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

February 15, 2021

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

Given the equations:
 p p
 3x2 + 6x + 1 = y + 3y 2 − 3
and
 2 2
ax + by + 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.

• 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

• 5615: Proposed by Pedro Henrique Oliveira Pantoja, University of Campina Grande, Brazil
Solve in < × < the system:
 √
3
√ √
 2x + 2 + 3 4 − x + 3 2 − x = 2
p
5
p
5
p
20 − 2y + 7 − y + 5 3y + 5 = 2

• 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

1
 where r is the radius of the insphere of the tetrahedron.

• 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 2 2
 
c b 
2 2 2
 
det  c
 2ca − b a 

2 2 2
 b a 2ab − c 

• 5618: Proposed by Ovidiu Furdui and Alina Sîntămărian, Technical University of Cluj-
Napoca, Cluj-Napoca, Romania
Let k > 0 be a real number. Calculate
 
2 1 1 1
lim n + + + ··· .
n→∞ n3 (n + k)3 (n + 2k)3

Solutions

• 5595: Proposed by Kenneth Korbin, New York, NY

Trapezoid ABCD with integer length sides is inscribed in a circle with diameter 233 . Side
AB = 4439. Find the other three sides.

Solution by Kee-Wai Lau, Hong Kong, China

We show that the other three sides are 2717, 4439, and 10051.
12167
Let O be the center and r = be the radius of the circle.
2
Case I: AB = 4439 and CD = 4439 are the legs of the trapezoid.
Let BC = s, AD = t, ∠AOB = ∠DOC = θ and ∠BP C = ϕ.
a) O lies outside the trapezoid.
205343
It suffices to consider the case t ≥ s. Then cos θ =
√ √ 279841
29336 42 ϕ 148035889 − s2 ϕ s
sin θ = , cos = and sin = , so that
279841 2 12167 2 12167
p
 ϕ  ϕ ϕ  29336 42 (148035889 − s2 ) + 205343s
t=2rsin θ + = 2r sin θ + cos θ sin = .
2 2 2 279841

Since s ≤ t ≤ 2r = 12167, we have 1 ≤ s ≤ 8927. By a computer search, we find that

s = 2717, t = 10051 is the only solution in integers.

b) O lies inside the trapezoid.

2
 p
29336 42(148035889 − s2 ) + 205343s
We still have t = , but here we have 8928 ≤ 12167.
279841
A computer search finds no integral solutions for t.

Case II: AB = 4439 is one of the parallel sides of the trapezoid.

Let AD = BC = x and CD = y.
a) O lies outside the trapezoid and 1 ≤ y < 4439
 p 
x 386x + 152 42(148035889 − x2 )
Here we have y = 4439 − , where 1 ≤ x ≤ 2258.
6436343
b) O lies outside the trapezoid and 4439 < y < 12167. Here we have
 p 
x 386x − 152 42(148035889 − x2 )
y = 4439 − , where 1 ≤ x ≤ 6856.
6436343
c) O lies inside or on the trapezoid so that 1 ≤ y ≤ 12167. We still have
 p 
x 386x − 152 42(148035889 − x2 )
y = 4439 − , but here we have 6857 ≤ x ≤ 11955.
6436343
A computer search for cases IIa), IIb) and IIc) finds no integral solutions for y.
This completes the solution.

Observations made by Ken Korbin, proposer of the problem:

Diameter = (23)(23)(23).
Sides of inscribed trapezoid are (4439, 2717, 4439, 10051).
4439 =(2)(19)(19)(23)-(23)(23)(23)
10051 = (19)(23)(23)
2717 = (3)(19)(23)(23) − (4)(19)(19)(19).
If the 19 is replaced by an 18, the inscribed trapezoid will have sides (2737, 5238, 2737, 9522).

Also solved by Ioannis D. Sfikas, National Technical University of Athens, Greece,

and the proposer.

3
• 5596: Proposed by Albert Natian, Los Angeles Valley College, Valley Glen, CA

z
a
b

x y
A
c
B

Let V be a vertex of a rectangular box. Let V A, V B√ and V C be the three edges meeting at
vertex V . Suppose the area of the triangle ABC is 6 26. The volume of the box is 144. And
the sum of the edges of the box is 76. Find the total surface area of the box.

Solution 1 by Titu Zvonaru, Comănesti, Romania

We have x + y + z = 19. Since 16[ABC] = 2a2 b + 2b2 c2 + 2c2 a2 − a4 − b4 − c4 , we obtain

16·36·26 = 2(x2 +y 2 )(y 2 +z 2 )+2(y 2 +z 2 )(z 2 +x2 )+2(z 2 +x2 )(x2 +y 2 )−(x2 +y 2 )2 −(y 2 +z 2 )2 −(z 2 +x2 )2

4(x2 y 2 + y 2 z 2 + z 2 x2 ) = 16 · 36 · 26
x2 y 2 + y 2 z 2 + z 2 x2 = 144 · 26
(xy + yz + zx)2 − 2xyz(x + y + z) = 144 · 26
(xy + yz + zx)2 = 2 · 144 · 19 + 144 · 26,
hence xy + yz + zx = 96 and the total surface area of the box is 2 · 96 = 192.

Solution 2 by Bruno Salgueiro Fanego, Viveiro, Spain

Let, as it appears in the figure, x = V A, y = V B and z = V C. Since the box is rectangular,
triangles 4V AB, 4V BC, and 4V AC are situated in such a way that forces V a right angle.
So the Pythagorean theorem, applied to the triangles mentioned above, implies that x2 + y 2 =
c2 , y 2 + z 2 = a2 and x2 + z 2 = b2 .
From the hypothesis in the problem, the area of the 4ABC is, (from Herons formula),
r
a + b + c −a + b + c a − b + c a + b − c √
· · · = 6 26, the volume of the box is xyz = 144,
2 2 2 2
and the sum of the edges of the box is 4x + 4y + 4z = 76.

4
Since

(a + b + c)(−a + b + c)(a − b + c)(a + b − c) = (b + c)2 − a2 a2 − (b − c)2

2bc + b2 + c2 − a2 2bc − b2 − c2 + a2 = (2bc)2 − (b2 + c2 − a2 )

=

= 4 x2 + z 2 x2 + y 2 − 4x4 = x2 y 2 + y 2 z 2 + x2 z 2 ,

the givens in the statement of the problem are equivalent to:


 x + y + z = 26
x2 y 2 + y 2 z 2 + x2 z 2 = 3744
xyz = 144.


This implies that (x, y, z) = (3, 4, 12) and hence the total surface area of the box is 2(xy +
yz + zx) = 192.

Solution 3 by Pratik Donga, Junagadh, India.

√
The volume of the box is xyz = 144 and the area of 4ABC = 6 26. The sum of the edges
is 4(x + y + z) = 76 ⇒ x + y + z = 19.
Total surface area S.A. = 2(xy + yz + zx).
V olume of the cuboid 144
Volume of the tetrahedron V ABC = = = 24. Volume of the
6 6
tetrahedron
1
V ABC = × height × area of the base
3
1
= × h × area of 4ABC = 24
3
3 × 24 12
⇒ h= √ =√
6 26 26
144
⇒ h2 = .
26

For the tetrahedron

1 1 1 1 2 x2 y 2 z 2 2 2 2 (144)2 × 26
= + + ⇒ h = ⇒ (xy) +(yz) +(zx) = = 3744.
h2 x2 y 2 z 2 x2 y 2 + y 2 z 2 + z 2 x2 144

Now, (S.A.)2 = 4(xy + yz + zx)2 = 4 (xy)2 + (yz)2 + (zx)2 + 2(xyz)(x + y + z)

so, (S.A.)2 = 4 (3744

√ + 2(144) × 19)) = 4(3744 + 5472) = 4 × 9216, and hence the total surface
area of the box is 4 × 9216 = 192.

Solution 4 Brian D. Beasley, Presbyterian College, Clinton, SC

We have xyz = 144 and x + y + z = 19. Then xy(19 − x − y) = 144, or
x2 y − 19xy + xy 2 + 144 = 0.
Next, using Heron’s formula, we obtain

5
 (a + b + c)(a + b − c)(a − b + c)(−a + b + c) = 14976.
Since a = y + z , b2 = x2 + z 2 , and c2 = x2 + y 2 , this yields d1 d2 d3 d4 = 14976, where
2 2 2
p p p
d1 = a + b + c = py 2 + (19 − x − y)2 + px2 + (19 − x − y)2 + px2 + y 2 ;
d2 = a + b − c = py 2 + (19 − x − y)2 + px2 + (19 − x − y)2 − px2 + y 2 ;
d3 = a − b + c = py 2 + (19 − x − y)2 − p x2 + (19 − x − y)2 + p x2 + y 2 ;
d4 = −a + b + c = − y + (19 − x − y) + x + (19 − x − y) + x2 + y 2 .
2 2 2 2

The intersection of these graphs in the first quadrant (see below) consists of six points, corre-
sponding to {x, y, z} = {3, 4, 12}. Hence the total surface area of the box is 2(xy + yz + zx) =
192.

Addenda. (1) The diagram below shows the graphs in the first quadrant of the equations
x2 y − 19xy + xy 2 + 144 = 0 (in red) and d1 d2 d3 d4 = 14976 (in blue).

√ √
(2) We note that {a, b, c} = {5, 3 17, 4 10}.

Editor0 s comment: Shortly after receiving the above solution Brian sent a note expanding on
the above question. His note;

Motivated by how the graphs of the two equations barely intersect in the first quadrant (in
our solution for the original problem), we pose the following questions:
√
1. Given a triangle area of 6 26 and a box edge sum of 76, what is the maximum possible

6
 box volume?
2. Given a box volume of 144 and a box edge sum of 76, what is the minimum possible triangle
area?
√
3. Given a box volume of 144 and a triangle area of 6 26, what is the maximum possible box
edge sum?

Solution. In each case, due to symmetry, we assume x = y in order to produce three points
(instead of six) where the two graphs intersect.

z = 19 and (a + b + c)(a + b − c)(a − b + c)(−a + b + c) = 14976 with x = y,

1. Given x + y + p
√
we have a = b = x2 + z 2 and c = 2x. Then using z = 19 − 2x yields
2x2 (18x2 − 304x + 1444) = 14976,
or 9x4 − 152x3 + 722x2 − 3744 = 0. Thus x ≈ 3.56611, so the maximum volume is xyz ≈
150.92443 (not too much larger than the originally given 144).

2. Given x + y + z = 19 and xyz = 144 with x = y, we have x2 (19 − 2x) = 144, or

2x3 − 19x2 + 144 = 0. Then x ≈ 3.44966, so the minimum √ triangle area is approximately
30.11067 (not too much smaller than the originally given 6 26 ≈ 30.59412).

3. Given xyz = 144 and (a + b + c)(a + b − c)(a p

− b + c)(−a + b + c) = 14976
√ with x = y, we
2 2
let S = x + y + z. Then 2x + z = S, so a = b = x + (S − 2x) and c = 2x. Thus
36x4 − 32Sx3 + 8S 2 x2 = 14976.
Using z = S − 2x = 144/x2 , we have S = (2x3 + 144)/x2 and hence
8(2x3 + 144)2
36x4 − 32x(2x3 + 144) + = 14976.
x2
Then x ≈ 3.38838, so the maximum box edge sum is 4S ≈ 77.27628 (not too much larger
than the originally given 76).

In summary, we salute the proposer of the original problem for finding constraints that not
only produce integer solutions for the box dimensions but also cannot be tweaked too much
more without losing any chance of a solution.

Also solved by Michel Bataille, Rouen, France; Kee-Wai Lau, Hong Kong, China;
Annabel Ma, (student), New Trier High School, Winnetka, Il; Ioannis D. Sfikas
National Technical University of Athens, Greece; Daniel Văcaru, Pitesti Romania;
David Stone and John Hawkins, Georgia Southern University, Statesboro, GA,
and the proposer.

• 5597: Proposed by Daniel Sitaru, National Economic College “Theodor Costescu,” Mehedinti,
Romania
If x, y, z > 0; xyz = 1 then:
 1 2  1 2  1 2
x+y− √ + y+z− √ + z+x− √ ≥3
z x y

7
Solutions 1, 2, and 3 by Henry Ricardo, Westchester Area Math Circle, NY
p √
Solution 1: First we note that Maclaurin’s inequality gives us (xy + yz + zx)/3 ≥ 3 xyz =
1, or xy + yz + zx ≥ 3. Then the AGM inequality yields
X  1 2
 X √
x+y− √ = (x + y − xy)2
z
cyclic cyclic
X √
≥ ( xy)2 = xy + yz + zx ≥ 3.
cyclic

Equality holds if and only if x = y = z = 1.

√
Solution 2: First we note that the AGM inequality gives us (x + y + z)/3 ≥ 3 xyz = 1, or
x + y + z ≥ 3. Then the AGM inequality and Radon’s inequality yield
X  1 2
 X √
x+y− √ = (x + y − xy)2
z
cyclic cyclic
X  x + y 2 [2(x + y + z)]2 62
≥ ≥ ≥ = 3.
2 12 12
cyclic

Equality holds if and only if x = y = z = 1.

Solution 3: Using the AGM inequality twice, we see that

X  1 2
 X √
x+y− √ = (x + y − xy)2
z
cyclic cyclic
X √
≥ ( xy)2
cyclic
p
3
≥ 3 (xyz)2 = 3.

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

By the AM-GM inequality, we have
√ √ p
2(x + y + z) − ( xy + yz + zx)

√ √ √ √ √ √
( x − y)2 + ( y − z)2 + ( z − x)2
= (x + y + z) +
2

≥ x+y+z
√
≥ 3 3 xyz

= 3.

8
 Hence by the Cauchy-Schwarz inequality, we have

1 2 1 2 1 2
     
x+y− √ + y+z− √ + z+x− √
z x y
     2
1 1 1 1
≥ x+y− √ + y+z− √ + z+x− √
3 z x y
1 √ √ √

2
= (x + y − xy) + (y + z − yz) + z + x − zx
3

1 √ √ √

2
= 2 (x + y + z) − xy + yz + zx
3
≥ 3, as required.

Comments from the solution of David Stone and John Hawkins of Georgia South-
ern University, Statesboro, GA: Like others they commented that A(x) is the Lagrange
i


Interpolating Polynomial passing through n + 1 given points i, 3 , i = 0. . . . n. They then
went on to say that “the value 3 in the problem is not necessary. That is if a > 1 is any real
number a we require that A(x) be a polynomial of degree n such that A(i) = ai for 0 ≤ i ≤ n,
the exact same proof shows that A(n + 1) = an+1 − (a − 1)n+1 .”
They continued: “This example demonstrates the problem with using the polynomial A(x)
to approximate the exponential function 3x . Even when A(x) passes through the n + 1 ‘nice’
points, i, 3i , i = 0, 1, . . . n, it misses the ‘next’ value, 3n+1 , by a long way. Of course, it is

beautiful that the amount of the miss is known to be 2n+1 .”

Also solved by Hatef I. Arshagi, Guilford Technical Community College, Jamestown,

NC; Dionne Bailey, Elsie Campbell, and Charles Diminnie, Angelo State Univer-
sity, San Angelo, TX; Michel Bataille, Rouen, France; Brian Bradie, Christo-
pher Newport University, Newport News, VA; Marin Chirciu, National Col-
lege “Zinca Golescu,” Pitesti, Romania; Bruno Salgueiro Fanego, Viveiro, Spain;
Pratik Donga, India; Oleh Faynshteyn, Leipzig, Germany; Igbal Z. Hasanli (stu-
dent, mentored by Yagub Aliyev), ADA University, Baku, Azerbaijan; Moti
Levy, Rehovot, Israel; Ángel Plaza, Universidad de Las Palmas de Gran Ca-
naria, Spain; Ioannis D. Sfikas National Technical University of Athens, Greece;
Albert Stadler, Herrliberg, Switzerland; Daniel Văcaru, Pitesti, Romania; Titu
Zvonaru, Comănesti, Romania, and the proposer.

• 5598: Proposed by José Luis Dı́az-Barrero, Barcelona Tech, Barcelona, Spain

Let A(x) be a polynomial of degree n such that A(i) = 3i for 0 ≤ i ≤ n. Find the value of
A(n + 1).

Solution 1 by Ulrich Abel, Technische Hochschule Mittelhessen, Germany

We consider the following generalization: Let a ∈ R and suppose that A (x) is a polynomial

9
of degree (at most) n such that A (i) = ai , for 0 ≤ i ≤ n. Then
n
X
A (n + 1) = ak Ln,k (n + 1)
k=0

with the Lagrange polynomials

n  
X n+1−j (n + 1)!/ (n + 1 − k) n−k n+1
Ln,k (n + 1) = = = (−1) .
j=0
k−j k! (−1)n−k (n − k)! k
j6=k

Hence,
n  
X n+1−k n+1 k
A (n + 1) = − (−1) a = an+1 − (a − 1)n+1 .
k
k=0

Solution 2 by Albert Stadler, Herrliberg, Switzerland By Lagrange’s interpolation

formula (see for instance
https://en.wikipedia.org./wiki/Langrangepolynomial),
n n n
X
i
Y n+1−j X (−1)n−1 (n + 1)!
A(n + 1) = 3 = 3i =
i−j i!(n + 1 − i)!
i=0 j=0 i=0
j6=t

n  
n
X n+1
i i
= (−1) (−1) 3 = (−1)n (1 − 3)n+1 + 3n+1 = 3n+1 − 2n+1 .
i
i=0

Solution 3 by Ángel Plaza, Universidad de Las Palmas de Gran Canaria, Spain

It may be shown by induction that
n  
22 23 2n X x k
A(x) = 1 + 2x + x(x − 1) + x(x − 1)(x − 2) + · · · + x(x − 1) · · · (x − n + 1) = 2 ,
2 3! n! k
k=0

n  
X n+1 k
and therefore A(n + 1) = 2 = 3n+1 − 2n+1 .
k
k=0

Solution 4 by Brian Bradie, Christopher Newport University, Newport News, VA

Using divided differences, the Newton form for the polynomial A(x) is
n k−1
X 2k Y
A(x) = 1 + (x − j).
k!
k=1 j=0

10
Thus,
n k−1
X 2k Y
A(n + 1) = 1 + (n + 1 − j)
k!
k=1 j=0
n  
k n+1
X
= 1+ 2
k
k=1
n+1
X n + 1
= 2k − 2n+1
k
k=0
n+1
= 3 − 2n+1 .

Solution 5 by Michel Bataille, Rouen, France

A(x) is the Lagrange polynomial associated to the values 3i taken for x = i (0 ≤ i ≤ n). This
polynomial is
n \ n
X
i x(x − 1) · · · (x − i + 1)(x − i)(x − i − 1) · · · (x − n) X i 1 P (x)
A(x) = 3 · n−i
= 3 · 0 ·
i!(−1) (n − i)! P (i) x − i
i=0 i=0

n
Y
where the hat indicates the omitted factor and P (x) = (x − i).
i=0
P (n + 1) (n + 1)!(n − i)!
Since = , we obtain
n+1−i (n + 1 − i)!
n
X 1 (n + 1)!(n − i)!
A(n + 1) = 3i · ·
i!(−1)n−i (n − i)! (n + 1 − i)!
i=0
n
X (n + 1)!
= (−1)n−i 3i ·
i!(n + 1 − i)!
i=0
n  
n
X n+1
= (−1) (−3)i
i
i=0
n+1
!
X n + 1 
= (−1)n (−3)i − (−3)n+1
i
i=0
= (−1) (1 − 3)n+1 − (−1)n+1 3n+1
n

and finally A(n + 1) = 3n+1 − 2n+1 .

Solution 6 by Dionne Bailey, Elsie Campbell, and Charles Diminnie, Angelo State
University, San Angelo, TX
Our solution makes use of the Binomial Theorem: For m ≥ 1 and a, b > 0,
       
m m m m m−1 m m−1 m m
(a + b) = a + a b + ... + ab + b ,
0 1 m−1 m

11
  
m m!
where = for k = 0, . . . , m.
k k! (m − k)!
We begin by introducing the new function
n
X 2k
f (x) = 1 + x (x − 1) · · · (x − k + 1)
k!
k=1

for n ≥ 1. Note first of all that f (x) is a polynomial of degree n and f (0) = 1 = 30 . Next,
when 0 < i < n, we have
i (i − 1) · · · (i − k + 1) = 0
for k ≥ i + 1. Hence,
n
X 2k
f (i) = 1 + i (i − 1) · · · (i − k + 1)
k!
k=1
i
X 2k
=1+ i (i − 1) · · · (i − k + 1)
k!
k=1
2 22 2i
=1+ i + i (i − 1) + . . . + i (i − 1) · · · (1)
1!  2!  
 i!
i i 2 i i
=1+ 2+ 2 + ... + 2
1 2 i
      
i i i i−1 i i−2 2 i i
= 1 + 1 2+ 1 2 + ... + 2
0 1 2 i
= (1 + 2)i
= 3i .

Finally,

2 22 2n
f (n) = 1 + n + n (n − 1) + . . . + n (n − 1) · · · (1)
 1! 2!    n!  
n n n n−1 n n−2 2 n n
= 1 + 1 2+ 1 2 + ... + 2
0 1 2 n
= (1 + 2)n
= 3n .

Since f (x) and A (x) are both polynomials of degree n and

f (i) = 3i = A (i) for 0 ≤ i ≤ n, it follows that f (x) = A (x) for all real x. As a result, for

12
n ≥ 1,

A (n + 1) = f (n + 1)
n
X 2k
=1+ (n + 1) (n) · · · (n + 2 − k)
k!
k=1
21 22 2n
=1+ (n + 1) + (n + 1) (n) + . . . + (n + 1) (n) · · · (2)
1!  2!
   n! 
n+1 1 n+1 2 n+1 n
=1+ 2 + 2 + ... + 2
1 2 n
       
n+1 1 n+1 2 n+1 n n + 1 n+1
=1+ 2 + 2 + ... + 2 + 2
1 2 n n+1
 
n + 1 n+1
− 2
n+1
       
n + 1 n+1 n+1 n 1 n + 1 n−1 2 n+1 1 n
= 1 + 1 2 + 1 2 + ... + 1 2
0 1 2 n
 
n + 1 n+1
+ 2 − 2n+1
n+1
= (1 + 2)n+1 − 2n+1
= 3n+1 − 2n+1 . (1)

To complete our claim, we note that when n = 0, A (x) is a polynomial of degree 0 (i.e., A (x)
is a constant polynomial) with A (0) = 30 = 1. It follows that A (x) = 1 for all real x and we
have
A (0 + 1) = A (1) = 1 = 3 − 2 = 30+1 − 20+1 .
This also demonstrates that (1) is true when n = 0.

Solution 7 by Ioannis D. Sfikas, National Technical University of Athens, Greece

Using [1] we have : ai = 3i , so:
n+1  
X i−1 n + 1 n+1−i
A(n + 1) = (−1) 3 = 3n+1 − 2n+1
i
i=1

Since:
n+1   n  
X n + 1 n+1−i X n + 1 n−i
(−1)i−1 3 = (−1)i 3
i i+1
i=1 i=0

n−1    
X i n n
= (−1) + 3n−i + (−1)n
i+1 i
i=0

n−1   n−1  
X i n n−i
X i n
= (−1) 3 + (i − 1) 3n−i + (−1)n−1
i+1 i
i=0 i=0

13
 n−1   n  
X i−1 n n−i+1 X i−1 n
= (−1) 3 − (−1) 3n−i + (−1)n + (−1)n−1
i i
i=1 i=0

n   n  
X i−1 n n−i+1 X i−1 n
= (−1) 3 − (−1) 3n−i
i i
i=1 i=0

= 3n+1 − 3 · 2n − 2n

= 3n+1 − 2n+1 .

[1] Alt, Arkady M. (2019). Numerical sequences and polynomials. Arhimede Mathematical
Journal, 6(2) : 114−120. http : //amj−math.com/wp−content/uploads/2020/02/AM J2019−
vol6iss2.pdf
[2] http://math.stackexchnge.com/questions/2161052/showing−3n− sum−k−0n−1k−binomnk4n-
k

Solution 9 by Albert Natian, Los Angeles Valley College, Valley Glen, California
Answer. A(n + 1) = 3n+1 − 2n+1 .

Justification. We begin with the following lemma whose proof is provided toward the end of
this solution.

Lemma.
n  
X n+i n+1 k
(−1) i = (n + 1)k for k = 0, 1, 2, · · · , n.
i
i=0

Now let’s suppose

A (i) = bi for 0≤i≤n
n
X
where b is a fixed number (e.g., b = 3) and A (x) = ak xk .
k=0

14
We have
n n n  
X X X n+1 k
A (n + 1) = ak (n + 1)k = ak (−1)n+i i
i
k=0 k=0 i=0
n  X n n  
n+i n + 1 n+i n + 1
X X
k
= (−1) ak i = (−1) A (i)
i i
i=0 k=0 i=0
n   n  
n+i n + 1 n+1
X n
X
= (−1) i
b = (−1) (−b)i
i i
i=0 i=0
n+1
"   #
n n+1
X n+1 i
= (−1) − (−b) + (−b)
i
i=0
h i
= (−1) (−1) b + (1 − b)n+1
n n n

= bn+1 − (b − 1)n+1 .

Generalization. Suppose A(x) is a polynomial of degree n such that

m
X
A (i) = βt bit for 0≤i≤n
t=1

where βt and bt are indexed fixed numbers. Then

m
X m
X
A (n + 1) = βt bn+1
t − βt (bt − 1)n+1 .
t=1 t=1

Corollary. Suppose A(x) is a polynomial of degree n such that

m
X
A (i) = ti for 0 ≤ i ≤ n.
t=1

Then
A (n + 1) = mn+1 .
Lemma.
n  
X n+i n+1 k
(−1) i = (n + 1)k for k = 0, 1, 2, · · · , n.
i
i=0

Proof. Clearly the above result holds for k = 0 and all n. The remainder of the proof is by
induction on n with k ≥ 1. For n = 0, 1, 2, the above statement (clearly) holds. Suppose the

15
 statement of the lemma above holds for n. Then
n+1   n+1  
X n+1+i n+1+1 k X n+1+i n + 1 + 1 k
(−1) i = (−1) i
i i
i=0 i=1
n+1    
X n+1 n+1
= (−1)n+1+i + ik
i i−1
i=1
"n+1   # "n+1   #
X n + 1 X n + 1
= (−1)n+1+i ik + (−1)n+1+i ik
i i−1
i=1 i=1
n
"   #
n+i n + 1 k
k
X
= (n + 1) − (−1) i +
i
i=1
"n+1   #
n+1+i n + 1 k
X
+ (−1) i
i−1
i=1
 
n  
h i X n+1
= (n + 1)k − (n + 1)k +  (−1)n+j (j + 1)k 
j
j=0
n   k  
X n+1 X k p
= (−1)n+j j
j p
j=0 p=0
k  X
n  
X k n+j n + 1
= (−1) jp
p j
p=0 j=0
k  
X k
= (n + 1)p = (n + 1 + 1)k .
p
p=0

Also solved by Bruno Salgueiro Fanego (two solutions), Viveiro, Spain; Kee-Wai Lau,
Hong Kong, China; Moti Levy, Rehovot, Israel, and the proposer.

• 5599: Proposed by Ovidiu Furdui and Alina Sîntămărian, Technical University of Cluj-Napoca,
Cluj-Napoca, Romania

Let n ≥ 2 be an integer. Calculate

Z π
2 sin x + cos x
2n−1 dx.
0 sin x + cos2n−1 x

Solution 1 by Seán M. Stewart, Bomaderry, NSW, Australia

Denote the integral to be found by In where n is a positive integer such that n > 2. Splitting the
integral as follows
Z π Z π
2 sin x 2 cos x
In = 2n−1 dx + 2n−1 dx,
0 sin x + cos2n−1 x 0 sin x + cos2n−1 x

16
 π
if in the first of these integrals we enforce a substitution of x 7→ − x one finds
2
Z π
2 cos x
In = 2 2n−1 dx.
0 sin x + cos2n−1 x
Rearranging the integrand we have
Z π
2 cos x
In = 2 dx
0 cos2n−1 x(1 + tan2n−1 x)
π
sec2n−2 x
Z
2
=2 dx
0 1 + tan2n−1 x
π
sec2n−4 x sec2 x
Z
2
=2 dx.
0 1 + tan2n−1 x
Letting u = tan x, as sec2n−4 x = (1 + u2 )n−2 one has
Z ∞
(1 + u2 )n−2
In = 2 du.
0 1 + u2n−1
From the binomial theorem, since
n−2
X 
2 n−2 n − 2 2k
(1 + u ) = u ,
k
k=0

the integral for In can be rewritten as

n−2 Z ∞
X n−2 u2k
In = 2 du.
k 0 1 + u2n−1
k=0

Identification of the integral that remains as a beta function is now made. Setting t = u2n−1 in the
integral we find
n−2   Z ∞ 2k+2−2n
2 X n−2 t 2n−1
In = dt
2n − 1 k 0 1+t
k=0
2k+1
t( 2n−1 )−1
n−2  Z ∞
2 X n−2
= 2k+1 2n−2k−2 dt
2n − 1 k 0 (1 + t)( 2n−1 )+( 2n−1 )
k=0
n−2    
2 X n−2 2k + 1 2n − 2k − 2
= B , (0)
2n − 1 k 2n − 1 2n − 1
k=0
n−2      
2 X n−2 2k + 1 2n − 2k − 2
= Γ Γ
2n − 1 k 2n − 1 2n − 1
k=0
n−2      
2 X n−2 2k + 1 2k + 1
= Γ Γ 1− (1)
2n − 1 k 2n − 1 2n − 1
k=0
n−2    
2π X n − 2 2k + 1
= cosec π. (2)
2n − 1 k 2n − 1
k=0

Explanation for the changes made are as follows:

17
 ∞
tx−1
Z
(1) Identification of the integral as a beta function, namely B(x, y) = dt, for
0 (1 + t)x+y
x, y > 0.
(2) Relationship between the beta function and gamma function of B(x, y) = Γ(x)Γ(y)/Γ(x+
y) has been used.
(3) Euler’s reflexion formula for the gamma function Γ(x)Γ(1 − x) = π/ sin(πx) has been
used.
So in conclusion the value for the integral is given by
n−2    
2π X n − 2 2k + 1
In = cosec π, n = 2, 3, 4, . . . .
2n − 1 k 2n − 1
k=0

Solution 2 by Albert Stadler, Herrliberg, Switzerland

1 dx
We perform a change of variables: y = tan x, 1 + y 2 = , dy = = (1 + y 2 )dx that results
cos2 x cos2 x
in

Z π Z π
2 sin x + cos x 4 sin x + cos x
dx = 2 dx =
0 sin2n−1 x 0 sin 2n−1
x + cos2n−1 x

π
1
(1 + y 2 )n−2 (1 + y)
Z Z
4 1 + tan x
= 2 dx = 2 dy.
0 cos 2n−2 x(1 + tan2n−1 x) 0 1 + y 2n−1

A partial fraction decomposition applied to the integrand yields

 n−2  
2πi(2k+1) πi(2k+1)
2n−2 1+e 2n−1 1+e 2n−1
(1 + y 2 )n−2 (1 + y) 1 X 1
= · =
1 + y 2n−1 2n − 1 y − e
πi(2k+1)
2n−1 e
πi(2k+1)(2n−2)
2n−1
k=0
   
πi(2k+1)
− 2n−1 n−2 π(n(2k + 1) π(2k + 1)
2n−2 e cos cos
2n−1 i X 2n − 1 2(2n − 1)
= (−1)k −πi(2k+1)
.
2n − 1 1 − ye 2n−1
k=0

Furthermore,
πi(2k+1)
e−
Z 1   1  
2n−1 −πi(2k+1) πi(2k+1)
− 2n−1

−πi(2k+1)
dy = − log 1 − ye 2n−1
 = − log 1 − e
 =
0 1 − ye 2n−1 0

  
π(2k + 1) πi(2k + 1) πi
= − log 2 sin + − .
4n − 2 4n − 2 2

We conclude Z π
2 sin x + cos x
2n−1 dx =
0 sin x + cos2n−1 x
18
 2n−2
2n i X
       
k n−2 π(2k + 1) π(2k + 1) π(2k + 1 πi(2k + 1) πi
(−1) cos cos − log 2 sin + − =
2n − 1 2n − 1 2n − 1 4−2 4n − 2 2
k=0

2n−2
2n X
    
π(2k + 1) π(2k + 1) π π(2k + 1)
= (−1)k cosn−2 cos − =
2n − 1 2n − 1 2(2n − 1) 2 4π − 2
k=0

2n−2
2n+1 π X
    
k n−2 π(2k + 1) π(2k + 1) n−1−k
= (−1) cos cos .
2n − 1 2n − 1 2(2n − 1) 2n − 1
k=0

Solution 3 by Moti Levy, Rehovot, Israel

Z π Z π
sin x + cos x
2 2 sin x
In := 2n−1 2n−1
dx = 2 2n−1 dx. (1)
0 sin x + cos x 0 sin x + cos2n−1 x
After changing the variable w = tan2 x, we get
(1 + w)n−2
Z ∞
In = 1 dw. (2)
0 1 + wn− 2
The following definite integral from Gradshteyn I , Ryzhik I Table Of Integrals, Series And Products
(7Ed , Elsevier, 2007), entry 3.241 is used:
Z ∞ µ−1
t 1 µ µ
dw = B , 1 − , (3)
0 1 + tν ν ν ν
Z 1
where B (x, y) := tx−1 (1 − t)y−1 dt is the Beta function.
0
Proof of (3):
By definition of the Beta function
µ  1 1 µ −1
Z
1 µ µ
B ,1 − = u ν (1 − u)− ν du
ν ν ν ν 0
tν du νtν−1
After changing the variable u = , =
1 + tν dt (tν + 1)2
Z  ν  µν −1 
1
1 ∞ µ νtν−1
Z 
1 µ µ t 1
u ν −1 (1 − u)− ν du = − dt
ν 0 ν 0 1+t ν 1+tν ν (tν + 1)2
Z ∞ µ−1
t
= dx.
0 1 + tν
Expanding (1 + w)n−2 gives,
∞ n−2
wm

n−2
Z X
In = 1 dw
0 m=0 m 1 + wn− 2
n−2
X n − 2 Z ∞ wm
= 1 dw
m 0 1 + wn− 2
m=0
n−2
!
X n−2


m m+1 m+1
= 1B 1 ,1 − 1 .
m=0
n − 2 n − 2 n − 2

19
Applying the reflection rule
π
B (x, 1 − x) = ,
sin (πx)
we conclude that
n−2 n−2


2π X m
In =  .
2n − 1 sin 2m+2
π
m=0 2n−1

Solution 4 by Ioannis D. Sfikas, National Technical University of Athens, Greece

We have
Z π
2 sin x + cos x
In = 2n−1 dx =
0 sin x + cos2n−1 x
Z π Z π
2 cos x 2 dx
2 2n−1 dx = 2 .
0 sin x + cos2n−1 x 0 (1 + tan2n−1 x)(cos2n−2 x)

We may assume: t = tan x. So,

Z +∞
(1 + t2 )n−2
In = 2 dt
0 1 + t2n−1

+∞ n−2 n−2
X n − 2 Z +∞
X n − 2 t2k
Z
1 2k
= 2 2n−1 t dt = 2 dt
0 1 + tt k k 0 1 + t2n−1
k=0 i=0

Since:
+∞
ta
Z
π
J(a, b) = dt = i , with
1 + tb
h
0 b sin (a+1)π
b

1
a,b ∈ < and b > a + 1 > 0. With the substitution = z, we have:
1 + tb
Z 1      
1 a+1 a+1
−1 1 a+1 a+1 1 a+1 a+1
J(a, b) = z b (1 − z) b dz = B ,1 − = Γ Γ 1−
b 0 b b b b b b

π
= h i.
(a+1)π
b sin b

So,we have
n−2 n−2


2π X k
In =  .
2n − 1 (2k + 1)π
k=0 sin
2n − 1

Also solved by Michel Bataille, Ruen, France; Kee-Wai Lau, Hong Kong, China; and
the proposer.

5600: Proposed by Seán M. Stewart, Bomaderry, NSW, Australia

20
Evaluate: Z π
log 1 + 2a cos x + a2 log 1 + 2b cos x + b2 dx,



0

if a, b ∈ < are such that the product ab with |a|, |b| < 1 satisfies the equation a2 b2 + ab = 1.

Solution 1 by Albert Stadler, Herrliberg, Switzerland

We claim that

Z π
log 1 + 2a cos x + a2 log 1 + 2b cos x + b2 dx = 2πLi2 (a, b),



(*)
0

where Li2 (x) is the dilogarithm (see for instance

https://en.wikipedia.og/wiki/Spence%27s function).
1 √  1 √ 
a2 b2 + ab = 1 implies that ab = −1 + 5 or ab = −1 − 5 . However |a|, |b| < 1, so
2 2
1 √ 
ab = −1 + + 5 It is known (see above reference) that
2
√ π2 √
   
1 2 1
Li2 (−1 + 5) = − ln (1 + 5) .
2 10 2
Therefore
π
π3 √
Z  
1
log 1 + 2a cos x + a2 log 1 + 2b cos x + b2 dx = − 2π ln2



(1 + 5) .
0 5 2

Proof of (∗) :
Z π
log(1 + 2a cos x + a2 ) log(1 + 2b cos x + b2 )dx =
0
Z 2π
1
= log(1 + 2a cos x + a2 ) log(1 + 2b cos x + b2 )dx =
2 0
Z 2π
1
log 1 + 2a(eix + e−ix ) + a2 log 1 + b(eix + e−ix ) + b2 dx =



=
2 0
Z 2π
1
log (1 + aeix )(1 + ae−ix ) log (1 + beix )(1 + be−ix ) dx =



=
2 0
Z 2π Z 2π
1 1
+ log(1 + aeix ) log(1 + beix )dx + log(1 + aeix ) log(1 + be−ix )dx+
2 0 2 0

1 2π 2π
Z Z
1
+ log(1 + ae−ix ) log(1 + beix )dx + log(1 + ae−ix ) log(1 + be−ix )dx+
2 0 2 0
∞ ∞
! !
2π
(−1)k+1 (−1)k+1
Z
1 X
k ikx
X
k ikx
= a e b e dx+
2 0 k k
k=1 k=1
∞ ∞
! !
2π
(−1)k+1 k ikx (−1)k+1 k −ikx
Z
1 X X
+ a e b e dx+
2 0 k k
k=1 k=1

21
 ∞ ∞
! !
2π
(−1)k+1 (−1)k+1
Z
1 X X
+ ak e−ikx bk eikx dx+
2 0 k k
k=1 k=1
∞ ∞
! !
2π
(−1)k+1 (−1)k+1
Z
1 X
k −ikx
X
k −ikx
+ a e b e dx =
2 0 k k
k=1 k=1
∞ ∞
X ak bk X ak bk
=0+π +π + 0 = 2πLi2 (ab).
k2 k2
k=1 k=1

interchange of summation and integration is permitted, since all series converge absolutely and
uniformity for 0 ≤ x ≤ 2π.

Solution 2 by Michel Bataille, Rouen, France

√ !!
π2 5−1
Let I be the required integral. We show that I = 2π − log2 .
10 2

We observe that
Z π Z a  Z b  Z aZ b
2(t + cos x) 2(u + cos x)
I= dt · du dx = 4 F (t, u) dt du
0 0 1 + 2t cos x + t2 0 1 + 2u cos x + u
2
0 0

where Z π
(t + cos x)(u + cos x)
F (t, u) = dx.
0 (1 + 2t cos x + t2 )(1 + 2u cos x + u2 )
To calculate F (t, u), we make use of the change of variables defined by y = tan(x/2), that is,
2dy 1 − y2
x = arctan(2y), dx = , cos x = . We obtain:
1 + y2 1 + y2
Z ∞
[y 2 (1 − t) − (1 + t)][y 2 (1 − u) − (1 + u)] dy 2
F (t, u) = 2 ·
2 (1 − t)2 + (1 + t)2 ][y 2 (1 − u)2 + (1 + u)2 ] y 2 + 1
= J(r, s)
0 [y (1 − t)(1 − u)
1+t 1+u
where r = > 0, s = > 0 (since |t| ≤ |a| < 1, |u| ≤ |b| < 1) and
1−t 1−u
Z ∞
(y 2 − r)(y 2 − s)
J(r, s) = dy.
0 (y 2 + r2 )(y 2 + s2 )(y 2 + 1)
π
Since J(r, s) = (see a quick proof at the end), we have
2(r + s)
2 π π
F (t, u) = ·  = .
(1 − t)(1 − u) 2 1+t + 1+u 2(1 − tu)
1−t 1−u

√
5−1
Noticing that from the hypotheses on ab, we must have ab = , we obtain
2
b a Z b Z ab
log(|1 − au|) log |1 − w|
Z Z
dt
I = 2π du = 2π − = −2π dw
0 0 1 − tu 0 u 0 w

22
hence
ab
log(1 − w)
Z
I = 2π − dw = 2πLi2 (ab)
0 w
x X xn ∞
ln(1 − w)
Z
where Li2 (x) = − dw = denotes the dilogarithm function. The claimed result
0 w n2
n=1
now follows from √ ! √ !
5−1 π2 5−1
Li2 = − log2
2 10 2
(see for example: D. Zagier, The Remarkable Dilogarithm, J. Math. and Phys. Sciences, 22(1988),
131-145).
π
Proof of J(r, s) = .
2(r + s)
If r 6= s, we have

(y 2 − r)(y 2 − s) 1 1 r(r2 + s) 1 s(r + s2 ) 1

2 2 2 2 2
= · + · + · 2
(y + r )(y + s )(y + 1) (r − 1)(s − 1) y + 1 (r − 1)(r − s ) y + r (s − 1)(s − r ) y + s2
2 2 2 2 2 2 2

Z ∞  ∞
dy 1 π
and the result readily follows from 2 2
= arctan(y/m) = for positive m.
0 y +m m 0 2m
If r = s, r 6= 1, the result similarly follows from the decomposition

(y 2 − r)2 r2 (r2 − 1) r2 − 2r
 
1 1
= − + 2
(y 2 + 1)((y 2 + r2 )2 (r − 1)2 y 2 + 1 (y 2 + r2 )2 y + r2
∞
1 ∞ 1 π/2
Z Z Z
dy dx π
and 2 2 2
= 3 2 2
= 3 cos2 u du = 3 .
0 (y + r ) r 0 (x + 1) r 0 4r
If r = s = 1, the decomposition

(y 2 − 1)2 4 4 1
2 3
= 2 3
− 2 2
+ 2
(y + 1) (y + 1) (y + 1) y +1
Z ∞ Z π/2
dy 3π
and = cos4 u du = readily show that the result is still valid.
0 (y + 1)3
2
0 16

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

√ !!
π2 5−1
We show that the given integral, denoted by I, equals 2π − log2 .
10 2
It is known ([2], p.237) that for |k| < 1 , we have
∞
X (−1)m−1 k m cos(mx).
log(1 + 2k cos x + k 2 ) = 2
m
m=1

Since
∞ X
∞ ∞ ∞
  ! !
X (−1)m+n am bn cos(mx) cos(nx)  X |a|m X |b|n
 ≤
 
mn m n

 
m=1 n=1 m=1 n=1

23
 = log(1 − |a|) log(1 − |b|) < ∞,

so interchanging the order of integration and summation, we have

∞ X ∞
Z π X !
(−1)m+n am bn cos(mx) cos(nx)
I = 4 dx
0 mn
m=1 n=1

∞ X
∞ π
(−1)m+n am bn
X Z
= 4 cos(mx) cos(nx)dx.
mn 0
m=1 n=1

It is well known that for positive integers m and n, we have

(
Z π 0, m 6= n
cos(mx) cos(nx)dx = π
0 , m = n.
2
∞
X (ab)m
Hence I = 2π ; Since |a|, b| < 1 satisfy the equation
m2
m=1
√
2 2 5−1
a b + ab = 1, so ab = . According to entry (2.6.12) in theorem 2.63 on p. 105 of [1], we
2
 √ m
∞ 5−1 √ !
X 2 π2 5 − 1
have = − log2 . Hence our claim for I.
m2 10 2
m=1

References:

1. G.E. Andrews, R. Askey, R. Roy: Special Functions, Cambridge University Press, 1999
2. Paul J. Nahin: Inside Interesting Integrals, Springer-Verlag, New York, 2015.

Solution 4 by Moti Levy, Rehovot, Israel

Let
Z π
log 1 + 2a cos x + a2 log 1 + 2b cos x + b2 dx,



J (a, b) :=
0
1
F (x, a) := .
1 + 2a cos x + a2
Then by differentiation under the integral sign,
∂2J
∂a∂bZ
π
(a + cos x) (b + cos x)
=4 2 2
dx
0 (1 + 2a cos x + a ) (1 + 2b cos x + b )
a2 − 1 b2 − 1 π
Z π

a2 − 1 π b2 − 1 π
Z Z Z
1
= dx + F (x, a) dx + F (x, a) dx + F (x, a) F (x, b) dx
0 ab ab 0 ab 0 ab 0
(1)

24
 x
The following definite integrals can be evaluated by substitution t = tan ,
2
Z π Z ∞
2
F (x, a) dx = dt
0 0 (a − 1) t + (a + 1)2
2 2
 ∞
2 1−a 1
= arctan t = π. (2)
1 − a2 1+a 0 1 − a2

Z π Z ∞
1 1 2
F (x, a) F (x, b) dx = 1−t2 2 dt
0 0 1+ a2 + 2a 1+t2 1+ b2 + 2b 1−t 1 + t2
1+t2
∞
t2 + 1
Z
2    dt
0 (a − 1)2 t2 + (a + 1)2 (b − 1)2 t2 + (b + 1)2
    ∞
1−a
−2a 1 − b2 arctan 1+a t + 2b 1 − a2 arctan 1−b



1+b t
= 2 2 2 2

(1 − a ) (1 − b ) (b − a + a b − ab )
0
(1 + ab)
= π. (3)
(1 − a ) (1 − b2 ) (1 − ab)
2

Substitution of (2) and (3) into (1) gives

∂2J 2π
= .
∂a∂b 1 − ab
Since
∂J
= f or b = 0,
∂a
J = 0 f or a = 0,

then Z bZ a
2π
J (a, b) = dudv = 2πLi2 (ab) ,
0 0 1 − uv
where Li2 (x) is the Dilogarithm function defined by
∞
X xk
Li2 (x) := , |x| < 1.
k2
k=1
√
2 2 1 5+1
The condition a b + ab = 1 implies that ab = , where φ is the golden ratio φ = .
φ 2
1
The value of the Dilogarithm at has been calculated using the properties of the Dilogarithm (see
φ
the entry Spence’s function in Wikipedia),

π2
 
1 3
Li2 = ζ (2) − ln2 φ = − ln2 φ.
φ 5 10
We conclude that the integral is equal to

π3
− 2π ln2 φ ∼
= 4.74629
5
25
Solution 5 by Ioannis D. Sfikas, National Technical University of Athens, Greece
The definite integral is the limit of the Riemann sum: The definite integral of a continuos function
Z b
f over the interval [a, b], denote by f (x)dx is the limit of a Riemann sum as the number of
a
subdivisions approaches infinity. That is:
Z b n
X
f (x)dx = lim ∆x · f (xi ),
a n→∞
i=1

b−a
where ∆x = and xi = a + ∆x · i. Let:
n
Z π Z π
F (a) = log(1 + 2a cos x + a2 )dx, F (b) = log(1 + 2b cos x + b2 )dx,
0 0
Z π
F (a, b) = log(1 + 2a cos x + a2 ) log(1 + 2b cos x + b2 )dx.
0

First, we have
1 + 2a cos x + a2 = (a + cos x)2 + sin2 x = a + eix  ,
 

1 + 2b cos x + b2 = (b + cos x)2 + sin2 x = |b + eix |2 .

Hence:
n
πX  k 2
 1
F (a) = lim log a + ei n  = π lim log |an + 1|2 = 0.

n→∞ n n→+∞ n
k=1

n
πX  k 2
 1
F (b) = lim in
log b + e  = π lim log |bn + 1|2 = 0

n→∞ n n→+∞ n
k=1

for a, b ∈ < are such that the product ab with |a|, |b| < 1, satisfies the equation a2 b2 + ab = 1.
Using the identities:
X cos(nx)
log(1 − 2a cos x + a2 ) = −2 an ,
n
n≥1
Z π
π
with x, a ∈ <, |a| < 1, and cos mx cos(nx)dx = δm,n where δ is the Kroneocker delta, we have:
0 2
n
πX  k 2
in
  k 2
in

F (a, b) = lim log a + e  log b + e 
 
n→+∞ n
k=1
Z n X cos(nx) X cos(mx)
=4 (−a)n (−b)m dx
0 n m
n≥1 m≥1

X (−a)n (−b)m Z π
= 4 cos(nx) cos(mx)dx
nm 0
m,n≥1
X (ab)n
= 2π = 2πLi2 (ab).
n2
n≥1

26
where Li2 x is the dilogarithm. Also, we have:
√
5−1
ab = = φ ≈ 0.61803,
2
where φ is the conjugate of the golden ratio. So, we have:
 2
π2

−1 1
Li2 (ab) = Li2 (φ) = − sinh ≈ 0.75539561953
10 2

where sinh−1 x the inverse hyperbolic sine. So, we have:

 2 !
π2

−1 1
F (a, b) = 2πLi2 (φ) = 2π − sinh ≈ 4.746291.
10 2

Also solved by the proposer.

M ea Culpa
The name of Ioannis D. Sfikas of the National Technical University of Athens, Greece
was inadvertently not listed as having solved problems 5589, 5590, and 5591.
The name of Albert Stadler of Herrliberg, Switzerland was also inadvertently omitted from
the list of those who solved 5590.

27