Book Homc2019

LIBRARY HOMC
HANOI OPEN MATHEMATICS

COMPETITION 2019
(SCIENTIFIC SEMINAR)
HANOI, 02-06/04/2019
HOMC 2019 - Scientific seminar, Hanoi, April 3, 2019
PREFACE
Founded in 2004 by the Hanoi Mathematical Society, Hanoi Open
Mathematics Competition (HOMC) is a mathematics competition held
every year for school students, and it has two sections: contestants
with 14 and 16 of age (equivalently, those in 8 and 10 grade by the
educa-tion classification in Viet Nam). As the original regular of
HOMC, all questions, problems, and contestants presentation should
be presented in English. Through those fourteen events, HOMC were
attracted par-ticipation of thousands of school students as well as
attention of a great number of high and secondary schools in Viet
Nam. From 2013 to 2016, hosting by the Department of Education and
Training of Ha Noi City, the HOMCs were posited in three cities: Ha
Noi, Dac Lac, and Cao Lanh with the participation of thousands of
students coming from 50 high and secondary schools in Viet Nam.
Taking some interaction between students as a part of competition,
the 2017 HOMC took place of Hanoi city and time of two days with 900
contestants from 30 schools in Viet Nam.
Supporting by the administrative Committee of Hanoi city, HOMC 2018
has been organized with two groups: Group A is the International HOMC
which had eight international teams: China-Zheng Huan, Ghana,
Indonesia, Malaysia, Myanma, Philippines, Poland, Thailand together with
two teams of the Hanoi capital; Group B is the Home HOMC which had
more than forty teams from provinces and schools in Viet Nam.
This proceedings is published and hosted by the Department of Edu-
cation and Training of Ha Noi City, which can be seen as a registered
part among the activities HOMC 2019 by opening and sharing with the
pur-poses of HOMC. Hopefully, every participant of each one in the orga-
nized HOMC’s will find their contribitions, and new comers in HOMC
2019 will take part their new cooperation with pleasure. We would like to
express our special thanks to Prof. Nguyen Minh Tuan, Prof. Nguyen
Huu Dien who gave their valuable suggestions and some proofreading.
1
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Tran Nam Dung, Gabriel Dospinescu, Mixing variables . . . . . . . . . . . . . . . . 4 Nguyen
Thanh Hong, On the counter-example about parallelogram . . . . . 18 Nguyen Van Loi,
The separation of elements into two parts of equal weight23 Nguyen Van Mau, On a
class of functional equations in plane geometry. . 36 Nguyen Van Ngoc, On the
Rearrangement Inequality . . . . . . . . . . . . . . . . . . 41 Dam Van Nhi, Some new
combinatorial identities . . . . . . . . . . . . . . . . . . . . . . 57
Nguyen Van Nho, Menelaus’s theorem and its applications in some Grade 8
problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
th
Vu Tien Viet, Proofs and generalizations of one inequality at the 36 IMO 78
Nguyen Dang Phat, Some International Olympiad problems in geometry 88
Tran Ngoc Nam, Selected problems from Mathematics and Youth

Magazine 2018.................................................................99
Nguyen Ngoc Han, Selected via mail competition questions from Children’s Fun
Maths Journal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Nguyen Hung Son, Invariant in Tiling problems . . . . . . . . . . . . . . . . . . . . . 124
2
Le Quy Thuong, Nguyen Minh Tuan, On the Shapiro’s cyclic inequality and some
relative issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3
MIXING VARIABLES
Tran Nam Dung
Hochiminh City University of Natural Science
and Gabriel Dospinescu
Romania
One of the main properties of most inequalities, especially polynomial
ones is the fact that equality holds when all or some variables are equal.
Mixing variables method is based on this property and its purpose is to
decrease the number of variables, reducing the problem to an easier form,
which can be proved using factorization, AM-GM, one-variable calculus or
by mathemat-ical induction.
1 Some examples
To prove the inequality
f (x1, x2, . . . , xn) 0 (1)
we can try to show that

f (x1, x2, . . . , xn) f ((x1 + x2)/2, (x1 + x2)/2, . . . , xn) (2)
or that p p
f (x1, x2, . . . , xn) f ( x1x2 xx ,...,x ), (3)
1 2 n
And then, if we succeded in proving this, we reduce proving (1) to
proving the inequality
f (x1, x2, x3 . . . , xn) = g(x1, x3, . . . , xn) 0 (4)
which has one less variables. Of ,course, inequalities (2) and (3) can
be totally wrong or can be true only with some additional condition. In
(2) and (3) we change only two variables, so usually it is not hard to
verify them. The following example will show that this technique can
be used both for easy inequalities and hard ones.
4
Example 1 (Kvant). Let a, b, c > 0. Show that we have following inequal-

ity: p
2 2 2 2 2 2 2
2(a + b + c ) + 3 3 a b c (a + b + c)
Solution. Let’s consider the function

p
2 2 2 2 2 2 2
f (a, b, c) = 2(a + b + c ) + 3 3 a b c (a + b + c) .
The expression of f itself suggests us to mix two variables into their ge-p
ometric mean (to keep 3 3 a2b2c2 unchanged). We have
p p 2 2 p 2 2
f (a, b, c) f (a, bc , bc ) = 2(b + c 2bc) + (a + 2 bc ) (a + b + c)
=(
p
b
p
c )2[(
pb
+ c )
2
2a]. p
So, if a = minfa, b, cg (and we can obviously suppose that) then we
have p p
f (a, b, c) f (a, bc , bc ).
Thus, to prove the initial inequality, it is sufficient to show that f (a, b, b)
0 for all a and b. But this inequality is equivalent to
p
2 2 2 4 2
2(a + 2b ) + 3 3 a b (a + 2b) ,
or
p3
4ab, a2 + 3 ab
2 4
2
which is true by AM-GM for a , 3 p3 2 4 , 3 p3 2 4 , 3p3 2 4 .
ab ab ab
The following example is much more difficult:
Example 2 (Vietnamese IMO Team selection Test, 1996). Let a, b, c

be ar-bitrary real numbers. Prove that
4 4 4 4 4 4 4
F(a, b, c) = (a + b) + (b + c) + (c + a) 7(a + b + c ) 0
Solution. We will show that with some additional condition
F(a, b, c) F(a, (b + c)/2, (b + c)/2).

5
Indeed, long but easy computations show that we have

F(a, b, c) F(a, (b + c)/2, (b + c)/2)
4 4 4
4 (a4 + b4 + c4)
= (a + b) + (b + c) + (c + a)
7
4 4 4 4
2(a + (b + c)/2) (b + c) + 4 (a + 2((b + c)/2)
7
4 4 4 4 4 4 4
= (a + b) + (c + a) 2(a + (b + c)/2) + c((b + c) /8 b c )
7 3
3 3 3 2 2 2 2 4 4 4
= a(4b + 4c (b + c) ) + 3a (2b + c (b + c) ) + (b + c (b + c) /8)
7
= 3a(b + c)(b 2 2 2 2 2 2
c)3a (b c) + (3/56)(b c) [7b + 7c + 10bc]
2 2 2 2
= 3a(a + b + c)(b c) + 3 (b c) [7b + 7c + 10bc].
56
The last term is nonnegative. If a, b, c have the same sign, then the
initial inequality is trivial. If a, b, c have different signs then at least one
from a, b, c has the same sign as a + b + c. WLOG, suppose that it is
a. From the last equality we have
F(a, b, c) F(a, (b + c)/2, (b + c)/2).
Thus, it is now sufficient to show that F(a, b, b) 0 for all a, b, or
4 44 4 4
2(a + b) + (2b) 7(a + 2b ) 0.
If b = 0 then inequality is trivial. If b 6= 0, then dividing the two sides of

4
the inequality by b and putting x = a/b, we reduce it to the equivalent
form:
4 4 4
2(x + 1) + 16 7(x + 2) 0
The last one can be proved by one-variable calculus. Consider
4 4 (x 4 + 2 ).
f (x) = 2(x + 1) + 16
7
6
We have
0 0
f (x) = 8(x + 1)
3 16 .f (x) = 0
x3
r
7
2
, x+1=3 7x
, x = 2.9294.
Thus
4 4
fmin = f ( 2.9294) = 2( 1.9294) + 16 (4/7)( 2.9294) 8/7 = 0.4924.
For free of conditions inequalities in general we can mix variables

into whatever we want: arithmetic mean, geometric mean, harmonic
mean, quadratic mean ... and in some cases, we can even define
special means. A beautiful example is the following famous (due to its
difficulty) prob-lem:
Example 3 (Iran Mathematical Olympiad 1996). Let a, b, c be positive

re-als. Prove that:
(ab + bc + ca) 1 1 1 9
2 2 2 4
(a + b) (b + c) (c + a)
Analysis.
We can mix a, b into (a + b)/2, but then both terms in L.S.H will be
changed and the result will be very bulky. Naturally, we want to keep
one term unchanged. For example, we will keep ab + bc + ca
unchanged. For this goal, we mix b and c into t such that
2
ab + bc + ca = at + t + ta.
If we solve this equation (in t) we get
q
t =a + (a + c)(a + b).
Let
2 2 2
f (a, b, c) = 1/(a + b) + 1/(b + c) + 1/(c + a) .
7
Our goal is to show that
f (a, b, c) f (a, t, t).
Investigating this inequality, we see that it holds only with some as-
sumption, for example for a = minfa, b, cg.
Solution. WLOG, we can assume that a = minfa, b, cg. Let
q
t =a + (a + c)(a + b),
g(a, b, c) = ab + bc + ca
and
2 2 2
f (a, b, c) = 1/(a + b) + 1/(b + c) + 1/(c + a) .
Then
g(a, b, c) = g(a, t, t).
We will show that
g(a, b, c) (a, t, t). (5)
Note that
q
a+t= (a + c)(a + b) ,
q t (b + c)/2
(a + c)(a + b)
Also, (5) is equivalent to
2 2 2
(1/(a + b) 1/(a + c)) 1/4t 1/(b + c) ,
which can also be written

2 2 2 2 2
(b c) /(a + b) (a + c) (b + c 2t)(b + c + 2t)/4t (b + c) .
Now,
q
b + c 2t = (a + b) + (a + c) 22 (a + c)(a + b)
p p
=( a+b a + c)
2 p p 2
= (b c) /( a+b+ a+c )
8
and it sufficient to show that

p p
2 2 2 2 2
( a + b + a + c) 4t (b + c) (a + b) (a + c) (b + c + 2t)
p
But this is trivial because (a + c)(a + b) 2t b + c.
Thus, all we have to do now is to prove that
g(a, t, t) f (a, t, t) 9/4.
This can be done by easy computation:
g(a, t, t) f (a, t, t) 9/4

, 2 2 2
4(2at + t )(2/(a + t) + 1/4t ) 9
2 2 2 2 2
,
(2at + t )(8t + (a + t) ) 9t (a + t)
2 2 2 2 2 2
,
(2at + t )(9t + a + 2at) 9t (a + 2at + t )
, 2
at(a t) 0.
Mixing variables can be used not only for free of conditions
inequalities, but also for inequalities with additional conditions. In these
cases, when we mix variables, we have to be sure that the conditions
still hold. So, we have less variants to choose.
Example 4 (Vietnamese Mathematical Olympiad 2002). Prove that for

ar-bitrary real numbers x, y, z, satisfying
2 2 2
x + y + z = 9,
the inequality
2(x + y + z) xyz + 10
holds.
Solution.
Let
f (x, y, z) = 2(x + y + z) xyz.
We have to show that
2 2 2
f (x, y, z) 10 when x + y + z = 9.
9
2 2 2
To leave the condition x + y + z = 9 unchanged, we will mix two
variables into their quadratic mean. Consider
q 2 2 q
2 2
)
f (x, (y + z )/2, (y + z )/2 f (x, y, z)
q 2 2 ) x(y2 + z2)/2
= 2(x + 2 (y + z )/2 2(x + y + z) + xyz
q 2 2 y z) x(y z) /2
2
= 2( (y + z )
2 q 2 2
+ y + z) x/2].
= (y z) [2/( 2(y + z )
If x, y, z > 0, we consider two cases. The first one:
1 x y z.
Then q
2 2 2
p
2(x + y + z) xyz 2 3(x + y + z ) 1=6 3 1<10.
The second one: 0 < x 1. Then q q
2(x + y + z) xyz < 2(x + y + z) 2x + 2 2 2

2(y + z ) = 2x + 2 2(9 x2) = g(x).
g0
g(1) = 10. x p g x
( ) = 2 ( )
x x
We have 2 2 /(9 > 0, so it follows that

)
2
If among numbers x, y, z there is a negative number, WLOG, suppose
that x < 0. Then
q 2 2
q
2 2
) f (x, y, z) 0
f (x, (y + z )/2, (y + z )/2
and we reduce our problem to proving that
q q ) 10
f (x, (y2 + z2)/2, 2 2
(y + z )/2
or q
or 2x + 2 2(9 2 x(9 2
x ) x )/2 10,
3
h(x) = x 5x + 4 q 2(9 2
x ) 20.
10
We have: q
0 2 2
h (x) = 3x 5 4x 2/(9 x ).
0
Solving the equation h (x) = 0 (where x < 0), we get x = 1. It is
maximal point of h, thus
h(x) h( 1) = 20.
2 2
Example 5 (Gabriel Dospinescu). Prove that if a, b, c > 0 verify a + b
2
+ c = 3, then
3 3 3
a (b + c) + b (c + a) + c (a + b) 6.
Solution.
Again, the purpose it to preserve the relation given in the problem. That is why we study the difference
f (a, b, c) f (a, t, t), where t = r b 2 c .
2
2 +
It is easy to compute the difference. It is
3 3 3 3 2 2
a (2t b c) + a(b + c 2t ) + t (2bc 2t ).
Let us take a = minf a, b, cg. Then

3 3 3 2 2 3 2 2 3 2
b +c 2t = (b + c)(b bc + c ) 2t 2t(b bc + c ) 2t = t(b c) .
So, the difference is at most
3 2 2 2
a (2t b c) + at(b c) t (b c) 0,
since 2t b c 0 and a t. Thus, it remains to prove that if

2 2
a + 2t = 3, then f (a, t, t) 6.
This can be written
3 3 4
a t + at + t 3
, 2 4
at(3 t ) 3 t
, 2 2 2 2 4 2
(3 2t )t (3 t ) (3 t )
11
and this one is true since it is equivalent to

2 2 4 2
3(t 1) (t 3t + 3) 0.
Thus, it isn’t always trivial to show the inequality for two numbers. We
will see that there is an easier solution for this problem, thought this is
the most natural solution.
We finish with an inequality proposed for the final round of the ro-
manian mathematical olympiad in 2003:
Example 6 (Mircea Lascu and Marian Tetiva). Prove that if a, b, c > 0

have product 1, then the following inequality holds:
2 2 2
a + b + c + 3 a + b + c + ab + bc + ca.
Solution. Now it’s obvious that we should consider the function:

2 2 2
f (a, b, c) = a + b + c + 3 a b c ab bc ac. It is also
easy to deduce that
p p p p 2 p p 2
f (a, b, c) f (a, bc, bc) = ( b + c) (( b + c) 1 a).
So, we would like to have the last term positive. Naturally, we choose
a = mon fa, b, cg. By AM-GM we have that
p p 2 p 4
( b + c) 4 bc = p a 4 1 + a, since a 1.
Thus, it is enough to prove that
f (a, p , p ) 0
bc bc
and this is in fact
1 12
a2 + p +2 a+2
a pa
and it is obviously true since
p
2
a+2a 2a + 1 a + 2.
12
n n n
(ab) (bc) (ca)
In considered examples, we mix variables into their means. We do

so, because equality holds when these two variables are equal. But for
some inequalities, this property does not hold, and we have to choose
other expression to mix into, rather means. Usually, in these cases we
use boundary values (look at the ends priciple).
Example 7 (Gabriel Dospinescu, Gheorghe Vranceanu Contest, 2003).
Let n be a natural number, n 2. Find the greatest value of the expres-
sion
+ +
1 ab 1 bc 1 ca
when a, b, c are arbitrary nonnegative reals satisfying a + b + c = 1.
Analysis. The investigation of the easiest case (n = 2) shows that the
maximum value is not achieved in the point (1/3, 1/3, 1/3), but in the
point (1/2, 1/2, 0). This fact suggests that we should mix a and b into a
+ b and 0. But in additional condition that c = maxfa, b, cg. When one
variable is equal to 0, the function becomes a simple one-variable
function.
Solution. Let’s assume that a b c. Put
n
f (a, b, c) = (ab) (bc)n (ca)n
+ + .
1 ab 1 bc 1 ca
We will show that f (a, b, c) f (0, a + b, c).
This is equivalent to
n n n n n (6)
(ab) (bc) (ca) (a + b ) c
+ +
1 ab 1 bc 1 ca 1 ac bc
We write consequently
n n n n 1 n n
(a + b ) c (a + nab + b )c
1 ac bc 1 ac bc
n n n n n n
a c a b b c
1 ac bc + 1 ac bc + 1 ac bc
n n n n n n
a b a c b c
+ +
1 ab 1 ac 1 bc .
13
and we get (6).

All we have to do now is to find the maximum value of
n
(bc)
f (0, b, c) = where b + c = 1.
1 bc
This is an easy exercise, since the function
n
h(x) = x /(1 x)
is an increasing function on (0, 1) and
2
bc (b + c) /4 = 1/4,
so max
n 1
f = h(1/4) = 1/(3.4 ).
This is the answer to the problem, since the value is attained for
1
example when a = b = 2 and c = 0.
As it can be easily deduced from the above, mixing variables is most
useful for 3-variable inequalities. The reason is simple: if we can mix
variables for a 3-variable inequality, then we get 2-variable one, which,
in most of cases can easily reduce to 1-variable (for example, a
homogenus 2-variable is equivalent to 1-variable inequality). When the
expression which needs to be evaluated has more than 3 variables,
using this method meets more difficulty, because every mixing does the
expression more complicated. Especially in the case of a general
inequality (with n vari-ables), the difficulties are more than obvious:
mixing variables change the general form and we cannot use induction,
which is the most ap-propriate approach. Mixing variables will be more
effective if we use it together with other arguments, for example, the
properties of continu-ous functions on a compact.
Example 8 (Gabriel Dospinescu). Prove that for positive reals x1, x2, . . .
, xn, whose product is 1 we have
1 + 1 + + 1 x1 + x2 + + xn .
1 + x1 1 + x2 1 + xn 4
14
Analysis. Let
f (x) = x + 1 1 .
4 x+1
We have to show that if
x1.x2 . . . .xn = 1 then f (x1) + f (x2) + + f (xn) 0.
p
The condition x1.x2 . . . .xn = 1 forces us to mix x1, x2 into x1, x2,
p
x1, x2. But after that? We can mix some pairs of variables, but if the
number of variables is odd? More detailed consideration shows that
k
mixing vari-ables brings us to result iff n = 2 . What we can do in other
cases? We can work backward!
If
minf f (x1) + f (x2) + + f (xn)g = f (x1 ) + f (x2 ) + + f (xn),
we will show that
x1 = x2 = = xn = 1,
and we get that min = 0. It is trivial if we can prove that if
p
x1 6= x2 then f (x1 ) + f (x2 ) > 2 f ( x1 x2 ).
But why does this minimum exist? Sometimes, we ignore this fact,
considering it trivial. But it is one of main question in this problem.
Some properties of continuous functions will help us.
Solution. Let
f (x) = x + 1 1 .
4 x+1
We will show that if
x1.x2 . . . xn = 1 then f (x1) + f (x2) + + f (xn) 0.
First, we will show that there exists a minimal value of

S(x1, . . . , xn) = f (x1) + f (x2) + + f (xn)
on the surface
+ n
f(x1, . . . , xn) 2 (R ) jx1.x2 . . . xn = 1g.
15
Note that
f (1) + + f (1) = 0 and 1/(1 + x1) + 1/(1 + x2) + + 1/(1 + xn) < n.
So, if
xi > 3n for some i then S(x1, . . . , xn) > 0.
Similarly, if
n n
xi < 1/(3n) then x1 . . . xn/xi > (3n) ,
and
1/n 1
x1 + + xn xi > (n 1)(x1 . . . xn/xi) > 3n
and
S(x1, . . . , xn) > 0.
Thus, we can find the minimal value in the compact:
n
C = f(x1, . . . , xn)j1/(3n) xi 3 n, x1 . . . x n = 1 g .
S(x1, . . . , xn) is continuous on C, so the first thing is done.

Secondly, let
minf f (x1) + f (x2) + + f (xn)g = f (x1 ) + f (x2 ) + + f (xn).
We will show that

x1 = x2 = = xn = 1.
1 6 2
Indeed, if it is not so, then we can find x = x . We will show that if
p xy ),
6
(7)
x = y then f (x) + f (y) > 2 f (
which leads to a contradiction. But
(7) p p ) 1/(1 + x) 1/(1 + y)
(x + y 2 xy )/4 > 2/(1 + xy
, p p
p ( x p 2 y) > [2(1 + x)(1 + y) (1 + xy )(2 + x + y)]/(1 + xy
)(1 + x)(1
p p p
,( px
p
y)2/4 > (
p
x
2
y) (1 xy)/(1 + xy)(1 + x)(1 + y).
,
which is true. Finally, we conclude that
f (x1) + f (x2) + + f (xn) f (1) + + f (1) = 0.
16
2 Studied Problems
Problem 1 (After Kvant problem). Prove that for arbitrary real numbers
x1, x2, . . . , xn we have
q
2 2 2 2 2 2 2
(n 1)(x1 + x2 + + xn ) + n n x1 x2 . . . xn (x1 + x2 + + xn) .
Problem 2 (PTNK Team selection test 2002). Find all positive integer k such
that inequality p
k k k
2(a + b + c ) + 3 abc 9
holds for all positive reals a, b, c satisfying a + b + c = 3.
Problem 3 (Vasile Cartoaje, Gazeta Matematica A). Let x, y, z be
3 3 3
positive reals satisfying x + y + z = 3. Show that
4 4 4 4 4 4
x y +y z +z x 3.
Problem 4 (Proposed for 2000 Moscow Olympiad). Prove that if x, y, z >
0 have product 1 then:
2 2 2
x +y +z +x+y+z 2(xy + yz + zx).
Problem 5 (1993 IMO Shortlist, proposed by Vietnam). Prove that if a,
b, c, d are nonnegative reals which add up to 1, then
1 + 176abcd
abc + bdc + cda + dab .

Problem 6. Prove that for all a, b, c > 0 we have the following inequality:
2 2 2
a + b + c + 2 + abc a + b + c + ab + bc + ca.
Problem 7. For positive real numbers a, b, c prove that
(a2 + b2)(b2 + c2)(c2 + a2)(ab + bc + ca)2 8a2b2c2(a2 + b2 + c2)2
Problem 8. For positive real numbers a, b, c prove that
3 3 3
a b c 3
3 + 3 + 3 8
(a + b) (b + c) (c + a)
Vietnamese Team selection test 2005
Problem 9. For arbitrary real numbers a, b, c, show that
6 6 6 16 6 6 6
(a + b ) + ( b + c ) + ( c + a ) 63(a + b + c ).
17
ON THE COUNTER-EXAMPLE ABOUT PARALLELOGRAM

Nguyen Thanh Hong, Hanoi University of Education
Combined the properties of a parallelogram, we propose some problems,

which seems useful on proving quadrilaterals are parallelograms. . .
1 Introduction
1.1 Definition
A parallelogram is a quadrilateral with both pairs of opposite sides
parellel.
1.2 Several properties of parallelograms

Theorem 1. If a quadrilateral is a parallelogram, then its opposite
sides are congruent.
Theorem 2. If a quadrilateral is a parallelogram, then its opposite

angles are congruent.
18
Theorem 3. If a quadrilateral is a parallelogram, then its consecutive

angles are supplementary.
Theorem 4. If a quadrilateral is a parallelogram, then its diagonals
bisect each other.
1.3 Proving quadrilaterals are parallelograms

Corresponding to the 4 above theorems, we have the following well-
known ways to prove that a quadrilateral is a parallelogram.
Theorem 5. If both pairs of opposite sides of a quadrilateral are

congruent, then the quadrilateral is a parallelogram
Theorem 6. If both pairs of opposite andgles of a quadrilateral are
congruent, then the quadrilateral is a parallelogram
Theorem 7. If an angle of a quadrilateral is supplementary to both of
its con-secutive angles, then the quadrilateral is a parallelogram.
Theorem 8. If the diagonals of a quadrilateral bisect each other, then
the quadri-lateral is a parallelogram.
2 A counter example about parallelogram?
Combined the properties of a parallelogram, we propose the follow-

ing problem, which seems useful on proving quadrilaterals are parallel-
ograms.
Problem 1. Given a quadrilateral ABCD such that AB = CD and \BAD
= \BCD. Does this quadrilateral must be a parallelogram?
We present below our proof.
19
Proof. Let H, K are respectively the foot of perpendiscular from B, D to

AD, BC. Since \BAD = \BCD, we see that either K, H belong to
segments AD, BC, or A, C belong to segments DH, BK respectively.
The two triangles ABH and CDK are congruent, then DK = HB and
CK = AH.
Consequently, the two triangles BHD and DKB are congruent,
hence BK = DH.
Thus, we obtain BC = AD, combining with the hypothesis AB = CD,
quadrilateral ABCD must be a parallelogram.
Unfortunately, the following construction will show a counter-example
of the above problem.
Given an isoscelec trapzoid ADBE. Let C be the symmetric point of

E across the side BD. Then we have AB = DE (diagonals of an
isoscelec trapzoid) and DE = DC. It shows that AB = CD.
On the other hand, from the symmetry we have \BCD = \BED.
Moreover, \BED = \DAB. Then \BAD = \BCD.
Finally, the quadrilateral ABCD satisfies the condition of the above
problem, but it is not a parallelogram.
20
3 Selected problems
Problem 2 (Iranian Geometry Olympiad 2015). In rectangle ABCD, the

points M, N, P, Q lie on AB, BC, CD, DA respectively such that the
area of triangles AQM, BMN, CNP, DPQ are equal. Prove that
quadrilateral MNPQ is parallelogram.
Problem 3 (Estonian 2005). Let ABCD be a parallelogram, M the mid-

point of AB and N the intersection of CD and the angle bisector
of \ABC. Prove that CM ? BN iff AN is the angle bisector of \DAB.
Problem 4 (JBMO 2001). Let ABCD be a parallelogram centered at O.
Let M and N be the midpoints of BO and CD. Prove that if the triangles
ABC and AMN are similar, then ABCD is a square
Problem 5 (JBMO 2001). Consider the triangle ABC with \A = 90
and \B 6= \C. A circle C(O, R) pass through B and C and intersect the
sides AB and AC in D and E, respectively. Let S be the foot of the
perpen-dicular from A to BC and let K be the intersection point of AS with
the segment DE. If M the midpoint of BC, prove that AKOM is a parallelo-
gram.
Problem 6 (Iranian Geometry Olympiad 2016). Suppose that ABCD is

a convex quadrilateral with no parallel sides. Make a parallelogram on
each two consecutive sides. Shows that among these 4 new points,
there is only one point inside the quadrilateral ABCD.
Problem 7. Consider a parallelogram ABCD such that \ABC = 80
and \ACD = 20 . A line passing through B meets the line AB at an angle
20 and intersects the line AC in the point R. A line passing through C
meets the line AC at an angle of 30 and intersects the line AB in the point
T. Find the measure of the angle determined by the lines TR and DC.
Problem 8. Let ABCD be a parallelogram. The circle c with diameter

AB passes through the midpoint of the side CD and through the point
D. Determine the measure of the angle \ABC.
21
Problem 9. Consider a convex quadrilateral ABCD and let M, N, P, Q

be the midpoints of the sides AB, BC, CD, DA respectively. Prove that
if 2(MN + PQ) = AB + BC + CD + DA, then ABCD is a parallelogram.
Problem 10. Given a parallelogram ABCD. N is the intersection point
of CD and the angle bisector of \ABC. Prove the following assertions
are equivalent
1) BN?AN
2) AB = 2BC
3) AN is angle bisector of \DAB
Problem 11 (Estonian math competitions 2002). Let ABCD be a
rhombus with \DAB = 60 . Let K, L be points on its sides AD and DC
and M a point on the diagonal AC such that KDLM is a parallelogram.
Prove that triangle BKL is quilateral.
Problem 12 (Estonian 2004-final round). Consider a convex quadrilat-
eral ABCD. Choose a point O in the plane and let K, L, M, N be the cir-
cumcenter of triangles AOB, BOC, COD, and DOA, respectively.
Prove that there exists exactly one point O in the plane such that
KLMN is a parallelogram.
References
[1] D. Branzei, L. Serdean, V. Serdean Junior Bankal Mathematical
Olympiads, Plus Publishing House, 2003, 146 pages.
[2] R. Geretschlage, J. Kalinowski, J. Svrcek, A Central European
Olympiad, the Mathematical Duel, vol.7, World Scientific Publishing,
2018, 281 pages
[3] website: Art of Problems Solving.
22
THE SEPARATION OF ELEMENTS INTO TWO PARTS OF EQUAL

WEIGHT
Nguyen Van Loi, Budapest - Hanoi
When dealing with sets where elements are attached distinct weights, the
question naturally arises: what options are there to divide the weights?
This question also has great relevance to everyday life. In this article, I will
con-tinue the research we began with Peter Erben [4] and then continued
with N. H. Dang and D. Th. Khang [3], placing the emphasis on a clearer
presen-tation, simpler forms of proof, as well as potential utilization.
1 The structure of elements with weight and its effect

on the distribution of weight
Basic knowledge can be found in [2] and [5].
Let A be a set with elements e1, e2, , ek, k is the number of elements. Each
element is assigned ei. Accordingly, the weight assigned to ei is denoted
s(ei), leaving no room for misunderstanding. To make calculations eas-ier,
we list the elements (weights) in a monotonous order.
In this article all weights are positive integers and all sets are finite. For ele-
ments s1, s2, , st from A, the sum s1 + s2 + + st is the sum of s(s1) + s(s2) +
+ s(st).
Given is n positive integer and set A. Set A can be expressed up to the
th
weights n if for any m n there is e1, e2, , et 2 A so that m = s1 + s2 + +
st, where si is the weight of ei.
If S is the sum of the elements in set A and set A can be expressed until
S, we can say that S is fully expressible. Note, if a set is expressible until
a point (> 0), then it obviously contains the weight unit. Thus, a set which
is fully expressible can be regarded as a measuring system in which we
can measure any weight between 1 and S.
Theorem 1. Given are set A and N weight. If the weight of elements in A are not
more than N and set A is expressible up to N, then set A is fully expressible.
23
Proof. Let A = fa1; a2; ; ak g be a set of the weights (here the elements
are taken as identical to their weights), a is denoted as a = a 1 + a2 + .... +
ak. By induction according to N, we shall prove the Theorem.
If N = 1, then all elements in set A weight one unit, thus the statement
is trivial.
Assuming the statement is true up to N, we need to show that the
state-ment is true to N + 1. If A can be expressed up to N and there
are no elements with the weight of N + 1, then by the rules of induction
the statement is true. Thus, we can assume that set A has a few
elements with the weight (N + 1). Let
N + 1 = a1 = a2 = .... = ai > ai+1 ai+2 .... ak
and b = ai+1 + ai+2 + .... + ak. It is obvious that if we are looking to ex-
press a weight s with elements from set A, then these elements cannot
contain one with the weight (N + 1). It follows that B = fa i+1; ai+2; ; akg
is a set which is expressible up to N and the total weight of its
elements is no greater than N. Let us take an arbitrary 0 < s a. If s b,
then s can be expressed with weights contained in B. If b < s a, then
there exist r, q 2 N so that s b = q(N + 1) r where 0 r < N.
If q i + 1 ) s (i + 1)(N + 1) + b r = a + (N + 1) r > a, which
would be contradictory. Thus, 0 q i, which means b r can be expressed
with q pieces of (N + 1) weights. At the same time, r can be expressed
with elements from B. These two sets are disjoint, thus s is expressible
with elements from A. Thus, A is fully expressible.
If A and B are two fully expressible sets, then the element with the
largest weight can either be found in A or B, thus the theorem is proved.
Corollary 1. If A and B are two fully expressible subsets of set S, then

C = A [ B is also fully expressible.
Let A be a set which contains the unit weight, thus A is expressible at
least up to 1. Let us denote P(A) as a fully expressible subset of set A.
Then P(A) is fully expressible and is the largest among the fully
expressible subsets. From this we can conclude:
24
Corollary 2. Let A be a set that contains the unit element. Let P(A) be
the largest fully expressible subset of A and the sum of elements of
P(A) is h. If the weight of any element in A is no larger than h, then
either A = P(A) or a h + 2, provided there is an a 2 A P(A).
Proof. If there is an s weight where s 2 A and s = h + 1, then P(A) [ fsg
is also fully expressible, which would contradict the statement that H is
the largest such subset.
In the following, let us denote A(M, k) a set where the total weight is M
and has k elements.
M M
Proposition 1. Let A(M, k); k 4 + 2 and ai 2. Then the average of
the weights is smaller than 4. If A contains a maximum of 3 elements
with weight less than 4, then
a) a1 = a2 = a3 = 1, a4 = a5 = . = ak 1 = 4 and ak = 5; or
b) a1 = a2 = 1, a3 = 2, a4 = a5 = . = ak = 4.
In these cases A(M, k) is fully expressible.
Note: Since M > k, it is easy to prove that M 3.
M M
Proof. M < 4( 4 + 2) 4k, then k < 4, which means the average of
the weights of its elements is smaller than 4.
On the other hand, let us assume that there are at most 3 elements in
A with weight less than 4. In this case, A has at least k 3 elements with
M
weight over 3. Thus, M a1 + a2 + a3 + 4(k 3) 3 + 4( 4 + 2 3) =
M 1. Since a1 a2 a3 ak (the weights are in ascending order), it
follows that
Either a1 = a2 = a3 = 1, a4 = a5 = . = ak 1 = 4 and ak = 5
Or a1 = a2 = 1, a3 = 2, a4 = a5 = . = ak = 4.
It follows from Theorem 1 that A(M,k) is fully expressible.
N
Theorem 2. Given A(2N, k), s(ai) = ai N and k 2 + 2. Let us denote
the largest fully expressible subset of A as P(A). If P(A) is expressible
up to at least 3, then:
25
N
1) Let S be the total weight of P(A), then S 2;
2) A is either fully expressible or A has exactly two elements with weight
N
no less than S + 2, thus no less than ( 2 + 2), and set A can be
separated into two subsets of equal weight (meaning their weight is N).
Proof. 1) Let S be the weight of P(A) (obviously S is a positive
integer). If A = P(A), then A is fully expressible and so A can be
separated into two parts of equal weight.
Let us assume that A is not fully expressible.
a) First we need to see that N > 4. Due to Proposition 1, set A has at
least 4 elements with weight less than 4. As P(A) can be expressed at
least up to 3, P(A) contains every element with weight no larger than 4.
Thus, if 2N 8, then the weight of the elements in A can be no larger than
4, which means they are in P(A). So P(A) = A, which is a contradiction of
the statement that A(M, k) is not fully expressible. Thus, N > 4.
b) If N > 4, let B = A P(A) = fb 1; b2; ; bpg. We can identify the
elements of B with their weights without misunderstanding, that is
b j = s(bj). Then 2N = S + b1 + b2 + + bp and p > 0. According to Corollary 2, bj S + 2 for
every 0 < j p. If the number of the elements in set P (A) is q, then k = q + p and S q. Then:
2N = S + b1 + b2 + .... + bp S + p(S + 2) = S + (k q)(S + 2)

Since S q ) 2N S + (k 2 1)S + 2k
S)(S + 2) = S + (k
2
=) S (k 1)S 2k + 2N > 0. Looking at the quadratic polynomial:
2 (t + 1)x + 2t 4
f (x) = x
2 4= 2<0
f (2) = 2 (t + 1) 2 + 2t
2 (t + 1)(t 1) + 2t 4 = 2 < 0
f (t 1) = (t 1)
This means that on the [2, t 1] interval f (x) < 0. Since S > 3 and
f (S) > 0, thus S t and T = N )S N .
2 2
2) Obviously if set A is fully expressible then it can be separated into two
subsets of equal weight. Thus we only need to look at the case when set
26
A is not fully expressible. It follows that A 6= P(A). In this case, due to

N
S 2,
N N N
2NS + p(S + 2) 2 + p( 2 + 2) = 2 (p + 1) + 2p
where p is a positive integer. From this we can find that 4N 4p

N (p + 1) ) 3 4p > p.
N
As p is an integer, p 2.
a) If p = 1 ) b1 = 2N S 2N (b1 2) ) b1 N + 1 it would
contradict the fact that the weight of the elements of A is no larger
than N. Thus, p 6= 1.
N
b) If p = 2, then b1; b2 S + 2 2 + 2. Furthermore
N b1 N ( N + 2) < N 2 S
2 2
Since P(A) is fully expressible, because of N S, N b1 can be
b1
2 2
expressed with the elements of P(A). Thus there exist h1, h2, , hl so that
N b1 = h1 + h2 + + hl meaning N = b1 + h1 + h2 + + hl.
This means that set A can be separated into two subsets of weight N.
N
Corollary 3. Given A(2N, k) where N 2, s(a i) = ai N and k 2 + 2, then
set A can be separated into two subsets the difference between the
weights of which is 0 or 2.
Proof. Let us add two elements of unit weight to set A and call the
N
new set A. The total weight of set A is 2N + 2 and has k + 2 2+2+
N+1
2> + 2 elements, and the weight of its elements is smaller than

(N + 1). Theorem 2 can be applied to prove that A can be separated into
two subsets of equal weight. Taking out the two elements we added, we
can prove the statement.
Examples
27
1) 1) Let A = f1, 1, 1, 6, 6g, then the weight of A is 16. The number

8
of elements is 6 = 2 + 2. A cannot be fully expressed (as 5 is
not expressible) and 1 + 1 + 6 = 1 + 1 + 6.
Counter-examples include instances when the criterion of
express-ibility up to 3 is not satisfied.
2) Let N be an odd number. K = N then A = f 2; 2; 2; ; 2 g. The total
N pieces of 2
weight of set A is 2N. In this case set A is not expressible.
| {z }
3) When N = 6m + 2(2N = 12m + 4), k = 4m + 2, let A = f1; 3; 3; ....; 3 g.
|
4m 1{z }
+ pieces of 3
In this case set A can only be separated into a subset the weight of
which is divisible by 3 and another which is not.
4) When N = 6m + 4(2N = 12m + 8), k = 4m + 3, let A = f2; 3; 3; ...; 3 g.
4m+2 pieces of 3
weight of
In this case set A can only be separated into a subset the | {z }
which is divisible by 3 and another which is not.
5) When N = 6m(2N = 12m), k = 3m + 1, let A = f 2; 2; ...; 2 ; 3; 6m
|
3m 1 {z }
pieces of 2
1g. In this case A cannot be separated into two equal parts either.
6) If N is odd, then A = f 2; 2; ...; 2 g. A cannot be separated into two
| {z }
n pieces of 2
equal parts.
From the above examples it can be seen that expressibility up to 3 is
very important.
N
Proposition 2. Given A(2N, k) where N 2, s(ai) = ai N and k 2 + 2.
Furthermore, let a1 = 2. Then, the sum of the weights of any two
elements in set A are not larger than N.
Proof. Let us order the elements in ascending order according to their
weight. This way it is enough to prove that ak 1 + ak N, which can be
28
seen through:
a1 + a2 + .... + ak 2 2(k 2) 2 N =N
2
From this it follows that ak 1 + ak N.
Theorem 3. Given A(2N, k) where N is even, s(ai) = aiN and k
N
2 + 2. Furthermore, let a1 = a2 = 2. Then, set A can be separated into two
subsets of equal weight.
Proof. Let us order the elements in two subsets, with the first being C =
fc1; c2; ; cu g, containing elements the weights of which are even numbers,
while the other is L = fl1; l2; ; lvg, containing odd weights. This way
A = C [ L and C \ L, u + v = k. Since 2N is an even number, v also
needs to be even. Let v = 2t and consider the (u + t) set.
B = fb1; b2; ; b(u + t)g where the bj elements are composed in the
follow-ing way:
c i
bi = 2 where i = 1, 2, , u, and
lj lv j
bu+j = where j = 1, 2, , t.
Due to Proposition 1, a3 a4 3. The two first elements of set B are 1, 1

and its third element is not larger than 3. It follows that B can be
expressed up to 3.
On the other hand, due to Proposition 2, the weight of elements in B cannot
N
be larger than 2 , and the total weight of B cannot be larger than
N either.
N
It can be proved that u + t 4 + 2 in the following way:
u+t=2+u 2 +v 2 + u + v 2 +u 2
2 2 2
N1 u 2 N u 2
2+ 2+ 2 2 =2+ 4+ 2
N u 2 N
From this: u + t 2 + 4 + 2 2 + 4.
29
All criteria of the Theorem 2: are satisfied, thus B can be separated

into two subsets of equal weight. Restoring the original values, we will
find proof of the separability of set A.
2 Applications
This section will focus on the quantity of elements in a set, looking into
how the process of separation can be made easier by increasing the
quantity of elements.
Proposition 3 (In case N divided by 6 yields a remainder of 2). Let N =

6m + 2 (m is a positive integer). Given A(2N, k) where s(a i) = ai N and
k 4m + 3, then set A can be separated into two subsets of equal
weight and the value of k cannot be reduced.
Proof. Since k = 4m + 33m + 3 = N + 2, due to Proposition
2
1, a33. If a3 = 3 ) a3 + a4 + .... + ak3(k 2) = 12m + 3 )
a
1 + a2 1, we get a contradiction. From that we can conclude that
a3 2 ) a1 a2 2. Applying Theorem 2 and 3, we can show that set
A can in fact be separated into two subsets of equal weight. Example
3 shows that the value of k = 4k + 3 cannot be reduced.
Proposition 4 (In case N divided by 6 yields a remainder of 4). Let N =

6m = 4 (m is a positive integer). Given A(2N, k) where s(a i) = ai N and
k 4m + 4, then set A can be separated into two subsets of equal
weight and the value of k cannot be reduced.
N
Proof. Since k = 4m + 4 3m + 4 = 2 + 2, due to Proposition 1, a 3 a4
3.
If a3 = 3 ) a3 + a4 + .... + ak 3(k 2) = 12m + 6 ) a1 + a2 2, then

a1 = a2 = 1.
Applying Theorem 2 and 3 (when a3 2), we can show that set A can be
separated into two subsets of equal weight. Example 4 shows that the
value of k = 4k + 4 cannot be reduced.
30
Proposition 5 (In case N is divisible by 6). Let N = 6m (m is a positive

integer). Given A(2N, k) where s(a i) = ai N. Es k 3m + 2, then set A
can be separated into two subsets of equal weight and the value of k
cannot be reduced.
6 m N
Note: k = 3m + 2 = 2 + 2 = 2 + 2.
Proof. Due to Proposition 1, a1 a2 a3 a4 3. There are the following
scenarios:
a) If A begins with (1, 1) or (1, 2) values, then A(2N, k) can be ex-
pressed up to 3, in which case Theorem 2 can be used.
b) If it begins with (2, 2) values, then Theorem 3 can be used.
c) There are two cases left when it begins either with (1, 3) or (2, 3).
In both cases a2 = a3 = a4 = 3 is true because of Proposition 1. In
this case we can classify the elements based on their weight.
C = fe j es weight is divisible by 3g = fc1; c2; ; cug
K1 = fa j as weight yields a remainder of 1 when divided by 3g =
fk1; k2; ; kvg
K2 = fa j as weight yields a remainder of 2 when divided by 3g =
fh1; h2; ; hwg
This way A = C [ K1 [ K2 and C \ K1 = C \ K1 = K1 \ K2 = ˘ (the element a
belongs in C or in K 1 or K2). Here k = u + v + w = 3m + 2. From K 1 and
K2 we can create a fourth set B so that b i = (ki + hi) until we ran out of
either v or w. For instance, in case v is larger, then we can take one
element each from K1 and K2 until we ran out of K2 elements, then we
take three elements from K 1 at the same time to create a new element.
Since 2N = 12m (divisible by 3), the task can be done without any
elements left over. Thus, we get sets C and B, the elements of which can
be divided by 3, and C contains at least 3 pieces of 3. The number of
vw
elements in set B is w + (a non-negative integer).
Let 3
e
E = f3 j e either from C or Bg
E is a set of elements with integers for weight. Leg the number of ele-
ments in set E be r, then r = u + w + v w = u + v + 2v +v +2w =
3 3 3 3 3
31
u + v +2w .
3 3 u
Since u has at least 3 elements, u 2 + . We get:
3
r = u + v +2w 2 +u +v +w 2 +3m = 2 + m
3 3 3 3 3 3
At the same time the weight of E is exactly 1 of the weight of A, thus
3
s(E) = 4m.
We need to show that the elements of E are not heavier than 2m. For this
it is enough to prove that ak 2 + ak 1 + ak 6m + 2 , ak + ak + ak
2 1
3
2
2m + 3.
Due to the construction, the elements are integers, so their weight is
less than 2m.
Truly: a1 + a2 + .... + ak 3 1 + 3(k 4) = 1 + 3(3m 2) = 9m 5.
All conditions of Theorem 2 are met. Thus, E can be separated into
two subsets of equal weight. We restore the original weights and
disassem-ble the elements from sets K1 and K2 then set A can be
separated into two subsets of equal weight, proving the Theorem.
Example 5 shows that the value of k = 3m + 2 cannot be reduced.
N
Note: This explains why we began with k 2+2 .
Theorem 4. Let N be a positive odd integer. Given A(2N, N + 1) where

s(ai) = ai N, then set A is fully expressible.
Proof. Apply Theorem 2 and 1, utilising the fact that the number of
elements is rather big (the average weight is smaller than 2). Let the
detailed proof be homework.
Summary:
Theorem 5. Given A(2N, k) where s(ai) = ai N. Set A can be separated

into two subsets of equal weight if:
1) K = N + 1 if N is odd;
32
2) K = 4m + 3 if N = 6m + 2;
3) K = 4m + 4 if N = 6m + 4;
4) K = 3m + 2 if N = 6m.
Lastly, let us mention the problem that began this whole paper:
Problem. There are 35 positive integers, none larger than 50. Their
sum is 100. Can they be separated into two different groups with equal
sums. Answer. Indeed they can, this is a case where N = 50 and k = 4
8 + 3 = 35.
3 Exercises
I found the following exercises a very good opportunity to craft prob-lems

for mathematics competitions as well as bases for further research.
1) There are 100 objects with weights expressed as positive integers,

with the total weight of 198. If n is an arbitrary number not larger
than 100, can a few objects be picked so that their total weight is n?
2) Given n positive integers, prove that it is always possible to pick a

few (in some cases even one) so that their sum is divisible by n.
3) Prove that if you cannot pick a few numbers from (n 1) positive

integers so that their sum is divisible by n, then these numbers all
yield the same remainder when divided by n.
4) Given A(2N, k), s(ai) = ai N and k N + 2. Denote P(A) as
2 2
the largest fully expressible subset of set A. Prove that if P(A) is
expressible up to 3, then A is fully expressible!
5) Given A(2N, k) where N2, s(ai) = aiN and k N + 2.
2
Furthermore, let a2 = 2. Prove that set A can be separated into two
subsets of equal weight.
33
6) There are 20 objects the weights of which can be expressed as pos-

itive integers that are not larger than 28. Their total weight is 56.
Prove that they can be separated into two groups of equal weight.
7) There are 30 objects with the weights being positive integers not
larger than 36 kg. Their total weight is 80 kg. Prove that they can
be separated into two groups of equal weight.
8) There are 20 objects with the weights being positive integers not
larger than 36. Their total weight is 72. Can they be separated into
two groups of equal weight?
9) There are 30 objects the weights of which are positive integers
but do not weigh more than 24. Their total weight is 72. Prove that
they can be arranged into three groups of equal total weight!
10) There are 24 objects that weigh positive integers but do not weigh
more than 24 kg. They weigh 72 in total. Can they be separated
into three groups of equal weight?
4 Afterword
When in 2014 I could not find a detailed proof for this problem, I sub-
mitted it to Komal Forum [4] (Problem 3906 07/2014) to help me find a
source for the solution. However, it was not meant to be. Later on, me
and Erben Peter worked out a solution for the problem through
painstaking work through the Forum.
I also submitted the problem to the Bai Toan Hay Loi Giai Dep on-line
mathematics forum [1], where it garnered great interest. The most
tenacious colleagues working on it (they were only students at the
time) were Nguyen Hai Dang and Nguyen Thanh Khang. We have
already published our joint work in the publication of the Ha Noi
Mathematics Association [3] (2015).
That publication concentrated on the quantitative summary of our re-
search results to provide direct assistance to students who are either
looking to continue our research or are working on similar topics.
34
The publication in your hands contains results that have been

reworked and proofs that are simpler and more compact.
I would like to seize this opportunity to once again thank my
colleagues for their co-operation.
References
[1] Online Forum Bai Toan Hay Loi Giai Dep .... (2014)
[2] Erds Pl, Abraham Ginzburg , Abraham Ziv Theorem in additive num-
ber Theory. Bull. Research Council, Israel, 10F; 41-43; 1961.
[3] HMS Proceeding Lai Chau Seminar, (2015), 73 88
[4] Komal Forum 3906 problema (2014) jul, Online Forum.
[5] Lovsz L., Pelikn J., Vesztergombi K Discrete mathematics.
Elementary and beyond, Springer-Verlag, 2003.
Ha Noi, 20/3/2019.
N. V. Loi
Email: loiscenter@gmail.com.
35
ON A CLASS OF FUNCTIONAL EQUATIONS IN PLANE GEOMETRY

Nguyen Van Mau
Hanoi University of Science
Let G be the set of all triples of positive numbers (A, B, C) such that
A + B + C = p,
i.e. every triple (A, B, C) 2 G is the angles of a triangle, and let F be the
set of all triples of positive numbers (a, b, c) satisfying
jb cj < a < b + c,
i.e. every triple (a, b, c) 2 F is the side lengths of a triangle.

Our main purpose is
- Determine all function f : (0, p) ! (0, p) such that ( f (A), f (B), f (C)) 2 G for
all (A, B, C) 2 G.
- Determine all function f : (0, ¥) ! (0, ¥) such that ( f (a), f (b), f (c)) 2 F
for all (a, b, c) 2 F.
1 Introduction
In the sequel, let G be the set of all triples of positive numbers (A, B, C)
such that A + B + C = p, i.e. every triple (A, B, C) 2 G is the angles of a
triangle, and denote by G0 the set of all triples of non-negative numbers
( A, B, C) such that A + B + C = p.
Let F be the set of all triples of positive numbers (a, b, c) such that
jb cj < a < b + c,
i.e. every triple (a, b, c) 2 F is the side lengths of a triangle.

The main purpose of the paper is to find the general solutions of the
following functional equations.
36
Question 1. Determine all functions f : (0, p) ! (0, p) such that ( f (A), f (B), f (C)) 2
Gfor all (A, B, C) 2 G.
Question 2. Determine all functions f : (0, ¥) ! (0, ¥) such that ( f (a), f (b), f (c)) 2
F for all (a, b, c) 2 F.
2 General solution of functional equations involving

an-gles of triangles
In this section, we consider Problem 1. We start by finding all differ-
entiable solutions.
Proposition 1. Suppose that f : (0, p) ! (0, p) is differentiable and that
f 1 2 2
satisfies ( f (A), f (B), f (C)) G for all (A, B, C) G. Then there exists
a 2 [ 2 , 1] such that
p(1 a)
f (x) = ax + .
3
Proof. We have
f (x) + f (y) + f (p x y) = p, 8x, y, x + y 2 (0, p). (1)
Taking derivative in x we obtain

0 0
f (x) f (p x y) = 0, 8x, y, x + y 2 (0, p). (2)
0
It follows that f is constant in (0, p), hence f (x) = ax + b for some constants
p(1 a)
a and b. Since f (x) + f (y) + f (p x y) = p we deduce that b = .
3
Since f only takes values in (0, p) we must have a 2 [ 1
2 , 1] .
We now show that the Proposition still holds without the

differentiability requirement.
Theorem 9. Suppose that f : ( 0, p) ! ( 0,p) satisfies ( f ( A ), f ( B ), f ( C )) 2
1
G for all (A, B, C) 2 G. Then there exists a 2 [ 2 , 1] such that

f (x) = ax + p(1 a) .
3
37
p 2 p p 2 p
Proof. Let g : ( 3 , 3 )!( 3 , 3 ) be defined by
g(x) = f (x + p p
)
3 3.
By (1),
g(x) + g(y) + g(z) = 0 if x + y + z = 0.
We deduce that g(0) = 0 and hence,
p 2 p p
g(x) + g(y) = g(x + y) for x, y 2 ( 3 , 3 ), x + y < 3
Since, g is also bounded from below, it follows that g must be linear on

p 2 p
( 3 , 3 ) (see [1]-[3]). Therefore, g(x) = ax for some a and hence,
p(1 a)
f (x) = ax + .
Since f only takes values in (0, p) we must have a 2 [ 1

2 , 1].
3 General solution of functional equations involving

side lengths of triangles
In this section we turn our attention to Problem 2. We first have

some easy observations.
f (x) = ax + b satisfies ( f (a), f (b), f (c)) 2 F for all (a, b, c) 2 F if

and only if a, b > 0 and a + b > 0.
1
f (x) = ax + b satisfies ( f (a), f (b), f (c)) 2 F for all (a, b, c) 2 F if
and only if a = 0, b > 0.
Let O be a cirle with diameter length 1, and M be the set of all tri-
angles inscribed in the cirle O. We next consider the case f maps M to
M.
38
Theorem 10. Any function f : (0, 1) ! (0, 1) such that ( f (a), f (b),
f (c)) 2 M for all (a, b, c) 2 M is of the form
f (x) = sin a arcsin x + p(1 a) (3)
,
3
1
for some a 2 [ 2
, 1].
Proof. By the sine theorem, a, b and g are the angles of a triangle if
and only if sin a, sin b, and sin g are the side-lengths of a triangle in M.
Appyling Theorem 15 to the function
g(x) = arcsin( f (arcsin x)),
we obtain the conclusion.
We now formulate our main result.
Theorem 11. Any function f : (0, ¥) ! (0, ¥) satisfying ( f (a), f (b),
f (c)) 2 F for all (a, b, c) 2 F is of the form
( 1 a)p 1
f (x) = u sin a arcsin q(x) + 3 , 2 6 a 6 1, u > 0, (4)
where q(x) is difined as follows; for every given triple (a, b, c) 2 F with the
x
circumscribed radius R then of q(x) = 2 R. Proof. Applying the above
additional discussion and theorem 10, it is easy to obtain the form
(4).
Remark 1. Some other types of functional equations in geometry were
considered by S. Galab [4].
References
[1] T. Acze’l, Lectures on functional equations and their applications,
Aca-demic Press, New York/San Francisco/London, m1966.
[2] M. Kuczma, B. Choczewski, R. Ger, Interative Functional Equa-
tions, Cambridge University Press, Cambridge/New York/Port
Chester/Melbourne/Sydney, 1990.
39
[3]P.K. Sahoo, T. Riedel, Mean Value Theorems and Functional Equations,

World Scientific, Singapore/New Jersey/London/HongKong, 1998.
[4] S. Galab, Functional equations in geometry, Prace Mat.,

NoCCXXIII, Zeszyt 14, 1969.
40
ON THE REARRANGEMENT INEQUALITY

Nguyen Van Ngoc
Thang Long University, Hanoi
Rearrangement inequality or permutation inequality is an elementary in-

equality and can be used to prove many famous inequalities. In this note
we reveal a nice result that provides a very simple but powerful inequality
that can be used for proving many inequalities.
1 Triples of real numbers

1.1 Difinition
Definition 1. Let’s consider two triples (a1, a2, a3) and (b1, b2, b2) of
real num-bers. The triples (a1, a2, a3) and (b1, b2, b2) are said to be:
similarly arranged if both are increasing (i. e. a1 a2 a3 and b1 b2 b3)
or both are decreasing (i. e. a1 a2 a3 and b1 b2 b3).
oppositely arranged if one is increasing and the other is decreasing.
Let triples (a1, a2, a3) and (b1, b2, b2) be similarly arranged or both be oppo-
sitely arranged, and (c1, c2, c3) be any permutation of numbers b1, b2, b3.
(i) S = a1b1 + a2b2 + a3b3 is called the Sorted sum of the numbers. (ii)
R = a1b3 + a2b2 + a3b1 is called the Reversed sum of the numbers. (iii)
P = a1c1 + a2c2 + a3c3 is called the Permutated sum of the numbers.
Example 9.
1. (1, 3, 5) and (2, 4, 6) are similarly arranged while (1, 3, 5) and (6, 4, 2)
are oppositely arranged.
1 1 1
2. If 0 < a b c, then (a, b, c) and ( a , b , c ) are oppositely arranged,
1 1 1
while (a, b, c) and (b+ c , c+ a , a+ b ) are similarly arranged.
3. If 0 < abc and m is a positive real number, then (a, b, c) and
m m m 1 1 1
(a , b , c ) are similarly arranged while (a, b, c) and (a m , b m , c m )
are oppo-sitely arranged.
41
4. If 0 < a b c, then (a, b, c) and (log a, log b, log c) are similarly ar-
ranged.
1.2 The Rearrangement Inequality

Theorem 12.
SPR. (5)
Proof. Let’s take two triples (a 1, a2, a3) and (b1, b2, b3) increasingly ar-
ranged and let (c1, c2, c3) be a permutation of (b1, b2, b3). Assume that
c1 c2.
Let Q = a1c1 + a2c2 + a3c3 and Q0 = a1c2 + a2c1 + a3c3. We have
Q
0
Q = a1c2 + a2c1 ac ac
1 1 2 2
= a2(c1 c2) a1(c1 c2)

= (c1 c2)(a2 a1) 0.
0
Hence Q Q. That is, interchanging c1 and c2 can only increase the
value of the sum Q. Therefore, if we interchange all pairs (c i, cj) so that
ci cj for i < j, the sum can only get larger. The lagest sum is that one
that corresponds to (b1, b2, b3), that is S = a1b1 + a2b2 + a3b3.
The above argument works similarly if (a1, a2, a3) and (b1, b2, b3)
are both decreasing.
The equality (5) occurs if and only if b1 = b2 = b3 or a1 = a2 = a3.
1.3 Problems
Problem 1. Let a, b, c be real numbers. Prove the inequality

5 5 5 4 4 4
a +b +c a b + b c + c a.
Solution. Without loss of generality we may assume that a b c, and

4 4 4
clearly that a b c . Now by the rearrangement inequality we get the
required inequality. Equality occurs iff a=b=c.
42
Problem 2. Let a, b, c, 2 R. Then
2 2 2
(i) a + b + c ab + bc + ca,
n n n n 1 n 1 n 1
(ii) a + b + c a b+b c+c a for every even positive integer n.
Solution. (i) is a particular case of (ii).Let’s show (ii). Assume a b c.
n 1 n 1 n 1
Since the triplets (a, b, c) and (a , b , c ) are similarly arranged
by using (5) we get
n 1 n 1 n 1 n 1 n 1 n 1
aa + bb + cc ab + bc + ca ,
which is just (ii).
Problem 3. If a, b, c >, 0 then
(i) a + b + c 1 1 1,
abc a
2+ b2 + c2
(ii) a2 b2 c2 b c a
+ + + + ,
2 2 2
b c a a b c
2 2 2
(iii) a b c a + b + c.
+ +
b c a
Solution. We may assume that a a b.
1 1 1 1 1 1
,
(i) The triplets , , and , are similarly arranged. Thus
a b c a b c
1
1 1 + 1 +1 1 11 + 1 1 + 1 1 ,
aa bb cc ab b c c a
that is (i). a b c a b c
(ii) The triplets ,
b c ,a and b , c , a are similarly arranged. Hence
a a +b b +c c a b + b c +c a ,
bb cc aa bc c a a b
which is (ii). 1 1 1
2 2 2
(iii) The triplets (a , b , c ) and a , b , c are opposetely arranged. Hence
2 1 2 1 2 1 1 1 1
a a+b b+c c a2 b + b2 c + c2 a,
that is, just (iii).

43
Problem 4. Let a, b, c be positive real numbers. Prove the inequality

2 2 2 2 2 2
a + c + b + a + c + b 2(a + b + c).
b c a
Solution. Since the given inequality is symmetic, without loss of
gener-ality we may assume that a b c. Then clearly
2
2 2
c and 1
a +b 1 1 .
a b c
By the Rearrangement inequality we have
2 2 2 2 2 2 2 2 2
a + b + c = a 1 + b 1 + c 1 a 1 + b 1 + c 1 = a + b + c, (6)
b c a b c a a b c
and
2 2 2 2 2 2 2 2 2
a + b + c = a 1 + b 1 + c 1 a 1 + b 1 + c 1 = a + b + c. (7)
c a b c a b a b c
Adding (1) and (2) yields the required inequality.
Problem 5. (Neshbit’s inequality). If a, b, c > 0, then
a + b + c 3 .
b+c c+a a+b 2
Solution. The triplets (a, b, c) and ( 1 , ,1 1 ) are similarly arranged
b+c c+a a+b
(we may assume a b c). Thus
1 1 1 1 1 1
ab + c + b c + a + c a + b ac + a + b a + b + c b + c ,
and also
1 1 1 1
ab + c+b c + a+c a + ba a + b+b b
1 1
+ c+c c + a. Adding these two inequalities we fine
a b c a b c
2 b + c+ c + a+ a + bc+ a+ a + b+ b + c
a b c
+ a+ b+ b + c+ c + a = 3.
Therefore, the required inequality is oblained. Equality holds if and
only if a = b = c.
44
Problem 6. (Neshbit’s generalized Inequality). If a, b, c > 0, and n is a

positive integer, then
n n n n 1 n 1
a b c an 1 +b +c
+ + .
b+c c+a a+b 2
n n n
Solution. Assume that a b c. Then tripltes (a , b , c ) and
1 1 1
b + c , c + a , a + b , are similarly arranged. Hence,
1 + bn 1 + cn 1 a
n 1 + bn 1 + cn 1
an b + c+a a+b c+a a+b b+c
c
n 1 n n n 1 + b n 1 + cn 1 .
a +b 1 +c 1 a
b+c c+a a+b a+b b+c c+a
Adding these two inequalities we fine
n n n n n n
an bn cn a +c b +a c +b
2 b+c+ c+ + a+b c+a
+
a+b
+
b+c . (8)
a
We have
n n
a +b 1 (an 1 + bn 1) , (a b)(an 1 bn 1) 0, 8a, b > 0, n 1.
a+b 2
Therefore the right side of (8) becomes
n n n n n n
a +c b +a c +b
+ + n 1 n 1 n 1 (9)
c+a a+b b+c a +b +c .
From (8) and (9) follows the our result. Equality holds if and only if
a = b = c.
Problem 7. Suppose a, b, c are the lengths of the sides of a triangle. Prove that
a + b + c 3.
b+c a c+a b a+b c
Solution. Suppose that the triplet (a, b, c) is increasing. Then the triplet
1 1 1
b + c a , c + a b , a + b c is increasing also. By the Rearrangement
Inequality we have
a + b + c b + c + a ,
b+c a c+a b a+b c b+c a c+a b a+b c
45
a b c c a b
+ + + +
b+c a c+a b a+b c b+c a c+a b a+b c .
Put a b c
Q= + +
b+c a c+a b a+b c.
Adding these inequalities we get
2Q b + c + c+a + a+b
b+c a c+a b a+b c
b+c a+a c+a b+b a+b c+c
= + +
b+c a c+a b a+b c
a b c + Q.
+
=3+ b+c a c+a b +
a+b c =3
which is equivalent to the desired inequality.
Problem 8. Let x, y, z > 0 be real numbers. Prove the inequality
2 2 2 2 2 2
x z y x z y
+ + 0.
y+z z+x x+y
Solution. We need to prove that
x2 y2 z2 z2 x2 y2
+ + + +
y+z z+x x+y y+z z+x x+y .
Without loss of generality we may assume that x y z(since the given
innequality is cyclic we also will consider case z y x). Then
clear 1 1 1
2 2 2
x y z and
y+z z+x x+y .
2 2 2 2 2 2
x y z z x y
+ + + +
y+z z+x x+y y+z z+x x + y,
as required.
If we assume z y x, then
2 2 2 1 1 1
z y x
and x + y x+z z+y.
46
By the Rearrangement inequality we obtain
2 2 2 2 1 + x2 1 2 1
x y z =z +y
+ +
y+z z+x x +y x+y y+z z+x
2 1 + x2 1 + y2 1
z
x+y z+x z+x
z2 x2 y2
= + + .
y+ z z+x x+y
Equality occurs if and only if x = y = z.
Problem 9. Let x, y, z > 0. Prove that
3 3 3
x y z x + y + z.
+ +
yz zx xy
Solution. Since the given inequality is symmetric we may assume that
x y z. Then
3 3 3 1 1 1
x y z and yz zx xy .

1 1 1
3 x 3 3 y z
+ + =x3 + y3 + z3
yz zx xy yz zx xy
2 2 2
x
3 1 +y 1 +z 1
3 3 x y z
= + + . (10)
yz yz zx y z x
We will prove that 2 2 2
x y z
x + y + z. (11)
+ +
y z x
2 2 2 1 1 1
Let x y z. Then x y z and z y x (since the inequality
(11) is cyclic we also need to consider the case z y x). By the
Rearrangement inequality we obtain
2 2 2 2 2 2
x +y +z x + y + z = x + y + z.
y z x x y z
47
The case when z y x is analogus to the previous case. Now by (10)

and (11) we have
3 3 3
x +y +z x + y + z.
yz zx xy
The equality occurs if and only if x = y = z.
Problem 10. Let a, b, c, d be positive real numbers such that a + b + c
+ d = 4. Prove the inequality
2 2 2 2
a bc + b cd + c da + d ab 4.
Solution. Let (x, y, z, t) be a permutation of (a, b, c, d) such that x y
z t. Then clearly xyz xyt xzt yzt. By the Rearrangement inequality
we obtain
2 2 2 2 (12)
x.xyz + y.xyt + z.xzt + t.yzt a bc + b cd + c da + d ab.
Since AM GM we deduce 2
(xy + xz + yt + zt)
x.xyz + y.xyt + z.xzt + t.yzt = (xy + zt)(xz + yt)
4 .
(13)
Since
(x + y + z + t)2
xy + xz + yt + zt = (x + z)(y + t) =44
by (13) we deduce that
x.xyz + y.xyt + z.xzt + t.yzt 4.
Finally by (12) we obtain

2 2 2 2
a bc + b cd + c da + d ab 4
and we are done. Equality holds iff a = b = c = d = 1 or a = 2b = c = 1,
d = 0(up to permutation).
Problem 11. (IMO, 1995). Let a, b, c be positive real numbers such
that abc=1. Prove that
1 1 1 3.
3 + 3 + 3
a (b + c) b (c + a) c (a + b) 2
48
1 1 1
Solution. Set x = a, y = x = b, z = x = c. Since abc = 1 we
obtain xyz = 1. With these new notations the required inequality
transforms into
2 2 2
x y z 3
+ +
y+z x+z x+y 2.
In virtue of the Neshbit’s generalized and Am-GM inequalities, we hve
2 2 2 x+y+z p 3
x y z 3 3 xyz
+ + = .
y+z x+z x+y 2 2 2
Equality holds if and only if x = y = z , a = b = c.
Problem 12. (Chbyshev’s inequality). If (a1, a2, a3) and (b1, b2, b3) are
sim-ilarly arranged, then
a + a+ a3
a1b1 + a2b2 + a3b3 1
b1 + b + b .
2 23
3 3 3
Solution. Using the Rearrangement inequality we get:
a1b1 + a2b2 + a3b3 = a1b1 + a2b2 + a3b3,
a1b1 + a2b2 + a3b3 a1b2 + a2b3 + a3b1,
a1b1 + a2b2 + a3b3 a1b3 + a2b1 + a3b2.
Adding these inequalities we obtain
3(a1b1 + a2b2 + a3b3) a1(b1 + b2 + b3) + a2(b1 + b2 + b3) + a3(b1 + b2
+ b3) = (a1 + a2 + a3)(b1 + b2 + b3),
which is the desired result. Equality holds if a1 = a2 = a3 or b1 = b2 =
b3. Problem 13. (1974, USA Math Olympiad). If a, b, c > 0, then
a+b+c
aabbcc (abc) .
3
Solution. We may assume that a b c. Then (a, b, c) and (log a, log b, log
c) are similar arranged. Using the Chebyshev’s inequality, we have
a+b+c
a log a + b log b + c log c (log a + logb + log c), 3
49
or +b+c
log(aabbcc) log(abc)a ,
3
which is the desired result.

Problem 14. (IMO, 1964). Let a, b and c be the sides of a triangle. Prove that
2 2 2
a (b + c a) + b (c + a b) + c (a + b c) 3abc.
Solution. We may assume that a b c. We first prove that c(a + b

c) b(c + a b) a(b + c a). Note that c(a + b c) b(c + a b) =
(b c)(b + c a) 0. The second inequality can be proved in the same
manner. By the Rearrangement inequality, we have
2 2 2
a (b + c a) + b (c + a b) + c (a + b c)
ba(b + c a) + cb(c + a b) + ac(a + b c),
2 2 2
a (b + c a) + b (c + a b) + c (a + b c)
ca(b + c a) + ab(c + a b) + bc(a + b c).
Adding these two inequalities, the right side simplifies to 6abc. The de-
sired inequality now follows.
Problem 15. (IMO, 1983). Let a, b and c be the sides of a triangle. Prove that
2 2 2
a b(ab) + b c(b c) + c a(c a) 0.
Solution. Wemay assume that a b, c. If a b c, we have a(b + c

a) b(c + a b) c(a + b c) as in Problem 14. By the Rearrangement
inequality,
1 1 1
c a(b + c a) + ab(c + a b) + bc(a + b c)
1 1 1
a a ( b + c a ) + b a ( c + a b ) + c a(a + b c) = a + b + c.
1 1 1
This simplifies to c a(b a) + ab(c b) + bc(a c) 0, which is equiv-
alent to the desired inequality. If a c b, then a(b + c a) c(a + b c) b(c +
a b).
50
Problem 16. Let a, b, c > 0. Prove that

8 8 8
a +b +c 1 1 1 .
3 3 3 + +
abc a b c
Problem 17. Leta b c and 0 x y z. Prove that
a b c a+b+c 3 a+b+c
+ + .
x y z p3 xyz x+y+z
Problem 18. Let a, b, c > 0. Prove that
p p p
ab + bc + ca a bc + b ac + c ab
Problem 19. Let a, b, c > 0 such that a2 + b2 + c2 3. Show that
2 2 2
a a a 3
+ +
b+c b+c b+c 2.
Problem 20. If a, b, c > 0, then
+b+c n n n n
a +b +c a
3 r 3 .
2 Generality
2.1 Definition
Definition 2. Let a1 a2 an and b1 b2 bn be any real number sequences.
(a) S(n) = a1b1 + a2b2 + + anbn is called the Sorted sum of the numbers.
(b) R(n) = a1bn + a2bn 1 + + anb1 is called the Reversed sum of the
num-bers.
(c) P(n) = a1c1 + a2c2 + + ancn is called the Permutated sum of the num-bers,
where c1, c2, . . . cn be any permutation of the numbers b1, b2, . . . , bn.
2.2 The Rearrangement Inequality

Theorem 13.
S(n) P(n) R(n). (14)
The equality (14) occurs if and only if b1 = b2 = . . . bn or a1 = a2 = = an.
51
Proof.
(i) S(n) P(n) : S(1) P(1) is obviously true. Assume S(k) P(k) is true for
0 0
some k 2 N. For P(k + 1), since c s are the permutations of the b s,
suppose bk+1 = ci and ck+1 = bj.
(a a )(b b) 0 a a + a b ab + a b
k+1 i k+1 j , i j k+1 k+1 i k+1 k+1 j
) ab + a b ac + a c .
i j k+1 k+1 i i k+1 k+1
So in P(k+1), we may switch c i and ck+1 to get a possibly large sum.

Af-ter switching of these terms, we come up with the inductive
hypothesis P(k). So, S(k + 1) P(k + 1) is also true. By the principle of
mathemat-ical induction, S(n) P(n) is true 8n 2 N.
(ii) The inequality P(n) R(n) follows easily from S(n) P(n) by replaing b 1
b2 . . . bn by bn bn 1 b1.
Corollary 1. Let a1, a2, . . . , an be real numbers and c1, c2, . . . , cn be

its permu-tation. Then
2 2 2
a 1 +a 2 +....+a n a1c1 + a2c2 + + ancn.
Corollary 2. Let a1, a2, . . . , an be positive real numbers and c1, c2, . . . , cn be its
permutation. Then
c c c
1 + 2 + + n n.
a a a
1 2 n
2.3 Classical Inequalities

2.3.1 Arithmetic Mean-Geometric Mean Inequality AM GM
Let x1, x2, . . . , xn be positive numbers. Then

pn . (15)
x1 + x2 + + xn
x x ...x
n 1 2 n
Equality holds if and only if x1 = x2 = = xn.

Proof. Let
p x1 x1x2 x1x2 . . . xn
G = n x1x2 . . . xn, a1 = , a2 = , . . . , an = = 1.
GGG
52
By Corollary 2,
na
1
+a2 +... an = x1 +x2 +...xn ,
an a1 a G G G
n 1
which is equivalent to the inequality (15).
2.3.2 Geometric Mean-Harmonic Mean Inequality GM HM
Let x1, x2, . . . , xn be positive numbers. Then

pn n . (16)
x1x2 . . . xn 1 1 1
+ x2 + +
x1 xn
Equality holds if and only if x1 = x2 = = xn.
Proof. Let G and a1, a2, . . . , an similarly as in the proof of AM-GM. By
Corollary 2,
n a1 + a2 +... an = G + G +... G ,
a2 a3 a1 x1 x2 xn
which is equivalent to the inequality (16).
2.3.3 Root Mean Square-Arithmetic Mean Inequality RMS AM
Let x1, x2, . . . , xn be real numbers. Then

x
s x1 2 x
n 1 2 n
x x + x n n . (17)
2 + 2 + + 2 + +
Equality holds if and only if x1 = x2 = = xn. Proof. By Corollary 7, we

have
2 2 2
x1 + x2 + + xn = x1x1 + x2x2 + + xn xn,
2 2 2 2
x1 + x2 + + xn x1x2 + x2x3 + + xn x1, x1 +
2 2
x2 + + xn x1x3 + x2x4 + + xn x2,
........................ .................................
2 2
x +x + + x2 x x + x x + + x x .
1 2 n 1 n 2 1 n n 1
53
Adding all inequalities together, we have

2 2 2 2
n(x1 + x2 + + xn ) (x1 + x2 + + xn) ,
which is equivalent to (17). The equality holds if and only if x1 = x2 =
= xn .
2.3.4 Cauchy-Bunyakowskii- Schwarz Inequality (CBS-Inequality)
Let a1, a2, . . . , an and b1, b2, . . . , bn be real number sequences. Then
2 2 2 2 2 2 2
(a 1 b 1 + a 2 b 2 + + a nbn) (a 1 +a 2 + + a n)(b1 + b2 + + bn ),
(18)
with equality if and only if for some constant k, a i = kbi, for 1 i n
or bi = kai, for 1in. Proof. If a1 = a2 == an = 0 or

b1 = b2 == bn = 0, the result is trivial. Otherwise, define P =
q ai qi
2 2 2 2 2 2
a1 + a2
+ +a n and Q = b1 + b2 + + bn . Since both are non-
b i n. By Corollary 7,
zero, we let xi = and xn+i =
P Q for 1
2 = a 1 + a 2 + + a n + b 1 + b 22 + + b n2
2 2 2 2
2 2
P Q
= 2 2 2
x1 + x2 + + xn
x1xn+1 + x2xn+2 + + xn x2n + xn+1x1 + xn+2x2 + + x2n xn
2(a b + a b + + a b )
= 1 1 2 2 n n ,
PQ
which is equivalent to the desired result. Equality holds if and only if
x = x
i n+1
for 1 i n, or ai Q = bi P for 1 i n.
2.3.5 Chebyshev’s Inequality
Let x1 x2 . . . xn and y1 y2 . . . yn be any numbers sequences.

Then
x1y1 + x2y2 + + xnyn x1 + x2 + + xn .y1 + y2 + + yn
n n n
x1yn + x2yn 1 + + xny1 . (19)
n
54
Proof. By Rearrangement inequality, we cyclically rotate xi and yi,
x1y1 + x2y2 + + xnyn = x1y1 + x2y2 + + xnyn

x1yn + x2yn 1 + + xny1,
x1y1 + x2y2 + + xnyn x1y2 + x2y3 + + xny1 x1yn + x2yn
1 + + xny1,
................... ......................,
x1y1 + x2y2 + + xnyn x1yn + x2yn 1 + + xny1 =x1yn
+ x2yn 1 + + xny1.
2
Adding up the inequalities and divide by n , we get our result.
2.4 Problems
Problem 21 (IMO, 1975). Let x1 x2 xn and y1 y2 yn be real numbers.

Let (z1, z2, . . . , zn) be a permutation of (y1, y2, . . . , yn). Prove that
2 2 2
(x1 y1) + (x2 y 2) + + (xn y n)
2 2 2
(x1 z1) + (x2 z 2) + + (xn zn) .
2 2 2 2 2 2
Solution. Note that we have y 1 + y 2 + + y n = z 1 + z 2 + + z n. After
expansion and simplification, the desired inequality is equivalent to
x1y1 + x2y2 + + xnyn x1z1 + x2z2 + + xnzn,
which follows from the Rearrangement inequality.
Problem 22 (IMO, 1978). Let a1, a2, . . . , an be distinct positive

integers. Prove that
a1 a2 an 1 1 1
1 2 + 2 2 + + n 2 1+ 2 ++ n .
Solution. Let (c1, c2, . . . , c n) be the permutation of (a 1, a2, . . . , an),

such that c1 c2 cn. Then ci i for 1 i n. By the Rearrangement
55
inequality,
a1 a2 an c1 c2 cn
1 2 + 2 2 + + n 2 1 2 + 2 2 + + n 2
c1 2 n
1 2 + 2 2 + + n 2
1 1 1
= 1+ 2+ + n .
References
[1] Zdravko Cvetkovski (2012), Inequalities: Theorems, Techniques
and Se-lected problems. Springer-Verlag Berlin Heidelberg.
[2] Samin Riasat (2008), Basic of Olympiad Inequalities.
[3] https://www.qc.edu.hk/math/Resource/AL/RearrangementRearrangement
Inequality. Yue Kwok Choy.
[4] https://www.math.ualberta.ca/pi/issue2/page21-23.pdf The Rear-
rangement. Inequality by Dragos Hrimiuc.
56
SOME NEW COMBINATORIAL IDENTITIES
Dam Van Nhi

Hanoi University of Education
We introduce some classes of systems of linear equations can be solved

by combinatorial identities.
1 Some systems of linear equations and combinatorial

identities
Example 1. Assume that a1, a2, . . . , an 2 R and ai + j 6= 0 with i, j =

1, 2, . . . , n. Solve the following linear system:
8 x1 + x2 + + xn = 1
1 a + 2 a1 + +
n a =1
x1 1 x2 xn 1
>
+ + +
> 1 + a2 2 + a2 n + a2
>
>
>
<
>...
> 1 + 2 + + n = 1. p(x)
1= n ,
+ + +
1 2 n
> an an an
Õ (i + x )
i=1
> x x x
>
>
>
: x x x
1 2 n
Proof. Consider f (x) = + + +
1+x 2+x n+x
where p(x) is a polynomial of degree n. Since f (ai) = 0, therefore p(ai)
= 0, i = 1, . . . , n. In view of this result, we get
f (x) = (x a1)(x a2) . . . (x an) .
(x + 1)(x + 2) . . . (x + n)
From x1 + x2 + + xn 1= (x a1) . . . (x an) we de-
1+x 2+x n+x (x + 1 ) . . . ( x + n )
57
duce
x1(x + 2)(x + 3) . . . (x + n) + x2(x + 1)(x + 3) . . . (x + n)
+ x3(x + 1)(x + 2)(x + 4) . . . (x + n) + x4(x + 1) . . . (x + n)
+
+ xn(x + 1)(x + 2) . . . (x + n 1) (x + 1)(x + 2) . . . (x + n)
= (x a1)(x a2) . . . (x an).
Upon direct computation, we get the solution of system of equations
8
n
> ( 1)n 1
Õ(1 + a i=1
) by x = 1
x1 = (n 1)! i
>
n
>>
>
( 1)n 2 (2+ a )
>
Õ
> i
> i=1
>
>x2 = by x = 2
>
>>
1!(n 2)!
> n
>>
above >
( )n 3 ( + )
>
1 Õ 3 ai
>
> i=1
>
<
>x3 = 2!(n 3)! by x = 3
>
> n
>
> n n
>
>...
>
Õ(n + ai)
> ( 1)
> i=1
>
>xn = by x = n.
>
> (n 1)!
>
>>
>
> n
>
>
:
Corollary 3. Set j(x) = Õ(x + ai). Then, there are the following identi-
ties: i=1
n i n n
(i)
å ( 1) ( i)j(i) = ( 1) n!.
i=0
(ii) n i n n n
1) ( i)i = ( 1) n!.
å(
i=0
(iii) n i n n+ i n
1) ( i)(i ) = ( 1) .
å(
i=0
Proof. (i) We invoke by Example 1 to deduce that:
n n i
( 1) j(i) (x a1)(x a2) . . . (x an)
å (x + i)(i 1)!(n i)! 1= (x + 1)(x + 2) . . . (x + n) .
i=1
58
n n i n+1 n
( 1) j(i) ( 1) j(0)
By x = 0 we get å 1= or å ( i n
i=1 i!(n i)! n! i=0
1) ( i)j(i) =
n n
( 1) n!.
= 0 in å ( i n n
(ii) Substituting a1 == an 1) ( i)j(i) = ( 1) n! we get
n i=0
i n n n
å ( 1) ( i)i = ( 1) n!.
i=0 n
i n n+ i n
(iii) By ai = i, i = 0, 1, . . . , n we obtain å ( 1) ( i)( i ) =( 1) .
i=0
Example 2. Solve the following system of linear eqations:
n
8 sxk + yk = 1
2 2
k=1 s + k s
< å s= 1,..., n; xk, yk R, k = 1, . . . , n
: 2
n
n 2 2
Õ (s + k )
and evaluate the sum S = s=1 .
å 2 2 2
k =1 Õ (k s )(1 + k )
s6=k
1 n p(x)
xk x + yk
Proof. Consider f (x) = x +å 2
x +k 2
= n , where p(x)
k=1 2 2
x Õ (x + k )
k=1
is a polynomial of degree 6 2n. Since f (s) = 0 therefore p(s) = 0 when
2 2 2 2 2 2
s = 1, . . . , n. We obtain f (x) = a(x 1 )(x 2 ) . . . (x n )
n . Since
2 2
x Õ (x + k )
k=1
n 2 2 2 2 2 2
1 xk x + yk a(x 1 )(x 2 ) . . . (x n )
x +å 2
x +k
2
= n we deduce the iden-
k=1 2 2
x Õ (x + k )
k=1
tity
2 2 2 2 2 2 2 2 2 2
(x + 1 ) . . . (x + n ) + x(x1x + y1)(x + 2 )(x + 3 ) . . . (x + n )
2 2 2 2 2 2
+ x(x2x + y2)(x + 1 )(x + 3 ) . . . (x + n ) +
2 2 2 2 2 2
+ x(xn x + yn)(x + 1 )(x + 2 ) . . . (x + (n 1) )
2 2 2 2 2 2
= a (x 1 )(x 2 ) . . . (x n )
59
8
>a = ( 1)n
>
by x = 0
>
> n
>
>
>
a( 1)n Õ (12 + k2)
>
k=1
>
>
> n by x = i
> 2
x1 + iy1 = Õ (k 1)
> k=2
>
>
n
>
>
>
a( 1)n Õ (22 + k2)
k=1
by x = 2i
>
>
>
> n
>
2 2
>
< Õ (k 2 )
and obtain the solution x2 + iy2 = k=1,k6=2
>
>
>
>
>
>
n
>
>...
>
a( 1)n Õ (n2 + k2)
k=1
by x = ni.
>
>
>
>
>
n 1
> 2 2
>
>
> x + iy =
Õ (k n )
> n n
>
>
>
>
>
:
n k=1
2 2
Õ (r + k )
k=1 with r = 1, . . . , n.
Hence y1 == yn = 0 and xr = n
2 2
Õ (k r )
k=1,k6=r
n 2 2 2 2 2 2
1 xk x a(x 1 )(x 2 ) . . . (x n)
Since
+å x2 + k2 = n we have the
x k=1 2 2
x Õ (x + k )
k=1
n
n n 2 2
xk Õ (s + k )
s=1
sum å 1 + k2 = 1 or å n
= 1 by replacing x = 1.
k=1 k=1 2 2 2
Õ (k s )(1 + k )
s6=k
Corollary 4. There is the following identity

n
n k 2 2 2 2n
n ( 1) k Õ (s + k )(n k)
2 å s=1
1 + k2
= (2n)!.
k=0
n 2 2
(s + k )
n Õ n
s=1 2 2
Proof. Sine å n = 1 by Example 2 and Õ (k s )=
k=1 2 2 2 s6=k
Õ (k s )(1 + k )
s6=k
60
n n n k (n k)!(n + k)!
Õ (k s) Õ (k + s) = ( 1) 2
we get
s6=k s6=k 2k
n
n 2 2 2
2k Õ (s + k )
å s=1 = 1.
k=1 ( 1)n k
(n k )! (n + k )! (1 + k
2
)
n k 2 n 2 2 2n
( 1) k (s + k )(
n Õ )
n k
By multiplicating (2n)! we obtain 2 å s=1 = (2n)!.
2
k=1 1+k
Example 3. Solute the following system of linear equations:

8 n n =1
xk yk
k=1 s k k=1 s +k s
< å +å
s= i, . . . , ni; xk, yk R
: n 2
n 2 2
Õ (s + k )
and evaluate the sum S = å s=1 .
2 2 2 2
k= 1 k (1 + k ) Õ (k s )
s6=k
Proof. Consider f (x) = 1 n n p(x) ,

xk yk
+å +å = n
x k=1 x k k=1 x+ k 2 2
x Õ (x k )
k=1
where p(x) is a polynomial of degree 6 2n. Since f (s) = 0 therefore
p(s) = 0 when s = i, . . . , ni. It is easy to show that
f (x) = a(x2 + 12)(x2 + 22) . . . (x2 +n )
2
n .
2 2
x Õ (x k )
k=1
1 n n 2 2 2 2 2 2
xk yk a(x + 1 )(x + 2 ) . . . (x + n )
Since +å +å = n there
x k=1 x k k=1 x+k 2 2
x Õ (x k )
k=1
61
is:
2 2 2 2 2 2 2 2 2 2
(x 1 ) . . . (x n ) + x1x(x + 1)(x 2 )(x 3 ) . . . (x n )
2 2 2 2 2 2
+ x2x(x + 2)(x 1 )(x 3 ) . . . (x n )+
2 2 2 2 2 2
+ xn x(x + n)(x 1 )(x 2 ) . . . (x (n 1 ) )
2 2 2 2 2 2
+ y1x(x 1)(x 2 )(x 3 ) . . . (x n )
2 2 2 2 2 2
+ y2x(x 2)(x 1 )(x 3 ) . . . (x n )+
2 2 2 2 2 2
+ yn x(x n)(x 1 )(x 2 ) . . . (x (n 1 ) )
2 2 2 2 2 2
= a(x + 1 )(x + 2 ) . . . (x + n ).
n n 2
k ). Then, we obtain xk and
2 2 2
Put j(x) = Õ (x + k ) and y(x) = Õ (x
k=1 k=1
yk as:
a= n
( 1) by x = 0
n
2 2
a Õ (1 + k )
k=1 by x = 1
x1 = n
2 2
2.1 Õ (1 k )
k=2
n
a Õ (22 + k2)
k=1
x2 = n by x = 2
2 2 2
2.2 Õ (2 k )
... k=1,k6=2
n
a Õ (n2 + k2)
k=1
xn = n 1
by x = n
2 2 2
2n Õ (n k )
k=1
n
2 2
a Õ (1 + k )
k=1 by x = 1
y1 = n
2 2
2.1 Õ (1 k )
k=2
62
n
a Õ (22 + k2)
k=1
y2 = n by x = 2
2 2 2
2.2 Õ (2 k )
... k=1,k6=2
n
a Õ (n2 + k2)
k=1
yn = n 1
by x = n.
2 2 2
2n Õ (n k )
k=1
1 n n 2 2 2 2 2 2
xk yk a(x + 1 )(x + 2 ) . . . (x + n )
From +å +å = n and
x k=1
xk k=1 x +k x Õ (x
2 2
k )
k=1
2n 2 2 2 2 2 2
xk = yk it follows 1 + 2x å xk a(x + 1 )(x + 2 ) . . . (x + n )
2 2 = .
k=1
x k n
2 2
Õ (x k )
n k=1
n 2 2
Õ (s + k )
By replacing x = i we obtain å 2
s=1
2 2
= 1.
2
k=1 k (1 + k ) Õ (k s )
s6=k
Corollary 5. There is the following identity

n
n k 2 2 2n
n ( 1) Õ (s + k )(n k)
2 s=1
å 2 = (2n)!.
k=1 1+k
n
n 2 2
Õ (s + k )
Since å s=1 = 1, by the Example 3 and
Proof. 2 2 2 2
k=1 k (1 + k ) Õ (k s )
n n s6=k (n k)!(n + k)!
n
Õ 2 2
s ) = Õ (k s) Õ (k + s) = ( we have
(k 1)n k
2
s6=k s6=k s6=k 2k
n
n 2 2
2 Õ (s + k )
å s=1 = 1.
k=1 ( 1) n k
(n k)!(n+ k)!(1 + k 2
)
63
n
n k 2 2 2n
n ( 1) Õ (s + k )( )
n k
We obtain 2 å s=1 = (2n)! by multiplicating with
2
(2n)!. k=1 1+k
Example 4. Solute the following system of lineat equations and evaluate the
following sum:
8 n sxk + yk = 1
(i) k=1 s2 + k2 s(s 2 + 1 2)(s 2 + 22 ) . . . (s 2 + n 2)
< å
s= 1,..., n; xk, yk R.
: n xk 2
(ii) T = å .
2 2
k=1 n +k
n xk x + y k 1 p(x)
Proof. Consider f (x) = å 2
x +k 2
n = n
x 2 2 2 2
k=1 Õ (x + k ) x Õ (x + k )
k=1 k=1
with polynomial p(x) of dgree 6 2n. Because f (s) = 0 therefore p(s) = 0
2 2 2 2 2 2
by replacing s = 1, . . . , n. Hence f (x) = a(x 1 )(x 2 ) . . . (x n )
n .
2 2
x Õ (x + k )
k=1
n 2 2 2 2
xk x + yk 1 a(x 1 ) . . . (x n )
Since å
x2 + k2 n = n we obtain
2 2 2 2
k=1 x Õ (x + k ) x Õ (x + k )
k=1 k=1
2 2 2 2 2 2
1 + x(x1x + y1)(x + 2 )(x + 3 ) . . . (x + n )
2 2 2 2 2 2
+ x(x2x + y2)(x + 1 )(x + 3 ) . . . (x + n ) +
+ x(xn x + yn)(x2 + 12)(x2 + 22) . . . (x2 + (n 1)2)
= a(x2 12)(x2 22) . . . (x2 n2) and
( 1)n 1
a= (n!)
2 by x = 0
n 2 2
a( 1) Õ
n (1 + k ) 1
k=1 + by x = i
x1 + iy1 = n n
2 2 2 2
Õ (k 1 ) Õ (k 1 )
k=2 k=2
64
n
n 2 2
a( 1) Õ (2 + k ) 1
k=1 + by x = 2i
x2 + iy2 = n n
2 2 2 2
Õ (k 2 ) Õ (k 2 )
... k=1,k6=2 k=1,k6=2
n
n 2 2
a( 1) Õ (n + k ) 1
k=1 + by x = ni.
xn + iyn = n 1 2 n 1
Õ (k
2 n ) Õ (k
2
n )
2
k=1 k=1
n
2 2
Õ (k + s ) 1
s=1 with k =
We have yk = 0, xk = 2 n
2 n
(n!) 2 k ) 2 2
Õ (s Õ (s k )
s=1,s6=k s=1,s6=k
1, . . . , n and obtain the identity:
n 1 n 1 2 2 2 2
xk x ( 1) (x 1 ) . . . (x n )
å x2 + k2 n = n .
2 2
k=1 x Õ (x + k ) (n!)2x Õ (x2 + k2)
k=1 k=1
By x = n there is n 1 .
xk
å n 2
+k 2
= n
2 2 2
k=1 n Õ (n + k )
k=1
Corollary 6. There is the following identity:
n
n 2 2
Õ (k + s ) ( 1)k 1 2 ( 2n ) (2n)!
h i 2k
k=0
s=1 2 1 2 n+k .
å = n
k=0
n 1
xk
Proof. By Example 4 we have å 1+k
2
= n or the identity
2
k=1 Õ (1 + k )
k=1
n
2 2
n Õ (k + s ) 1 1 1
h
k=1 (n!)2 Õ (s
2
s=1
k2) Õ (s2 k2)
i 1+k = Õ (1 + k2)
2
å n n n .
s=1,s6=k s=1,s6=k k=1
65
n ( 1) k 1 1
Since Õ (s2 2
k )= 2 (n k)!(n + k)! and T = we
2k n
s=1,s6=k 2
Õ (1 + k )
obtain k=1
n
2 2
n Õ (k + s ) 1 1
h
T k=1 s=1 i
=
(n!)2 Õ (s2 k2) Õ (s2 k2) 1+k
n n 2
å
s=1,s6=k s=1,s6=k
n
2 2
n h Õ (k + s ) 1 i ( 1 )k 1
2k
2
s=1
k=1 (n!) (n k)!(n + k)! (n k)!(n + k)! 1+k
= å 2 2
n 2n
2 2
n Õ (k + s )( ) ( 2n ) ( 1)k 1
2k
2
h
k=1
(n!) (2n)!
n+k i
(2n)! 1 +k
s=1 2 n+k 2 .
= å
n 2 2
n (k + s ) k 1 2 2n
Õ ( 1) 2k ( ) (2n)!
k=1 h i
=
(n!) 2 1 + k2 Õ (1 + k 2)
Hence å s=1 1 n+k n .

k=1
References
[1] D. Faddeev et I. Sominski, Recueil D’Exercices D’Agebre
Superieure, Editions Mir-Moscou 1977.
[2] R. Merris, Combinatorics, PWS publishing company 20 Park
Plaza, Boston, MA 02116-4324.
[3] J. Rivaud, Exercices D’Algebre 1, Paris Librairie Vuibert 1964.
66
MENELAUS’S THEOREM AND ITS APPLICATIONS

IN SOME GRADE 8 PROBLEMS
Nguyen Van Nho

Danang city
We would like to start with a problem in [1].

Problem 1 ([1], Example 41, pp. 90). Let ABC be a triangle with AC >
AB. Arbitrary points D, E are taken on sides AB, AC, respectively, such
that BD = CE. Let K be the intersection point of the two lines DE, BC.
KE
Prove that the ratio KD does not depend on how to choose points
D and E.
Vu Huu Binh [1] gave three methods for solving this problem.
Method 1. For making the ratio equal to KE : KD, we draw straight line
DG parallel to AC (G on BC). According to Thales’s theorem, we have
KE KC KE EC
KD = KG , KD = DG .
EC BD KE BD
Since BD = CE, replacing DG by DG , hence KD = DG , which is
AB
equal to AC , this happens because DG k AC. Consequently, KE :
KD does not depend on how to choose points D and E.
Method 2. Drawing EH k AB (H on BC), we have
KE EH EH AB
KD = BD = EC = AC .
67
Method 3. Drawing DM k BC (M on AC), we have
KE EC KE EC BD AB
ED = EM ) KD = MC = MC = AC .
In these three ways, we used Thales’s theorem and its

consequence, after drawing the appropriate parallel lines.
However, after drawing the shape according to the problem with-out
having to draw the necessary parallel lines, if we are familiar with
Menelaus’s theorem then the result will be seen immediately, with the
attention that BD = CE. We will continue to discuss this after
introducing the Menelaus’s theorem.
In fact, this theorem is proved by Thales’s theorem and the relation
of sides of similar triangles, and in excellent student exams, you have
the right to use it without proof. Now, I would like to introduce this
theorem.
68
2 Menelaus’s theorem
2.1 Observation
A transversal may meet two sides of a triangle and the third side
which is produced (Fig.4), or all three sides which are produced (Fig.5).
P, Q, R are called the traces of transversal PQR on the sides BC, CA,
AB of the triangle ABC. The trace (P) of the transversal on a side (BC)
and the two vertices (B, C) lying on that side determine two segments
(PB, PC). The content of Menelaus’s theorem gives the beautiful rela-
tionship about the six segments (AR, RB, BP, PC, CQ, QA).
2.2 Menelaus’s theorem

Theorem 14. Given triangle ABC. On straight lines containing the
sides BC, CA, AB, we take the corresponding points P, Q, R so that
each point does not coincide with vertices of the triangle, and there are
no more than two points on the sides of the triangle. Then, three points
P, Q, R are collinear if and only
BR .AQ . PC = 1. (1 )
AR QC BP
Proof.
69
(a) Let P, Q, R are collinear (in other words, P, Q, R are the traces of
transversal PQR on the sides BC, CA, AB of the triangle ABC). Let the
parallel through A to the opposite side BC meet PQR in L. From the two
pairs of similar triangles ALQ and CPQ, BRP and ARL we have
AL = AQ , AL = CP.AQ, BR = BP , BR.AL = 1.
CP CQ CQ AR AL AR BP
Replacing AL in the first equation into the second equation is what
must be proved.
* Remark. In the second case of the transversal PQR, you can rely
on Figure 7 to proceed similarly.
(b) Reversely, suppose (1) occurs. Call Q’ as the intersection point

of RP and side AC. Then, according to the above, we also have:
BR . AQ0 . PC = 1. (2 )
0
AR Q C BP
0
From (1) and (2) we deduce AQ AQ , that means Q Q’, which
0 =
Q C QC
proves the reverse proposition.
2.3 Remark
There are several different proofs of Menelaus’ theorem. In what
fol-lows, we give briefly the second proof for (a) by using ratio of two
trian-gles.
70
BR
Denoted the area of the triangle XYZ by [XYZ], we have AR =
[ PBR] AQ [ PAR] PC [ PCR ]
[ PAR ], QC = [ PCR ] and BP = [ PBR] ,
BR AQ PC
hence AR . QC. BP = 1.
2.4 Solving Problem 1 by using Menelaus’s theorem
Going back to Problem 1, we see that, the transversal BCK of the

tri-angle ADE meet all three sides which are produced (Fig.9, like
Fig.5). Therefore, we have
DB AC EK
AB. CE .KD = 1, and the conclusion is inferred immediately,
since DB = CE.
3 Applications of Menelaus’s theorem in some grade-

8-problems
Problem 2 (Hanoi Open Mathematical Competition 2016, Junior

Section, Q.11). Let be given a triangle ABC, and let I be the middle
point of BC. The straight line d passing I intersects AB, AC at M, N,
respectively. The straight line d0 (6= d) passing I intersects AB, AC at
Q, P, respectively. Suppose M, P are on the same side of BC and MP,
NQ intersect BC at E and F, respectively. Prove that IE = IF.
Hints. Applying Menelaus’s theorem to the cases: transversals EMP,

FNQ, MIN, QIP of the same triangle ABC (Fig.10), we get the following
71
equations in turn:
EB . PC .MA = 1, (3 )
EC PA MB
FC .QB .N A = 1, (4 )
FB QA NC
MB . N A . IC = 1, (5 )
MA NC IB
PA .IC . QB = 1. (6 )
PC IB QA
Using (3), (4), (5), (6) and the fact IB = IC, you can deduce EB = FC ,
EC FB
then IE = IF.
Problem 3 (Exam for excellent students of grade 8 at Nguyen Khuyen

school, Danang City, 2007). The diagonals MP and NQ of a quadrilat-
eral MNPQ meet at O in such a way that MO = OP and QO = 2ON.
Suppose that H and K are points on OP and NP respectively such that
MP NK
OH = KP = 3. Show that the points Q, H and K are collinear.
Solution. From the assumption QO = 2ON we have (Fig.11)
OQ = 2 . (7 )
QN 3
Since MO = OP (PO = 1 MP) and MP = 3, hence
2 OH
PH = PO HO = 1 .MP 1 =1 . (8 )
HO HO 2 HO 2
72
Now, consider triangle ONP and the points Q, H and K. By (7), (8)
NK
and using the assumption
KP = 3, we obtain
NK PH OQ 1 2
.
KP HO . QN = 3. 2. 3 = 1.
Therefore, by the converse of Menelaus’ theorem, Q, H and K are
collinear.
Problem 4 (From [2], Problem 1.12, pp. 28). Let ABC be a triangle and
three points A1, B1, C1, respectively be on the three sides BC, CA, AB such
that the lines AA1, BB1, CC1 intersect at O. Suppose that three pairs of lines
AB and A1B1, BC and B1C1, CA and C1 A1 alternately intersect at three
points C2, A2, and B2. Prove that C2, A2, and B2 are collinear.
Solution. Applying Menelaus’s Theorem to triangles and points:

* OAB and A1, B1, C2, we have
AA1 .OB1 . BC2 = 1. (9 )
OA1 BB1 AC2
* OBC and B1, C1, A2, we have
OC1 . BB1 . CA2 = 1. (10)
CC1 OB1 BA2
* OAC and A1, C1, B2, we have
OA1 .CC1 . AB2 = 1. (11)
AA1 OC1 CB2
73
By multiplying the relations (9), (10), and (11) side by side, we obtain
BC CA AB
2 . 2. 2 = 1.
AC2 BA2 CB2
Hence, by the converse of Menelaus’ theorem, C2, A2, and B2 are collinear.
Problem 5 (Problem 4, from the Entrance exam for Vnh Phc

specialized high school ). Let ABC be an acute triangle, AB < AC. Let
D, E, F be the foot of the height of the triangle dropping from A, B, C,
respectively. Call P the intersection point of the lines BC and EF.
Straight line through D which is parallel to EF cuts straight lines AB,
AC, CF at Q, R, S, re-spectively.
PB DB
Prove that PC = DC and D is the midpoint of QS.
Note. That is the content of sentence b) in the problem. I ignore the
two sentences a) and c) because the content is in grade 9 knowledge,
which is not appropriate here.
Solution. Since AB < AC, Q is on the opposite ray of ray BA and R is

inside of line segment CA, so, Q, C are on the same side to the
straight line BR (Fig.13).
From pairs of similar triangles DHB and EHA, DHC and FHA we
74
DB HB DC HC
have AE = H A, AF = H A.
DB =AE . HB = AE .FB . (12)
DC AF HC AF EC
Applying Menelaus’s theorem to the triangle ABC with the transver-
sal PEF, we obtain
PB CE AF PB AE FB
= .
PC .EA . FB = 1 ,
PC (13) AF EC .
(12) and (13) give PB DB
= . (14)
PC DC
DQ BD DS CD
We have QR k EF, so: PF = BP , PF = CP . Combining
these with (14) we obtain DQ = DS, that means D is the midpoint of QS.
Problem 6 (6th 30-4 Mathematical Olympiad, April 2000, Vietnam).
Let the triangle ABC have an area S0 = 1. Points M, N, P are on sides
MB NC
BC, CA, AB, respectively, such that MC = k1 , N A = k 2,
PA
PB = k3, (k1 , k2 , k3 < 1).
Calculate the area of the triangle which is created by three intersec-
tion points of the three line segments AM, BN and CP.
Solution. Let EIF be the triangle which is created by three intersection

points of the three line segments AM, BN and CP (Fig.14), we have
k
[BCN] = CN = 2 .
S0 CA k2 + 1
Having S0 = 1, so [BCN] = k2 .
k2 + 1
75
[BCF] BF BF k2
We also have = , hence [BCF] = .
[BCN] BN BN k2 + 1
(Denoted the area of the triangle XYZ by [XYZ]). Applying Menelaus’s
theorem to the triangle ABN with the transversal PCF, we
FB CN PA
obtain FN . CA . PB = 1, hence
FB 1 + k2 BF 1 + k2
FN = k2k3 =
) BN 1 + k2 + k2k3 ,
therefore 1 + k2 k2 k2
[BCF]= . = .
1 + k2 + k2k3 k2 + 1 1 + k2 + k2k3
k1
Similarly, we have [ACI] = k3 , [ABE] = .
1 + k3 + k1k3 1 + k1 + k1k2
Therefore, + +
[EIF] = 1 1 + k2
+ k2k3 1 + k3 + k1k3 1 + k1 + k1k2 .
k k k
2 3 1
Finally, we give some exercises for good students of grade 8 to

solve themselves. Hopefully, through this article, you will master how
to ap-ply Menelaus theorem to solve appropriate problems.
Problem 7. Let ABC be a triangle. On the opposite ray of ray CB, we
get the point M such that BM : MC = 3. On the line segment CA we
take the point N so that NA : CN = 3. Let P be the intersection point of
MN and AB. Prove that P is the midpoint of AB.
Problem 8. Let MNP be a triangle with angle N = 90 , NP = 3cm and
MN = 4cm. S is a point on the segment MP such that MS = 1cm, and T
is the midpoint of MN. The line ST meets the line NP at H. Find the
length of the segment NH.
Problem 9. Prove that the three external angle bisectors of a triangle
intersect their opposite sides at three collinear points.
References
[1] Vu Huu Binh, Some development topics of Geometry 8, Vietnam

Educa-tion Publishing House, 2000, Vietnamese.
76
[2]Nguyen Van Nho, The theorems in Plane Geometry through Olympic

exams, Vietnam Education Publishing House, 2004, Vietnamese.
Nguyen Van Nho

Danang, March, 2019
77
PROOFS AND GENERALIZATIONS OF ONE INEQUALITY AT THE

36TH IMO
Vu Tien Viet
Hanoi Mathematical Society
We introduce some solutions and generalizations of one International

Olympiad problem (36th IMO).
1 Second problem of IMO 1995

nd th
It is known that the 2 problem given at the 36 IMO held at
Toronto (Canada, 1995) was formulated as follows:
Problem 1. Let a, b, c be positive real numbers with abc = 1. Prove that
1 1 1 3 (1)
3 + 3 + 3
a (b + c) b (c + a) c (a + b) 2
2 Proofs
Proof 1. It follows from the condition abc = 1 that
1 2 2 1 2 2 1 2 2
bc c a ab
=, 3 =
=
3
a (b + c) a(b + c) b (c + a) b(c + a) , c3(a + b) c(a + b)
Noe, the inequality (1) is equivalent to
2 2 2 2 2 2 3
b c ca a b
+ +
a(b + c) b(c + a) c(a + b) 2
By Cauchy-Schwarz inequality
2 2 2 2 2 2
(a 1 +a 2 + a 3)(b1 + b2 + b3 ) (a1b1 + a2b2 + a3b3)2,
78
we have
[a(b + c) + b(c + a) + c(a + b)] 2 2 2 2+ + a2b2
b c c a
h
a
(b + c)b(c
+ )
2a
( + )i
ca b
or (ab + bc + ca) ,
2 2 2 2 2 2
b c c a ab 1 (ab + bc + ca).
+ +
a(b + c) b(c + a) c(a + b) 2
By AM-GM inequality, we have
2 2
bc
2 2
ca
2 2
ab 1 p 3
3 3 2 2 2
a(b + c) + b(c + a) + c(a + b) 2 (ab + bc + ca) 2 abc = 2
Proof 2. It follows from the condition abc = 1 that
1 2 2 1 2 2 1 2 2
bc ca ab
= , 3 =
=
3
a (b + c) a(b + c) b (c + a) b(c + a) , c3(a + b) c(a + b)
Noe, the inequality (1) is equivalent to
2 2 2 2 2 2 3
b c ca ab
+ +
a(b + c) b(c + a) c(a + b) 2
Let
b2c2 c2a2 a2b2 b2c2 c2a2 a2b2 = S.
+ + = + +
a(b + c) b(c + a) c(a + b) ab + ac bc + ba ca + cb
By Cauchy-Schwarz inequality
2 2 2 2 2 2
(a 1 +a 2 + a 3)(b1 + b2 + b3 ) (a1b1 + a2b2 + a3b3)2,
we have
2
[(ab + ac) + (bc + ba) + (ca + cb)]S (ab + bc + ca)
= (ab + bc + ca)(ab + bc + ca).
2(ab + bc + ca)S (ab + bc + ca)(ab + bc + ca)
p
2 2 2
(ab + bc + ca)3 3 a b c = 3(ab + bc + ca).
79
Hence S 3The desired conclusion follows.
2 1
Proof 3. Note that for a > 0 we have a + 1 2, or a 2 . We thus
a a
have
1 1 2 1 a 2( + ) 1 ab +bc
= =
b c
3
a (b + c) 2a a2(b + c) 2a 2 2 a 4
Similarly
1 1 bc + ca 1 1 ca + cb
3
b (c + a) b 4 , c3(a + b) c 4
Adding the above inequalities yields
1 1 1 1 1 1 1
+ + + +
2 (ab + bc + ca)
3 3 3
a (b + c) b (c + a) c (a + b)
b c a
= 1 (ab + bc + ca).
2
Finaly, the AM-GM inequality leads us to the
1 1 1 1 p 3
3 3
2 22
a (b + c) + b (c + a) +c (a + b) 2 (ab + bc + ca) ab c = 2
3 3 3 2
Proof 4. Inequality (1) is equivalent to
2 2 2 2 2 2 3
bc c a a b
+ +
a(b + c) b(c + a) c(a + b) 2
On the other hand, we have for l > 0,
2 2 p
b c
a(b + c) + la(b + c) 2 lbc,
2 2
c a p
b(c + a) + lb(c + a) 2 lca,
2 2
a b p
c(a + b) + lc(a + b) 2 lab.
80

2 2 2 2 2 2 p
bc c a ab
+ +
a(b + c) b(c + a) c(a + b) (2 l 2l)(ab + bc + ca).
b2c2 c 2 a2 a2b2 p
a(b + c) + b(c + a) +c(a + b) (2 l 2l)(ab + bc + ca)
p p 2 2 2 p l l).
6( l l) 3 a b c = 6(
Choosing l = 1 gives
4
2 2 2 2 2 2
bc ca a b 3
+ +
a(b + c) b(c + a) c(a + b) 2
Proof 5. Lemma 1. Let a, b be real numbers and x, y be positive real
num-bers.
Then, we have
2 (*)
a2 b2 (a + b )
+
x y x+y
We see that, inequality ( ) equivalent to
2 2 2 2 2 2 2
a y(x + y) + b x(x + y) (a + b) xy , a y + b x
2
2abxy , (ay bx) 0.
Let a, b, g be real numbers and x, y, z be positive real numbers.
Then, by inequality ( ) we have
2 2 2
a b g (a + b)2 g
2
(a + b + g)2 (**)
+ + +
x y z x+y z x+y+z
Using the inequality ( ) and AM-GM inequality, we have
2 2 2 2 2 2
1 1 1 b c ca a b
3 + 3 + 3 = + +
a (b + c) b (c + a) c (a + b) a(b + c) b(c + a) c(a + b)
(ab + bc + ca)2 1 3p3 3
= (ab + bc + ca) a2b2c2 =
2(ab + bc + ca) 222
81
Cch 6. Lemma 2. Let real numbers a1 a1 and b1 b2.

Then, we have (a1 a2)(b1 b2) 0, or a1b1 + a2b2 a1b2 + a2b1.
Let real numbers a1 a2 a3 and b1 b2 b3.
Then, we have (a2 a3)(b1 b3) 0, or a2b1 + a3b3 a2b3 + a3b1,
by above a1b1 + a2b2 a1b2 + a2b1,
lead to a1b1 + a2b2 + a3b3 a1b2 + a2b3 + a3b1.
On other hand (a1 a3)(b2 b3) 0, or a1b2 + a3b3 a1b3 + a3b2,
by above a1b1 + a2b2 a1b2 + a2b1,
lead to a1b1 + a2b2 + a3b3 a1b3 + a2b1 + a3b2.
Put x = bc > 0, y = ca > 0, z = ab > 0, since abc = 1,
2 2 2
a b c = 1, xyz = 1. The inequality (1) equivalent to
2 2 2 2 2 2
1 1 1 b c c a ab
3 + 3 + 3 = + +
2 2 2
x y z 3
= +
+
y+z z+x x+y 2
By symmetry, we may assume that x y z. Then
x y z .
y+z z+x x+y
Using the above lemma 2 gives
2 2 2
x y z y z x
+ +
y+z z+x x+y x z+x +y x+y +z y+z,
2 2 2
x y z z x y
+ +
y+z z+x x+y x x+y +y y+z+z z+x
2 2
x + y + z
2 1 (x + y + z ) 3p = 3
3
xyz
y+z z+x x +y 2 2 2
Proof 7. Put x = bc > 0, y = ca > 0, z = ab > 0, since abc = 1,
2 2 2
a b c = 1, xyz = 1. The inequality (1) equivalent to
2 2 2 2 2 2
1 1 1 b c c a ab
3 + 3 + 3 = + +
2 2 2
x y z 3
= + +
y+z z+x x+y 2
82
2 2 2 1 1 1 .
x y z ,
y+z z+x x+y
x2 y2 z2 x2 y2 z2
+ + +
+
y+z z+x x+y z+x x+y y+z ,

2 2 2 2 2
x
2 y z x y z
+ + +
+
y+z z+x x+y x+y y+z z+x

2 2 2 1 x2 + y2 y2 + z2 2 2
x y z z +x
+ + + +
y+z z+x x+y 2 x+y y+z z+x .
2
a +b
2 a+b
Clearly, we have a + b 2 for a > 0, b > 0. Then
2 2 2 2 2
x +y + y +z + z +x
2 x+y+z p = 3.
3 3
xyz
x+y y+z z+x
2 2 2
Proof 8. Since abc = 1, a b c = 1. The inequality (1) equivalent to
2 2 2 2 2 2 2 2 2 2 2 2 3
bc ca a b bc ca ab
+ + = + +
a(b + c) b(c + a) c(a + b) ab + ac bc + ba ca + cb 2
Consider the function
bc 2 ca p 2
f (x) = p x pab + ac + p xbc + ba
ab + ac bc + ba
ab 2
+ p x pca + cb =
ca + cb
2 2 2 2
= 2c2 ca ab
+
b
+ x2
ab + ac bc + ba ca + cb
2(ab + bc + ca)x + 2(ab + bc + ca).
0 0, or
Clearly f (x) 0, 8x 2 R. From there, we have D
2 2 2 2
2c2 ca ab (ab + bc + ca) 0.
2 ab + ac +bc + ba + ca + cb
b
2
(ab + bc + ca)
83
Inferred
2 2
bc ca
2 2 2 2
ab 1 p 3
3 3 2 2 2
ab + ac + bc + ba + ca + cb 2 (ab + bc + ca) 2 abc = 2
2 2 2
Proof 9. Since abc = 1, a b c = 1. The inequality (1) equivalent to
3
2 2 2 2 bc 2 2 ca ab
+ +
a(b + c) b(c + a) c(a + b) 2
On space Oxyz consider vectors
q
OA~( a(b + c) , q b(c + a) , qc(a + b)),
~ bc ca ab
OB( p , , ).
a(b + c) p b(c + a) p c(a + b)
~ ~
We have OA OB = ab + bc + ca and [
~ ~ ~ ~ ~ ~
OA OB = jOAj jOBj cos AOB jOAj jOBj =
1
1 2c2 c2a2 a2b2 2
+ bc + ba + ca + cb
b
= [2(ab + bc + ca)] 2
ab + ac .
Inferred
2 2
bc ca
2 2 2 2
ab 1 p 3
3 3 2 2 2
ab + ac + bc + ba + ca + cb 2 (ab + bc + ca) 2 abc = 2
Proof 10. Put 1 + 1 + 1 = a 1 + b 1 + c 1 = t > 0, we have
a b c 2 2 c 2
a b
1 1 1
3 + 3 + 3 = 1 1+ 1 1 + 1 1
a (b + c) b (c + a) c (a + b) b +c c +a a +b
2
a
2 b c 2
= + +
t t t
1 1 1 a b c
2
x
Consider the function f (x) = 00
t x , (0 < x < t). We have f (x) =
2
2t
> 0, lead to f (x) is convex function.
3
(t x)
84
Using Jensen’s inequality, we have
1 1 1 1 1 1 1 1 1 1
f (a ) + f (b ) + f (c ) 3 f ( a +b +c
)= ( + + )
r
3 2 a b c
3 3
1 3
2 abc = 2
So we have to prove the inequality.
Proof 11. Put x = bc > 0, y = ca > 0, z = ab > 0, s = ab + bc + ca =
x + y + z.
We have xyz = 1 and inequality (1) equivalent to =
+
1 +1 + 1 2 +y2 2 =
x z
3 3 3
a (b + c) b (c + a) c (a + b) y+z z+x x+y
=x2 +y2 + z2 3
s x s y s z 2
Clearly
0< s x = y+z <s y = z+x < 1,
1
,
2s 2(x + y + z) 2(x + y + z)
0
<
2s
0< s z = x+y
< 1, s x + s y + s z = 1.
2s 2(x + y + z) 2s 2s 2s
Consider random variable X with
P(X = x ) = s x ,P(X= y )= s y ,P(X= z ) = s z.
s s z 2s
x 2s s y 2s
The, we have
=
EX = x s x + y s y + z s z 1 ,
s x 2s s y 2s s z 2s 2
x 2 s x y 2 s y 2 2
s s s s s z 2 s z
s2
2
E(X ) = x 2 + y 2 + z 2 =
1+y +z x
= 2s s x 2 s y 2 s z
2 y z ,
2 2 1 x 1
0 Var(X) = E(X ) (EX) = + +
2s s x s y s z 4
85
Inferred
2
+ y2 + z2 s =x + y + z 3 p3 =3
x
xyz
s x s y s z 2 2 2 2
3 Generalizations
Problem 1. Let a, b, c be positive real numbers with abc = 1 and let
m 3.
Prove that
1 1 1 3 (2)
m + m + m
a (b + c) b (c + a) c (a + b) 2
1 1 1
Proof. Put x = a, y = b, z = c. We have xyz = 1 and inequality (2)
equivalent to
1 1 1 3
xm 1 ym 1 zm 1
m + m + m = + +
a (b + c) b (c + a) c (a + b) y+z z+x x+y 2
x
m 2
y
m 2
z
m 2
, 1 1 1 , xm 2 ym 2 zm 2
y+z z+x x+y y+z z+x x+y

xm 1 ym 1 zm 1 ym 2 zm 2 xm 2
+ +
y+z z+x x+y x z+x+y x+y +z y+z ,
xm 1 ym 1 zm 1 zm 2 xm 2 ym 2
+ +
y+z z+x x+y x x+y+y y+z +z z+x
1 (xm 2
+y
m 2
+z
m 2
)
xm 1 ym 1 zm 1
+ +
y+z z+x x+y 2
3 q3 3
2 xm 2ym 2zm 2 = 2
86
The inequality (2) is proved. In special case when m = 3, the inequality

(2) would to the inequality (1).
Problem 2. Let a1, a2, ..., an (n 3) be positive real numbers with
a1a2...an = 1 and let m 3. Prove that
å 1 n(n 1) (3)
1 i n,i6=k,l i m 1 1 i<j n,i6=k,l i
j
(n + 1)(n 2)
1 k<l n
Õ a å aa
Hint. Applying the generalized Radon’s inequality.
In a special case when n = 3, a1 = a, a2 = b, a3 = c and m = 3 the

inequality (3) would to the inequality (1).
87
SOME INTERNATIONAL OLYMPIAD PROBLEMS

IN GEOMETRY
Nguyen Dang Phat

Hanoi University of Education
We introduce some known International Olympiad problems in geometry

with short comments.
1 Some remarks
Problem 1. In the plane we are given two circles intersecting at X and
Y. Prove that there exist four points with the following property :
For every circle touching the two given circles at A and B, and
meet-ing the line XY at C and D, each of the lines AC, AD, BC, BD
passes through one of those four points.
Comment. The problem was received in the following formulation
Suppose m is the radical axis of the given circles G1 and G2 in the
plane and n is the set consisting of all circles L touching G 1 and G2
both internally or both externally and intersecting m. n 6= ˘ : Suppose
L 2 n, touching Gi in Ai and intersecting m in Bi, i 2 f1, 2g.
Prove, there exists a set W of four points in the plane, such that for
every L 2 n every line Ai Bj, i, j 2 f1, 2g is incident with at least one
point in W.
Problem 2. Two circles G1 and G2 intersect at M and N. Let AB be the
line tangent to these circles at A and B, respectively, so that M lies closer
to AB than N. Let CD be the line parallel to AB and passing through M,
with C on circle G1 and D on G2. Lines AC and BD meet at E; lines AN
and CD meet at P; lines BN and CD meet at Q. Show that EP = EQ.
Comment. The proposer has also suggested an alternative version of
the problem: Under the same assumptions, prove that NE is the
bisector of the angle CND.
88
Problem 3. Let O be the circumcentre and H the orthocentre of an

acute triangle ABC. Show that there exist points D, E, F on sides BC,
CA, AB respectively, such that OD + DH = OE + EH = OF + FH and
the lines AD, BE, CF are concurrent.
Comment (by the proposer). If P is a variable point on BC, then the
po-sition of P for which OP + PH is a minimum is given by P = D.
Problem 4. Let A1 A2 . . . An be a convex polygon, n 4. Prove that A1

A2 . . . An is cyclic if and only if to each vertex A j one can assign a pair
(bj; cj) of real numbers, j = 1, 2, . . . , n, so that
Ai Aj = bjci bicj for all i, j with 1 i<j n. (*)
Comment. The numbers bj, cj are not uniquely determined. Here is one
more proof of the necessity. Let A1, A2, . . . , An be arranged counterclock-
wisely around a circle G with centre O and radius R. Choose a system of
rectangular coordinates with origin at O so that the positive x-axis in-
tersects G at a point U between An and A1. Let 2aj be the measure of the
oriented angle UOAj (j = 1, 2, . . . , n), that is : the angle through which U
has to be rotated in counterclockwise direction around O so that it co-
incides with Aj. For any pair of indices i, j with i < j, we have Ai Aj =
1
2R sin 2 \AiOAj, where \AiOAj is the respective central (non-oriented)
angle. Note that \AiOAj is equal to 2(aj ai), if 0 < 2(aj ai) 180 ,
and to 360 2(aj ai), if 180 < 2(aj ai) < 360 . In both cases,
Ai Aj = 2R sin(aj ai) = 2R sin aj cos ai 2R cos aj sin ai.
So we can define bj = u sin aj, cj = v cos aj for j = 1, 2, . . . , n, where u
and v are arbitrary numbers such that uv = 2R.
Problem 5. The tangents at B and A to the circumcircle of an acute-
angled triangle ABC meet the tangent at C at T and U respectively. AT
meets BC at P, and Q is the midpoint of AP; BU meets CA at R, and S
is the midpoint of BR.
Prove that \ABQ = \BAS. Determine, in terms of ratios of side-
lengths, the triangles for which this angle is a maximum.
89
Comment (by the proposer). An alternative method for finding ymax is

2 2 2 2
: From (1), we obtain 9(1 sin C) = (4 y sin C) , sin C(y + 9)
8y sin C + 7 = 0. Since this equation in sin C has real roots,
2 2 2
64y 4.7(y + 9), so y 7, y p 7 as before; equality when sin C =
p
7 3
4 , cos C = 4.
Comment. The original formulation of the problem statement did not
assume that triangle ABC was acute, and required that the size of the
maximum angle be computed.
Problem 6. Let ABCD be a convex quadrilateral with AB not parallel to
CD, and let X be a point inside ABCD such that \ADX = \BCX < 90 and
\DAX = \CBX < 90 . If Y is the point of intersection of the perpendicular
bisectors of AB and CD, prove that \AYB = 2\ADX.
Comment. The proposer has also supplied one more proof of the fact
that the point W coincides with Y, using inversion.
Comment. In its original formulation, the problem statement allowed X
to be an arbitrary point of the plane, not necessarily inside ABCD, and
did not require that the angles ADX, BCX, DAX, CBX be acute.
Without these assumptions the conclusion of the problem requires
considering oriented angles, possibly exceeding 360 .
Problem 7. Let ABCDEF be a convex hexagon such that AB =
BC, CD = DE, EF = FA. Prove that
BC DE FA 3
+ + .
BE DA FC 2
When does the equality occur?
Comment. The problem was proposed for a cyclic hexagon with the
use of Ptolemy’s equality.
Problem 8. Four different points A, B, C, D are chosen on a circle G

so that the triangle BCD is not right-angled. Prove that:
a) The perpendicular bisectors of AB and AC meet the line AD at certain
points W and V, respectively, and that the lines CV and BW meet at a
90
certain point T.
b) The length of one of the line segments AD, BT and CT is the sum of
the lengths of the other two.
Comment. The problem was slightly changed and a geometric
solution to it found. The author’s solution makes use of trigonometry.
Original formulation. In triangle ABC the angle at A is the smallest. A

line through A meets the circumcircle again at the point U lying on the
arc BC opposite to A. The perpendicular bisectors of CA and AB meet
AU at V and W, respectively and the lines CV, BW meet at T. Show
that AU = TB+TC.
Problem 9. Let AH1, BH1, CH3 be the altitudes of an acute-angled trian-
gle ABC. Its incircle touches the sides BC, CA, AB at T 1, T2, T3, respec-
tively. Consider the symmetric images of the lines H 1 H2, H2 H3, H1 H3
with respect to the lines T1T2, T2T3, T1T3. Prove that these images form
a triangle whose vertices lie on the incircle of the triangle ABC.
Comment. The last proof is motivated by a simple observation. It is easy
to compute the angles between AB and the lines H 1 H2, T1T2, which are
2ja bj and ja bj, respectively. It follows that the mirror image of H 1 H2
across T1T2 is parallel to AB. A similar conclusion holds for the other two
mirror images, so the triangle 4. formed by the three reflec-tions has
sides parallel to those of 4ABC. Hence there is a homothety h taking
4ABC to 4. Now, the claim is that 4. is inscribed in the incircle of 4ABC.
This can be true only if h takes the circumcircle of 4ABC to its incircle,
which suggests defining h as in the solution above.
Problem 10. Let ABCDEF be a convex hexagon such
that \B + \D + \F = 360 and
AB CD EF
=1.
BC DE FA
Prove that
BC AE FD
=1.
CA EF DB
91
Comment. Considering the arguments of the complex numbers on

both sides of the equality
b c a e f d
c e = 1,
a f b d
it shows that \BDF = \AEF + \ACB.
2 Other problems
Problem 11. Let A1 be the center of the square inscribed in acute tri-
angle ABC with two vertices of the square on side BC. Thus one of the
two remaining vertices of the square is on side AB and the other is on
AC. Points B1, C1 are defined in a similar way for inscribed squares
with two vertices on sides AC and AB, respective1y. Prove that lines
AA1, BB1, CC1 are concurrent.
Problem 12. In acute triangle ABC with circumcenter O and altitude
AP,
\C > \B + 30 . Prove that \A + \COP < 90 .
Problem 13. Let ABC be a triangle with centroid G. Determine, with
proof, the position of the point P in the plane of ABC such that AP.AG
+ BP.BG + CP.CG is a minimum, and express this minimum value in
terms of the side lengths of ABC.
0
Problem 14. Let M be a point in the interior of triangle ABC. Let A lie
0 0 0
on BC with MA perpendicular to BC. Define B on CA and C on AB
similarly. Define
0 0 0
MA .MB .MC
p(M) = .
MA.MB.MC
Determine, with proof, the location of M such that p(M) is maximal. Let
m(ABC) denote this maximum value. For which triangles ABC is the
value of m(ABC) maximal?
Problem 15. Let ABC be an acute triangle. Let DAC, EAB, and FBC
be isosceles triangles exterior to ABC, with DA = DC, EA = EB and
92
FB = FC, such that
\ADC = 2\BAC, \BEA = 2\ABC, \CFB = 2\ACB.

0 0
Let D be the intersection of lines DB and EF, let E be the
0
intersection of EC and DF, and let F be the intersection of FA and
DE. Find, with proof, the value of the sum.
DB EC FA
+ + .
DD0 EE0 FF0
Problem 16. Let ABC be a triangle and P an exterior point in the plane
of the triangle. Suppose AP, BP, CP meet the sides BC, CA, AB (or
ex-tensions thereof) in D, E, F, respectively.Suppose further that the
areas of triangles PBD, PCE, PAF are all equal. Prove that each of
these areas is equal to the area of triangle ABC itself.
Problem 17. Let O be an interior point of acute triangle ABC. Let A 1 lie
on BC with OA1 perpendicular to BC. Define B1 on CA and C1 on AB
similarly. Prove that O is the circumcenter of ABC if and only if the
perimeter of A1B1C1 is not less than any one of the perimeters of
AB1C1, BC1 A1, and CA1B1.
Problem 18. Let ABC be a triangle with \BAC = 60 . Let AP

bisect \BAC and let BQ bisect \ABC, with P on BC and Q on AC. If AB +
BP = AQ + QB, what are the angles of the triangle ?
Problem 19. Ten gangsters are standing on a flat surface, and the dis-
tances between them are all distinct. At twelve o’clock, when the
church bells start chiming, each of them shoots at the one among the
other nine gangsters who is the nearest, At least, how many gangsters
will be killed?
Problem 20. Let ABC be a triangle and M be an interior point. Prove

that
minfMA, MB, MCg + MA + MB + MC < AB + AC + BC.
93
Problem 21. A set S of points from the space will be called completely
symmetric if it has at least three elements and fulfils the condition :
for every two distinct points A, B from S the perpendicular besector
plane of the segment AB is a plane of simmetry for S.
Prove that if a completely symmetric set is finite then it cosists of
the vertices of either of a regular polygon, or a regular tetrahedron or a
reg-ular octahedron.
Problem 22. For a triangle T = ABC we take the point X on the side
AX 4
( AB) such that XB = 5 the point Y on the segment (CX) such that CY =
2YX and, if possible, the point Z on the ray (CA such that
[ [ S
CXZ = 180 ABC. We denote by the set of all triangles T for which
XYZ[ = 45 .
Prove that all the triangle form S are similar and find the measure of
their smallest angle.
Problem 23. Let ABC be a triangle, W its incircle and W a, Wb, Wc
three circles orthogonal toW passing through (B, C), (A, C) and (A, B)
0
respec-tively. The circles Wa and Wb meet again in C : in the same
0 0
way we obtain the points B and A . Prove that the circumradius of
0 0 0
A B C is half the radius of W.
Problem 24. Two circles W1 and W2 touch internally the circle W in M
and N and the center of W2 is on W1. The common chord of the circle
W1 and W2 intersects W in A and B. MA and MB intersects W 1 in C
and D. Prove that W2 is tangent to CD.
Lemma. The circle k1 touches internally circle k at A and touches one
of k’s chords MN at B. Let C be the mid-point of k’s arc MN which does
2
not contain A. Then the points A, B, C are collinear and CA.CB = CM .
Proof. [Proof of the lemma] (fig 89a) The homothety with center A
which transforms k1 into k transforms MN into a tangent at k parallel to
MN, i.e. into the tangent at C to k, so A, B, C are collinear.
\ \
4
For the second part notice that N MC CAM, therefore ACM v 4MCB,
2
whence CA.CB = CM .
94
Problem 25. The point M is inside the convex quadrilateral ABCD,

such that
\\\ \[\ MA = MC, AMB = MAD + MCD and CMD = MCB + MAB.
Prove that AB.CM = BC.MD and BM.AD = MA.CD (Fig 90a).

Problem 26. Points A, B, C divide the circumcircle W of the triangle
ABC into three arcs. Let X be a variable point on the arc AB and O 1,
O2 be the incenter of the triangles CAX and ABX. Prove that the
circumcircle of the triangle XO1O2 intersects W in a fixed point.
Problem 27. A convex quadrilateral ABCD has perpendicular diago-
nals. The perpendicular bisectors of AB and CD meet at a unique point
P inside ABCD. Prove that ABCD is cyclic if and only if triangles ABP
and CDP have equal areas.
Problem 28. Let ABCD be a cyclic quadrilateral. Let E and F be
variable points on the sides AB and CD, respectively, such that AE :
EB = CF : FD. Let P be the point on the segment EF such that PE : PF
= AB : CD. Prove that the ratio between the areas of triangles APD
and BPC does not depend on the choice of E and F.
Problem 29. Let I be the incenter of triangle ABC. Let K, L and M be
the points of tangency of the incircle of ABC with AB, BC and CA
respec-tively. The line t passes through B and is parallel to KL. The
lines MK and ML intersect t respectively at the points R and S. Prove
that \RIS is acute.
Problem 30. Let M and N be points inside triangle ABC such
that \MAB = \N AC and \MBA = \NBC.
Prove that
AM.AN BM.BN CM.CN
+ + = 1.
AB.AC BA.BC CA.CB
Problem 31. Let ABC be a triangle. H its orthocenter, O its
circumcenter, and R its circumradius. Let D be the reflection of A
across BC, E be that of B across CA, and F that of C across AB. Prove
that D, E and F are collinear if and only if OH = 2R.
95
Problem 32. Let ABC be a triangle such that \ACB = 2\ABC. Lel D be
the point on the side BC such that CD = 2BD. The segment AD is
extended to E so that AD = DE. Prove that
\ECB + 180 = 2\EBC.
Problem 33. Let ABC be a triangle such that \A = 90 and \B < \C. The
tangent at A to its circumcircle w meets the line BC at D. Let E be the
reflection of A across BC, X the foot of perpendicular from A to BE,
and Y the midpoint of AX. Let the line BY meet w again at Z. Prove
that the line BD is tangent to the circumcircle of triangle ADZ.
Problem 34. Let A1 A2 A3 be a non-isosceles triangle with the incentre
I. Let
Ci, i = 1, 2, 3, be the smaller circle through I tangent to A i Ai+1 and Ai
Ai+2 (the addition of indices being mod 3). Let B i, i = 1, 2, 3, be the
second point of intersection of C i+1 and Ci+2. Prove that the
circumcen-tres of the triangles A1B1 I, A2B2 I, A3B3 I are collinear,
Problem 35. In an acute-angled triangle ABC, let AD.BE be altitudes and
AP, BQ internal bisectors. Denote by I and O the incentre and the
circumcentre of the triangle, respectively. Prove that the points D, E and
I are collinear if and only if the points P, Q and O are collinear.
Problem 36. The altitudes through the vertices A, B, C of an acute-
angled triangle ABC meet the opposite sides at D, E, F, respectively.
The line through D parallel to EF meets the lines AC and AB at Q and
R, re-spectively. The line EF meets BC at P. Prove that the
circumcircle of the triangle PQR passes through the midpoint of BC.
Problem 37. Let D be an internal point on the side BC of a triangle
ABC. The line AD meets the circumcircle of ABC again at X. Let P and
Q be the feet of the perpendiculars from X to AB and AC, respectively,
and let g be the circle with diameter XD. Prove that the line PQ is
tangent to g if and only if AB = AC.
Problem 38. Let ABCD be a convex quadrilateral and O be the
intersec-tion of its diagonals AC and BD. If
OA sin \A + OC sin \C = OB sin \B + OD sin \D,
96
prove that ABCD is cyclic.

Problem 39. The bisectors of angles A, B, C of a triangle ABC meet
its circumcircle again at the points K, L, M, respectively. Let R be an
internal point on the side AB. The points P and Q are difined by the
conditions: RP is parallel to AK and BP is perpendicular to BL, RQ is
parallel to BL and AQ is perpendicular to AK. Show that the lines KP,
LQ, MR have a point in common.
Problem 40. Let triangle ABC have orthocenter H, and let P be a point
on its circumcircle. Let E be the foot of the altitude BH, let PAQB and
PARC be parallelograms, and let AQ meet HR in X. Prove that EX is
parallel to AP.
Problem 41. Let P be a point inside 4ABC such that
\APB \C = \APC \B.
Let D, E be the incenters of 4APB, 4APC, respectively. Show that AP,

BD and CE meet in a point.
Problem 42. Let ABC be an acute-angled triangle with BC > CA. Let
O be its circumcenter, H its orthocenter, and F the foot of its altitude
CH. Let the perpendicular to OF at F meet the side CA at P. Prove that
\FHP = \BAC.
Problem 43. Let 4ABC be an equilateral triangle and let P be a point
in its interior. Let the lines AP, BP, CP meet the sides BC, CA, AB at
the points A1, B1, C1 respectively. Prove that
A1B1.B1C1.C1 A1 A1B.B1C.C1 A.
Problem 44. Let ABCDEF be a convex hexagon such that AB is

parallel to DE, BC is parallel to EF and CD is parallel to AF. Let R A,
RC, RE denote the circumradii of triangles FAB, BCD, DEF
respectively, and let P denote the perimeter of the hexagon. Prove that
P
R +R +R .
A C E 2
97
Problem 45. Let the sides of two rectangles be fa, bg and fc, dg
respec-tively, with
a < c d < b and ab < cd. Prove that the first rectangle can be placed
within the second one if and only if
2 2 2 2
(b a ) (bd ac) + (bc ad)2.
Problem 46. Let ABC be an acute-angled triangle with circumcenter O
0
and circumradius R. Let AO meet the circle BOC again in A , let BO
0
meet the circle COA again in B and let CO meet the circle AOB again
0
in C . Prove that
0 0 0 3
OA .OB .OC 8R .
When does equality hold ?
Problem 47. Let ABCD be a convex quadrilateral and let R A, RB, RC and
RD denote the circumradii of the triangles DAB, ABC, BCD and CDA
respectively. Prove that RA + RC > RB + RD if and only if \A + \C >
\B + \D.
Problem 48. On the plane are given a point O and a polygon F (not
necessarily convex). Let P denote the perimeter of F, D the sum of the
distances from O to the vertices of F, and H the sum of the distances
2
2 2 P
from O to the lines containing the sides of F. Prove that D H .
4
98
SELECTED PROBLEMS FROM MATHEMATICS

AND YOUTH MAGAZINE 2018
Tran Ngoc Nam

Mathematics and Youth Magazine
1 For Secondary School
Problem 1 (1/488 (For 6th grade)). Given a pentagon ABCDE.

Assume that BC is parallel to AD, CD is parallel to BE, DE is parallel to
AC, and AE is parallel to BD. Show that AB is parallel to CE.
Problem 2 (2/488 (For 7th grade)). Prove that

1 +1 + 1 + + 1 2019
3 5 4035 >
1 +1 + 1 + + 1 4036
2 3 2018
Problem 3 (3/488). Given two triples (a, b, c); (x, y, z), none of them
contains all 0’s, such that
a + b + c = x + y + z = ax + by + cz = 0.
Prove that the expression

2 2
(b + c ) (y + z )
P=
ab + bc + ca + xy + yz + zx
is a constant.
Problem 4 (4/488). Outside a triangle ABC , draw triangles ABD, BCE,

[ [ 0 [ [ [
CAF such that ADB = BEC = CFAd = 90 , ABD = CBE = CAF = a.
Prove that DF = AE.
99
Problem 5 (5/488). Show that the following sum is a positive integer

1 1 + 1 1 2
S=1+ 2 ++ 1+2 + + +
2017 2017
+2 + + 2017 2 + + 2017 2
1 1 1
Problem 6 (1/489 (Grade 6)). Find all pairs of integers (x, y) satisfying
2 2018y
x +x=3 + 1.
0
Problem 7 (2/489 (Grade 7)). Given a triangle ABC with Bb = 45 ,
0 \
Cb = 30 . Let BM be one of the medians of ABC. Find the angle AMB.

Problem 8 (3/489). Given real numbers x, y satisfying 0 < x, y < 1. ).
Find the minimum value of the expression
2 2 2xy x y+1
F=x +y
+
4xy
Problem 9 (4/489). Given a circle (O) with a diameter AB. On (O) pick
a point C (C is different from A and B). Draw CH perpendicular to AB
at H. Choose M and N on the line segments CH and BC, respectively,
such that MN is parallel to AB. Through N draw a line perpendicular to
BC. This line intersects the ray AM at D. On the line DO choose two
points F and K such that O is the midpoint of FK. The lines AF and AK
respectively intersect (O) at P and Q. Prove that D, P, Q are colinear.
3 2
Problem 10 (5/489). Suppose that the polynomial f (x) = x + ax + bx +
c has 3 non-negative real solutions. Find the maximal real number a
3
so that f (x) a(x a) , 8x 0.
Problem 11 (1/490 (For 6th grade)). The natural number a is coprime
with 210. Dividing a by 210 we get the remainder r satisying 1 < r <
120. Prove that r is prime.
Problem 12 (2/490 (For 7th grade)). Given non-zero numbers a, b, c,
d satisfying
2 2 3 3 3
b = ac; c = bd; b + 27c + 8d 6= 0.
100
Show that
a 3 3 3
=3 a + 27b
3
+ 8c
3
d b + 27c + 8d
Problem 13 (3/490). Find all natural solutions of the equation 3xyz
5yz + 3x + 3z = 5.
Problem 14 (4/490). Given a half circle with the center O, the diameter
AB, and the radius OD perpendicular to AB. A point C is moving on the
_
arc BD. The line AC interdects OD at M. Prove that the circumcenter I
of the triangle DMC always belongs to a fixed line.
3 3
Problem 15 (5/490). Let x, y be real numbers such that x + y = 2.
2 2 9
Find the minimum value of the expression P = x + y + x + y
Problem 16 (1/491 (For 6th grade)). Consider all pairs of integers x, y
2017 2018
which are greater than 1 and satisfy x =y . Find the pair with
the smallest possible value for y.
Problem 17 (2/491 (For 7th grade)). Given an isosceles triangle ABC with the
0
apex A. Suppose that Ab = 108 , BC = a, AC = b. Outside ABC, construct the
[ 0
isosceles triangle ABD with the apex and 36 .
A BAD =
Find the perimeter of the triangle ABD in terms of a and b.
Problem 18 (3/491). Find positive integers n such that if the positive

integer a is a divisor of n then a + 2is also a divisor of n + 2.
Problem 19 (4/491). Given a triangle ABC inscribed the circle (O) with
diameter AC. Draw a line which is perpendicular to AC at A and inter-
sect BC at K. Choose a point T on the minor AB (T is different from A
and B). The line KT intersects (O) at the second point P. On the tangent
line to the circle (O) at the point T choose two points I and J such that
KIA and KAJ are isosceles triangles with the apex K. Show that
a) TdIP = TKdJ.
b) The circle (O) and the circumcircle of the triangle KPJ are
tangent to each other.
101
Problem 20 (5/491). Find integral solutions of the equation

2 2
y + 2y = 4x y + 8x + 7.
Problem 21 (1/492 (For 6th grade)). Let
1 1 1 1 1 1
M = 10 + 20 + 35 + 56 + 84 + 120 +...
1
a) Is the fraction 15400 a term of M? Why?
b) Compute the sum of the 8 first terms of M?
Problem 22 (2/492 (For 7th grade)). Given a triangle ABC with AB ¡ AC.
[
The angle bisector of BAC intersects the perpendicular bisector of BC
at M. Let H, K, and I respectively be the perpendicular projections of M
on AB, AC, and BC. Prove that H, I, K is collinear.
Problem 23 (3/492). Find integral solutions of the
3 3 2
equation x = 4y + x y + y + 13.
Problem 24 (4/492). Given a quadrilater ABCD inscribed in the circle
with the diameter AC. Let M be the point on AB such that AM = AD. The
lines DM and BC intersects at N, and the AN intersects the circle at K.
Let H be the perpendicular projection of D on AC. Assume that AB
intersects NH and CK P and Q. Show
1 1 1
that MP = MA + MQ
Problem 25 (5/492). Solve the system of equations
8 3 3
3x + 6x + 2 = 2y
<
3 3
3y + 6y + 2 = 2z
: 3 3
3z + 6z + 2 = 2x
Problem 26 (1/493 (For 6th grade)). Find 2018 numbers so that each
of them is the square of the sum of all remaining numbers.
Problem 27 (2/493 (For 7th grade)). Find the following sum
2 2 2 2
S = (1 + 2.3 + 3.5 + + 101.201) + (1 + 2 + 3 + + 100 ).
102
Problem 28 (3/493). Find all pairs of positive integers (m, n) such that
3 m
n 5n + 10 = 2 .
Problem 29 (4/493). Given a triangle ABC with BC = a , AC = b , AB = c
0 5
, and 3B + 2C = 180 . Prove that b + c 4 a.
b (5/493) Solve the system of equations
.
Problem 30 b p
x 2 y + x y+2=0
2
p p p =0
x + 8y + 4 x 8 y 4 xy
Problem 31 (1/494 (For 6th grade)). Find all natural numbers m, n and
m n m+n
primes p satisfying each of the following equalities 1) p + p = p ;
m n mn
2) p + p = p .
Problem 32 (2/494 (For 7th grade)). Given a triangle ABC. Let M, N,
and P respectively be the midpoints of AB, AC, and BC. Let O be the
intersection between CM and PN, I be the intersection between AO
and BC, and D be the intersection between MI and AC. Show that AI,
BD, MP are concurrent.
1 4p 2x
Problem 33 (3/494). Solve the equation x = 2.
2x + 1 2
x +1
Problem 34 (4/494). Given a right triangle ABC with the right angle A.
Let AH be the altitude. On the opposite ray of the ray HA pick an
arbitrary point D (D 6= H). Through D draw the line perpendicular to
BD. That line intersects AC at E. Let K be the perpendicular projection
of E on AH. Show that DK has a fixed length when D varies.
2 2
Problem 35 (5/494). Given real numbers x, y such that x + y = 1.
Find the minimum and maximum values of the expression
p p
T= 4 + 5x + 4 + 5y.
Problem 36 (1/495 (For 6th grade)). Show that it is impossible to write
92018
2 as a sum of n consecutive positive integers for any n 2 N, n 2.
Problem 37 (2/495 (For 7th grade)). Find the last twelve digits of the
1040
number 5 .
103
Problem 38 (3/495). Find integral solutions of the following systems of

equations
2 2 2 2
3a + 2ab + 3b = 12 ; b) (z 3)( + y ) 2xy = 0
a) 2 2 2 x
a +b =c
x+y=z
Problem 39 (4/495). Given a right triangle ABC with the right angle
[ [ 1 [
A. In the angle BAC draw the rays Ax, Ay such that CAx = 2 ABC;
1 [ [
BAy = 2 ACB .The ray Ax intersects the angle bisector of ACB at Q, the
d AQ
AK
ray Ay intersects BC at K. Compute the ratio
Problem 40 (5/495). Solve the equation 3 r3 x 2 2x + 2 + 2x = 5.

2x 1
Problem 41 (1/496 (For 6th grade)). Find 3-digit numbers so that each
of them is 9 times the sum of the squares of its digits.
Problem 42 (2/496 (For 7th grade)). Given a triangle ABC
0 0
with Bb = 45 , Cb = 30 . Outside ABC we choose D so that
[ [ 0
DBC = DCB = 15 . Prove that the triangle ABD is equilateral.

Problem 43 (3/496). Given 3 non-negative numbers x, y, and z. Show
2
that [(x + y)(y + z)(z + x)] xyz(2x + y + z)(2y + z + x)(2z + x + y).
Problem 44 (4/496). Given a semicircle (O; R) with the diameter AB
and the tangent Ax. On the ray Ax choose a point K so that AK = R.
The circle with center K and radius R intersects KB at I. The circle with
center B and radius BI intersects (O) at E. The ray BE meets Ax at C.
Show that the perpendicular bisector of BC is tangent to (O).

x
8 3 y 2 xy 1 yp 2 2
xy) = y5
1 2 2 2
> 2( 2 2 )+ (xy + 1)(3

x (1 x ) + x + + (x y) =xy
2 4
r !
<
>
104
Problem 46 (1/497 (For 6th grade)). Find all positive integers n such
2
that there exist prime numbers p, q satisfying 1! + 2! + 3! + + n! = p +
2
q + 5895 . Notice that n! = 1.2.3 . . . n.
Problem 47 (2/497 (For 7th grade)). Given an acute angle ABC with
the altitude AH, the median BM, the angle bisector CK. Show that if
HMK is an equilateral triangle then so is ABC.
n
Problem 48 (3/497). Suppose that n is a positive integer so that 3 +
n n n
7 is divisible by 11. Find the remainder in the divison of 2 + 17 +
n
2018 2 by 11.
Problem 49 (4/497). Given a circle (O) and suppose that BC is a fixed

chord of (O). Let A be a point moving on the major arc BC; and M, N
respectively the midpoints of AB and AC. Show that each of the
0 0
altitudes MM and N N of DAMNcontains some fixed point.
Problem 50 (5/497). Solve the equation
p
2 2
13x 5x 13 = (16x 11) 2x 3.
2016 2016 2016
Problem 51 (1/498 (For 6th grade)). Let A = 1 +2 +3 ++
2016 2016
2015 + 2016 . Show that A is not a perfect square.
Problem 52 (2/498 (For 7th grade)). Find all natural numbers k, m, n
so that 2.k! = m! 2.n! where n! = 1.2.3 . . . n; 0! = 1.
Problem 53 (3/498). Find integral solutions (x; y) of the equation
q p p
2
y y +2 . x+ 4 + 2x = 2.
Problem 54 (4/498). Given a triangle ABC with the angles satisfying Ab :

Bb : Cb = 8 : 3 : 1. Let AD be the angle bisector of the angle A (D is on
1 1 4
BC). Assume furthermore that BD2 + BC2 = 3 Compute the length of
AD.
p
3 3
Problem 55 (5/498). Solve the equation 2x + 20 = 9x 3 x 7.
105
2 For High School

5 3 2 3 2
2x 2x y x y + 10 x + y 5y = 0
2
(x + 1 ) y 5 =y 3x + x 2
p 2p
10 8p
p
Problem 57 (7/488). Prove that x0= cos 21 + cos 21 + cos 21 is a solu-
3 2
tion of the equation 4x + 2x 7x 5=0.
Problem 58 (8/488). In any triangle ABC, show that
cos (A B) + cos (B C) + cos (C A) 1 a + b +b + c +c + a
2 c a b
Problem 59 (9/488). Given 6 positive numbers a, b, c, x, y, z and
x y z
assume that x + y + z = 1. Show that ax + by + cz a .b .c .
p
Problem 60 (6/489). Solve the equation (1 2 sin x)(cos 2x + sin 2x) =
1
2
Problem 61 (7/489). Given the following system of equations
8
yzxy+ yz+ zx ) = a
> (+
zx(z + x y) = b
>
>
> x+y+z
>
>
<
(x + y ) =c
xy
> x + y + zz
>
>
>
>
>
:
where a, b, c are positive parameters.
a) Show that the system always has a positive solution.
b) Solve the system when a = 2, b = 5, c = 10.
Problem 62 (8/489). Suppose that a, b, c are the lengths of 3 sides of a
triangle a, b, c. Prove that ab + bc +
1 + 2r with R, r
ca
2 2 2 2 2 2
a +b b +c c +a 2 R
respectively are the inradius and the circumradius of the triangle.
106
Problem 63 (9/489). For any integer n which is greater than 3, let P =

p p p
3 2
p 2 p
60 120 (24n +24n ) n +n 12 8
3. 4 ... (n n )
n 1. Show that 3 P< 3.
Problem 64 (6/490). Given 3 positive numbers a, b, c satisfying abc = 1.
Prove that
1 1 1 1.
5 5 2 + 5 5 2 + 5 5 2
a +b +c b +c +a c +a +b
a cos p
q p
Problem 65 (7/490). Find all positive integers a, b such that 8+ 32+ p 768 =
b
Problem 66 (8/490). Given a triangle ABC. Let (K) be the circle pass-
ing through A, C and is tangent to AB and let (L) be the circle passing
through A, B and is tangent to AC. Assume that (K) intersects (L) at
another point D which is different from A. Assume that AK, AL respec-
tively intersect DB, DC at E, and F. Let M, N respectively be the mid-
points of BE, CF. Prove that A, M, N are colinear.
Problem 67 (9/490). Given real numbers a, b, c such that

2 2 2 2 2 2
2(a b + b c + c a ) a4 + b4 + c4.
Prove that
jb + c aj + jc + a bj + ja + b cj + ja + b + cj
= 2(jaj + jbj + jcj).
Problem 68 (6/491). Sove the system of equations
p p p
x2 + 4 + y2 + 4 + z2 + 4 = p 2(x + y + z)
log2x + log2y + log2z = 3
Problem 69 (7/491). Show that
v
p
lim
n !¥ u
1+2
s
1+3 1+
r
+ 1 + (n 1) 1 + n = 3.
u
q
t
107
Problem 70 (8/491). Letma, mb, mc be the lengths of the medians of

the triangle with the perimeter 2. Show that
p3 3
maxf1 ; 3 rg ma + mb + mc < p 2,
where r is the inradius of the triangle.
Problem 71 (9/491). Assume that a, b, c are three non-negative numbers
such that a + b + c = 3. Find the maximum value of the expression P =
p p p
3 3
a b + 1 + b c + 1 + c a3 + 1.
Problem 72 (8/492). Given a triangle ABC inscribed in the circle (O).
The tangent lines of (O) at B and C intersect at P. The line which goes
through A and is parallel to BP intersects BC at M. The line which
goes through A and is parallel to BC intersects BP at N. Suppose that I
is the intersections between AP and MN. Prove that four points B, I, O,
C lie on a circle.
Problem 73 (9/492). Given the equation
3 2
x + mx + n = 0. ()
Find m, n so the the equation (*) has three distinct non-zero real roots
u, v, t satisfying
4 4 4
u v t
3 2n + 3 2n
+3 = 3.
u v t 2n
Problem 74 (6/493). Given three positive numbers a, b, c satisfying a
+ b + c = 3. Show that
1 + 1 + 1 3
2 2 2 2 2 2 5
(a + b) + c (b + c) + a (c + a ) + b
y 1=p 2
x 1= 9 + 12y
9 + 12x 6x
4 6y
p4 2
108
0 0 0
Problem 76 (8/493). Given a right prism with equilateral bases ABC.A B C .
Let a be the angle between the line BC’ and the plane (A’BC). Prove that p
sin a 2 3 3.
Problem 77 (9/493). Given positive numbers a, b. Show that
b a
1 min(a, b) 2 a ln b + ln a
21
max(a, b) 2
1 max(a, b) 1 . 2
min(a, b)
Problem 78 (6/494). Find conditions on a, b besides a > b 1 so that

the system
2
x = (a y)(a + y + 2)
2
y = (b x)(b + x + 2)
has unique solution.
Problem 79 (7/494). Given positive real numbers a, b, c. Prove that
b(2a b) + c(2b c) + a(2c a) 3
a(b + c) b(c + a) c(a + b) 2
Problem 80 (8/494). Let A, B, and C denote the angles of an arbitrary
triangle ABC. Show that
sin A + sin B + sin C 3
tan B tan C tan A 2
Problem 81 (9/494). Find all pairs of positive integers (x; y) satisfying
3 3 2 2
x + y = x + 72xy + y .
Problem 82 (6/495). Given real numbers x, y, z such that xyz = 1,
show that
1 + xy 2 + 1 + zy 2 + 1 + xz 2 4 .
x y z 3
2018 2016
Problem 83 (7/495). Suppose that the polynomial P(x) = x ax +
a (a is a real parameter) has 2018 real solutions. Show that there exists at p
least one solution x0 of P(x) with jx0j 2.
109
Problem 84 (8/495). Given a pyramid S.ABCD with the base ABCD is a

parallelogram, and SA = SB = SC = a, AB = 2a, BC = 3a. Let SD = x p
(0 < x < a 14). Find x in terms of a so that the product AC.SD obtains its
maximum value.
Problem 85 (9/495). Suppose that x, y, z are non-negative numbers
sat-isfying x + y + z = 1. Find the maximum value of the expression
P = xy + yz + zx + 2 (M m)3
9
where M = maxfx, y, zg and m = minfx, y, zg.
Problem 86 (6/496). Find real solutions of the following system of
equa-tions
8 1 +1 +1 = 1
x y z xyz
> 8 3
< x + y + z = xyz + 27 (x + y + z) x
: Solve the equation log 2 = log 5 3.
Problem 87
(7/496)
x
>.
Problem 88 (8/496). Show that the sums of the squares of the
distances from the midpoints of the edges of a triangle to its
orthocenter and cir-cumcenter are equal.
Problem 89 (9/496). For any positive integer k, denote the square of
the sum of all of its digits by f1(k). Let fn+1(k) = f1( fn(k)) (n = 1, 2, . . .).
1990
Find f2018 2 . Solve the system of equations
Problem 90 .
(6/497)
8
p
2
p2
> 4x + 5 + 4y + 5 = 6 jxyj
< 1212
2x
8 + 4 +
> y + 2y8 x4 = 16
:
Problem 91 (7/497). Show that the following inequality holds for all
positive numbers a, b, c
2 2 2 1
a +b +c 2(ab + bc + ca)+
s
3 3 3 3 3 3
+ 2(a + b + c)(a b + b c + c a )
(a + b)(b + c)(c + a)
110
Problem 92 (8/497). Given triangle ABC with the incenter I, the

centroid G. There lines AG, BG, CG respectively intersect the
circumcircle of ABC at A2, B2.C2. Show that
GA2 + GB2 + GC2 IA + IB + IC.
Problem 93 (9/497). Given continuous functions f, g: [a, b] ! [a, b] such
that f (g(x)) = g( f (x)) 8x 2 [a, b]; where a, b are real numbers. Show
that the equation f (x) = g(x) has at least one solution.
8 = + 1 y
2 2 2+x 3
2(x + 4y 8) y
2
p p y +x 3+1+y
>
>
< p
>
> p
4
: 24 y +5=x
Problem 95 (7/498). Solve the equation
r
p log2x2 3
2 4
1 + 2log x + 4 3
log x
16 4 8
+ = 0.
2
log x 2 log x3 + 2
2
3 2
Problem 96 (8/498). Given a triangle ABC (AB < AC), inscribed in a
given circle (O), with the point A can be varied and the points B, C are
fixed, and two points A and O are always on the same side determined
0
by BC. A circle (O ) is internally tangent to (O) at T (T is outside the
triangle ABC) and is tangent to the sides AB, AC respectively at P, Q.
0
The line PQ intersects BC at R. The lines TB, TC meet again (O )
respectively at E, F (E 6= T, F 6= T). Prove that
a) The line EF is parallel to the line BC.
b) The line RT always passes through a fixed point when A varies.
Problem 97 (9/498). Let a, b, c be positive numbers such that
a + b + c + 2 = 2abc. Prove that
a+2 + b+2 + c+2 2.
p 2 p 2 p 2
6(a + 2) 6(b + 2) 6(c + 2)
111
3 Towards Mathematical Olympiad

Problem 98 (10/488). Given an infinite sequence of positive integers
a1 < a2 < < an < such that ai+1 ai 8 for all i = 1, 3, . . . .
For each n, letsn = a1 + a2 + + an Show that for each n, there are at least
two complete squares inside the half-open interval [s n, sn+1).
Problem 99 (11/488). Given two positive sequences (an)n 0 and (bn)n 0
p
which are determined as follows a0 = 3, b0 = 2 and
8 +b = a
an n 1 2 n+1
+a
< 1 n+1
2
an + 1 = bn
:
for al n = 0, 1, 2, . . . . Show that they converges and find the limits.
Problem 100 (12/488). Given a triangle ABC with AB 6= AC .A circle
(O) passing through B, C intersects the line segments AB and AC at M
and N respectively. Let P be the intersection between BN and CM. Let Q
_
be the midpoint of the arc BC which does not contain M, N . Let K be
the incenter of PBC. Show that KQ always goes through a fixed point
when (O) varies.
m n
Problem 101 (10/489). Find natural numbers n so that 4 + 2 + 29
can-not be a perfect square for any natural number m.
Problem 102 (11/489). The sequence (an) is given as follows:
2
a1 = 1 ; an+1 = (an 1) , n = 1, 2, . . . .
2 2 an
a) Find lim an . p
b) Show that a1 + a2 + + an 1 2 for n = 1, 2, . . . .
n 2
Problem 103 (12/489). Given a triangle ABC inscribed in a cirle (O). A
point P varies on (O) but is different from A, B, and C. Choose M, N
respectively on PB, PC so that AMPN is a parallelogram.
a) Prove that there exists a fixed point which is equidistant from M
and N.
112
b) Prove that the Euler cirle of AMN always goes through a fixed
point.
Problem 104 (10/490). Find all triples of positive integers (a, b, c)
a b c
such that 2 + 5 = 7 .
Problem 105 (11/490). The sequence (un) is determined as follows
u1 = 14, u2 = 20, u3 = 32, un+2 = 4un+1 8un + 8un 1
2018
with n 2. Show that u2018 = 5.2 .
Problem 106 (12/489). Given a triangle ABC with (O) is the

circumcircle and I is the incenter. Let D be the second intersection of
AI and (O). Let E be the intersection between BC and the line passing
through I and perpendicular to AI. Assume that K, L respectively are
the intersections between BC, DE and the line passing through I and
perpendicular to OI. Prove that K I = KL.
2n
Problem 107 (10/491). For every n 2 N ,let Fn = 2 + 1. For eachn 2
n+1
N, let q be a prime divisor of Fn, show that 2 j q 1. Furthermore, if n
n+2
2,show even more that 2 j q 1.
Problem 108 (11/491). Find all functions f : R ! R satisfying f
(x) f (y) + f (xy) + f (x) + f (y) = f (x + y) + 2xy
for all x, y 2 R.
Problem 109 (12/491). Given a convex hexagon ABCDEF
circumscrib-ing a circle (O). Assume that O is the circumcenter of the
triangle ACE. Prove that the circumcircles of the triangles OAD, OBE
and OCF has another common point besides O.
Problem 110 (6/492). Solve the inequality
p
3 2
x + 6x + 9x x + 4 2.
Problem 111 (7/492). For a, b, c are positive numbers with the
product is equal to 1, find the minimum value of the expression
1 1 1
P= + + .
a2017 + a2015 + 1 b2017 + b2015 + 1 c2017 + c2015 + 1
113
Problem 112 (10/492). Let p be an odd prime and a1, a2, . . . , ap is an

arithmetic progression with the common difference d which is not di-
visible by p. Prove that
p
.. 2
Õ(ai + a1a2 . . . ap) .p .
i=1
Problem 113 (11/492). Find all functions f : R ! R such that

2018 2018
f (x + f (y)) = f (x + y ) + f (y f (y)), 8x, y 2 R.
Problem 114 (12/492). Given an acute triangle ABC with the altitudes
BE, CF. Let ST be a chord of the circumcircle of AEF. Two circles
which go through S, T is tangent to BC respectively at P and Q. Prove
that the intersection between PE, QF lies on the circumcircle of AEF.
Problem 115 (10/493, Daniel Sitaru and Leonard Giugiuc (Romania)).
Find the maximal positive number k so that the following inequality
2 2 2
a + b + c + k(a + b + c) 3 + k(ab + bc + ca)
holds true for all positive numbers a, b, c satisfying abc = 1.
Problem 116 (11/493). Given the sequence (xn)
(
x2 = x3 = 1
3 2 3
(n + 1)(n 2)xn+1 = (n n n)xn (n 1) xn 1 8n 3
Find all indices n so that xnis an integer.
Problem 117 (12/493). Given a cyclic quadrilateral ABCD. Let K be
the intersection between AC and BD. Let M, N, P, and Q respectively
be the perpendicular projection of K on AB, BC, CD, and DA. And
then, let X, Y, Z, and T respectively be the perpendicular projection of
K on MN, NP, PQ, QM. Prove that AXCZ and BYDT have equal areas.
Problem 118 (10/494). For any integer n, show that
p p !n p p !n 2
an = 3 + 5 7 + 3 5 + 3 5 7 3 5 +
10 2 10 2 5
is a perfect square.
114
Problem 119 (11/494). A class has n students attending n 1 clubs. Show

that we can choose a group of at least two students so that, for each
club, there are an even number of students in that group attend it.
Problem 120 (12/494). Given a square ABCD and P is an arbitrary
point on the side AB. Let (I1), (I2)respectively be the incircles of ADP,
CBP. Assume that DI1, CI2intersect AB respectively at E, F. The line
through E which is parallel to AC intersects BD at M and the line
through F which is parallel to BD intersects AC at N. Show that MN is
a common tangent to (I1) and (I2).
p ],n
Problem 121 (10/495). Consider the sequence (an): an = n + [ 3 n
p
positive integers ( [ 3 n ] is the integral part of p3 n ). Suppose there ex-
ists a positive integer k such that the terms ak; ak+1; . . . ; ak+p are p + 1
consecutive natural numbers with p = 6015 2006 + 1. Show that
9 7
k > 8.10 + 6.10 .
Problem 122 (11/495). Find the least number k so that in any subset of k
elements of 1, 2, . . . , 25 we can always find at least a Pythagorean triple.
Problem 123 (12/495). Let ABC be a triangle inscribed in a circle (O).
Suppose that AH is the altitude and the line AO intersects BC at D. Let
K be the second intersection of the circumcircle of ADC and the
circum-circle of AHB. Suppose that the circumcircle of KHD intersects
(O) at M and N. Let X be the intersection of MN and BC.
Show that X A = XK.
Problem 124 (10/496). The sequence (xn) (n 2 N ) is defined as follows
1 xn
x1 = 0, xn+1 = , 8n 2 N .
4
Show that the limit of the sequence exists and find it.
Problem 125 (11/496). Find all functions f : R ! R such that f
4
(x) f (y) 9 xy = f (x + y), 8x, y 2 R.
Problem 126 (12/496). Given a quadrilateral ABCD inscribed in a circle
(O). AC intersects BD at E, AD intersects BC at F. Let (O1) be the circle
which is tangent to the rays EA, EB and is internally tangent to (O); (O2)
115
the circle which is tangent to the rays FA, FB and is externally tangent to
(O) at a point on the arc AB which does not contain C, D. Prove that the
intersection of two external tangents to (O1) and (O2) lies on the circle
(O).
Problem 127 (10/497). Find all pairs of integers (m, n) so that
5 5
both m n + 2019 and mn + 2019 are cubes of integers.
Problem 128 (11/497). Let S = f1, 2, 3, . . . , 2019 g and suppose that
S1, S2, . . . , S410 are subsets of S such that for every i 2 S, there exist
exactly 40 of them containing i. Prove that there exist different indices
j, k, l 2 f 1, 2, . . . , 410 g such that Sj \ Sk \ Sl 1.
Problem 129 (12/497). Given a cyclic quadrilateral ABCD with E is the

intersection between AC and BD. The circumcircles of EAD and EBC
intersect at another point F (besides E). The perpendicular bisectors of
BD, AC respectively intersect FB, FA at K, L. Show that KL goes
through the midpoint of AB.
Problem 130 (10/498). a) Given rational numbers a1, a2, . . . , an (n 2
+ m m m
Z ). Show that if, for any positive integer m, a 1 + a2 + + a n is an
interger then a1, a2, . . . , an are integers.
b) Is the above conclusion still correct if we only assume
that a1, a2, . . . , an are real?
Problem 131 (11/498). people come and sit on the bench, one by one.
The first person can choose any seat. The next people firstly try to
avoid to sit next to the previous ones. If they cannot avoid, then they
can choose any empty seat. How many such arrangements are there?
Problem 132 (12/498). Given a triangle ABC. Let ma, mb, mc respectively
be the lengths of the medians corresponding to the sides BC = a,
AC = b, AB = c. Show that
1 1 1 p
(ab + bc + ca) + + 2 3(m + m +
a b c a b
m ).
c
When does the equality hold?
116
SELECTED VIA MAIL COMPETITION QUESTIONS FROM

CHILDREN’S FUN MATHS JOURNAL
Nguyen Ngoc Han
Children’s Fun Maths Journal
1 1 1
Problem 1 (Nguyen Ngoc Hung, CFMJ 1(187)). Let A = 1! + 2! + 3! +
1 2017
+ 2018! . Compare A with 1152. (Notation: n! = 1.2.3 . . . n).
Problem 2 (Ta Thap, 2(187)). Two boxs contain 100 marbles and 123
marbles respectively. Hong and Ha play taking arbitrary number of
marbles in one box. The winner is the one who gets the last marble.
Hong takes the marbles first. Show the way helping Hong to win the
game.
Problem 3 (Nguyen Duc Tan, 3(187)). Find prime numbers a, b, c,
2 2
such that a + 5ab + b = 7c.
Problem 4 (Luu ly Tuong, CFMJ 4(187)). Let ABC be an equilateral
triangle. M is inside triangle ABC such that MA : MB : MC = 3 : 4 : 5.
AH
Draw AH perpendicular to BM at H. Find the ratio CM .
Problem 5 (Lai Quang Tho, CFMJ 1(188+189)). Find the integer x
x 1
such that 9 x + 7 is the square of a fraction.
Problem 6 (Truong Quang An, CFMJ 2(188+189)). Find digits x, y, z,
2
t, u such that (xy + ztu) = xyztu.
Problem 7 (Nguyen Duc Tan, CFMJ 3(188+189)). Find the smallest
pos-itive integer equal to 3 times the cube of a whole number and 4
times the exponent of fourth of another whole number.
Problem 8 (Thai Nhat Phuong, CFMJ 4(188+189)). Let ABC and ADC
[[[[
be triangles such that BAC, BCA, DAC, DCA are acute angles and B,
D are on the opposite side of AC.
2 2 2 2
Given that AB + CD = AD + BC , prove AC ? BD.
117
Problem 9 (Nguyen Ngoc Hung, CFMJ 5(188+189)). Find all pairs of

2 2
positive integers (a; b) such that 9a b 5a + 5b be a square number and
2019 2018
a = 2020 b .
Problem 10 (Lai Truong Sinh, CFMJ 6(188+189)). On Oxy co-

ordinate, given lines with equation y = (m 1)x + 2m 3. In these lines,
we choose 3 arbitrary lines. In these angles formed by two lines of
these 3 lines which are no common interior points, we choose the
angle with the greatest measure x. Find the smallest value of x.
Problem 11 (Vu Dinh Hoa, CFMJ 7(188+189)). A bench has 4 seats.
4 people sit in this bench. The first person sits randomly. The next
three people choose the seats with no adjacent people. If they cannot
find the seat they want, they will choose the vacant seats. How many
ways to arrange the seats for these four people are there?
Problem 12 (Le Tran Quoc Canh, CFMJ 8(188+189)). Let ABC be a
right-angled triangle at C with AC = b, CB = a; AB = c. AE, BF are
medians of triangle ABC. AE = m; BF = n. Let r be a radius of the
2
r
inscribe circle of triangle ABC. Find the greatest value of A = m2 + n2 .

Problem 13 (Le Son Tung, CFMJ 1(190)). Find all prime numbers a,
b, c such that ac b + c and ca + b are also prime numbers.
Problem 14 (Vo Xuan Minh, CFMJ 2(190)). Find prime numbers a, b, c
such that a < b < c and b a, c b, c b + a are also prime numbers.
Problem 15 (Cao MInh Quang, CFMJ 3(190)). Find all positive integers
m, n such that
a) 3m n! = 1;
b) 3m n! = 2.
Problem 16 (Tran Quang Hung, CFMJ 4(190)). Let ABC be an isosce-
les triangle with AB = AC. P and Q are inside triangle ABC such that
\APB = \AQC and AP = AQ. Prove that PQ k BC .
Problem 17 (Nguyen NGoc Hung, CFMJ 5(190)). Given the sum
A= 1 + 1 + 1 + + 1 .
3 3 3 3
2 +3 3 +4 4 +5 2018 + 2019
118
1
Compare A and 6.
Problem 18 (Mai Van Nam, CFMJ 6(190)). Solve the following equa-
tion
p
3 3 2
2x + 2 = x + 9x + 26x + 28.
Problem 19 (Vu Dinh Hoa, CFMJ 7(190)). Find all natural numbers n
3 such that we can fill in an n n grid with real numbers satisfying the
following conditions simultaneously:
1) The sum of all numbers in an arbitrary 2 2 grid is positive.
2) The sum of all numbers in an arbitrary 3 3 grid is negative.
Problem 20 (Nguyen Tan Ngoc, CFMJ 8(190)). Let ABCD be a
o
quadri-lateral inscribed in circle (O) such that \BCD < 90 , the diagonal
AC intersects BD at the midpoint M of BD. Draw line b through D,
perpen-dicular to DC that cuts the perpendicular bisector of BD at E.
AB cuts CD at F. Show that BC is perpendicular to EF.
Problem 21 (Nguyen Ha Ha Uyen, CFMJ 1(191)). Given that
1 2 3 99 100
M = 1 + 2 + 3 + + 99 + 100 . Show
that M has 201 digits and find the first 2 digits of M.
Problem 22 (Cao Ngoc Toan, CFMJ 2(191)). Given that
12 3 4 2019
S = 2018 1+ 2+ 3++ 2018 .
Prove that S is not a natural number.

Problem 23 (Le Son Tung, CFMJ 3(191). Find all the 2-digit numbers
having the form ab such that ab, ba, (a + 1)b, (b + 1)a are all 2-digit
prime numbers.
Problem 24 (Tran Quang Hung, CFMJ 4(191). Given acute triangle
ABC, we construct two right-angled isosceles triangles ACE with
hypotenuse AC and ABF with hypotenuse AB (both of them are outside
triangle ABC). Then, we construct another right-angled isosceles DEF
with hy-potenuse EF such that A, D and are on the same plane side of
EF. Show that AD is perpendicular to BC.
119
Problem 25 (Nguyen Ngoc Hung, CFMJ 5(191). Given real numbers x

and y satisfying
3 3
x + y = 1948
(x + y)(x + 1)(y + 1) = 2017.
Find the sum x + y.
Problem 26 (Vo Quoc Ba Can, CFMJ 6(191). Given real numbers a and
4 3 2
b such that the equation x + ax + 3x + 2bx + 2 = 0 has at least one
2 2
real root. Prove that a + 2b 12.
Problem 27 (Thai Nhat Phuong, CFMJ 7(191). Solve the following
sys-tem of equations
x+y+z=3
p p p p
x y+y z+z x= xyz + 2.
Problem 28 (Doan Van Truc, CFMJ 8(191). Given an acute triangle
ABC (AB < AC). D and E are on side AB and side AC respectively so
that BD = CE. Prove that the centroid G of triangle ADE would move
along a fixed line when points D and E move along the segments AB
and AC, respectively.
Problem 29 (Truong Quang An, CFMJ 1(192+193). Let S(n) be the
sum of the digits of positive integer n. Find the smallest value of
positive integer n so that S(n).S(n + 1) = 87.
Problem 30 (Nguyen Ngoc Hung, CFMJ 2(192+193). Find all possible
2 2
doublets of natural numbers (a; b) satisfying 1000 a + b = 1001 b +
a.
Problem 31 (Nguyen Duc Tan, CFMJ 3(192+193). A box contains 219
marbles, Hong and Ha take turns to pick up the marbles which are ex-
ponents of 2 (20; 21; 22; . . . ). The person to pick up the last marble is
the winner.
If Hong is the first one to pick, give a scenario in which Hong is the
winner with the fewest number of turns.
120
Problem 32 (Tran Quang Hung, CFMJ 4(192+193). Triangle ABC, we

construct an equilateral triangle APB and an isosceles triangle ACE
o
(EC = EA) outside DABC such that \CEA = 120 . Then, we construct
o
isosceles triangle BCD (DB = DC) with \BDC = 120 and A and D are
on the same side of BC.
Finally, we construct another isosceles triangle DEF (DE = DF)
o
with \EDF = 120 and F and B are on the same side of DE.
Show that PF = CE.
Problem 33 (Lai Quang Tho, CFMJ 5(192+193). Solve the following
system of equations
2 3 2
2 + 2xy + 2xy = 3
x y y3 + xy = 1.
x2
Where x < 0, y > 0.
Problem 34 (Vo Quoc Ba Can, CFMJ 6(192+193). Given three non-zero
real numbers a, b and c. Show that
2 2 2
a bc b ca c ab
2 2 2 + 2 2 2 + 2 2 2 0.
a + 2b + 3c b + 2c + 3a c + 2a + 3b
Problem 35 (Vu Dinh Hoa, CFMJ 7(192+193). How many natural 9-
digit numbers are there such that its decimal representation does not
contain any of these digits 1, 2, 3, 4, 5 and 6?
Problem 36 (Vu Cong Minh, CFMJ 8(192+193). Isosceles triangle
o
ABC (CA = CB) with \C = 15 . Suppose that there exists point D inside
o
D
ABC such thatp \ADB = 105 and AD = 2BD.
Show that 5.AD.BC = 2CD.AB.
Problem 37 (Ta Thap, CFMJ 1(194). There are two types of trucks: 4-
tonne trucks and 11-tonne trucks. How many of each types of trucks
would be needed to transport 58 tonnes of goods if they are all fully
loaded?
Problem 38 (Nguyen Ngoc Hung, CFMJ 2(194). Given that
A = 33 53 + 73 93 + 113 133 + 153 173 + + 1993 .
1 3 6 10 15 21 28 36 4950
Compare A with 814.
121
Problem 39 (Thai Nhat Phuong, CFMJ 3(194). Five numbers 1, 2, 3,

4, 5 are arbitrarily grouped into 2 smaller groups. Prove that there
always exists one group in which the difference between two members
of the group is equal to another number in that group.
Problem 40 (Nguyen Ba Dang, CFMJ 4(194). Given triangles ABC
and BCD with common side BC. A and D are on different sides of BC
and it is known that:
o o o o
\ABC = 36 , \CBD = 30 , \BAD = 81 , \CAD = 27 .
Find the measure of angles in triangle ACD.
Problem 41 (Nguyen Tien Lam, CFMJ 5(194). A positive integer n is
named interesting number if there exists other positive integers x, y, z,
t satisfying:
2 2
x +y
n= .
z2 + t2
a) Prove that there are an infinite number of interesting numbers.
b) Is 2019 an interesting number? Give an explanation.
Problem 42 (Luu Ly Tuong, CFMJ 6(194). Solve the equation
p
2
(x + 3 ) x x + 48 = x 24.
Problem 43 (Luu Ly Tuong, CFMJ 7(194). n is a positive integer (n 3).
a) Find the greatest positive integer k(n) so that among n arbitrary

distinct positive integers not exceeding k(n), there always exists 3
num-bers (not necessarily distinct) in which the sum of two of them is
equal to the other one.
b) Find the greatest positive integer h(n) so that among n arbitrary
distinct positive integers not exceeding h(n), there always exists 3 dis-
tinct numbers in which the sum of two of them is equal to the other one.
Problem 44 (Huynh Thanh Tam, CFMJ 8(194). Two circles of different
0
sizes with centers O and O intersect at points A and B. We construct
a common external tangent passing the two circles at C and D respec-
tively so that B is inside triangle ACD, then, draw AM perpendicular to
122
CD (where M is the point of intersection). From B, we construct a line

perpendicular to AB meeting AM at H. Prove that H is the orthocenter
of triangle ACD.
123
ON THE SHAPIRO’S CYCLIC INEQUALITY AND SOME RELATIVE

ISSUES
Le Quy Thuong, Nguyen Minh Tuan
In this paper we prove an interesting extension of the Shapiro’s cyclic in-

equality for four, and five variables, and formulate a generalization of the
well-known Shapiro’s cyclic inequality. The method used in the proofs of
the theorems in the paper concerns the positive quadratic forms. Some
stud-ied open problems which can be seen as a real generalization of the
Shapiro’s cyclic inequality are given at the end of Section 2. Moreover,
some different proofs of P(3) are provided in Section 3, and the relative
inequalities are given in Sections 4, 5.
1 Introduction
In 1954 Harold Seymour Shapiro proposed the inequality for a

cyclic sum in n variables as follows
x + x2 + + xn 1 + xn n , (1)
1
x2 + x3 x3 + x 4 xn + x1 x1 + x 2 2
where xi 0, xi + xi+1 > 0 and xi+n = xi for i 2 N. Although the prob-lem (1) was settled in
1989 by Troesch [13], the history of long year proofs of this inequality was interesting, and the certain
problems remain (see [1, 2, 3, 4, 5, 6, 7, 13]). Motivated by the directions of generalizations and proofs of (1),
we consider the following inequality
P(n, p, q) := x1 + x2 + + x + xn
n 1
px2 + qx3 px3 + qx4 pxn + qx1 px1 + qx2

n , (2)
p+q
132
where p, q 0 and p + q > 0. It is clear that the inequality (2) is true for n
= 3. Indeed, by the Cauchy inequality, we have
r qx (px + qx ) + r
(x1 + x2 + x3)2 = px2 + qx3 1 2 3 px3 + qx1 q x2(px3 + qx1)
x1 2 x2
r q
+ px1 + qx2 x3( px1 + qx2) P( 3, p, q)( p + q)( x1x2 + x 2x 3 + x3x1).
x3
It follows that
2
P(3, p, q) (x1 + x2 + x3) 3 .
(p + q)(x1x2 + x2x3 + x3x1) p+q
Obviously, (2) is true for every n 4 if p = 0 or q = 0.
In this note, by studying the inequality (2) in the case n = 4, we
show that it is true when p q, and false when p < q. Moreover, we give
a sufficient condition of p, q under which the inequality (2) is true in the
case n = 5. It is worth saying that if p < q, then the inequality (2) is
false for every even n 4. Two open questions are discussed at the end
of the next section. For short, for any n 3 and p = q = 1, let P(n) stand
simply for P(n, 1, 1) (see [8, 9, 11, 12, 14]).
2 Main result
Without loss generality of (2) we assume that p + q = 1. The
inequal-ity (2) for n = 4 now is of the form
P(4, p, q) = x1 + x2 + x3 + x4 4. (3)
px2 + qx3 px3 + qx4 px4 + qx1 px1 + qx2
Theorem 15. The inequality (3) is true for p q, and it is false for p < q.
Proof. By the Cauchy inequality we have
2 3 2 3 4 3 4 1 4 1 2
(x1+x2 + x3 + x4)
Hence 1
(px 2 (px (px (px
P(4, p, q)x + qx) + x + qx ) + x + qx ) + x + qx ) .
2
(x1 + x2 + x3 + x4)
P(4, p, q) .
px1x2 + 2qx1x3 + px1x4 + px2x3 + 2qx2x4 + px3x4
133
It is an equality if and only if

px2 + qx3 = px3 + qx4 = px4 + qx1 = px1 + qx2. (4)
Consider the following quadratic form

2
w(x1, x2, x3, x4) = (x1 + x2 + x3 + x4)
4(px1x2 + 2qx1x3 + px1x4 + px2x3 + 2qx2x4 + px3x4).
By a simple calculation we obtain the canonical quadratic form w as

follows
2 2 2
w(t1, t2 , t3, t4) = t1 + 4pqt2 + 4q(2p 1) t3 , (5)
p
where
t1 = x1 + (1 2p)x2 + (1 4q)x3 + (1 2p)x4,
1 2p q
t2 = x2 + p x3 p x 4,
t3 = x3 x 4.
1
It is easily seen that if p q, i.e. p , then w 0 for all t1, t2, t3 2 R. 2
Now let us consider the cases when w vanishes. This depends con-
1
siderably on the comparison of p with q. If p = q, i.e. p = 2 , then the
quadratic form w attains 0 at t1 = x1 x3 = 0 and t2 = x2 x4 = 0. By
(4) we assert that P(4, p, q) = 4 whenever x1 = x3 and x2 = x4. Also, if p
1
> 2 , then w vanishes if and only if
t1 = x1 + (1 2p)x2 + (1 4q)x3 + (1 2p)x4 = 0,
1 2p q
t2 = x2 + p x3 p x4 = 0,
t3 = x3 x4 = 0.
Combining these facts with (4) we conclude that P(4, p, q) = 4 when x1
= x2 = x3 = x4.
134
Now we give a counter-example to the inequality (3) in the case p < q,

i.e. p < 12 . Let x1 = x3 = a, x2 = x4 = b, and a 6= b. We shall prove that
+ + + = 2
pb + qa pa + qb pb + qa pa + qb pb + qa + pa + qb < 4.
a b a b a b
(6)
It is obvious that
2 2 2 2 q)ab > 0 , p(1 2
(6) , p(2q 1)(a + b ) + 2(p + q 2p)(a b) > 0.
1
The last inequality is evident as a 6= b and p < , so (6) follows. 2
Remark 2. Let A denote the matrix of the quadratic form w in the

canonical base of the real vector space R4. Namely,
0 1 1 2p 1 4q 1 1
2p
A = B 1 2p 1 1 2 p 1 4q C.
1 4q 1 B 1 2p
2p 1 1 2p
1 4q 1 2p 1 C
@ A
Let D1, D2, D3 and D4 be the principal minors of orders 1, 2, 3 and 4

respec-tively of A. By direct calculation we obtain
2
D1 = 1, D2 = 4pq, D3 = 16q (2p 1), D4 = 0.
Then w is positive if and only if Di 0 for every i = 1, 2, 3, 4. We find the
first part of Theorem 15.
Thanks to the idea of using positive quadratic form we now study

the inequality (2) in the case n = 5. It is sufficient to consider the case
p + q = 1. By the Cauchy inequality we reduce our work to the
following inequality
5
j(x1, . . . , x5) = å xi2 + (2 5p)x1x2 + (2 5q)x1x3 + (2 5q)x1x4
i=1
+ (2 5p)x1x5 + (2 5p)x2x3 + (2 5q)x2x4 + (2 5q)x2x5 + (2 5p)x3x4
+ (2 5q)x3x5 + (2 5p)x4x5 0.
135
The matrix of j in an appropriate system of basic vectors is of the form
0 1
1 2 5p 2 2 5p 2 5q 2 5q
B 2 2 5p 2 5q 2 5q 2 5p C
B= 2 5q 2 5p 2 2 5p 2 5q
2
B 2 5q 2 5q 2 5p 2 2 5p C
B
B 2 5p 2 5q 2 5q 2 5p 2
C C
@ A
which has the principal minors

5p(4 5p) 25q(5pq 1) 2
125(1 5pq)
D1 =1, D2 = , D3 = , D4 = , D5=0.
4 4 16
This implies that the necessary and sufficient condition for the
positivity of the quadratic form j is
p p
5 5p 5 + 5.
10 10
We thus obtain a sufficient condition under which the inequality (2)
holds for n = 5.
p p
5 5 5+ 5
Theorem 16. If 10 p 10 , then (2) is true for n = 5.
Remark 3. Consider the inequality (2) in the case n 4, n is even, and p

< q. According to the proof of the second part of Theorem 15 this
inequality is false. Indeed, we choose x1 = x3 = = a, x2 = x4 = = b. By
n
the above counter-example we conclude P(n, p, q) < p+ q .
Open Problem. (a) Find pairs of non-negative numbers p, q so that

the in-equality (2) is true for every n 4.
(b) For certain n 5, which sufficient condition of the pair p, q so that
the inequality (2) is true.
We observe that if p = 0 or q = 0, then it is the case for every n 4,

and Theorem 16 is the detailed answer for n = 5.
136
3 Proof of P(3)
In this section we present some different proofs for P(3).
Problem 1 (Nesbitt’s inequality). Let a, b, c > 0 be given. Prove the

inequal-ity
a + b + c 3 . (7)
b+c a+c a+b 2
Proof 1. Consider the following expressions
S= a + b + c ,
b+c a+c a+b
M= b + c + a ,
b+c a+c a+b
N= c + a + b .
b+c a+c a+b
Evidently, M + N = 3. Using the AM-GM inequality we have
M+S=a + b + b + c + c + a 3,
b+c c+a a+b
N+S=c + a + a + b + b + c 3.
b+c c+a a+b
Hence,
M+N+2S 6.
This implies
2S 3.
The equality occurs if and only if a = b = c.

137
Proof 2. By calculating and applying the AM-GM inequality, we have
b+c + a + c+ a + b= b + c+ 1 + c + a+1 + a + b+1 3
a b c a b c
= (a + b + c) b + c + c + a+ a + b 3
1 1 1
+
= 2 (b + c ) + ( c + a ) + ( a + b ) b + c c+a + a + b 3
1 1 1 1
3
r3
3 (b + c)(c + a)(a + b) 3 c+ a
+
a+ b
q
1 1
3
= 9 3= ,
2 2
as desiered. The equality occurs if and only if a = b = c.
Proof 3. Without loss of generality, we can assume that a b c. Then

1 1 1 .
b+c c+a a+b
By the Chebyshev’s inequality, we have
b + c +a + c +
a+b 3 (a + b + c) b + c + c + a + a + b
a b c 1 1 1 1
b+c +
= 6 ((b + c) + (c + a) + (a + b)) c+a + a + b
1 1 1 1
1 3
6.9 = 2.
The inequality is proved. The equality occurs if and only if a = b = c.
Proof 4. The left-hand side of (7) can be seen as a homogeneous

function of variables as follows
x y z
f (x, y, z) = y + z+ z + x+ x + y.
It is easily seen that f is a homogeneous function of zero-degree,
0
i.e. f (tx, ty, tz) = t f (x, y, z) = f (x, y, z),
138
for every x, y, z and for all t > 0. Then, one can assume that a + b + c
= 1. It follows 0 < a, b, c < 1. The inequality now can be seen as the
following type
a + b + c 3.
1 a 1 b 1 c 2
Consider the function x
f (x) = .
1 x
We have 2
(x) = > 0,
f0 3
(x 1)
for all x 2 (0, 1). We deduce that f is a convex function defined on (0,
1). By the Jensen’s inequality, we have
f (a) + f (b) + f (c) 3 f a+b+c = 3 f (1 3
)= .
3 3 2
Proof 5. Put
x = b + c,
y = c + a,
z = a + b.
It follows
a=y+z x,
2
b=z+x y,
2
c = x + y z.
2
The inequality is now equivalent to the following ones:
y + z x z + x y x + y z 3,
+ +
2x 2y 2z 2
y+z x z+x y x+y z 3.
+ +
x y z
139
Put
x
u= y ,
y
v= z,
The last inequality becomes

1 1 1
u +w 1+ v + u 1 + w + v 1 3.
Equivalently,
1 1 1
( u + u) + ( v + v) + (w + w) 6, where
u, v, w are positive numbers. This inequality holds as
1
t+ t 2,
for all t > 0. Thus, P(3) is proved. The equality occurs if and only if
a = b = c.
Proof 6. Put
a + b + c = k.
It follows 0 < a, b, c < k. We shall prove the following inequality
a b c
+ + 1
T =k a k b k c .
3 2
Indeed, consider the function
t
f (t) = k t.
As Proof 4 shows that f is a convex function on interval (0, k). By the
Jensen’s inequality, we have
a+b+c k 1
3 3
T =2k =2.
k a+b+c
3 3
140
Proof 7. By the Cauchy-Schwarz inequality, we have

2 p
(a + b + c )
pa pc 2
q b q q #
= " p b + c . a(b + c) + p c + a . b(c + a) + p a + b . c(a + b)
2 b+c + a + c + a + b (ab + bc + ca).
a b c
Hence, a b c 2 3
(a + b + c )
.
+ +
b+c a+c a+b 2(ab + bc + ca) 2
Proof 8. Put
a b c
f (a, b, c) = b + c+ a + c+ a + b,
t = a + b.
2
We see that
f (a, b, c) = a2 + b2 + c(a + b) c
+
2
ab + c + c(a + b) a + b
2 + c = f (t, t, c).
2t + 2tc
2 2 2t
t + c + 2ct
This implies
0 0 0 3
f (a, b, c) f (t , t , t ) = 2,
where
a + b + c
t0 = .
3
141
Proof 9. By the Cauchy-Schwarz inequality, we have

a b c 2 2 2
a b c
+
b+c a+c a+b
+ =ab + ac +bc + ba +ca + cb
(a + b + c)2
2(ab + bc + ac) .
It is easy to prove that
2
(a + b + c ) 3(ab + bc + ac).
Then,
a + b + c 3.
b+c a+c a+b 2
Proof 10. It is easy to prove the following inequality for every x > 0
x 8x 1 .
4x + 1
By this, we have
a 8 1 8a b c
a
b+
c
4 = .
b+ a +1 4(a + b + c)
c b+
c
Similarly,
b 8b 1 8b c a
c+
a
c+a 4b + 1 = 4(a + b + c) .
c+
a
c 8c 1 8c a b
a+
b
4c = .
a+b a+b
+1 4(a + b + c)
We deduce
a b c 6(a + b + c) 3
+ + = .
b+c a+c a+b 4(a + b + c) 2
142
Proof 11. We shall prove that

3
a 3a 2 . (8)
3 3 3
b+c 2(a + b + c 2 )
2 2
Equivalently,
3 3 3 1
2(a 2 + b 2 + c 2 ) 3a 2 (b + c).
Indeed, by the AM-GM inequality, we have
3 3 3 1
a2 + b 2 + b 2 3a 2 b
3 3 3 1
a +c +c 2 3a c.
2 2 2
Hence, the above inequality holds. Similarly, we have two more

inequal-ities:
3
b 3b 2
(9)
3 3 3
c + a 2(a + b + c 2 ) 2 2
3
c 3c 2 . (10)
3 3 3
a + b 2(a 2 + b + c 2 ) 2
From the inequalities (8), (9), (10) we derive

3 3 3
a b c 3(a 2 + b2 + c 2 ) 3
+ + 3 = .
3 3
b+c a+c a+b

2
+b +c ) 2 2 2
2(a
Proof 12. The inequality is of the following form
2a(a + b)(a + c) + 2b(b + c)(b + a) + 2c(c + a)(c + b)

3(a + b)(b + c)(c + a).
Equivalently,
3 23 2 3 2
a +b +c a b + b c + c a.
By the AM-GM inequality, we have
3 3 3 2
a +a +b 3a b.
143
3 3 3 2
b +b +c 3b c.
3 3 3 2
c +c +a 3c a.
It follows
3 3 3 2 2 2
a +b +c a b + b c + c a.
Proof 13. Without loss of generality, we can assume that

a + b + c = 1.
b+ + 4 2s = 3a,
b + c. 4
c
a 9a(b + c) a 9a(b + c)
c+ + 4 2s = 3b,
a c+a .
4
b 9b(c + a) b 9b(c + a)
a+b+ 4 2s = 3c.
.
a+b 4
c 9c(a + b) c 9c(a + b)
This implies
+
+
b + c+ 4 c+a 4
a 9a(b + c) b 9b(c + a)
+ a + b+ 4 3.
c 9c(a + b)
Equivalently,
a + b + c 3 9 (ab + bc + ca). (11)
b+c a+c a+b 2

In other side, we can prove that
ab + bc + ca 1 (a + b + c)2 = 1. (12)
3 3
144
From (11) and (12) it follows

a + b + c 3.
b+c a+c a+b 2
Proof 14. As above we can assume a b c. Put

x = a,
c
y = b.
c
It follows x y 1. The inequality becomes
a b 1 3
c c
+
+1 + +b 2 .
b a a
+1
c c c c
Equivalently, x y 3 1
+ .
y+1 x+1 2 x+y
Applying the AM-GM inequality, we have
x + 1 + y + 1 2.
y+1 x+1
Or, y 1 1
x + 2 .
y+1 x+1 x+1 y+1
We shall prove that
2 1 1 3 1 .
x+1 y+1 2 x+y

Equivalently,
1 1 1 1 ,
2 y+1 x+1 x+y
y 1 y 1
.
2(y + 1) (x + 1)(x + y)
145
The last inequality is true, since x y 1. The equality occurs if and only if
a = b = c.
Proof 15. As the above proof, one can assume that c = 1. Moreover,
we can assume that a b 1. We shall prove that
a b 1 3
+ +
b+1 a+1 a+b 2 .
Put
A = a + b, B = ab.
The inequality becomes
2 2 1 3
a +b +a+b
+
(a + 1)(b + 1) a+b 2 .
Equivalently,
2
A 2B+A
1 3,
+
A+B+1 A 2
3 2 A + 2 B(7A 2).
2A A
Hence,
7A 2 > 2(a + b 1) > 0,
and
2 2
A = (a + b) 4ab = 4B.
We now prove the following inequality
3 2 2
4(2A A A + 2) A (7A 2).
Or, 2
2A 4A+8 0.
3
A
Equivalently,
2
( A 2) (A + 2) 0.
The last inequality holds, as A 2. The proof is completed. The equality
occurs if and only if a = b = c.
146
Proof 16. Put a b c

M(3) := + + .
b + c a+ c a+b
By the Cauchy-Schwarz inequality, we have
2
(a + b + c)
" r r # 2
= r b+c
.
q
a(b + c) + .q
c+a
b(c + a) + a+ b
.
q
c(a + b)
a b c
h i
M(3) a(b + c) + b(c + a) + c(a + b) .
It follows
M(3) 2 2 2 2
(a + b + c ) a +b +c
2(ab + bc + ca) = 1 + 2(ab + bc + ca) .
We can prove that
2 2 2
a +b +c 1.
2(ab + bc + ca) 2
This follows
M(3) 3
2.
Proof 17. (Nesbitt, 1903). The inequality is equivalent to the following

inequality
a+b+c a+b+c a+b+c 9,
b+c + a+c + a+b 2
( a + b + c) b + c+ a + c + a + b2,
1 1 1 9
h i + 9.
(b + c) + (a + c) + (a + b) b + c a+c + a + b
1 1 1
The last inequality holds by the Cauchy-Schwarz inequality. In particu-

147
lar, we have +
h
(b + c) + (a + c) + (a + b) b + c
i a+c + a + b
1 1 1
q i
=h (b + c) + (a + c) + (a + b)
2 q 2 q 2
s s s
" (b + c)2 + (a + c)2 + (a + b)2 #
1 1 1
2
"q 2 s q 2 s q 2s #
(b + c) . (b + c)2 + (a + c) . (a + c)2 + (a + b) . (a + b)2
2 1 1 1
=3 =9.

Proof 18. This proof is based on the following:
a + b + c
b+c a+c a+b
2 2 2
3 1 ( b) (a c ) (b c )
= + + + . a
2 2 (a + c)(b + c) (a + b)(b + c) (a + b)(a + c)

By this we deduce that the inequality is proved. The equality occurs if
and only if a = b = c.
Proof 19. (Nesbitt, 1903). Put
x = a + b,
y = b + c,
z = c + a.
The inequality becomes
x+z y y+z x x+y z 3.
+ +
2y 2x 2z 2
Equivalently,
x+ z y + z x + y 6
y + x + z 1.
148
Or,
x y y z x z
y + x+ z+ y + z + x 6.
The last inequality holds, since :
a b
b + a 2,
for every a, b > 0. The equality occurs if and only if a = b = c.
Proof 20. By the inequality of AM-HM we have
x+y+z 3 ,
1
3 x
+1y +1z
for every x, y, z > 0. Equivalently,
1 1 1
(x + y + z ) x+ y+ z 9.
Put
x = b + c,
y = c + a,
z = a + b.
Insert this in the above inequality we derive the expected proof. The
equality occurs if and only if a = b = c.
Proof 21. The inequality becomes
a b c 9
b + c + 1 + c + a + 1 +a + b + 12.
Without loss of generality, we can assume that a + b + c = 1. We shall
prove
a+b+c b+c+a c+a+b 9
+ +
b+c c+a a+b 2.
Equivalently,
1 1 1 9.
+ +
b+c c+a a+b 2
149
The last inequality can be proved by applying the AM-HM inequality as

follows:
1 + 1 + a+b1 3 3
b+c c+a
3 a + b + b + c + c + a =2.
The proof is completed. The equality occurs if and only if a = b = c.
4 Proofs of P(4), P(5), and P(6)
This section presents the Shapiro’s cyclic inequality for n = 4, 5, 6.

0
The known proofs for other n s are very complicated, and these are
out of the scope of this work. However, the reader can refer the
references of this work.
Proposition P(4). Assume that a, b, c, d are arbitrarily non-negative

real numbers. Then,
a + b + c + d 2. (13)
b+c c+d d+a a+b
We shall present two proofs for P(4).

Proof 1. Consider the following expression
S= a + b + c + d ,
b+c c+d d+a a+b
M= b + c + d + a ,
b+c c+d d+a a+b
N= c + d + a + b .
b+c c+d d+a a+b
Evedently,
M+N=4.
150
By the AM-GM inequality we have
M+S=a + b + b + c + c + d + d + a,
b+c c+d d+a a+b
N+S=a + c + b + d + c + a + d + b,
b+c c+d d+a a+b
=a + c + c + a + b+d + d+b
b+c d+a c+d a+b
4(a + c) + 4(b + d)
= 4.
a+b+c+d a+b+c+d
We deduce
M+N+2S 8.
Thus,
S 2.
The equality occurs if and only if a = b = c = d.
Proof 2. Put
a b c d
M(4) := b + c+ c + d + d + a + a + b.
By applying the Cauchy-Schwarz inequality, which is similarly to the
proof of M(3) as presented above, we obtain
2
(a + b + c + d )
M (4 ) . (14)
ab + ac + bc + bd + cd + ca + da + db
We then have
ab + ac + bc + bd + cd + ca + da + db = (a + c)(b + d) + 2(ac + bd)

1 2 1 2
(a + c)(b + d) + 2(a + c) + 2 (b + d )
= 1 (a + b + c + d )2 . (15)
2
Combining (14) and (15) we receive M(4) 2. The equality occurs if and
only if a = b = c = d.
151
Proposition P(5). Let a, b, c, d, e be arbitrarily real numbers. Then

a b c d e 5
+ + + +
b+c c+d d+e e+a a+b 2. (16)
Proof. By the Cauchy-Schwarz inequality, we have
a b c d e
+ +
b+c c+d d+e e+a a+b
+ +
2 2 2 2 2
a b c d e
= ab + ac +bc + bd +cd + ce +de + da +ea + eb
2 .
(a + b + c + d + e )
ab + bc + cd + de + ea + ac + ce + eb + bd + da
We shall prove that
2 5 .
(a + b + c + d + e )
ab + bc + cd + cd + de + ea + ac + ce + eb + bd + da 2
Indeed, this inequality is equivalent to the followings:
2 2 2 2
2(a + b + c + d )
(ab + bc + cd + cd + de + ea + ac + ce + eb + bd + da),
2 2 2 2 2 2
(a b) + (b c) + (c d) + (d e) + (e a) + (a c) +
2 2 2 2
+ (c e) + (e b) + (b d) + (d a) 0.
The equality occurs if and only if a = b = c = d = e.
Proposition P(6). Let a, b, c, d, e, f be arbitrarily real numbers. Then
a b c d e f 3. (17)
+ + + + +
b+c c+d d+e e+f f+a a+b
Proof. Applying the Cauchy-Schwarz inequality, we have
a b c d e f
+ + + + +
b+c c+d d+e e+f f+a a+b
2 2 2 2 2 2
a b c d e f
= + + + + +
ab + ac bc + bd cd + ce de + da e f + ea f a + f b
2 . (18)
(a + b + c + d + e + f )
ab + bc + cd + de + e f + f a + ac + ce + ea + bd + d f + f b
152
Denote
S := ab + bc + cd + de + e f + f a + ac + ce + ea + bd + d f + f
b. We then have
2
2S = (a + b + c + d + e + f )
2 2 2 2 2 2
(a + b + c + d + e + f + 2ad + 2bd + 2c f ).
2 2 2 2 2 2
a +b +c +d +e +f + 2ad + 2bd + 2c f
= (a + d)2 + (b + e)2 + (d + f )
2
1 (a + b + c + d + e + f ) 2 . (19)
3
Hence,
2
2S 3(a + b + c + d + e + f )2.
Combining (18) and (19) we receive M(6) 3. The equality occurs if and
only if a = b = c = d = e = f .
5 Generalizations of P(3)
Problem 2. Let a, b, c be arbitrarily real numbers. Prove that
r r r
a b c 3 , (20)
r r + r r + r r p+q
pb + qc pc + qa pa + qb
where p, q > 0.
Proof. By the Cauchy-Schwarz inequality, we have
r + r + r
a b c
r r r r r r
a2r b2r c2r
= r r r r + r r r r + r r r r
pa b + qa c pb c + qb a pc a + qb c
= r r r 2 3 .
(a + b + c )
r r r r r r p+q
(p + q)(a b + b c + c a )
153
Corollary 7. Let a, b, c > 0 be given. Prove that

r + r + r 3 , (21)
a b c
r r2 r r r r
where p, q are positive numbers satisfying p + q = 2.
Proof. By the Cauchy-Schwarz, we have
r + br + r
a c
r r r r r r
pb + qcpc + qa pa + qb
= a2r + b2r + c2r
r r r r r r r r r r r r
pa b + qa c pb c + qb a pc a + qb c
(ar + br + cr)2 3 .
= r r r r r r
2(a b + b c + c a ) 2
We remark that P(3) can be rewritten in the following form

a1 b1 c1 a0 b0 c0
+ + + + (22)
1 1 1 1 1 1 0 0 0 0 0 0,
b +c c +a a +b b +c c +a a +b
0 0 0
where a = b = c = 1. Let
t t t
F(t) = a b c
+ + , (23)
t t t t t t
b +c c +a a +b
stand for the function of variable t 2 (0, +¥). Then, F(t) is monotone.
We then have
F(1) F(0).
Figure 1 is the curve of F(t) defined by (23) for the case a = 2, b = 3,

c = 4. We see that the function is monotone on the interval (0, +¥).
We have the following problems.
Problem 3. Let a, b, c be given positive numbers, and a > b 0. Prove that
aa ba ca ab bb cb
+ + + + .(24)
ba + ca ca + aa + ba bb + cb cb + ab + bb
aa ab
154
F(0)
O x
Figure 1: Function F(t) is defined by (23)
Proof. Calculating the derivative of F as

ln b)
0 t t t t 2ct + at + bt (bt
F (t) = a b (a b )(ln a
+ ct)(at + bt)
t t t
2a + b + c
t t t t
+ b c (b c )(ln b
ln c)
(ct + at)(bt + ct)
t t t t t ln
+ c a (c a )(ln c t t
2b + c + a
a) .
(at + bt)(ct + at)
0
It follows F (t) 0 for every t 0. This implies that F is monotone on the
interval (0, ¥). We deduce that if a b, then
aa ba ca ab bb cb
++ + +
ba + c a ca + aa + ba bb + cb cb + ab + bb .
aa ab
The inequality is proved.
Corollary 8 below is an inequality which can be seen as an

immediate consequence of the above.
Corollary 8. If a, b, c are non-negative numbers, then
2 2 2 a b c
a b c
2 2 + 2 2 + 2 2 + +
b + c c + a a + b.
b +c c +a a +b
However, we can prove the following problem by using Problem 3.
Problem 4. If a, b, c 0, then
a3 +
b3 +
c3 a2 +
b2 +
c2
.
b3 + c3 c3 + a 3 a3 + b3 b2 + c2 c2 + a2 a2 + b2
155
A further derection of the generalization of P(3) as follows.

Problem 5. Let a, b, c > 0 be given, and k 1. We have
k k k
a b c a k 1 + b k 1 + ck 1 .
+ +
b+c c+a a+b 2
In particular, this inequality is homogeneous of (k 1)-degree, and we
leave the proof of it to the reader.
A generalization for n in this derection is the following problem.
Problem 6. Let a1, . . . , an be n positive numbers with n 3, and p, q
> 0.
Then, for every k 2 we have:
a1
k k k 1 k
a2 an an
pa2 + qa3 + pa3 + qa4 + + pan + qa1 + pa1 + qa2
k 1 k 1 k 1 k 1 .
a1 + a2 + + an 1 + an
p+q
Proof. Put
k k k k
M := a1 a2 an 1 an
+ + + + .
pa2 + qa3 pa3 + qa4 pan + qa1 pa1 + qa2
We denote
an+1 := a1, an+2 := a2.
k 1
For each one of a , i = 1, 2, . . . , n we have
i
s ak 1
q
aik 1 = a
i
k 1
pa + qa i (pai+1 + qai+2).
i+ 1 i+2
k 1 k 1 k 1 k 1 2
a1 + a2 + + an 1 + an
"
k 2 k 2 k 2 k 2
M p(a1 a2 + a 2 a3 + + a n 1an +a n a 1)
#
k 2 k 2 k 2 k 2
+ q (a 1 a3 + a 2 a4 + + a n 1a1 +a n a 2 ). (25)
156
By proving similar to the homogeneous inequalities as above, we

obtain the following inequalities:
k 2 k 2 k 2 k 2
a1 a2 + a2 a3 + + an 1an + an a1
k 1 k 1 k 1 k (26)
a1 + a2 + + an 1 + an
1 ,
k 2 k 2 k 2 k 2
a1 a3 + a2 a4 + + an 1a1 + an a2
k 1 k 1 k 1 k 1 (27)
a1 + a2 + + an 1 + an .
Combining (25), (26), and (27) we receive
k 1 k 1 k 1 k 1 2
a1 + a2 + + an 1 + an
k 1 k 1 k 1 k 1
M(p + q)(a1 + a2 ++a n 1 +a n ).
Therefore, the inequality is proved.
Remark 4. When k = 1 and p = q = 1, the inequalityin Problem 6 turns

out the Shapiro’s cycle one. Thus it does not true for every n 3. However,
it may hold for some n which can be seen as an open problem.
We have the following corollary.
Corollary 9. Assume that a1, a2, . . . , an are positive numbers (n 3)
and k 2. Then
k
a1 + + ank 1 + ank
+ a2
k
a2 + a 3 a 3 + a4 an + a 1 a1 + a 2
k 1 k 1 k 1 k 1
a1 + a2 + + an 1 + an
2 .
We can provide now a generalization of P(3).
Problem 7. Let a, b, c be given positive numbers, and a 0. Then
b+c a + c+a a + a+b a .
2a
a b c 3
157
Proof. Put
M := b + c a c+a a+
a + b
a
a b c
+ .

s
M 3 (a + b)(b + c)(c + a) .
a
abc
In other side, we can prove that
Equivalently, (a + b)(b + c)(c + a) 8abc.
abc 1
(a + b)(b + c)(c + a) 8.
From the assumption a 0 it follows a a
abc 1
8 .
(a + b)(b + c)(c + a)
Hence,
b+c a c+a a+ a + b a
a b c 3
.
+ 2a
The inequality is proved.
Problem 8. Let a, b, c > 0 be given, and let a 1. Then

b+c a c+a a+ a
a + b
a b c 3
.
+ 2a
Proof. By the Bernoulli’s inequality, we have
b+c a + a 1 ab + c,
2a 2a
c+a a
+ a 1 ac + a ,
2b 2b
a+b a
+ a 1 a a + b,
2c 2c
(a 1 ) b + c+ c + a + a + b (a 1)3.
2a 2b 2c
158
Combining the sides of the above inequalities we deduce the expected

proof.
Problem 9. Assume that a, b, c > 0 and which satisfy abc = 1. Let a
1 be
given. Then aa ba ca 3
+ + .
b+c c+a a+b 2
Proof. Without loss of generality, we assume that a b c. Then
a b c ,
b+c c+a a+b
1 1 1
aa ba ca .
By the Chebyshev’s inequality, we have

aa ba ca
+ +
b+c c+a a+b1 a b c
a a 1 + ba 1 + ca 1 .
+ +
3 b+c c+a a+b
Applying the AM-GM inequality and having in mind that abc = 1, we
obtain
p = 3.
aa 1 + ba 1 + ca 1 3 aa 1ba 1ca 1
By the Shapiro’s cycle inequality P(3), we have

a b c 3 .
+ +
b+c c+a a+b 2
The proof is completed.
An another proof of Problem 9 is the following. We have:
a2 = r q a(b + c),
aa
b+c
1+a
b2 = r q b(c + a),
c+a
1+a ba
r q
c2 = a+b
c(a + b.
1+a ca
159
Therefore, we apply the Cauchy-Schwarz inequality to receive
1 a a a
a1 2 a +b 2
a
+ c1 2 a a b c
+ + + 2
h i
a(b + c) + b(c + a) + c(a + b) b + c + c + a + a + b .

The proof would be completed by the following inequality
a 1 2 + b1 2a + c1 2a 2
a
+ + + 3(ab + bc + ca). (28)
For the proof of (28), we have

1+a 1+a 1+a
+b +c a + b + c. a 2 2 2 (29)
Indeed, by the AM-GM inequality and note that abc = 1 we have
a+b+c 3 p = 3. (30)
3 abc
In other side, by the Bernoulli’s inequality it holds
1+a
1+a 1+
a
a 2 + 2 1 2 .a (31)
b 1+a + 1 + a 1 1 + .b (32)
a
2
2 2
c 1+a
+1+a 1 1 + .c. (33)
a
2
2 2
Combining the inequality (31), (32), and (33), and using (30) we obtain
a 1+a2 + b 1+a 1+a
2 + c 2
1+ a a + b + c + 3 3 + 3
a
2 2 +
ha 1 i 3 3
= (a + b + c) + a+b+c +3 a
2 2
(a + b + c ) + a 1 .3 + 3 3 + 3a = a + b + c.
2 2
a
Thus, (29) is proved. It follows that 2
( a + b + c) .
2a
a1 + b 1 2a + c 1 2 (34)
+ + + 2
160
In other side, we have

2 (35)
(a + b + c ) 3(ab + bc + ca).
By the inequalities (34) and (35) we deduce (28). The problem is proved.
Problem 10. Let a, b, c > 0 be satisfied abc = 1. Assume that a 2. Then
1 1 1 3
+ +
aa(b + c) ba(c + a) ca(a + b) 2.
Proof. By changing the variables
1
x = a,
1
y = b,
1
z= c,
and
0
a = a + 1.
the inequality can be seen as an immediate consequence of Problem 9.
Note that we also have xyz = 1. The inequality is proved.
Furthermore, the inequality (36) below is a special case of the
above one.
Problem 11 (IMO, 1995). Let x, y, z be given positive numbers
satisfying the condition xyz = 1. Prove the inequality
1 1 1 3 . (36)
3 + 3 + 3
a (b + c) b (c + a) c (a + b) 2
References
[1] P. J. Bushell. Shapiro’s cyclic sum. Bull. London Math. Soc.,
26:564– 574, 1994.
[2] P. J. Bushell and J. B. McLeod. Shapiro’s cyclic inequality for even
n. J. of Inequal. & Appl., 7(3):331–348, 2002.
161
[3] P. H. Diananda. On a cyclic sum. Proc. Glasgow Math. Assoc.,

6:11–13, 1963.
[4] V. G. Drinfeld. A cyclic inequality. Math. Notes, 9:68–71, 1971.
[5] A. M. Fink. Shapiro’s inequality. In G. V. Milovanovic, editor,
Recent Progress in inequality, Mathematics and Its Applications,
part 13, pages 241–4248. Kluwer Academic Publishers, Dordrecht/
Boston/London, first edition, 1997.
n
[6] L. J. Mordell. On the inequality år =1 xr/(xr+1 + xr+2) n/2 and some
others. Abh. Math. Se. Univ. Hamburg, 22:229–240, 1958.
n
[7] L. J. Mordell. Note on the inequality år =1 xr/(xr+1 + xr+2) n/2 and
some others. J. London Math. Soc., 37:176–178, 1962.
[8] B. G. Pachpatte, Mathematical Inequalities, North-Holland
Mathemat-ical Library, V. 67 (first ed.). Amsterdam, The
Netherlands: Elsevier, 2005.
[9] R. A. Rankin, A cyclic inequality, Proc. Edinburgh Math. Soc. (2),
V. 12 (1961), pp. 139-147.
[10] R. T. Rockafellar, Convex analysis, Princeton University Press,
Princeton, 1970.
[11] H. S. Shapiro, Advanced problem # 4603, Amer. Math. Monthly,
61 (1954), p. 571.
[12] J. M. Steele, The Cauchy-Schwarz Master Class: An Introduction
to the Art of Mathematical Inequalities, Cambridge University Press,
Cam-bridge, 2004.
[13] B. A. Troesch. The validity of Shapiro’s cyclic inequality. Math.
Comp., 53:657–664, 1989.
[14] N. M. Tuan and L. Q. Thuong, On an extension of Shapiros cyclic
in-equality, J. Inequalities Appl. (2009), no. 12, 1–5.
162

Book Homc2019

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Book Homc2019

Uploaded by

Copyright:

Available Formats

LIBRARY HOMC

HANOI OPEN MATHEMATICS

Tran Nam Dung, Gabriel Dospinescu, Mixing variables . . . . . . . . . . . . . . . . 4 Nguyen

Thanh Hong, On the counter-example about parallelogram . . . . . 18 Nguyen Van Loi,

class of functional equations in plane geometry. . 36 Nguyen Van Ngoc, On the

Rearrangement Inequality . . . . . . . . . . . . . . . . . . 41 Dam Van Nhi, Some new

Nguyen Dang Phat, Some International Olympiad problems in geometry 88

Tran Ngoc Nam, Selected problems from Mathematics and Youth

Nguyen Hung Son, Invariant in Tiling problems . . . . . . . . . . . . . . . . . . . . . 124

we can try to show that

Example 1 (Kvant). Let a, b, c > 0. Show that we have following inequal-

Solution. Let’s consider the function

Example 2 (Vietnamese IMO Team selection Test, 1996). Let a, b, c

Solution. We will show that with some additional condition

F(a, b, c) F(a, (b + c)/2, (b + c)/2).

Indeed, long but easy computations show that we have

F(a, b, c) F(a, (b + c)/2, (b + c)/2).

Thus, it is now sufficient to show that F(a, b, b) 0 for all a, b, or

If b = 0 then inequality is trivial. If b 6= 0, then dividing the two sides of

For free of conditions inequalities in general we can mix variables

Example 3 (Iran Mathematical Olympiad 1996). Let a, b, c be positive

Our goal is to show that

f (a, b, c) f (a, t, t).

which can also be written

and it sufficient to show that

This can be done by easy computation:

g(a, t, t) f (a, t, t) 9/4

Example 4 (Vietnamese Mathematical Olympiad 2002). Prove that for

The second one: 0 < x 1. Then q q

2(x + y + z) xyz < 2(x + y + z) 2x + 2 2 2

We have 2 2 /(9 > 0, so it follows that

Let us take a = minf a, b, cg. Then

since 2t b c 0 and a t. Thus, it remains to prove that if

and this one is true since it is equivalent to

Example 6 (Mircea Lascu and Marian Tetiva). Prove that if a, b, c > 0

Solution. Now it’s obvious that we should consider the function:

In considered examples, we mix variables into their means. We do

and we get (6).

First, we will show that there exists a minimal value of

S(x1, . . . , xn) is continuous on C, so the first thing is done.

minf f (x1) + f (x2) + + f (xn)g = f (x1 ) + f (x2 ) + + f (xn).

We will show that

f (x1) + f (x2) + + f (xn) f (1) + + f (1) = 0.

abc + bdc + cda + dab .

ON THE COUNTER-EXAMPLE ABOUT PARALLELOGRAM

Combined the properties of a parallelogram, we propose some problems,

1.2 Several properties of parallelograms

Theorem 2. If a quadrilateral is a parallelogram, then its opposite

Theorem 3. If a quadrilateral is a parallelogram, then its consecutive

1.3 Proving quadrilaterals are parallelograms

Theorem 5. If both pairs of opposite sides of a quadrilateral are

2 A counter example about parallelogram?

Combined the properties of a parallelogram, we propose the follow-

Proof. Let H, K are respectively the foot of perpendiscular from B, D to

Given an isoscelec trapzoid ADBE. Let C be the symmetric point of

Problem 2 (Iranian Geometry Olympiad 2015). In rectangle ABCD, the

Problem 3 (Estonian 2005). Let ABCD be a parallelogram, M the mid-

Problem 6 (Iranian Geometry Olympiad 2016). Suppose that ABCD is

Problem 8. Let ABCD be a parallelogram. The circle c with diameter

Problem 9. Consider a convex quadrilateral ABCD and let M, N, P, Q

THE SEPARATION OF ELEMENTS INTO TWO PARTS OF EQUAL

1 The structure of elements with weight and its effect

N + 1 = a1 = a2 = .... = ai > ai+1 ai+2 .... ak