A1 A grasshopper starts at the origin in the coordinate plane and makes a sequence of hops.

Each
hop has length 5, and after each hop the grasshopper is at a point whose coordinates are both
integers; thus, there are 12 possible locations for the grasshopper after the first hop. What is
the smallest number of hops needed for the grasshopper to reach the point (2021, 2021)?
Answer: 578.
Solution: Each hop can be described by a displacement vector hp, qi with p2 + q 2 = 25 ; the
twelve possible vectors are

h3, 4i; h−3, 4i; h3, −4i; h−3, −4i; h4, 3i; h−4, 3i; h4, −3i; h−4, −3i; h5, 0i; h−5, 0i; h0, 5i; h0, −5i.

One way to write the total displacement as a sum of 578 of these vectors is

h2021, 2021i = 288 · h3, 4i + 288 · h4, 3i + h0, 5i + h5, 0i.

To show that it cannot be done with fewer, note that each hop can increase the sum of the
grasshopper’s coordinates by at most 3 + 4 = 7. Because this sum has to reach

2021 + 2021 = 4042 = 7 · (577) + 3 ,

at least 578 hops are needed.

1
A2 For every positive real number x, let

r1
g(x) = lim (x + 1)r+1 − xr+1 .
r→0

g(x)
Find lim .
x→∞ x
Answer: e.
Solution: Note that for r > −1 and any positive x, we have (x + 1)r+1 − xr+1 > 0. Thus, by
the continuity of the logarithm,

 1

log g(x) = lim log ((x + 1)r+1 − xr+1 ) r
r→0
1
log (x + 1)r+1 − xr+1 .


= lim
r→0 r
Applying L’Hôpital’s rule, we get

(x + 1)r+1 log(x + 1) − xr+1 log x

log g(x) = lim
r→0 (x + 1)r+1 − xr+1
(x + 1) log(x + 1) − x log x
= log (x + 1)x+1 x−x ,


= so
(x + 1) − x
 x
1
g(x) = (x + 1)x+1 x−x = (x + 1) 1 + .
x

Finally,  x
g(x) x+1 1
lim = lim 1+ = 1 · e = e.
x→∞ x x→∞ x x

2
A3 Determine all positive integers N for which the sphere

x2 + y 2 + z 2 = N

has an inscribed regular tetrahedron whose vertices have integer coordinates.

Answer: A necessary and sufficient condition is that N be of the form N = 3m2 , where m is
a positive integer.
Solution 1: To see that the condition is sufficient, note that the four points

(−m, −m, −m), (m, m, −m), (m, −m, m), (−m, m, m)

are the vertices of a regular tetrahedron inscribed in the sphere x2 + y 2 + z 2 = 3m2 .

To show that the condition is necessary, we will use two lemmas:

Lemma 1. If T is a tetrahedron whose vertices have integer coordinates, then its volume is
of the form V (T ) = D/6 for some integer D.

Lemma 2. The √ volume of a regular tetrahedron T inscribed in a sphere of radius R is given

8 3R3
by V (T ) = .
27

Assuming that the sphere x2 + y 2 + z 2 = N has an inscribed regular tetrahedron T whose √

vertices have integer coordinates, we can combine the results of these lemmas (for R = N )
to get √
16 3 N 3/2 16N √
D = 6 V (T ) = = 3N .
9 9
√
Because D is an integer, it follows that 3N is a rational number. Thus the prime factorization
of N must contain an odd number of factors 3 and an even number of factors p for any other
prime p ; therefore, N = 3m2 for some positive integer m.

Proof of Lemma 1: Let P , Q, R, and S be the vertices of the tetrahedron. As these are all
lattice points, the three vectors

P~Q = hq1 , q2 , q3 i, P~R = hr1 , r2 , r3 i, P~S = hs1 , s2 , s3 i

have integer components. We can use a triple product to express the volume as
 
q1 q2 q3
1 ~ ~ ~ 1
V = |P Q · (P R × P S)| = det r1
 r2 r3  ,
6 6
s1 s2 s3

and as the determinant is an integer, we are done.

Proof of Lemma 2: If a regular tetrahedron T is inscribed in a sphere of radius R, we can
choose a coordinate system in which the vertices of T are given by

P = (−d, −d, −d), Q = (d, d, −d), R = (d, −d, d), S = (−d, d, d)

with 3d2 = R2 . We then have

P~Q = h0, 2d, 2di, P~R = h2d, 0, 2di, P~S = h2d, 2d, 0i,

3
and the volume is
√
 
0 2d 2d
1 ~ ~ ~ 1 8 3 8 3R3
V = |P Q · (P R × P S)| = det  2d 0 2d  = d = .
6 6 3 27
2d 2d 0

Solution 2: As in Solution 1, the condition on N is sufficient. To show that it is necessary,

note that if
vi = (xi , yi , zi ), i = 1, 2, 3, 4
are the vertices of a regular tetrahedron inscribed in the sphere x2 + y 2 + z 2 = N , we can add
the four antipodal points

wi = (−xi , −yi , −zi ), i = 1, 2, 3, 4

to get the eight vertices of a cube. (This is easily seen by choosing an alternate coordinate
system as in the proof of Lemma
√ 2 above.) Because this cube is inscribed in thep sphere, its
space diagonals have length 2 N ; therefore, each edge of the cube has length 2 N/3 and its
p 3
volume is 8 N/3 . But the volume of the cube is the determinant of three vectors with
integer coordinates, so it is an integer, and as in Solution 1 it follows that N = 3m2 for some
positive integer m.

4
A4 Let
1 + 2x2 1 + y2
ZZ  
I(R) = 4 2 2 4
− dx dy .
1 + x + 6x y + y 2 + x4 + y 4
x2 +y 2 ≤R2

Find
lim I(R) ,
R→∞

or show that this limit does not exist.

√
π 2 log 2
Answer: The limit exists and equals .
2
Solution: First we symmetrize the integrand. Let

1 + 2x2 1 + y2
f (x, y) = − , so that
1+ x4 2 2
+ 6x y + y 4 2 + x4 + y 4

2 + 2(x2 + y 2 ) 2 + x2 + y 2
f (x, y) + f (y, x) = − and thus
1 + x4 + 6x2 y 2 + y 4 2 + x4 + y 4

2 + 2(x2 + y 2 ) 2 + x2 + y 2
ZZ
2 I(R) = − dx dy.
1 + x4 + 6x2 y 2 + y 4 2 + x4 + y 4
x2 +y 2 ≤R2

Now consider the “first part” of this double integral, say

2 + 2(x2 + y 2 )
ZZ
J(R) = dx dy.
1 + x4 + 6x2 y 2 + y 4
x2 +y 2 ≤R2

Let u = x − y and v = x + y. Then

u2 +v 2 = (x+y)2 +(x−y)2 = 2(x2 +y 2 ), u4 +v 4 = (x+y)4 +(x−y)4 = 2x4 +2y 4 +12x2 y 2

∂(u, v) 1 −1
and = = 2 , so
∂(x, y) 1 1

2 + 2(x2 + y 2 )
ZZ
J(R) = 2 dx dy
2 + 2x4 + 12x2 y 2 + 2y 4
x2 +y 2 ≤R2

2 + u2 + v 2
ZZ
= du dv.
2 + u4 + v 4
u2 +v 2 ≤2R2

Note that if we rename the variables in this last integral x, y instead of u, v, the integrand
will be the same as the integrand of the “second part” of the double integral for 2 I(R) above.
Thus we can recombine the parts to get

2 + x2 + y 2
ZZ
2 I(R) = dx dy.
2 + x4 + y 4
R2 <x2 +y 2 ≤2R2

Converting to polar coordinates, we get

√
2π R 2
2 + r2
Z Z
2 I(R) = r dr dt.
t=0 r=R 2 + r4 (cos4 t + sin4 t)

5
As R → ∞, throughout the range of integration r also goes to infinity and

2r + r3 1
4 = + O(1/r3 ) ,
2+ r4 (cos4 t + sin t) r(cos t + sin4 t)
4

where the error term makes a vanishing contribution to the integral. So

"Z √ # Z
R 2 2π 
dr dt
2 I(R) ∼ 4 4 .
r=R r t=0 cos t + sin t

Now √
Z R 2
dr √ √ 1
= log(R 2) − log(R) = log( 2) = 2 log 2
r=R r
and
Z 2π Z 2π Z 2π
dt dt 2 dt
= =
t=0 cos4 t + sin4 t t=0 1 − 2 sin2 t cos2 t t=0 2 − sin2 (2t)
Z2π
2 dt
= .
t=0 2 cos (2t) + sin2 (2t)
2

π
The integrand is periodic with period and is also even, so we can proceed as follows:
2
Z 2π Z π/4 Z π/4
dt 2 dt 2 dt
= 4 = 8
t=0 cos4 t + sin4 t t=−π/4 2 cos2 (2t) + sin2 (2t) t=0 2 cos2 (2t) + sin2 (2t)
Z π/4
16 sec2 (2t) dt
=
t=0 2 + tan2 (2t)
Z ∞ Z ∞
8 dw 4 dw
= 2
= 2
w=0 2 + w w=0 1 + w /2
√ √ ∞ √
= 4 2 tan−1 (w/ 2) = 2π 2 .
w=0

So √
1 √ √ π 2 log 2
lim 2 I(R) = ( log 2)(2π 2) = π 2 log 2 , and lim I(R) = .
R→∞ 2 R→∞ 2

6
A5 Let A be the set of all integers n such that 1 ≤ n ≤ 2021 and gcd(n, 2021) = 1. For every
nonnegative integer j, let X
S(j) = nj .
n∈A

Determine all values of j such that S(j) is a multiple of 2021.

Answer: All j that are not multiples of 42 or 46.

Solution: Note that modulo 2021, the set A consists precisely of the elements of the mul-
tiplicative group. Multiplying by an element of that group permutes the elements, so if x is
relatively prime to 2021, then
X X
xj · S(j) = (xn)j ≡ mj ≡ S(j) (mod 2021).
n∈A m∈A

Therefore,
(xj − 1) S(j) ≡ 0 (mod 2021).

Also note that 2021 = 2025 − 4 = 452 − 22 = 43 · 47 gives the prime factorization of 2021. Let
x be a primitive root modulo 43 (that is, an integer between 1 and 42 that is a generator of
the cyclic group (Z/43 Z)∗ , which is the multiplicative group of the field with 43 elements).
Then xj − 1 ≡ 0 (mod 43) if and only if j is a multiple of 42 ; also, x is relatively prime to
2021. In particular, if j is not a multiple of 42 we have

(xj − 1) S(j) ≡ 0 (mod 2021) ⇒ (xj − 1) S(j) ≡ 0 (mod 43) ⇒ S(j) ≡ 0 (mod 43) .

Similarly, if y is a primitive root modulo 47 and y 6= 43, we have

(y j − 1) S(j) ≡ 0 (mod 2021) ⇒ (y j − 1) S(j) ≡ 0 (mod 47) ⇒ S(j) ≡ 0 (mod 47)

whenever j is not a multiple of 46. So if j is not a multiple of 42 or 46, then S(j) is a multiple
of both 43 and 47, hence of 2021.

Conversely, suppose that j is a multiple of 42. Then nj ≡ 1 (mod 43) for all n in the sum,
and S(j) is therefore not a multiple of 43 (or of 2021), as
X
S(j) ≡ 1 = 42 · 46 · 1 ≡ −3 ≡ 40 (mod 43).
n∈A

Similarly, if j is a multiple of 46, then S(j) ≡ 5 (mod 47).

7
A6 Let P (x) be a polynomial whose coefficients are all either 0 or 1. Suppose that P (x) can be
written as the product of two nonconstant polynomials with integer coefficients. Does it follow
that P (2) is a composite integer?
Solution: Yes, we will show that P (2) must be composite. Let P (x) = F (x)G(x) have
degree N , where F (x) and G(x) are nonconstant polynomials with integer coefficients, and
suppose that P (2) = p were prime. Then either F (2) or G(2) would be a unit, so without
loss of generality we may assume that G(2) = 1. We have N ≥ 2 because F (x) and G(x) are
nonconstant, and we can write
N
X
P (x) = σn xn ,
n=0
where each σn is either 0 or 1 and σN = 1.
Because the coefficients of P (x) are nonnegative integers, P (x) cannot have a positive real
root, so G(x) cannot have a positive real root either. Thus, as G(2) is positive, G(x) must be
positive for all x > 0. In particular, because σN = 1 is the product of the leading coefficients
of F (x) and G(x), the polynomial G(x) must be monic. Let r1 , . . . , rk be the (complex, not
necessarily distinct) roots of G(x), so that
k
Y
G(x) = (x − rj ) .
j=1

Consider the integer G(1). Because G(x) > 0 for x > 0, we have G(1) ≥ 1 = G(2). In
particular, |G(1)| ≥ |G(2)|, and using the factorization of G(x) we get
k
Y k
Y
|1 − rj | ≥ |2 − rj | .
j=1 j=1

It follows that G(x) must have at least one root ρ with |ρ − 1| ≥ |ρ − 2|, which is equivalent
to Re(ρ) ≥ 23 . This implies that |ρ| ≥ 32 ; note that ρ is also a root of P (x).
First consider the case N = 2. Dividing P (ρ) = 0 by ρ yields
σ0
ρ + σ1 + = 0.
ρ
3
We have Re(ρ) ≥ 2 > 0 and hence
   
1 ρ̄ 1
Re = Re 2
= 2 Re (ρ) > 0.
ρ |ρ| |ρ|
But then  
σ0
Re(ρ) ≤ Re ρ + σ1 + = 0,
ρ
which is a contradiction. (Alternatively, one can check the four possible polynomials P (x) of
degree 2 with coefficients from {0, 1}.)
For N > 2, we again divide P (ρ) = 0 by ρN −1 , which now yields
σN −2 σN −3 σ0
ρ + σN −1 + = − 2 − · · · − N −1 .
ρ ρ ρ

Once again, the terms on the left have nonnegative real parts. The triangle inequality gives
 
σN −2 σN −2 σN −3 σ0
Re(ρ) ≤ Re ρ + σN −1 + ≤ ρ + σN −1 + ≤ + · · · + N −1 ,
ρ ρ |ρ|2 |ρ|

8
and we can estimate the sum on the right using an infinite geometric series:
σN −3 σ0 1 1
+ · · · + N −1 ≤ 2 + · · · + N −1
|ρ|2 |ρ| |ρ| |ρ|
 
1 1 1
≤ 2 1+ + + ···
|ρ| |ρ| |ρ|2
1 1 1
= 2· = .
|ρ| 1 − 1/|ρ| |ρ|(|ρ| − 1)

1 3
But is a decreasing function of x for x > 1, and |ρ| ≥ Re(ρ) ≥ , so we get
x(x − 1) 2
3 1 1 4
≤ Re(ρ) ≤ ≤ 3 3 = ,
2 |ρ|(|ρ| − 1) 2(2 − 1) 3

a contradiction.

Remark. There are polynomials like x7 + x2 + x + 1 = (x + 1)(x2 + 1)(x4 − x3 + 1) or

x7 + x3 + x2 + x + 1 (which is irreducible) which have roots with real part greater than 1. The
polynomial x11 + x3 + x2 + x + 1 has a root r with |r − 2| < 1. Hence one needs to take some
care with this argument.