Solutions

J1. Solve in real numbers the system of equations:

_
r
4
+ 2j
3
r =
1
4
+ 3
_
3
j
4
+ 2r
3
j =
1
4
3
_
3.
Proposed by Dr. Titu Andreescu, University of Texas at Dallas
Solution by Bogdan Enescu, "B.P. Hasdeu" National College, Romania
Adding up the two equations yields
r
4
+ 2r
3
r +
1
4
+j
4
+ 2j
3
j +
1
4
= 0,
or, equivalently,
_
r
2
+r
1
2
_
2
+
_
j
2
+j
1
2
_
2
= 0.
We deduce that r, j
_
1
p
3
4
,
1+
p
3
4
_
. Observe that if c is a root of the
equation t
2
+t
1
2
= 0, then c
2
= c+
1
2
, c
3
= c
2
+

2
=
31
2
, and, nally,
c
4
= 2c +
3
4
. Replacing in the rst equation of the system gives
r +j =
_
3,
hence we must have r =
1
p
3
4
and j =
1+
p
3
4
.
Also solved by Bin Zhao, YunYuan HuaZhong University of Technology and Sci-
ence,Wuhan,CHINA and the proposer.
F
J3. Consider the sequence
a
n
=
_
1 +
_
1 +
1
:
_
2
+
_
1 +
_
1
1
:
_
2
, : _ 1.
Prove that
1
a
1
+
1
a
2
+. . . +
1
a
20
is an integer.
Proposed by Dr. Titu Andreescu, University of Texas at Dallas
Solution by by Jos Luis Daz-Barrero, Universidad Politcnica de Catalua,
Barcelona, Spain
Mathematical reflections 2, (2006) 1
Since
1
a
n
=
1
_
1 +
_
1 +
1
:
_
2
+
_
1 +
_
1
1
:
_
2
=
:
4
_
_
_
1 +
_
1 +
1
:
_
2

_
1 +
_
1
1
:
_
2
_
_
,
then
1
a
1
+
1
a
2
+. . . +
1
a
20
=
_
_
5
4

1
4
_
+
_
_
13
4

_
5
4
_
+
_
5
4

_
13
4
_
+
_
_
41
4

5
4
_
+
_
_
61
4

_
41
4
_
+. . .
+
_
_
685
4

_
613
4
_
+
_
_
761
4

_
685
4
_
+
_
29
4

_
761
4
_
=
29
4

1
4
= 7
and we are done.
Also solved by Bin Zhao, YunYuan HuaZhong University of Technology and Sci-
ence,Wuhan,CHINA and the proposer.
F
J5. Let r, j, . be positive real numbers such that rj. = 1. Show that the
following inequality holds:
1
(r + 1)
2
+j
2
+ 1
+
1
(j + 1)
2
+.
2
+ 1
+
1
(. + 1)
2
+r
2
+ 1
_
1
2
.
Proposed by Dr. Cristinel Mortici, Valahia University, Trgoviste, Romania
Solution by Bin Zhao, YunYuan HuaZhong University of Technology and
Science,Wuhan,CHINA
We have
1
(r + 1)
2
+j
2
+ 1
=
1
2 +r
2
+j
2
+ 2r
_
1
2 (1 +r +rj)
Since rj. = 1, we substitute
r =
/
a
, j =
c
/
, . =
a
c
Mathematical reflections 2, (2006) 2
to obtain
1
2 (1 +r +rj)
=
a
2 (a +/ +c)
.
Writing the other two similar inequalities and adding them up gives the desired
result.
Also solved by the proposer.
F
S1. Prove that the triangle 1C is right-angled if and only if
cos

2
cos
1
2
cos
C
2
sin

2
sin
1
2
sin
C
2
=
1
2
.
Proposed by Dr. Titu Andreescu, University of Texas at Dallas
Solution by Daniel Campos Salas
Let

2
=

4
+j, with [j[ <

4
. Then the given identity turns into
_
_
2
2
cos j
_
2
2
sinj
_
cos
1
2
cos
C
2

_
_
2
2
cos j +
_
2
2
sinj
_
sin
1
2
sin
C
2
=
1
2
.
Rearranging the terms it follows that this result holds if and only if
cos j
_
cos
1
2
cos
C
2
sin
1
2
sin
C
2
_
sinj
_
cos
1
2
cos
C
2
+ sin
1
2
sin
C
2
_
=
_
2
2
=cos j cos
_
1 +C
2
_
sinj cos
_
1 C
2
_
=
_
2
2
=cos j cos
_

4
j
_
sinj cos
_
1 C
2
_
=
_
2
2
=
_
2
2
_
cos
2
j + cos j sinj
_
sinj cos
_
1 C
2
_
=
_
2
2
Using the identity cos
2
c = 1 sin
2
c, it follows that the equation holds if
and only if
sinj
_
_
2
2
(cos j sinj) cos
_
1 C
2
_
_
= 0
=sinj
_
cos
_

4
+j
_
cos
_
1 C
2
__
= 0
Mathematical reflections 2, (2006) 3
=sinj
_
cos

2
cos
_
1 C
2
__
= 0
=sinj sin
_
+1 C
4
_
sin
_
+1 C
4
_
= 0
=sin
_
2
4
_
sin
_
21
4
_
sin
_
2C
4
_
= 0,
which gives the desired result.
Also solved by Bin Zhao, YunYuan HuaZhong University of Technology and Sci-
ence,Wuhan,CHINA and the proposer. Partially solved by Karsten Bohlen.
F
S5. Let a and / be two real numbers such that a
p
/
p
is a positive integer
for each prime number j. Prove that a and / are integer numbers.
Proposed by Nairi Sedrakian, Yierevan, Armenia.
Solution by the author
First we will prove that numbers a, / are rational. The case a/ = 0 is simple.
Consider now a/ ,= 0. Then we have
_
a
5
/
5
_
2

_
a
7
/
7
_ _
a
3
/
3
_
= a
3
/
3
_
a
2
/
2
_
2
,
hence a
3
/
3
is a rational number. From the latter and the identity
_
a
7
/
7
_
2

_
a
11
/
11
_ _
a
3
/
3
_
= a
3
/
3
_
a
2
/
2
_
2
_
a
2
+/
2
_
2
we deduce, that the number
_
a
2
+/
2
_
2
=
_
a
2
/
2
_
2
+4a
2
/
2
is a rational number,
and so a
2
/
2
is rational, from which follows that a/ =
a
3
/
3
a
2
/
2
is also rational
number.
Since
_
a
5
/
5
_ _
a
11
/
11
_

_
a
13
/
13
_ _
a
3
/
3
_
= a
3
/
3
_
a
2
/
2
_
2
_
a
2
+/
2
_ _
a
4
+/
4
_
,
then
_
a
2
+/
2
_ _
a
4
+/
4
_
=
_
a
3
/
3
_
2
+ 2a
3
/
3
+a
2
/
2
_
a
2
+/
2
_
is a rational number, from which we deduce that a
2
+ /
2
is a rational number.
Finally the numbers a / =
a
3
/
3
a
2
+a/ +/
2
and a + / =
a
2
/
2
a /
are rational
numbers, and hence a and / are rational.
Mathematical reflections 2, (2006) 4
Let a =
:
:
and / =
/
:
, where :, / Z, : N and : is minimal. Then for
any prime number j, :
p
divides :
p
/
p
. Now since : is minimal we nd that
(:, :) = (/, :) = 1. If : 1, then there exists a prime number so that
divides :. We have
_
:
3
/
3
_ _
:
2
+/
2
_

_
:
3
+/
3
_ _
:
2
/
2
_
= 2:
2
/
2
(:/),
consequently
2
divides 2(:/), that is, divides :/.
If j , then :
p
/
p
= (:/)
_
:
p1
+:
p2
/ + +/
p1
_
is divisible by

p
and
_
:
p1
+:
p2
/ + +/
p1
_
= j/
p1
(mod), so that
_
:
p1
+:
p2
/ + +/
p1
_
isnt divided by , hence
p
divides :/, which is impossible because j, (j )
is any prime number, while :/ ,= 0.
Thus : = 1, and hence a and / are integers.
Also solved by Iurie Boreico, student, Chisinau, Moldova
F
S6. Let a, /, c be positive real numbers such that a +/ +c = 3. Prove that
1. If a _ / _ 1 _ c, then
1
a +/
+
1
/ +c
+
1
c +a
_
1
a + 1
+
1
/ + 1
+
1
c + 1
;
2. If a _ 1 _ / _ c, then
1
a +/
+
1
/ +c
+
1
c +a
_
1
a + 1
+
1
/ + 1
+
1
c + 1
;
Proposed by Marian Tetiva, Brlad, Romania
Solution by the author
Using the condition a +/ +c = 3, we have

_
1
a +/

1
c + 1
_
=

c + 1 (a +/)
(a +/) (c + 1)
= 2

c 1
(a +/) (c + 1)
.
Mathematical reflections 2, (2006) 5
But
a 1
(/ +c) (a + 1)
+
/ 1
(a +c) (/ + 1)
+
c 1
(a +/) (c + 1)
=
a 1
(/ +c) (a + 1)
+
/ 1
(a +c) (/ + 1)

a +/ 2
(a +/) (c + 1)
=
a 1
(/ +c) (a + 1)

a 1
(a +/) (c + 1)
+
/ 1
(a +c) (/ + 1)

/ 1
(a +/) (c + 1)
=
(a 1) (/ 1) (c a)
(/ +c) (a + 1) (a +/) (c + 1)
+
(a 1) (/ 1) (c /)
(a +c) (/ + 1) (a +/) (c + 1)
.
The conclusion follows observing that c a _ 0, c / _ 0 and if a _ / _ 1, then
(a 1) (/ 1) _ 0 while if a _ 1 _ /, then (a 1) (/ 1) _ 0.
Also solved by Bin Zhao, YunYuan HuaZhong University of Technology and Sci-
ence,Wuhan,CHINA
F
U1. Evaluate
1
_
0
3
_
2r
3
3r
2
r + 1dr.
Proposed by Dr. Titu Andreescu, University of Texas at Dallas
Solution by Bin Zhao, YunYuan HuaZhong University of Technology and
Science,Wuhan,CHINA
Let
)(r) =
3
_
2r
3
3r
2
r + 1
First we will prove:
)(r) = )(1 r)
This is because
3
_
2r
3
3r
2
r + 1 =
3
_
2 (1 r)
3
3 (1 r)
2
(1 r) + 1
== 2r
3
3r
2
r + 1 =
_
2 (1 r)
3
3 (1 r)
2
(1 r) + 1
_
,
which is obvious by expanding.
Thus:
2
_
1
0
3
_
2r
3
3r
2
r + 1dr = 2
_
1
0
)(r)dr
Mathematical reflections 2, (2006) 6
=
_
1
0
)(r)dr +
_
1
0
)(1 r)dr =
_
1
0
()(r) +)(1 r)) dr = 0
So
_
1
0
3
_
2r
3
3r
2
r + 1dr = 0
Also solved by Iurie Boreico, student, Chisinau, Moldova, and the author.
F
U2. Solve in real numbers the equation
6
x
+ 1 = 8
x
27
x1
.
Proposed by Dr. Titu Andreescu, University of Texas at Dallas
Solution by the author
If we denote a = 1, / = 2
x
, c = 3
x1
, the equation becomes
a
3
+/
3
+c
3
3a/c = 0.
The latter factors as
a
3
+/
3
+c
3
3a/c = (a +/ +c)
_
a
2
+/
2
+c
2
a/ /c ac
_
,
and the second paranthesis equals zero only if a = / = c, which is impossible.
Therefore the initial equation is equivalent to
1 2
x
+ 3
x1
= 0.
Write this as
3
x1
2
x1
= 2
x1
1,
and for each r consider the function ) (t) = t
x1
, t 0. It follows from La-
granges theorem the existence of the numbers c (2, 3) and , (1, 2) such
that ) (3) ) (2) = )
0
(c) and ) (2) ) (1) = )
0
(,) . Since )
0
(t) = (r 1) t
x2
,
we obtain
(r 1) c
x2
= (r 1) ,
x2
.
Clearly, this implies either r = 1 or (since c ,= ,) r = 2.
Also solved by Iurie Boreico, student, Chisinau, Moldova
F
Mathematical reflections 2, (2006) 7
U5. Let : _ 2 be a positive integer. Prove that there is no rational function
1(r) such that
1
2
s
+
1
3
s
+. . . +
1
:
s
= 1(:) ,
for all positive integers : _ 2.
Proposed by Mihai Piticari, Cmpulung, and Dorin Andrica, Babes-Bolyai
University, Cluj Napoca, Romania
Solution by Iurie Boreico, student, Chisinau, Moldova
Suppose there exists such a function 1(:). Let
o(:) =
1

i=2
1
i
s
1(:),
thus o(:) =

1
i=n+1
1
i
s
. Let o(r) =
P(x)
Q(x)
, qcd(1, Q) = 1.
Then o(: 1) o(:) =
1
n
s
, or
:
s
(1(: 1)Q(:) 1(:)Q(: 1) = Q(: 1)Q(:)).
This relation must therefore hold for any r, as it holds for innitely many r.
Particularly, if o(r) is not dened, then o(r 1) is also not dened so if r is a
root of Q then so is r 1. This would mean that Q has innitely many roots,
which is not the case so Q is constant. However in this case o is a polynomial,
but every non-constant polynomial is unbounded, which is not the case for o(:).
Also solved by the authors
F
O1. A circle centered at O is tangent to all sides of the convex quadrilateral
1C1. The rays 1 and C1 intersect at 1, the rays 1 and 1C intersect at
1. The points A, 1 are considered on the line segments O, OC, respectively.
Prove that \A11 =
1
2
\1C if and only if \A11 =
1
2
\1C.
Proposed by Pavlo Pylyavskyy, graduate student, MIT
Solution by Iurie Boreico, student, Chisinau, Moldova
Since we have
\1O = \C1O =
1
2
\1C,
the equality \A11 =
1
2
\1C is equivalent to \A1O = \1 1C and equiva-
lent to
sin\1A
sin\A1O
=
sin\O11
sin\1 1C
.
Mathematical reflections 2, (2006) 8
Using the sine law in triangles A1, AO1, O11, 1 1C the latter is equivalent
to
AO
A
1 O
1 C
1 C1
1O
2
= 1.
Analogously, \A11 =
1
2
\1C is equivalent to
AO
A
1 O
1 C
1 C1
1O
2
= 1.
Thus, to prove the conclusion we need to prove that
1 C1
1O
2
=
1 C1
1O
2
.
But this follows from the sine law applied in triangles 1O, 1OC, 1O, and
1OC.
Also solved by Yufei Zhao, Don Mills Collegiate Institute, Toronto, Canada, and
the author.
F
O2. Find all positive integers : such that the set
= 1, 3, 5, . . . , 2: 1
can be partitioned into 12 subsets, the sum of elements in each subset being the
same.
Proposed by Marian Tetiva, Brlad, Romania
Solution by Yufei Zhao, Don Mills Collegiate Institute, Toronto, Canada
The sum of the elements in is 1 + 3 + 5 + + (2: 1) = :
2
. Thus, in
each of the 12 subsets, the sum of the elements is :
2
,12. In order for :
2
,12 to
be an integer, : has to be a multiple of 6. So let : = 6/, where / is a positive
integer. Then, the sum of elements in each of the 12 subsets is 3/
2
.
Note that since the element 2:1 = 12/ 1 has to be in one of the subsets,
and the sum of all the elements in that subset is 3/
2
, we must have 12/1 _ 3/
2
.
Thus 3(/ 2)
2
_ 11, implying that / _ 4.
First, let us show that / ,= 5. Suppose that / = 5, then = 1, 3, 5, . . . , 59
has 30 elements, and the sum of elements in each of the 12 subsets must be 75.
Thus each subset must have more than one element. Considering the parity,
we see that each subset must have an odd number of elements. Therefore, each
of the 12 subsets must contain at least 3 elements, which is impossible since
[[ = 30.
Now, we will show examples of the partition for / = 4, 6, 7, 9.
When / = 4, = 1, 3, . . . , 47. We can take the subsets 1
i
= 2i 1, 49
2i, i = 1, 2, . . . , 12. Each subset has a sum of 48.
Mathematical reflections 2, (2006) 9
When / = 6, = 1, 3, . . . , 71. We can take the subsets 1
i
= 35 +
2i, 73 2i, i = 1, 2, . . . , 9, 1
10
= 23, 25, 29, 31, 1
11
= 19, 21, 33, 35, and
1
12
= 1, 3, 5, 7, 9, 11, 13, 15, 17, 27. Each subset has a sum of 108.
When / = 7, = 1, 3, . . . , 83. We can take the subsets 1
i
= 4i 3, 65
2i, 852i, i = 1, 2, . . . , 10, 1
11
= 3, 27, 35, 39, 43 and 1
12
= 7, 11, 15, 19, 23, 31, 41.
Each subset has a sum of 147.
When / = 9, = 1, 3, . . . , 107. We can take the subsets 1
i
= 23 +
8i, 1094i, 1114i, i = 1, 2, . . . , 7, 1
8
= 1, 41, 65, 67, 69, 1
9
= 3, 15, 73, 75, 77,
1
10
= 29, 37, 57, 59, 61, 1
11
= 5, 9, 17, 23, 25, 27, 33, 51, 53, 1
12
= 7, 11, 13, 19, 21, 35, 43, 45, 49.
Each subset has a sum of 243.
Now, we show that if / = /
0
yields a partition, then so does / = /
0
+ 4.
Suppose that 1
i
, i = 1, 2, . . . , 12 is a partition for = 1, 3, . . . , 12/
0
1.
Then the new set
0
= 1, 3, 5, . . . , 12(/
0
+ 4) 1 can be partitioned into the
following 12 subsets: 1
0
i
= 1
i
' 12/
0
1 + 2i, 12/
0
+ 49 2i, i = 1, 2, . . . , 12.
For each i, the sum of the elements in 1
0
i
is greater than that of 1
i
by 24/
0
+48.
Since all the 1
i
s have the same sum as each other, so do the 1
0
i
s.
Combining the above results, we see that / can be any number of the form
4 + 4:, 6 + 4:, 7 + 4:, or 9 + 4:, where : is a non-negative integer. That
is, / can be any positive integer other than 1, 2, 3, 5 (and we have shown that
these values are invalid).
Therefore, : = 6/, where / is either 4 or any positive integer greater than 5.
Also solved by Iurie Boreico, student, Chisinau, Moldova and the author.
F
O3. Prove that there are innitely many prime numbers j with the following
property: in the main period of the decimal representation of
1
p
, the number of
1s plus the number of 3s equals the number of 6s plus the number of 8s.
Proposed by Adrian Zahariuc, student, Romania
Solution by Yufei Zhao, Don Mills Collegiate Institute, Toronto, Canada
Let j be a prime divisor of 10
n
+ 1, where : N. We will show that in the
main period of the decimal expansion of 1,j, the number of times that the digit
i appears equals to the number of times that the digit (9 i) appears, for each
i = 0, 1, 2, . . . , 9.
Let 10
n
+ 1 = ja, where a is a positive integer. Note that a has at most
: digits. Let the decimal representation of a be a
1
a
2
a
3
a
n
(ll with leading
zeros if necessary). Note that a
n
,= 0 since 10
n
+ 1 is not divisible by 10. Now,
1
j
=
a(10
n
1)
10
2n
1
=
[10
n
(a 1)] + [(10
n
1) (/ 1)]
10
2n
1
Observe that the decimal representation of [10
n
(a 1)] +[(10
n
1) (/ 1)] is
a
1
a
2
a
n1
(a
n
1)(9 a
1
)(9 a
2
) (9 a
n1
)(10 a
n
)
Mathematical reflections 2, (2006) 10
Thus, the decimal expansion of 1,j is
0.a
1
a
2
a
n1
(a
n
1)(9 a
1
)(9 a
2
) (9 a
n1
)(10 a
n
)
(Although the above representation may include a multiple of the minimal pe-
riod, it does not aect our purpose in comparing the number of times that each
digit appears). Note that for each appearance of the digit i, the digit 9 i
appears in the corresponding position in the other half of the period. Thus
there is a one-to-one correspondence between the appearances of the digit i and
that of 9 i in the period. It follows that those two digits always appear an
equal number of times.
Now, we need to show that there are innitely many primes that is the divisor
of 10
n
+1 for some : N. Consider the set containing all numbers of the form
10
n
+1 where : is a prime greater than 11. Suppose that 10
m
+1, 10
n
+1 ,
and let j (if it exists) be a prime common divisor of the two numbers. Then
10
m
= 1 (mod j) and 10
n
= 1 (mod j). It follows that 10
2m
= 1 (mod j)
and 10
2n
= 1 (mod j). Since gcd(2:, 2:) = 2, we have 1 = 10
gcd(2m;2n)
= 100
(mod j). So j [ 99. Note that 3 does not divide 10
n
+ 1 (since the sum of its
digits is 2). Therefore, we can only have j = 11. So the only prime that can
divide into more than one element in is 11.
Let 10
n
+ 1 . Note that : is odd. Now,
10
n
+ 1
10 + 1
= 10
n1
10
n2
+ 10
n3
+ 10
2
10 + 1
= (1)
n1
(1)
n2
+ (1)
n3
+ (1)
2
(1) + 1 (mod 11)
= : (mod 11)
Since : is a prime greater than 11, we see that
10
n
+1
11
is not divisible by 11.
It follows that the elements of 1 =
10
n
+1
11
[ : 11 is prime are mutually
relatively prime. Therefore, there are innitely many prime numbers that divide
into some element of 1. Hence, there are innitely many prime numbers that
divide 10
n
+ 1 for some positive integer :.
The result follows immediately.
Also solved by Iurie Boreico, Chisinau, Moldova, and the author.
F
O4. Let 1 be a diameter of the circle and let C be a point on the circle,
dierent from and 1. Denote by 1 the projection of C on 1 and let . be
a circle tangent to 1, C1, and , touching at A. Prove that the angle
bisectors of \A1 and \C1 meet on 1.
Proposed by Liubomir Chiriac, student, Moldova
Solution by Iurie Boreico, student, Chisinau, Moldova
Mathematical reflections 2, (2006) 11
If 1 is the point of tangency of the small circle with 1 then a well-known
states that A1 bisects \A1. Therefore we must prove that C1 bisects \C1
which is equivalent to
11 =
C11
C1 +C
=
C11(C C1)
C
2
C1
2
=
C1(C C1)
1
.
With no loss of generality, let 1 (O) where O is the center of the big
circle, and let O = 1O = 1, 1 = :. If J is the center of the small circle
and r is its radius then 11 = r. Since the two circles are tangent, we get
OJ
2
= (1 r)
2
or r
2
+(r +1 :)
2
= (1 r)
2
. Solving the quadratic equation,
we get 11 = r =
_
2(2 :) (2 :). From the other side,
C1 =
_
111 =
_
:(2 :), C =
_
11 =
_
2:
so
C1(C C1)
1
=
_
:(2 :)(
_
2:
_
:(2 :))
:
=
_
2(2 :)(2:) = 11.
Also solved by Yufei Zhao, Don Mills Collegiate Institute, Toronto, Canada, and
the author.
F
O6. Let r, j, . be nonnegative real numbers. Prove the inequality
r
4
(j +.) +j
4
(. +r) +.
4
(r +j) _
1
12
(r +j +.)
5
.
Proposed by Vasile Crtoaje, Ploiesti, Romania
Solution by Yufei Zhao, Don Mills Collegiate Institute, Toronto, Canada
If r = j = . = 0, then the inequality obviously holds. So assume that
not all of the three variables are zero. Since the inequality is homogeneous and
symmetric, we may assume wolog that r + j + . = 1 and r _ j _ .. It is now
equivalent to prove the inequality
r
4
(1 r) +j
4
(1 j) +.
4
(1 .) _
1
12
(r +j +.)
5
Let )(r) = r
4
(1 r). Note that its second derivative )
002
20r
3
is positive
in (0,
4
5
). Thus )(r) is convex in [0,
4
5
]. Since r _ j _ . and r + j + . = 1,
we must have
1
2
_ j + . _ j _ .. Hence j, ., j + . [0,
4
5
]. By Karamatas
Majorization Inequality,
j
4
(1j)+.
4
(1.) = )(j)+)(.) _ )(j+.)+)(0) = (j+.)
4
(1j.) = (1r)
4
r
Mathematical reflections 2, (2006) 12
Therefore,
r
4
(1 r) +j
4
(1 j) +.
4
(1 .) _ r
4
(1 r) + (1 r)
4
r
= r(1 r)((1 r)
3
+r
3
)
= r(1 r)(1 3r + 3r
2
)
=
1
3
[3r(1 r)] [1 3r(1 r)]
_
1
3

1
4
=
1
12
(r +j +.)
5
The last inequality was due to AM-GM. And the original inequality is proven.
To nd the equality conditions, we just need 3r(1 r) = 1 3r(1 r) (for
the AM-GM step to have equality). It follows that the equality condition is
(r, j, .) =
_
1
2
+
p
3
6
,
1
2

p
3
6
, 0
_
(or some permutation/dilation thereof).
Also solved by Iurie Boreico, student, Chisinau, Moldova, and the author.
Mathematical reflections 2, (2006) 13