
1· + 2 · + ··· + n · = .
n n n 2

Now imagine we transform the sum A to the sum B, one step at a time, by swapping
adjacent elements. Every time we do a swap of two neighboring u ≤ v, the sum decreases
by
(iu + (i + 1)v) − (iv + (i + 1)u) = v − u ≤ n + 1.
We want to prove we land in the interval
 
n+1 n+1
I= − ,
2 2

at some point during this transformation. But since B ≤ n+1 2 ≤ A (since 2 was the
n+1

average) and our step sizes were at most the length of the interval I, this is clear.

6
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§2 Solutions to Day 2
§2.1 IMO 1997/4
Available online at https://aops.com/community/p611.

Problem statement

An n × n matrix whose entries come from the set S = {1, 2, . . . , 2n − 1} is called a

silver matrix if, for each i = 1, 2, . . . , n, the i-th row and the i-th column together
contain all elements of S. Show that:

(a) there is no silver matrix for n = 1997;

(b) silver matrices exist for infinitely many values of n.

¶ Solution to (a) . Define a cross to be the union of the ith row and ith column. Every
cell of the matrix not on the diagonal is contained in exactly two crosses, while each cell
on the diagonal is contained in one cross.
On the other hand, if a silver matrix existed for n = 1997, then each element of S
appears in all 1997 crosses. Since 1997 is odd, each number s ∈ S must appear on the
diagonal an odd number of times. (For example, s could appear on the diagonal once
and off-diagonal 998 times, or on the diagonal three times and off-diagonal 997 times,
etc.) In particular, each number s appears at least once on the diagonal.
However, |S| = 3993 while there are only 1997 diagonal cells. This is a contradiction.

¶ Solution to (b) . We construct a silver matrix Me for n = 2e for each e ≥ 1. We

write the first three explicitly for concreteness:
 
1 2
M1 =
3 1
 
1 2 4 5
3 1 6 7
M2 =  7 5 1 2


6 4 3 1
 
1 2 4 5 8 9 11 12
 3 1 6 7 10 15 13 14
 
 7 5 1 2 14 12 8 9
 
 6 4 3 1 13 11 10 15
M3 =  15 9 11 12 1 2

 4 5 
10 8 13 14 3 1 6 7
 
14 12 15 9 7 5 1 2
13 11 10 8 6 4 3 1
The construction is described recursively as follows. Let
Me−1 + (2e − 1)
 
0 Me−1
Me = .
Me−1 + (2e − 1) Me−1
Then to get from Me0 to Me , replace half of the 2e ’s with 2e+1 − 1: in the northeast
quadrant, the even-indexed ones, and in the southwest quadrant, the odd-indexed ones.

7
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

Remark. In fact, it turns out silver matrices exist for all even dimensions. A claimed proof
is outlined at https://aops.com/community/p7375020.

8
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§2.2 IMO 1997/5

Available online at https://aops.com/community/p3845.

Problem statement

Find all pairs (a, b) of positive integers satisfying

2
ab = ba .

The answer is (1, 1), (16, 2) and (27, 3).

We assume a, b > 1 for convenience. Let T denote the set of non perfect powers other
than 1.

Claim — Every integer greater than 1 is uniquely of the form tn for some t ∈ T ,
n ∈ N.

Proof. Clear.

Let a = sm , b = tn .
n )2 m
sm·(t = tn·s .
Hence s = t and we have
n
m · t2n = n · tm =⇒ t2n−m = .
m
Let n = te m and 2 · te m − m = e, or

e + m = 2te · m.

We resolve this equation by casework

• If e > 0, then 2te · m > 2e · m > e + m.
• If e = 0 we have m = n and m = 2m, contradiction.
• If e = −1 we apparently have
2 t
· m = m − 1 =⇒ m =
t t−2
so (t, m) = (3, 3) or (t, m) = (4, 2).
• If e = −2 we apparently have
2 2 2t2
· m = m − 2 =⇒ m = = .
t2 1 − 2/t2 t2 − 2
This gives (t, m) = (2, 2).
• If e ≤ −3 then let k = −e ≥ 3, so the equation is
2m k · tk 2k
m−k = k
⇐⇒ m = k
=k+ k .
t t −2 t −2
However, for k ≥ 3 and t ≥ 2, we always have 2k ≤ tk − 2, with equality only when
(t, k) = (2, 3); this means m = 4, which is not a new solution.

9
IMO 1997 Solution Notes web.evanchen.cc, updated 30 January 2025

§2.3 IMO 1997/6

Available online at https://aops.com/community/p356713.

Problem statement

For each positive integer n, let f (n) denote the number of ways of representing n as
a sum of powers of 2 with nonnegative integer exponents. Representations which
differ only in the ordering of their summands are considered to be the same. For
instance, f (4) = 4, because the number 4 can be represented in the following four
ways: 4; 2 + 2; 2 + 1 + 1; 1 + 1 + 1 + 1.
n2 n2
Prove that for any integer n ≥ 3 we have 2 4 < f (2n ) < 2 2 .

It’s clear that f is non-decreasing. By sorting by the number of 1’s we used, we have the
equation
       
N N N
f (N ) = f +f −1 +f − 2 + · · · + f (1) + f (0). (F)
2 2 2

¶ Upper bound. We now prove the upper bound by induction. Indeed, the base case is
trivial and for the inductive step we simply use (F):
(n−1)2 n2
− 12
f (2n ) = f (2n−1 ) + f (2n−1 − 1) + · · · < 2n−1 f (2n−1 ) < 2n−1 · 2 2 =2 2 .

¶ Lower bound. First, we contend that f is convex. We’ll first prove this in the even
case to save ourselves some annoyance:

Claim (f is basically convex) — If 2 | a + b then we have f (2a) + f (2b) ≥ 2f (a + b).

Proof. Since f (2k + 1) = f (2k), we will only prove the first equation. Assume WLOG
a ≥ b and use (F) on all three f expressions here; after subtracting repeated terms, the
inequality then rewrites as
X X
f (x) ≥ f (x).
(a+b)/2≤x≤a b≤x≤(a+b)/2

This is true since there are an equal number of terms on each side and f is nondecreasing.

Claim — For each 1 ≤ k < 2n−1 , we have

f (2n−1 − k) + f (k + 1) ≥ 2f (2n−2 )

Proof. Use the fact that f (2t + 1) = f (2t) for all t and then apply convexity as above.

Now we can carry out the induction:

(n−2)2 n2
f (2n ) = f (2n−1 ) + f (2n−1 − 1) + · · · > 2n−1 f (2n−2 ) + f (0) > 2n−1 2 4 =2 4 .

10