Problems Ted Eisenberg, Section Editor

*********************************************************
This section of the Journal offers readers an opportunity to exchange interesting mathematical
problems and solutions. Please send them to Ted Eisenberg, Department of Mathematics,
Ben-Gurion University, Beer-Sheva, Israel or fax to: 972-86-477-648. Questions concerning
proposals and/or solutions can be sent e-mail to <eisenbt@013.net>. Solutions to previously
stated problems can be seen at <http://www.ssma.org/publications>.

————————————————————–

Solutions to the problems stated in this issue should be posted before

May 15, 2021

• 5631: Proposed by Kenneth Korbin, New York, NY

Trapezoid ABCD with integer length sides is inscribed in a circle with diameter P 3 where
P is a prime number great than 14. Base AB =7P2 . Express the the lengths of the other
three sides in terms P .

• 5632: Proposed by Toyesh Prakash Sharma (Student), St. C.F. Andrews School, Agra,
India

Show that
√  ◦  ◦  ◦  ◦
1 − 2 sin(89/2)◦ 1 2 3 89
√ < tan2 · tan2 · tan2 · · · tan2 .
1 + 2 sin(89/2)◦ 2 2 2 2

• 5633: Proposed by Goran Conar, Varaz̆ din, Croatia

Calculate: √
n
lim sinh n + tanh n.
n→+∞

• 5634: Proposed by Daniel Sitaru, National Economic College,“Theodor Costescu” Drobeta

Turna-Severin, Romania

a) Find all real numbers x such that tan 3x = tan 2x + tan x.

b) Find: Z  π  π
Ω= tan x + tan 3x tan 2x − dx.
3 3

• 5635: Proposed by José Luis Dı́az-Barrero, Barcelona Tech, Barcelona, Spain

8
17
Assuming that 1 − 3x + 3x2 1 − x − x2 + x3 + 3x4 + 3x5 =

1
 a0 + a1 x + a2 x2 + . . . + a100 x100 + a101 x101 calculate the value of
a1 + a2 + a3 + . . . + a49 + a50 and a0 + a4 + a8 + . . . + a100 .

• 5636: Proposed by Ovidiu Furdui Technical University of Cluj-Napoca, Cluj-Napoca, Ro-

mania

Prove that
∞ ∞
1 2
 
X 1 1 X 1
n! e − 1 − − − · · · − =e .
1! 2! n! n · n!
n=0 n=1

Solutions

• 5613: Proposed by Kenneth Korbin, New York, NY

Given the equations:
 √ p
 3x2 + 6x + 1 = y + 3y 2 − 3
and
 2
ax + by 2 + cxy + dx + ey + f = 0.
Find integers (a, b, c, d, e, f ) so that infinitely many pairs of positive integers (x, y) satisfy
both equations.

Solution 1 by Albert Stadler, Herrliberg, Switzerland

We have p p
3x2 + 6x + 1 − 3y 2 − 3 = y
and obtain p p
3x2 + 6x + 1 + 3y 2 − 3 − y 2 = 2 3x2 + 6x + 1 3y 2 − 3.
We square again and obtain:

0 = (3x2 + 6x + 2y 2 − 2)2 − 4(3x2 + 6x + 1)(3y 2 − 3) =

= (3x2 + 2y 2 − 6xy + 6x − 6y + 4)(3x2 + 2y 2 + 6xy + 6x + 6y + 4).

So either
(i) 3x2 + 2y 2 − 6xy + 6x − 6y + 4 = 0
or
(ii) 3x2 + 2y 2 + 6xy + 6x + 6y + 4 = 0.

Obviously (ii) has no solutions in positive integers x and y. So we must have (a, b, c, d, e, f ) =
(3, 2, −6, 6, −6, 4), and it remains to prove that there are infinitely many pairs of positive
integers (x, y) that satisfy (i). Case (i) is equivalent to

y 2 − 3(x − y + 1)2 = 1.

2
The Pell equation u2 − 3v 2 = 1 has infinitely many solutions in positive integers (u, v).
They are given by (u, v) = (uj , vj ), j ≥ 2 where
√ √ √ √
(2 + 3)j + (2 − 3)j (2 + 3)j − (2 − 3)j
uj = , vj = √ .
2 2 3
If xj = uj + vj − 1, y = uj , then (x, y) = (xj , yj ) is a solution of y 2 − 3(x − y + 1)2 = 1.
Finally,
p p q q
3x + 6x + 1−y− 3y − 3 = 3(uj + vy − 1) + 6(uj + vj − 1) + 1−uj − 3u2j − 3 =
2 2 2

q q
3u2j + 3vj2 + 6uj vj − 2 − uj − 3vj = u2j + 9vj2 + 6uj vj − uj − 3vj =
q
= (uj + 3vj )2 − uj − 3vj = 0.

Solution 2 by Trey Smith, Angelo State University, San Angelo, TX

√
Noting that 3x2 + 6x + 1 = 3(x + 1)2 − 2, and letting z = x + 1 and q = 3z 2 − 2, we
have the associated equation
q 2 − 3z 2 = −2.
p
Letting p = 3y 2 − 3 yields the equation

p2 − 3y 2 = −3.

We make the following claim: If z1 , y1 , p1 , q1 satisfy the following:

1. q 2 − 3z 2 = −2
2. p2 − 3y 2 = −3
3. p = 3z − 3y
4. q = 3z − 2y
5. q = p + y
then so will z2 , y2 , p2 , and q2 where

z2 = 2z1 + q1
y2 = 2y1 + p1
p2 = 3y1 + 2p1
q2 = 3z1 + 2q1 .
The proof of the claim follows:
1.
(q2 )2 − 3(z2 )2 = (3z1 + 2q1 )2 − 3(2z1 + q1 )2

= 9(z1 )2 + 12(z1 )(q1 ) + 4(q1 )2 − 12(z1 )2 − 12(z1 )(q1 ) − 3(q1 )2

= (q1 )2 − 3(z1 )2

= −2.

3
 2.
(p2 )2 − 3(y2 )2 = (3y1 + 2p1 )2 − 3(2y1 + p1 )2

= 9(y1 )2 + 12(y1 )(p1 ) + 4(p1 )2 − 12(y1 )2 − 12(y1 )(p1 ) − 3(p1 )2

= (p1 )2 − 3(y1 )2

= −3.

3.
p2 = 3y1 + 2p1

= 3(q1 − p1 ) + 2(3z1 − 3y1 )

= 3q1 − 3p1 + 6z1 − 6y1

= 3(2z1 + q1 ) − 3(2y1 + p1 )

= 3z2 − 3y2 .
4.
q2 = 3z1 + 2q1

= 3z1 + q1 + q1

= 3z1 + (3z1 − 2y1 ) + q1

= 6z1 − 2y1 + q1

= 6z1 − 4y1 + q1 + 2y1

= 6z1 − 4y1 + q1 + 2(q1 − p1 )

= 6z1 + 3q1 − 4y1 − 2p1

= 3(2z1 + q1 ) − 2(2y1 + p1 )

= 3z2 − 2y2 .
5.
q2 = 3z1 + 2q1

= p1 + 3y1 + 2p1 + 2y1

= 3y1 + 2p1 + 2y1 + p1

= p2 + y2 .
Starting with z1 = 1, y1 = 1, p1 = 0, and q1 = 1 – which do satisfy all five equations –
we have an infinite number of positive integer solutions for the five equations. Then for
any such solution (z, y) (with z > 1) we know that, for x = z − 1, (x, y) will be a positive

4
integer solution to the equation
p p
3x2 + 6x + 1 = y + 3y 2 − 3.

We obtain our coefficient values using the fact that if (z, y, p, q) is obtained using the
above recursion, then,

q =p+y

=⇒ q 2 = (p + y)2

=⇒ q 2 = p2 + 2py + y 2

=⇒ 3z 2 − 2 = 3y 2 − 3 + 2(3z − 3y)y + y 2

=⇒ 3z 2 − 2 = −2y 2 + 6zy − 3

=⇒ 3(x + 1)2 + 2y 2 − 6(x + 1)y + 1 = 0

=⇒ 3x2 + 2y 2 − 6xy + 6x − 6y + 4 = 0.
Letting n be any real number, the coefficients are

[a, b, c, d, e, f ] = [3n, 2n, −6n, 6n, −6n, 4n].

Obviously, n can be chosen to insure integer coefficients.

Note: the actual recursive formulas for x and y that will work with the two equations are

x2 = 5x1 − 2y1 + 4
y2 = 3x1 − y1 + 3,
where the initial pair is x1 = 0 and y1 = 1. These are easily obtained using the four
recursive formulas and the claim.

Solution 3 by Albert Natian, Los Angeles Valley College, Valley Glen, Cali-
fornia

Answer: (a = 3, b = 2, c = −6, d = 6, e = −6, f = 4).

Broadly (a = 3µ, b = 2µ, c = −6µ, d = 6µ, e = −6µ, f = 4µ) with µ ∈ Z.
Derivation:
By way of Brahmagupta’s identity

(ax + Hby)2 − H (ay + bx)2 = a2 − Hb2 x2 − Hy 2 ,

we
√ will develop formulas
p that generate infinitely many positive integers x and y for which
3x2 + 6x + 1 and 3y 2 − 3 are integers. First set u = x+1 so that 3x2√
+6x+1 = 3u2 −2.
We would like the quantity 3u2 − 2 to be a perfect square so that 3x2 + 6x + 1 or
√
3u2 − 2 is an integer. That is, for some integer v, we want
p
3u2 − 2 = v, 3u2 − 2 = v 2 , v 2 − 3u2 = −2.

5
So, we ask: For what integer values of u and v does the preceding Diophantine equation
hold? A general form of the above Diophantine equation is

x2 − Hy 2 = k

where all the variables are integers, with H > 0. Suppose we know of two triplets
(x1 , y1 , k1 ) and (x2 , y2 , k2 ) of integers that satisfy the equations

x21 − Hy12 = k1 and x22 − Hy22 = k2 .

Then, by Brahmagupta’s identity

(x1 x2 + Hy1 y2 )2 − H (x1 y2 + x2 y1 )2 = x21 − Hy12 x22 − Hy22 = k1 k2 ,

the triplet
(x1 x2 + Hy1 y2 , x1 y2 + x2 y1 , k1 k2 )
satisfies the equation
X 2 − HY 2 = k1 k2 .
So, in particular, if the triplets (x0 , y0 , 1) and (xn , yn , k) of integers satisfy the equations

x20 − Hy02 = 1 and x2n − Hyn2 = k,

then the triplet

(xn+1 , yn+1 , k) := (x0 xn + Hy0 yn , x0 yn + xn y0 , k)

of integers satisfies the equation

x2n+1 − Hyn+1
2
= k.

This leads to a recursion:  

xn+1 = x0 xn + Hy0 yn
,
yn+1 = y0 xn + x0 yn
      n  
xn+1 x0 Hy0 xn x0 Hy0 x1
= =
yn+1 y0 x0 yn y0 xt0 y1

where (x1 , y1 ) satisfies x21 − Hy12 = k.

Via diagonalization (assuming distinct eigenvalues), we have

 n
x0 Hy0
  =
y0 x0
 √ h √  n  √  ni h √ n  √ n i 
H x 0 + y0 H + x 0 − y0 H H x0 + y0 H − x0 − y0 H
1  
√  .
2 H  √ n  √ n √ h √ n  √ n i 
x0 + y0 H − x0 − y0 H H x0 + y0 H + x0 − y0 H
Thus

1
h √  √ n  √  √ n i 
 xn+1 =

 2 x1 + y1 H x0 + y0 H + x1 − y1 H x0 − y0 H 


.
 h √  √ n  √  √ n i 
 yn+1 = √1 x1 + y1 H x0 + y0 H − x1 − y1 H x0 − y0 H
 

2 H

6
Recalling the equation v 2 − 3u2 = −2, we see that (v0 = 2, u0 = 1) satisfies v 2 − 3u2 = 1
and (v1 = 1, u1 = 1) satisfies v 2 − 3u2 = −2 . Thus

1
h √
√
n √
√
n i 


 v n+1 = 2 1 + 3 2 + 3 + 1 − 3 2 − 3 


.
 1
h √
√
n √
√
n i 
 un+1 = √ 1+ 3 2+ 3 − 1− 3 2− 3
 

2 3

√ 1
√
2
Since 2 ± 3= 2 1 ± 3 , then the preceding can be written as

1
h √
2n−1 √
2n−1 i 


 v n = 2n 1 + 3 + 1 − 3 


.
 1√
h √
2n−1 √
2n−1 i 
 un = 1+ 3 − 1− 3
 

2n 3
p
Now we turn to the expression 3y 2 − 3, which we would like to be an integer, say, z.
Thus p
3y 2 − 3 = z, 3y 2 − 3 = z 2 , z 2 − 3y 2 = −3.
The triplets (z0 = 2, y0 = 1, 1) and (z1 = 3, y1 = 2, −3) satisfy

z02 − 3y02 = 1 and z12 − 3y12 = −3.

Thus  h √
√
m √
√
m i 
1
 zm+1 =

 2 3+2 3 2+ 3 + 3−2 3 2− 3 


 1
h √
√
m √
√
m i 
 ym+1 = √ 3+2 3 2+ 3 − 3−2 3 2− 3
 

2 3

which can be written as

 √
3
h √
2m √
2m i 
 zm =

 2m+1
1+ 3 − 1− 3 


.
 1
h √
2m √
2m i 
 ym = 1+ 3 + 1− 3
 

2m+1

Inserting the above results into the original equation

p p p p
3x2 + 6x + 1 = y + 3y 2 − 3 or 3u2 − 2 = y + 3y 2 − 3

or
v = y + z,
we have
vn = ym + zm ,

√ 2n−1  √ 2n−1 √ 2m  √ 2m

   
1  1
1+ 3 + 1− 3 = m+1 1 + 3 + 1− 3 +
2n 2
√ 
√ 2m  √ 2m

3
+ m+1 1 + 3 − 1− 3 ,
2

7
 √ 2n−1  √ 2n−1 √ 2m+1  √ 2m+1
   
1  1
1+ 3 + 1− 3 = m+1 1 + 3 + 1− 3 ,
2n 2

√ 2n−1  √ 2n−1 √ 2(m+1)−1  √ 2(m+1)−1

   
1  1
1+ 3 + 1− 3 = m+1 1 + 3 + 1− 3 ,
2n 2
which holds if n = m + 1.

Since u = x + 1 or x = u − 1, then the given equation

ax2 + by 2 + cxy + dx + ey + f = 0

becomes
au2 + by 2 + cuy + (d − 2a) u + (e − c) y + (a − d + f ) = 0.

Claim: For a = 3µ, b = 2µ, c = −6µ, d = 6µ, e = −6µ, f = 4µ with µ ∈ Z, there

exist infinitely many pairs of positive integers (x, y) that satisfy the system
 √ p 
 3x2 + 6x + 1 = y + 3y 2 − 3 
.
 2 2
ax + by + cxy + dx + ey + f = 0


Proof: Since d − 2a = 0, e − c = 0 and a − d + f = 1, then the equation

au2 + by 2 + cuy + (d − 2a) u + (e − c) y + (a − d + f ) = 0

becomes
3u2 + 2y 2 − 6uy + 1 = 0.
√ √
Letting p = 1 + 3 and q = 1 − 3, we insert the above values for ym and um+1 into the
latter equation and write

 2  2
   
1  2m+1 2m+1 1  2m 2m
3 √ p −q +2 p +q
2m+1 3 2m+1
  
1 1  2m
√ p2m+1 − q 2m+1 2m
  
−6 p + q + 1 = 0.
2m+1 3 2m+1
We will show the latter equation holds for all positive integer values of m. Taking advan-
tage of the fact that pq = −2, we simplify the above as
3 2
p4m+2 + 22m+2 + q 4m+2 + 2m+2 p4m + 22m+1 + q 4m



3·2 2m+2 2
6
p4m+1 + 22m p − 22m q − q 4m+1 + 1 = 0


− √
22m+2 3
both sides of which we multiply by 22m+2 to get

p4m+2 + 22m+2 + q 4m+2 + 2 p4m + 22m+1 + q 4m

√
−2 3 p4m+1 + 22m p − 22m q − q 4m+1 + 22m+2 = 0,

8
 p4m+2 + 22m+2 + q 4m+2 + 2 p4m + 22m+1 + q 4m



√
−2 3 p4m+1 + [p − q] · 22m − q 4m+1 + 22m+2 = 0,

p4m+2 + 22m+2 + q 4m+2 + 2 p4m + 22m+1 + q 4m

√  h √ i 
−2 3 p4m+1 + 2 3 · 22m − q 4m+1 + 22m+2 = 0,

h √ i h √ √ i
p4m+2 + 2p4m − 2 3p4m+1 + 22m+2 + 2 · 22m+1 − 2 3 · 2 3 · 22m
h √ i
+ q 4m+2 + 2q 4m + 2 3q 4m+1 + 22m+2 = 0. (?)

Since √  √ 
p4m+2 + 2p4m − 2 3p4m+1 = p4m p2 + 2 − 2 3p = 0
and √  √ 
q 4m+2 + 2q 4m + 2 3q 4m+1 = q 4m q 2 + 2 + 2 3q = 0,
then the above equation (?) is simplified to
h √ √ i
22m+2 + 2 · 22m+1 − 2 3 · 2 3 · 22m + 22m+2 = 0

which obviously holds for all integer values of m. This completes the proof.

Solution 4 by Kee-Wai Lau, Hong Kong, China

We show that we can take (a, b, c, d, e, f ) = (3, 2, −6, 6, −6, 4).
√ √ √ √
(1 + 3)(2 + 3)k − (1 − 3)(2 − 3)k
For positive integers k let xk = √ − 1 and yk =
√ k √ k 2 3
(2 + 3) + (2 − 3)
. It is easy to see that both xk and yk are positive integers.
2
It can be checked readily that
√ √ √ √
(1 + 3)(2 + 3)k + (1 − 3)(2 − 3)k
q q
2 2
3xk + 6xk + 1+ = yk + 3yk − 3 =
2
and
3x2k + 2yk2 − 6xk yk + 6xk − 6yk + 4 = 0.
Hence our claim.

Solution 5 by Charles Burnette, Xavier University of Louisiana (New Orleans,

LA)

Let (x, y) be a pair of positive integer solutions to the first equation. If we square both
sides of the first equation and isolate the radical term remaining its right side, we get
p
3x2 − 4y 2 + 6x + 4 = 2y 3y 2 − 3. (1)

9
Squaring (1) yields
9x4 + 16y 4 − 24x2 y 2 + 36x3 − 48xy 2 + 60x2 − 32y 2 + 48x + 16 = 12y 4 − 12y 2 . (2)
This simplifies to
9x4 + 4y 4 − 24x2 y 2 + 36x3 − 48xy 2 + 60x2 + 12y 2 + 48x + 16 = 0. (3)
The left side of (3) can be reorganized to beget a difference of two squares:
(9x4 + 4y 4 + 12x2 y 2 + 36x3 + 24xy 2 + 60x2 + 48y 2 + 48x + 16) − (36x2 y 2 + 72xy 2 + 36y 2 )
= (3x2 + 2y 2 + 6x + 4)2 − (6xy + 6y)2 .
We therefore end up with the equation
(3x2 + 2y 2 + 6xy + 6x + 6y + 4)(3x2 + 2y 2 − 6xy + 6x − 6y + 4) = 0. (4)
In assuming that x and y are positive, we must then have
3x2 + 2y 2 − 6xy + 6x − 6y + 4 = 0. (5)
Because the above steps are reversible provided that
1p 2
1 ≤ |y| ≤ 3x + 6x + 4,
2
it suffices to show that (5) has infinitely many positive integer solutions in this range.
A Legendre transformation changes (5) into a Pell-type diophantine equation. Indeed,
u2 − 3v 2 = −2, (6)
with u = 3x − 2y + 3 and v = x + 1, is equivalent to doubling (5). Equation (6) has a
particular solution of (1, 1), and since 3 is not a perfect square, the corresponding Pell
resolvent u2 − 3v 2 = 1 and, consequently, (6) have infinitely many integer solutions in
each quadrant. Furthermore, if (u, v) is an integer solution to (6), then u and v must
share the same parity. As a result, x and
3x + 3 − u 3v − u
y= =
2 2
provide integer solutions of (5) whenever (u, v) is an integer solution of (6). If we also
specifically restrict our attention to positive solutions for u and v so that x ≥ 1 and
p p
u = 3v 2 − 2 = 3x2 + 6x + 1,
then
√ √
3x + 3 − 3x2 + 6x + 1 3x + 3 − 9x2 + 6x + 1 3x + 3 − (3x + 1)
y= ≥ = = 1,
2 2 2
and
√
3x + 3 − 3x2 + 6x + 1
y=
√ 2 √
9x + 18x + 9 − 3x2 + 6x + 1
2
=
√ 2 √
9x2 + 18x + 3x2 + 6x + 4 − 3x2 + 6x + 1
≤
√ √2
12x2 + 24x + 4 − 3x2 + 6x + 1
=
√ 2√
2
3x + 6x + 1 3x2 + 6x + 4
= ≤ ,
2 2

10
 as required. We can now conclude that
a = 3, b = 2, c = −6, d = 6, e = −6, f = 4.

Solution 6 by David Stone and John Hawkins, Georgia Southern University,

Statesboro, GA
Editor0 s Comment: The solution by the above authors contains some nice graphics and
interesting comments to ponder, but I received it too close to the deadline for submitting
papers for me to recopy it into the format of the other solutions in this issue. However,
I believe their solution to be very instructive and so I have included it as an appendix to
this March 2021 issue of the column. It listed as a separate pdf file.

Comments about 5613 from the author, Kenneth Korbin:

The roots of the first equation are:

(x, y) = (10, 7), (152, 97), (2130, 1351), · · · ,

x = (10, 152, 2130, · · · ) with
xN +1 = 14xN − xN −1 + 12.

y = (7, 97, 1351, · · · ) with

yN +1 = 14yN − yN −1 .

x − y = (3, 55, 779, 1063, · · · ),

The second equation can be written as: 3x2 + 2y 2 − 6xy + 6x − 6y + 4 = 0

Also solved by Peter Fulop, Gyomro, Hungary, and the proposer.

• 5614: Proposed by Michael Brozinsky, Central Islip, NY

Solve:
       
θ θ θ θ
cos2 θ + 6 cos(θ) cos + 9 cos2 = sin2 θ − 6 sin θ sin + 9 sin2 .
3 3 3 3
Solution 1 by Dionne Bailey, Elsie Campbell, and Charles Diminnie, Angelo
State University, San Angelo, TX
We will make repeated use of the following basic trig identities:
a. cos (2x) = cos2 x − sin2 x
b. cos (x ± y) = cos x cos y ∓ sin x sin y
c. sin (x ± y) = sin x cos y ± cos x sin y
d. cos (3x) = 4 cos3 x − 3 cos x
e. sin (3x) = 3 sin x − 4 sin3 x.
To begin, the given equation can be re-written in the form
         
2 2

θ θ 2 θ 2 θ
cos θ − sin θ + 6 cos θ cos + sin θ sin + 9 cos − sin = 0.
3 3 3 3
Using identities a and b above, this equation becomes
   
θ 2θ
cos (2θ) + 6 cos θ − + 9 cos = 0,
3 3

11
i.e.,  
2θ
cos (2θ) + 15 cos = 0.
3
 
2θ
Since 2θ = 3 , identity d implies that
3
      
2θ 3 2θ 2θ
cos (2θ) = cos 3 = 4 cos − 3 cos
3 3 3

and our equation reduces to

     
3 2θ 2θ 2θ
4 cos − 3 cos + 15 cos =0
3 3 3
   
3 2θ 2θ
4 cos + 12 cos =0
3 3
    
2θ 2θ
cos cos2 + 3 = 0.
3 3
 
2 2θ
Because cos + 3 > 0, we are left with
3
 
2θ
cos =0
3

whose solutions are

2θ π
= + nπ,
3 2
i.e.,  
3π 3π
θ= +n (1)
4 2
for all n ∈ Z.
To complete our solution, we must check whether (1) contains any extraneous solutions.
To aid in this task, we note first that identities d and e yield
      2
2 θ 2 θ θ
cos θ + 6 cos θ cos + 9 cos = cos θ + 3 cos
3 3 3
      2
3 θ θ θ
= 4 cos − 3 cos + 3 cos
3 3 3
 
θ
= 16 cos6 (2)
3

and
      2
2 θ 2 θ θ
sin θ − 6 sin θ sin + 9 sin = sin θ − 3 sin
3 3 3
      2
θ 3 θ θ
= 3 sin − 4 sin − 3 sin
3 3 3
 
6 θ
= 16 sin . (3)
3

12
  
3π 3π θ π π 
If θ = +n , then = + n and identity b implies that
4 2 3 4 2
 
θ hπ  π i
cos = cos +n
3 4 2
π   π π   π
= cos cos n − sin sin n
√ h4   2 4 2
2 π  π i
= cos n − sin n .
2 2 2
If n = 2k, we get
  √
θ 2
cos = [cos (kπ) − sin (kπ)]
3 2
√ h
2 i
= (−1)k − 0
2 √
k 2
= (−1) (4)
2
while n = 2k + 1 yields
  √ h
θ 2 π πi
cos = cos (2k + 1) − sin (2k + 1)
3 2 2 2
√ h
2  π i
= 0 − sin kπ +
2√ 2
2 h  π  π i
=− sin (kπ) cos + cos (kπ) sin
√2 2 2
2
=− [0 + cos (kπ)]
√2
2
=− (−1)k
2 √
k+1 2
= (−1) (5)
2
If n = 2k, conditions (2) and (4) imply that
     
2 θ 2 θ 6 θ
cos θ + 6 cos θ cos + 9 cos = 16 cos
3 3 3
" √ #6
k 2
= 16 (−1)
2
 
1
= (16)
8
= 2.

13
If n = 2k + 1, conditions (2) and (5) give
     
2 θ 2 θ 6 θ
cos θ + 6 cos θ cos + 9 cos = 16 cos
3 3 3
" √ #6
2
= 16 (−1)k+1
2
 
1
= 16
8
= 2.
 
3π 3π
Hence, for all n ∈ Z, θ = +n forces
4 2
   
2 θ 2 θ
cos θ + 6 cos θ cos + 9 cos = 2.
3 3
 
3π 3π θ π π 
As before, if θ = +n , then = + n and identity c gives
4 2 3 4 2
 
θ hπ  π i
sin = sin +n
3 4 2
π  π  π  π 
= sin cos n + cos sin n
√ h4   2 4 2
2 π  π i
= cos n + sin n .
2 2 2
If n = 2k, then
  √
θ 2
sin = [cos (kπ) + sin (kπ)]
3 2
√ h
2 i
= (−1)k + 0
2 √
k 2
= (−1) ,
2
θ π π 
while n = 2k + 1 leads to = + (2k + 1) and identity c gives
3 4 2
  √ h
θ 2  π  π i
sin = cos (2k + 1) + sin (2k + 1)
3 2 2 2
√ h
2  π i
= 0 + sin kπ +
√2 h 2
2 π   π i
= sin (kπ) cos + cos (kπ) sin
√2 h 2 2
2 i
= 0 + (−1)k
2 √
k 2
= (−1) .
2

14
In both situations, condition (3) yields
     
2 θ 2 θ 6 θ
sin θ − 6 sin θ sin + 9 sin = 16 sin
3 3 3
" √ #6
2
= 16 (−1)k
2
 
1
= 16
8
= 2.
 
3π 3π
Thus, for all n ∈ Z, θ = +n makes
4 2
     
2 θ 2 θ 6 θ
cos θ + 6 cos θ cos + 9 cos = 16 cos
3 3 3
=2
 
6 θ
= 16 sin
3
   
2 θ 2 θ
= sin θ − 6 sin θ sin + 9 sin
3 3

and (1) contains no extraneous solutions. This completes our solution.

Solution 2 by David A. Huckaby, Angelo State University, San Angelo, TX

We have

       
θ θ θ θ
cos2 θ − sin2 θ + 6 cos θ cos + 6 sin θ sin + 9 cos2 − 9 sin2 =0
3 3 3 3
   
2θ 2θ
cos 2θ + 6 cos + 9 cos =0
3 3
 
2θ
cos 2θ + 15 cos =0
3
     
3 2θ 2θ 2θ
4 cos − 3 cos + 15 cos =0
3 3 3
   
2θ3 2θ
4 cos + 12 cos =0
3 3
    
2θ 2 2θ
4 cos cos +3 =0
3 3

So cos 2θ 2θ π 3π 3π


3 = 0, whence 3 = 2 + πn for any integer n, so that θ = 4 + 2 n for any
integer n.

Solution 3 by Titu Zvonaru, Comănesti, Romania,

15
We have
 
θ θ
cos θ = cos 4 cos3 − 3
3 3
   
θ θ θ θ θ θ θ
⇒ cos2 θ + 6 cos θ cos + 9 cos2 = cos2 16 cos4 + 9 − 24 cos2 + 6 cos2 4 cos2 −3 +
3 3 3 3 3 3 3
 
θ θ θ θ
cos2 16 cos4 + 9 − 24 cos2 + 24 cos2 − 18 + 9
3 3 3 3

θ
= 16 cos6
3
 
θ 2 θ
sin θ = sin 3 − 4 sin =
3 3
   
θ θ θ 4 θ 2θ 2 θ 2 θ
= sin θ − 6 sin θ sin + 9 sin2 = sin2
2
16 sin + 9 − 24 − 6 sin 3 − 4 sin
3 3 3 3 3 3 3
θ
+ 9 sin2 =
3
 
θ2 θ
4 θ θ θ
= sin 16 sin + 9 − 24 sin2 + 24 sin2 − 18 + 9 = 16 sin6 .
3 3 3 3 3

We obtain the equation,

θ θ θ
tan6 = 1 ⇒ tan3 = 1, tan3 = −1
3 3 3
3π
hence, θ = 4 + 3kπ, θ = − 3π
4 + 3kπ where k is an integer.

Solution by 4 by Michel Bataille, Rouen, France

From the formulas cos 2x = cos2 x − sin2 x, cos a cos b + sin a sin b = cos(a − b), we deduce
that the given equation is equivalent to
2θ 2θ
cos 2θ + 6 cos + 9 cos = 0. (1)
3 3
2θ
Using cos 3x = 4 cos3 x − 3 cos x, (1) can be written as 4 cos3 3 − 3 cos 2θ 2θ
3 + 15 cos 3 = 0,
that is,  
2θ 2θ
4 cos 3 + cos2 = 0.
3 3
2θ
Since 3 + cos2 > 0 for all real θ, the latter reduces to cos 2θ
3 3 = 0. Thus, the solutions
3π 3kπ
are the numbers + , k ∈ Z.
4 2
Solution 5 by David E. Manes, Oneonta, NY

16
 3π 3nπ
The solution of the equation is θ = + , where n is any integer.
4 2
The given equation is equivalent to
         
θ θ θ θ
(cos2 θ − sin2 θ) + 6 cos(θ) cos + sin(θ) sin + 9 cos2 − sin2 = 0.
3 3 3 3

The addition formula for cosine and the double-angle formula (cos 2a = cos2 a − sin2 a)
imply
        
2θ 2θ 2θ 2θ
cos(2θ) + 6 cos + 9 cos = cos(2θ) + cos + 14 cos = 0.
3 3 3 3

By the sum-to-product formula, cos(2θ) + cos(2θ/3) = 2 cos(4θ/3) · cos(2θ/3). Therefore,

the equation reduces to
          
4θ 2θ 2θ 2θ 4θ
2 cos · cos + 14 cos = 2 cos cos + 7 = 0.
3 3 3 3 3
   
2θ 4θ
Hence, cos = 0 since cos + 7 6= 0 for any value of θ. Therefore,
3 3
2θ π 3π 3nπ
= + nπ or θ = + ,
3 2 4 2
for any integer n. This completes the solution.

Solution 6 by Albert Stadler Herrliberg, Switzerland

θ
Let z = cos 3θ + i sin 3θ = ei 3 . Then z 3 = eiθ = cos θ + i sin θ. The above equation is
equivalent to each of the following lines:

θ 2 θ 2
   
cos θ + 3 cos = sin θ − 3 sin ,
3 3

!2 !2
1
z3 + z3 z+ 1
z
z 3 − z13 z − z1
+3 = −3 ,
2 2 2i 2i
2
3 2
  
1 3 1
z3 + + 3z + + z 3 − 3 − 3z + = 0,
z3 z z z

1 15
z6 + 6
+ 15z 2 + 2 = 0,
z z
3
12
  
2 1 2
z + 2
+ 12 z + = 0.
z z
√ √
The equation u3 + 12u = 0 has the roots u = 0, u = 2i 3, u = −2i 3.
πi
1
z2 + z2
= 0, has the four roots z = e 4 ij , j = 0, 1, 2, 3.
√  √   √ 
z 2 + z12 = 2i 3 has the four roots z = (1 + i) 1+2 3 , (−1 − i) 1+2 3 ,
 √   √ 
(1-i) −1+2 3 , (−1 + i) −1+2 3 , none of which has modulus 1.

17
 √  √   √ 
z2 + 1
= −2i 3has the four roots z = (1 + i) −1+2 3 , (−1 − i) 1+2 3 ,
z2
 √   √ 
(1 − i) 1+2 3 , (−1 + i) 1+2 3 , none of which has modulus 1.
θ πi
j
Therefore the roots of the given equation satisfy ei 3 =e 4 i ,j=0,1,2,3 and thus the given
equation has the roots θ = 3π
4 (2j + 1), where j is an arbitrary integer.
Comment: the Mathematica command
TrigFactor[Cos[x]2 +6Cos[x]Cos[x/3]+9Cos[x/3]2 -Sin[x]2 -6Sin[x]Sin[x/3]+9Sin[x/3]2 )]
produces the output
  
-8Sin π4 − x3 Sin π4 + x3
     2x   2x 
− 2 + Sin 3 2 + Sin 3 .

It’s clear that ±2+sin((2x/3) = 0 has no realist,since | sin(y)| ≤ 1 for all real y. Therefore
the roots of the given equation are the roots of sin(π/4 − x/3) sin(π/4 + x/3) = 0, who
3π
are given by θ = (2j + 1) where j is an arbitrary integer.
4
Solution 7 by Daniel Văcaru, Pitesti, Romania
Equivalently:
cos2 ϑ + 6 cos ϑ cos ϑ3 + 9 cos2 ϑ3 = sin2 ϑ − 6 sin ϑ sin ϑ3 + 9 sin2 ϑ3 ⇔ cos2 ϑ − sin2 ϑ +
6 cos ϑ cos ϑ3 + 6 sin ϑ sin ϑ3 + 9 cos2 ϑ3 − 9 sin2 ϑ3 = 0,
which is
 
ϑ 2ϑ 2ϑ 2ϑ 2ϑ
cos 2ϑ+6 cos ϑ − +9 cos = 0 ⇔ cos 2ϑ+6 cos +9 cos = 0 ⇔ cos 2ϑ+15 cos = 0 (1) .
3 3 3 3 3

We use cos 3x = 4 cos3 x−3 cos x, and (1) becomes 4 cos3 2ϑ 2ϑ 2ϑ 2 2ϑ + 12 =

3 +12 cos 3 = 0 ⇔ cos 3 4 cos 3
0 (2).
We obtain cos 2ϑ 2ϑ 3 π


3 = 0, 3 ∈ {± arccos 0 + 2kπk ∈ Z} ⇔ ϑ ∈ 2 ± 2 + 2kπ k ∈ Z =
± 3π
4 + 3Zπ.

Solution 8 by HyunBin Yoo, South Korea

Move all of the terms to the left-hand
! side.
!! ! !!
θ θ θ θ
(cos2 θ − sin2 θ) + 9 cos2 − sin2 + 6 cos(θ) cos + sin(θ) sin =0
3 3 3 3

Use the angle sum identities

! to further !simplify the equation. ! !
θ θ θ 2 2
cos(θ+θ)+9 cos + +6 cos θ − = cos(2θ)+9 cos θ +6 cos θ = cos(2θ)+
3 3 3 3 3
15 cos(2θ) = 0
!
2 2
Notice that 2θ is the triple of θ. So we can express cos(2θ) in terms of cos θ using
3 3
the angle sum identities.
! ! ! ! ! ! ! !
2 4 2 4 2 4 2 2
cos(2θ) = cos θ+ θ = cos θ cos θ −sin θ cos θ = cos θ 2 cos2 θ −1 −
3 3 3 3 3 3 3 3

18
 ! ! !! ! ! ! !
2 2 2 3
2 2 2
2 2
sin θ 2 sin θ cos θ = 2 cos θ − cos θ − 2 sin θ cos θ =
3 3 3 3 3 3 3
! !! ! ! !! !
2 2 2 2 2 2
2 cos3 θ − 1 + 2 sin2 θ cos θ = 2 cos3 θ − 3 − 2 cos2 θ cos θ =
3 3 3 3 3 3
! !
2 2
4 cos3 θ − 3 cos θ
3 3
! !
2 2
Substituting the cos(2θ) above, we get 4 cos3 θ
+ 12 cos θ
= 0. Factoring it
3 3
! ! ! !
2 2 2
results in 4 cos3 θ cos2 θ + 3 = 0. Since cos2 θ + 3 can never be zero, the
3 3 3
!
2
only solution is cos θ = 0.
3

2 π
θ = 2nπ ± (n ∈ Z)
3 2
3
θ = 3nπ ± π(n ∈ Z)
4
Also solved by Brian D. Beasley, Presbyterian College, Clinton, SC; Anthony
Bevelacqua, University of North Dakota, Grand Falls, ND; Brian Bradie,
Christopher Newport University, Newport News,VA; Charles Burnette, Xavier
University of Louisiana (New Orleans, LA); Bruno Salgueiro Fanego, Viveiro
Spain; Peter Fulop, Gyomro, Hungary; Paul M. Harms, North Newton, KS;
Kee-Wai Lau, Hong Kong, China; Albert Natian, Los Angeles Valley Col-
lege,Valley Glen, CA; SQ Mathematical Problem Solving Group,Yogyakarta,
Indonesia; Seán M. Stewart, Bomaderry, NSW, Australia; David Stone and
John Hawkins, Georgia Southern University, Statesboro, GA, and the pro-
poser.

• 5615: Proposed by Pedro Henrique Oliveira Pantoja, University of Campina Grande,

Brazil
Solve in < × < the system:
 √
3
√
3
√
3
 2x + 2 + 4 − x + 2 − x = 2
√ √ √
5
20 − 2y + 5
7−y+ 5
3y + 5 = 2

Solution 1 by Michel Bataille, Rouen, France

As a lemma, we first show the following: Let a, b, c be real numbers such that a3 +b3 +c3 =
(a + b + c)3 = 8 (resp. a5 + b5 + c5 = (a + b + c)5 = 32). Then, one of a, b, c is equal to 2.
Proof. We introduce the polynomial P (X) = (X −a)(X −b)(X −c) = X 3 −2X 2 +mX −p
where p = abc, m = ab + bc + ca. From the identity

a3 + b3 + c3 − 3abc = (a + b + c)((a + b + c)2 − 3(ab + bc + ca)),

19
we deduce that 8 = 3abc + 2(4 − 3(ab + bc + ca)), that is, p = 2m. As a result,
P (X) = X 3 − 2X 2 + mX − 2m = (X 2 + m)(X − 2),
hence one of the roots a, b, c of P is 2.
In the case when a5 + b5 + c5 = (a + b + c)5 = 32, Newton’s formulas successively give
a2 + b2 + c2 = 4 − 2m, a3 + b3 + c3 = 8 + 3p − 6m, a4 + b4 + c4 = 16 + 2m2 + 8p − 16m and
a5 + b5 + c5 = 32 + 20p − 40m + 10m2 − 5pm and the latter yields (p − 2m)(4 − m) = 0.
However, we have m 6= 4 since otherwise P 0 (X) = 3X 2 − 4X + 4 is positive and P , as a
strictly monotone function, could not have three real roots. Thus, we again have p = 2m
and we conclude as above. √ √ √
Turning to the problem, if x is a solution, let us set a = 3 2x + 2, b = 3 4 − x, c = 3 2 − x.
Then we see that a + b + c = 2 and a3 + b3 + c3 = 8. From the lemma, we obtain a = 2 or
b = 2 or c = 2, hence x = 3 or x = −4 or x = −6. Conversely, it is readily checked that
3, −4, −6 are indeed solutions. Thus, the solutions to the first equation are 3, −4, −6.
√
Similarly, if y is a solution to the second equation, this time setting a = 5 20 − 2y, b =
√ √
5
7 − y, c = 5 3y + 5, we obtain a + b + c = 2 and a5 + b5 + c5 = 32. The lemma gives
that a or b or c is equal to 2, which leads to y = −6 or y = −25 or y = 9. Conversely
each of these three numbers is a solution. Thus, the solutions to the second equation are
−6, −25 and 9.

Solution 2 by David E. Manes, Oneonta, NY

By inspection, the solutions (x, y) of the system are: (3, −6), (3, 9), (3, −25), (−4, −6),
(−4, 9), (−4, −25), (−6, −6), (−6, 9) and (−6, −25). One verifies that each of these nine
ordered pairs is a solution of the system. They are the only solutions.
For the first equation in the system, consider the equivalent equation (2x + 2)1/3 + (4 −
x)1/3 = 2−(2−x)1/3 . Raising each side of the equation to the third power and simplifying,
one obtains the following
h i h i
1/3 1/3 1/3 1/3 1/3
3(2x + 2) (4 − x) 2 − (2 − x) = −3 · 2(2 − x) 2 − (2 − x) .

Noting that (2x + 2)1/3 + (4 − x)1/3 = 2 − (2 − x)1/3 . Therefore,

(2x + 2)1/3 (4 − x)1/3 = −2(2 − x)1/3
provided 2 − (2 − x)1/3 6= 0. If 2 − (2 − x)1/3 = 0, then (2 − x)1/3 = 2 and x = −6,
a solution for the first equation in the system. Cubing each side of the above displayed
equation, one obtains (2x + 2)(4 − x) = −8(2 − x). The resulting quadratic equation
x2 + x − 12 = 0 has roots x = −4 and x = 3, both of which are solutions to the first
equation in the system.
For the second equation in the system, consider the equation
h i5 h i5
(20 − 2y)1/5 + (7 − y)1/5 = 2 − (3y + 5)1/5 .

Then expanding and simplifying, one obtains

h i h     i
2 − (3y + 5)1/5 · (uv)1/5 u2/5 + (uv)1/5 + v 2/5 + 2 w1/5 22 − 2w1/5 + w2/5 = 0,

where u = 20 − 2y, v = 7 − y and w = 3y + 5. Therefore, 2 − (3y + 5)1/5 = 0 implies

y = 9. Similarly, [(7 − y)1/5 + (3y + 5)1/5 ]5 = [2 − (20 − 2y)1/5 ]5 implies
h i h    i
2 − (20 − 2y)1/5 · (uv)1/5 u2/5 + (uv)1/5 + v 2/5 + 2w1/5 22 − 2w1/5 + w2/5 = 0,

20
where u = 7 − y, v = 3y + 5 and w = 20 − 2y. Therefore, 2 − (20 − 2y)1/5 = 0 implies
y = −6. Finally, if [(20 − 2y)1/5 + (3y + 5)1/5 ]5 = [2 − (7 − y)1/5 ]5 , then
h    i
[2 − (7 − y)1/5 ] · (uv)1/5 u2/5 + (uv)1/5 + v 2/5 + 2w1/5 22 − 2w1/5 + w2/5 = 0,

where u = 20 − 2y, v = 3y + 5 and w = 7 − y. Therefore, 2 − (7 − y)1/5 = 0 implies

y = −25. This completes the solution.

Solution 3 by Kee-Wai Lau, Hong Kong, China

We show that the solutions are x = −6, −4, 3 and y = −25, −6, 9.
√ √ √ √
For real numbers a and b, let s = 3 a + 3 b and t = 5 a + 5 b.
By direct expansion, we can prove readily that

s9 − 3(a + b)s6 + 3(a2 − 7ab + b2 )s3 − (a + b)3 = 0, (1)

and

t25 − 5(a + b)t20 + 5(2a2 − 121ab + 2b2 )t15 − 5(a + b)(2a2 + 379ab + 2b2 )t10 +
+ 5(a4 − 121a3 b + 381a2 b2 − 121ab3 + b4 )t5 − (a + b)5 = 0. (2)

By the substitution x = 2 + (s − 2) 3 , we see that the first equation of the problem

p p
can be written as 3 6 + 2(s − 2)3 + 3 2 − (s − 2)3 = s. We put a = 6 + 2(s − 23 ) and
b = 2 − (s − 2)3 into (1) and after simplification and factorization we obtain

s3 ((s − 2)3 + 6)(s − 3)(s2 − 3s + 3) = 0.

√
3
Since s is real, so s = 0, 2 − 6, 3, giving x = −6, −4, 3.
By the substitution
p y = 7 + (t −p2)5 , we see that the second equation of the problem can
be written as 5 6 − 2(t − 2)5 + 5 26 + 3(t − 2)5 = t. We put a = 6 − 2(t − 2)5
and b = 26 + 3(t − 2)5 into (2) and after simplification and factorization we obtain

t5 ((t − 2)5 + 13)((t − 2)5 − 2)f (t) = 0,

where

f (t) = 7t10 −130t9 +1140t8 −6000t7 +20880t6 −49982t5 +83180t4 −94640t3 +70240t2 −30560t+5924.

By using the software

√ Mathematica,
√ we find that all the roots of f (t) = 0 are not real.
Hence, t = 0, 2 − 5 13, 2 + 5 2 giving y = −25, −6, 9.

Solution 4 by Peter Fulop, Gyomro, Hungary

√
3
√
3
√
3
2x + 2 + 4−x+ 2−x=2 (1)

p
5
p p
20 − 2y + 5 7 − y + 5 3y + 5 = 2 (2)

Solving equation (1):

21
 √
3
√
3
√
3
Rearrange the equation (1): 4−x+ 2−x=2− 2x + 2
Let’s raise to the third power:
√ √ √ √ √ √ 
2 3 2x + 2 − 2 + 3 2x + 2 = 3 2 − x 3 4 − x 3 2 − x + 3 4 − x
| {z }
√
2− 3 2x+2
After the cancellations:

 √
3
 √ √ √ 
2− 2x + 2 2 3 2x + 2 + 3 2 − x 3 4 − x = 0 (3)

Equation (3) gives two equations both are equal to zero.

√
3
Solving the 2 − 2x + 2 = 0 we get x1 = 3
√ √ √
and the other 2 3 2x + 2 = − 3 2 − x 3 4 − x and square it: x2 + 10x + 24 = 0
Resulting two other roots: x2 = −4 and x3 = −6
Solving equation (2):
√ √ √
Let a = 5 20 − 2y; b = 5 7 − y; c = 5 3y + 5 and d = 2.
Based on (2) we have: a + b = c − d raise it on fifth power and
let’s realize that a5 + b5 = d5 − c5 we get:

5ab(a3 + b3 ) + 10a2 b2 (a + b) = 5cd(c3 − d3 ) − 10c2 d2 (c − d) (4)

Using the appropriate identity on both sides of (4):

x3 ± y 3 = (x ± y)(x2 ∓ xy + y 2 )
We get:

ab(a + b)(a2 − ab + b2 ) + 2a2 b2 (a + b) = −cd(a + b)(a2 + ab + b2 ) + 2c2 d2 (a + b) (5)

It can be seen that a + b = 0 provides first root.

√ √
5
20 − 2y = − 5 7 − y namely: 20 − 2y = y − 7
y1 = 9
h i h i
Rearranging (5): ab (a + b)2 − ab = −cd (c − d)2 + cd
and using (c − d)2 = (a + b)2 we get:

(a + b)2 (ab + cd) = (ab + cd)(ab − cd) (6)

(ab + cd) = 0 provides the second two roots:

√ √ √
Namely 5 20 − 2y 5 7 − y = −2 5 3y + 5 gives us the following quadratic equation:

y 2 + 31y + 150 = 0. (7)

22
 We get further two roots:
y2 = −6 and y3 = −25
The remaining part of (6):

(a + b)2 = ab − cd (8)

b2 + (a − 1)b − a2 − a + 2 = 0
 
1 p
2
b1,2 = 1 − a ± −3a + 2a − 7
2

b should be real number but the discriminant less than zero so (8) will not give real a, b.
On the other hand: a2 + ab + b2 = −4a + 2a + 2b and it is known that
a3 − b3
= a2 + ab + b2 ; hence we obtain:
a−b
a(a2 − 2a + 4) = b(b2 − 2b + 4)
a = b does not provide a solution.
Finally the roots are:

x1 = 3 , x2 = −4, x3 = −6

y1 = 9 , y2 = −6, y3 = −25

Also solved by Charles Burnette, Xavier University of Louisiana (New Or-

leans, LA); SQ Mathematical Problem Solving Group, Yogyakarta, Indone-
sia; Albert Stadler, Herrliberg, Switzerland; David Stone and John Hawkins,
Georgia Southern University, Statesboro, GA; Daniel Văcaru, Pitesti, Roma-
nia, and the proposer.

• 5616: Proposed by D.M. Bătinetu-Giurgiu “Matei Basarb” National College, Bucharest

and Neculai Stanciu, “George Emil Palade” Secondary School Buzău, Romania
Prove that in all tetrahedrons [ABCD] the following inequality holds:
s s s s
1 3 hb hc 1 3 hc hd 1 3 hd ha 1 3 ha hb 1
2
+ 2 + 2
+ 2 ≥ ,
ha ha hb hb hc hc hd hd r

where r is the radius of the insphere of the tetrahedron.

Solution 1 by Moti Levy, Rehovot, Israel

Let us substitute ha , hb , hc and hd as follows:
1 1 1 1
x := , y := , z := , t := .
ha hb hc hd

23
 1
= 1r , hence
P
It is known that cyc ha

1
x+y+z+t= .
r
Therefore, the original inequality is equivalent to

3x2 3y 2 3z 2 3t2
+ + + ≥ x + y + z + t.
x+y+z y+z+t z+t+x t+x+y
The inequality is homogenous, hence we may assume x+y +z +t = 1. With this constraint
the inequality becomes :

x2 y2 z2 t2 1
+ + + ≥ , x + y + z + t = 1.
1−t 1−x 1−y 1−z 3
Now we apply Radon’s inequality

x2 y2 z2 t2
+ + +
1−t 1−x 1−y 1−z
(x + y + z + t)2 1
≥ = .
(1 − t + 1 − x + 1 − y + 1 − z) 3

Remark: Radon’s inequality states:

If xk , ak > 0, k ∈ {1, 2. . . . , n} , p > 0, then

xp+1
1 xp+1
2 xn+1
2 (x1 + x2 + · · · + xn )p+1
+ p + ··· + n ≥ .
ap1 a2 a2 (a1 + a2 + · · · + an )p

Solution 2 by Michel Bataille, Rouen, France

Regrettably, this problem is not new. It is part (b) of the College Mathematics Journal
problem 1018 proposed by the same authors in May 2014. A solution was published in
the May 2015 issue (Vol. 46, No 3).
Here is the solution that I sent in 2014.
If Fa , Fb , Fc , Fd are the areas of faces BCD, CDA, DAB, ABC, respectively, of tetrahe-
dron ABCD and V its volume, we have
1 1 1 1 1
V = ha · Fa = hb · Fb = hc · Fc = hd · Fd = (r · Fa + r · Fb + r · Fc + r · Fd )
3 3 3 3 3
and we deduce that
1 1 1 1 1
+ + + = .
ha hb hc hd r
1 1 1 1
Now, with α = ha , β = hb , γ = hc , δ = hd , the left-hand side L of the inequality rewrites
as
α2 β2 γ2 δ2
√
3
+ √
3
+ √
3
+ √
3
.
αβγ βγδ γδα δαβ
From the arithmetic mean-geometric mean inequality, we deduce

α2 β2 γ2 δ2
 
L≥3 + + + . (1)
α+β+γ β+γ+δ γ+δ+α δ+α+β

24
Using the Cauchy-Schwarz inequality, we obtain

α2 β2 γ2 δ2
 
+ + + (3(α + β + γ + δ) ≥ (α + β + γ + δ)2 .
α+β+γ β+γ+δ γ+δ+α δ+α+β
(2)
From (1) and (2) it now follows that
1
L≥α+β+γ+δ = ,
r
as required.

Editor0 s comment: Normally it does not happen that an identical problem appears in
two different journals. But in this case it happened and the fault is most likely mine.
I have a stack of problems submitted from many individuals from around the world.
Sometimes these problems have been submitted to our column several years earlier before
being selected for publication. At that time my system did not have a date of submission
written on every proposed problem. So it could well be that I had said to the authors
of the problem that I liked it and that it would be published in some future issue of
the column. But I never specify which issue and sometimes the authors might have to
wait years before seeing it in print. That might be what happened here; and if so, I
accept the blame for this “fashla.” Future editors will have to develop a better protocol
for acknowledging submissions.

Solution 3 by Albert Stadler, Herrliberg, Switzerland

Let ∆ be the volume of the tetrahedron and let ∆A , ∆B , ∆C be the area of the face oppo-
site the vertex A, B, C, respectively. Then (see for instance http://en.wikipedia.org/wiki/Tetrahedron)

1 1 1 1 1
∆ = hA ∆A = hB ∆B = hC ∆C = hD ∆D = (∆A + ∆B + ∆C + ∆D ) r.
3 3 3 3 3

The inequality is therefore equivalent to each of the following lines:

s
X 1 thB hC 1
3
2
≥ ,
hA (hA ) r
cycl

s
X 3∆ hB hC 3∆
3
2
≥ ,
hA (hA ) r
cycl
s
X 3 (∆A )2
∆A ≥ ∆A + ∆B + ∆C + ∆D ,
∆B ∆C
cycl
s
X 3 (∆A )3
∆A ≥ ∆A + ∆B + ∆C + ∆D ,
∆A ∆B ∆C
cycl

X
p p
∆2A 3
∆D ) ≥ (∆A + ∆B + ∆C + ∆D ) 3 ∆A ∆B ∆C ∆D .
cycl

However the last inequality is true, since by the AM-GM inequality,

25
 X p
32
X  119 2 3
p 26 2 3
p 29 2 3
p 11 p 
2 3
(∆A ) ∆D = (∆A ) ∆D + (∆B ) ∆A + (∆C ) ∆B + (∆C ) ∆C ≥
185 185 185 185
cycl cycl

X 119 1 119 26 1 26 29 1 29 11 1 11
≥ (∆A )2· 185 (∆D ) 3 · 185 (∆B )2· 185 (∆A ) 3 · 185 (∆C )2· 185 (∆B ) 3 · 185 (∆D )2· 185 (∆C ) 3 · 185 =
cycl

p
3
(∆A + ∆B + ∆C + ∆D ) ∆A ∆B ∆C ∆D .

Solution 4 by Kee-Wai Lau, Hong Kong, China

By the inequality of Cauchy-Schwarz we have
s s s s
1 3 hb hc 1 3 hc hd 1 3 hd ha 1 3 ha hb
+ + + =
ha h2a hb h2b hc h2c hd h2d

1 p 3 1 p 3 1 p 3 1 p 3
= h a hb hc + hb hc h d + hc h d ha + hd ha hb ≥
h2a h2b h2c h2d

 2
1 1 1 1
ha + hb + hc + hd
≥ q q q q
1 1 1 1
3
ha hb hc + 3
hb hc hd + 3
hc hd ha + 3
hd ha hb

By the AM-GM inequality, we see that the denominator of the last expression does not
exceed

       
1 1 1 1 1 1 1 1 1 1 1 1 1
+ + + + + + + + + + +
3 ha hb hc hb hc hd hc hd ha hd ha hb

1 1 1 1
= + + + .
ha hb hc hd

Now the inequality of the problem follows from the well known fact that
1 1 1 1 1
+ + + = .
ha hb hc hd r
Also solved by the proposers

• 5617: Proposed by José Luis Dı́az-Barrero, Barcelona Tech, Barcelona, Spain

Let a, b, c be the roots of the equation x3 + rx + s = 0. Without the aid of a computer,
calculate
2bc − a2 c2 b2
 

 2 2 2

det  c
 2ca − b a 

 b2 a2 2
2ab − c 

Solution 1 by Titu Zvonaru, Comănesti, Romania

26
We have a + b + c = 0. Adding the second column and the third column to the first
column, we obtain;

 2bc − a2 c2 b2  (b + c)2 − a2 c2 b2
   
 
   
det  c2 2ca − b2 a2  = det  (c + a)2 − b2 2ca − b2
  a2 =

 b2 a 2 2ab − c2   (a + b)2 − c2 a2 2ab − c2 

c2 b2 c2 b2
   
 (a + b + c)(b + c − a)   0 
 2
  
det  (c + a + b)(c + a − b) 2ca − b
 a2  = det  0 2ca − b2
  a2  = 0.

 (a + b + c)(a + b − c) a2 2ab − c2   0 a2 2ab − c2 

Solution 2 by Brian Bradie, Christopher Newport University, Newport

News, VA

First,

 2bc − a2 c2 b2
 

 2 2

det  c 2ca − b a2 

 b2 a2 2ab − c2 
is equal to

(2bc − a2 )(2ca − b2 )(2ab − c2 ) + 2a2 b2 c2 − (2b4 ca − b6 + 2a4 bc − a6 + 2c4 ab − c6 )

= 9a2 b2 c2 − 4abc(a3 + b3 + c3 ) + 2(a3 b3 + b3 c3 + c3 a3 ) − 2abc(a3 + b3 + c3 + a6 + b6 + c6
= 9a2 b2 c2 − 6abc(a3 + b3 + c3 ) + (a3 + b3 + c3 )2 .

Next, because a, b, c are the roots of the equation x3 + rx + s = 0, it follows that

a + b + c = 0, ab + bc + ca = r, and abc = −s.

Moreover,
a3 + b3 + c3 = −r(a + b + c) − 3s = −3s.
Thus,

 2bc − a2 c2 b2
 

 = 9s2 − 6(−s)(−3s) + 9s2 = 0.
 2 2

det 
 c 2ca − b a2 
 b2 a2 2ab − c2 

Solution 3 by Bruno Salgueiro Fanego, Viveiro, Spain

Expanding the determinate in the normal way we obtain:
2bc − a2 c2 b2 
 

 
 2 2 2

det  c
 2ca − b a =

 
 
 b2 a2 2ab − c2 

27
= (2bc−a2 )(2ca−b2 )(2ab−c2 )+c2 a2 b2 +b2 a2 c2 − b2 (2ca − b2 )b2 + c2 c2 (2ab − c2 ) + (2bc − a2 )a2 a.
After simplification we obtain:

(a + b + c))3 − 3(a + b)2 c − 3(a + b))c2 − 3ab(a + b + c))2

2
2
(a+b+c)2 (a + b + c)2 − 3(a + b)c − 3ab = (a+b+c)2 (a + b + c)2 − 3(ab + bc + ca) .

Since a, b, and c are the roots of x3 + rx + s = 0, we know that a + b + c = 0. So

2bc − a2 2 2
 
 c b 

a b + }c 2
+{z
   

2
(a + b + c)2 − 3(ab + bc + ca) = 0.
 2 2 2

det  c
 2ca − b a =
 |
  0
 
 b2 a2 2ab − c2 

Solution 4 by Daniel Vácaru, Pitesti, Romania

We add all the lines to the first line, and we obtain

 2
 b + 2bc + c2 − a2 c2 + 2ca + a2 − b2 a2 + 2ab + b2 − c2 

 2 
det a + 2ab + b2 − c2 a2 =

 b2 a2 2ab − c2 
(b + c)2 − a2 (a + c)2 − b2 (a + b)2 − c2 
 
 
=det  c2 2ca − b2 a2 =

 b2 a2 2ab − c2 
 
(b + c + a)(b + c − a) (c + a + b)(c + a − c) (a + b + c)(a + b − c) 
 
=det  c2 2ca − b2 a2 

 b2 a 2 2ab − c2 
By Viéa’s formula, a + b + c = 0, the whole first line is 0, implying that
2bc − a2 c2 b2
 

 2 2

det  c 2ca − b a2  = 0.

 b2 a2 2ab − c2 

Also solved by Dionne Bailey, Elsie Campbell, and Charles Diminnie, Angelo
State University, San Angelo, TX; Michel Bataille, Rouen France; Charles
Burnette, Xavier University of Louisiana (New Orleans, LA); Pratik Donga,
Junagadh, India; Michal N. Fried Ben-Gurion University of the Negev,
Beer-Sheva, Israel; Peter Fulop, Gyomro, Hungary; G. C. Greubel, Newport
News, VA; Paul M. Harms, North Newton, KS; Kee-Wai Lau, Hong Kong,
China; Moti Levy, Rehovot, Israel; David E. Manes, Oneonta, NY; Trey
Smith, Angelo State University, San Angelo, TX; SQ Mathematical Problem
Solving Group, Yogyakrta, Indonesia; Albert Stadler, Herrliberg,
Switzerland; Seán M.Stewart, Bomaderry, NSW, Australia; David Stone and
John Hawkins, Georgia Southern University, Statesboro, GA, and the
proposer.

5618: Proposed by Ovidiu Furdui and Alina Sîntămărian, Technical University of

Cluj-Napoca, Cluj-Napoca, Romania

28
Let k > 0 be a real number. Calculate
 
2 1 1 1
lim n + + + ··· .
n→∞ n3 (n + k)3 (n + 2k)3
Solution 1 Brian Bradie, Christopher Newport University, Newport News,
VA
First,
∞ ∞
1 1 1 X 1 1 X 1
3
+ 3
+ 3
+ · · · = 3
= 3 k 3
,
n (n + k) (n + 2k) (n + jk) n (1 + j · n )
j=0 j=0

so
  ∞ ∞
2 1 1 1 1X 1 1X 1
n + + + ··· = = .
n 3 (n + k) 3 (n + 2k)3 n
j=0
k
(1 + j · n ) 3 n
j=0
(1 + k · nj )3

Now, recognize that

∞
1X 1
n
j=0
(1 + k · nj )3
is a left-endpoint approximation to
Z ∞
1
dx.
0 (1 + kx)3
Thus,
  Z ∞
2 1 1 1 1
lim n 3
+ 3
+ + ··· = dx
n→∞ n (n + k) (n + 2k)3 0 (1 + kx)3
  ∞
1
= −

2k(1 + kx)2 

0
1
= .
2k

Solution 2 by Michel Bataille, Rouen, France

∞
1 1
We claim that lim n2 Sn =
P
Let Sn = (n+jk)3
. 2k .
j=0 n→∞

1
Since the function x 7→ (n+xk)3
is decreasing on (0, ∞), we have
Z j+1
1 dx 1
≤ ≤
(n + (j + 1)k)3 j (n + xk)3 (n + jk)3
for j = 0, 1, 2, . . .. It follows that for any positive integer J,
Z J+1 J Z J
dx X 1 1 dx
3
≤ 3
≤ 3+ . (1)
0 (n + xk) (1 + jk) n 0 (n + xk)3
j=0

From Z J Z ∞
dx dx 1
≤ =
0 (n + xk)3 0 (n + xk)3 2kn2

29
we deduce that
J
X 1 1 1
3
≤ 3+ .
(1 + jk) n 2kn2
j=0
∞
P 1 1 1
Therefore, the series (n+jk)3
is convergent and its sum Sn satisfies Sn ≤ n3
+ 2kn 2 . In
j=0
addition, letting J → ∞ in the left inequality of (1), we obtain
Z ∞
dx
≤ Sn
0 (n + xk)3
and finally we see that
1 1 1
2
≤ Sn ≤ 3 + .
2kn n 2kn2
The claimed result directly follows from the Squeeze Theorem.

Solution 3 by Albert Natian, Los Angeles Valley College, Valley Glen,

California
1
Answer. .
2k
Computation. Suppose (without loss of generality) that n > k. It’s immediate that for
any integer j ≥ 0:
Z n+(j+1)k Z n+jk
dx k dx
3
< 3 < 3
.
n+jk x (n + jk) n+(j−1)k x
Summing up the latter inequalities, we get
Z ∞ ∞ Z ∞ ∞ Z n+jk Z ∞
1 dx X n+(j+1)k dx X 1 X dx dx 1
= = < k 3 < = = 2,
2n2 n x 3
n+jk x 3
(n + jk) n+(j−1)k x 3
n−k x 3
2 (n − k)
j=0 j=0 j=0
∞
1 X 1 n2
< n2 <
2k
j=0
(n + jk)3 2k (n − k)2
which (by Squeeze Theorem) implies
∞
X 1 1
lim n2 3 = 2k
n→∞ (n + jk)
j=0

1 1 n2 1
since limn→∞ 2k = 2k and limn→∞ 2k(n−k)2
= 2k .
A Generalization. Suppose (αj )∞
j=−1 is an increasing sequence of positive real
numbers. Then
∞
X αj − αj−1 1
lim nν ν+1 = .
n→∞ (n + αj ) ν
j=0

Solution 4 by Moti Levy, Rehovot, Israel

 
21 1 1
lim n + + + ···
n→∞ n3 (n + k)3 (n + 2k)3
∞ Z ∞
1X 1 1 1
= lim 3 = 3 = , if k > 0.
n→∞ n 2k
 
j=0 1 + k
j 0 (1 + kx)
n

30
Solution 5 by Ulrich Abel,Technische Hochschule Mittelhessen, Friedberg,
Germany
Let k > 0. We have
 
2 1 1 1
n + + + ···
n3 (n + k)3 (n + 2k)3
∞ Z ∞
X 1
= n 2
t2 e−(n+jk)t dt
Γ (3) 0
j=0
2 Z ∞
n t2
= e−nt dt
2 0 1 − e−kt
n2
 
1 −3

1
= +O n → (n → ∞) ,
2 kn2 2k
by Watson’s lemma for Laplace integrals, since
t2 t t2
+ O t3


−kt
= + (t → 0) .
1−e k 2

Solution 6 by G.C. Greubel, Newport News, VA

Consider the series
∞
X 1
Sk =
(n + j k)3
j=0
∞ Z
1 X ∞ −nt−jkt 2
= e t dt
2
j=0 0
1 ∞ e−nt t2
Z
= dt
2 0 1 − e−kt
Z ∞ −(n/k) u 2
1 e u
= du
2 k3 0 1 − e−u
1  n
= ζ 3, .
k3 k
The Hurwitz zeta function has the asymptotic expansion, for a → ∞,
∞
a1−s 1 X B2j (s)2j−1
ζ(s, a) ≈ − +
s − 1 2 as (2j)! a2j+s−1
j=1

which leads to
∞
X B2j (3)2j−1 k 2j+2
 n k2 k3
ζ 3, ≈ − +
k 2 n2 2 n3 (2j)! n2j+2
j=1

and
n2  n 
lim n2 Sk = lim ζ 3,
n→∞ n→∞ k 3 k
  
1 1 k 1
= lim − + 2
+O
n→∞ 2 k 2n 4n n4
1
= .
2k

31
Also solved by Charles Burnette, Xavier University of Louisiana (New
Orleans, LA); Pratik Donga, Junagedh, India; Albert Stadler, Herrliberg,
Switzerland; Seán Stewart, Bomaderry, NSW Australia; HyunBin Yoo,
South Korea, and the proposer.

M ea Culpa

COVID-19 has played havoc with the snail-mail in most countries. Paul M. Harms of
North Newton, KS mailed his solution to 5605 on January 1, 2021; it arrived six
weeks later. His solution is correct and he should be credited with having solved 5605.

32