Professional Documents
Culture Documents
(S CIENTIFIC S EMINAR )
H ANOI , 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
Nguyen Van Loi, The separation of elements into two parts of equal weight23
Nguyen Van Nho, Menelaus’s theorem and its applications in some Grade 8
problems..............................................................................................................67
Vu Tien Viet, Proofs and generalizations of one inequality at the 36th IMO 78
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
1 Some examples
To prove the inequality
√
f ( a, b, c) = 2( a + b + c ) + 3
2 2 2
3 a2b2c2 − (a + b + c)2.
The expression of f itself√suggests us to mix two variables into their ge-
ometric mean (to keep 3 3 a2b2c2 unchanged). We have
√ √ √
f (a, b, c) − f (a, bc, bc) = 2(b + c − 2bc) + (a + 2 bc)2 − (a + b + c)2
2 2
√ √ 2 √ √
= ( b − c) [( b + c)2 − 2a].
So, if a = min{a, b, c} (and we can obviously suppose that) then we
have √ √
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
√
2( a + 2b ) + 3
2 2
3 a2 b4 ≥ ( a + 2b)2 ,
or
√
a2 + 3 3 a2b4 ≥ 4ab,
√3 √3 a2b4.
2
which is true by AM-GM for a , 3 a 2 b4 , 3 a 2 b4 ,
√3
3
The following example is much more difficult:
47
F(a, b, c) = (a + b)4 + (b + c)4 + (c + a)4 − (a4 + b4 + c4) ≥ 0
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
7
4
− 2(a + (b + c)/2)4 − (b + c)4 + (a4 + 2((b + c)/2)4
7
4
= (a + b)4 + (c + a)4 − 2(a + (b + c)/2)4 + c((b 7+ c)4/8 − b4 − c4)
3
= a(4b3 + 4c3 − (b + c)3) + 3a2(2b2 + c2 − (b + c)2 ) + ( b + c − (b + c) /8)
4 4 4
7 2 + 10bc]
= 3a(b + c)(b − c)2 − 3a2(b − c)2 + (3/56)(b − c)2[7b2 + 7c
3
= 3a(a + b + c)(b − c)2 + (b c)2[7b2 + 7c2 + 10bc].
56
The last term is nonnegative. If a, b, c have the same sign, then the
from c has
a, b, is
inequality the If
trivial. same
a, b,sign
c have + b + c.signs
as adifferent WLOG,thensuppose that From
at least one it is a.the
initial
last
equality we have
7
47 4
2(x + 1) + 16 − (x + 2) ≥ 0
4
4
f (x) = 2(x + 1)4 + 16 − (x4 + 2).
7
We have
16
f J ( x ) = 8( x + 1)3 − x3. f J ( x) = 0
7
3 2
⇔ x+1=. 7x
Thus ⇔ x = −2.9294.
1 1 1 9
( ab + bc + ca). Σ≥
( a + b )2 (b + c) 2 ( c + a )2 4
Analysis.
We can mix a, b into (a + b)/2, but then both terms in L.S.H will be
one term and
changed unchanged. Forwill
the result example, webulky.
be very will keep ab + bcwe
Naturally, + cawant
unchanged.
to keep
For this goal, we mix b and c into t such that
ab + bc + ca = at + t2 + ta.
If we solve this equation (in t) we get
t = − a + .( a + c)( a + b).
Let
sumption, forweexample
this inequality, holds aonly=withmin
see that it for {a, as-b,
some c}. Investigating
t = − a + .( a + c)( a + b),
g(a, b, c) = ab + bc + ca
and
f (a, b, c) = 1/(a + b)2 + 1/(b + c)2 + 1/(c + a)2.
Then
g(a, b, c) = g(a, t, t).
We will show that
g(a, b, c) ≥ (a, t, t). (5)
Note that
a + t = .( a + c)( a + b),
.( a + c)( a + b) ≤ t ≤ (b + c)/2
Also, (5) is equivalent to
( 1/
can also be written( a + b ) − 1/ ( a + c )) 2
≥ 1/4t 2
− 1/ ( b + c ) 2
, which
x2 + y2 + z2 = 9,
the inequality
Solution.
Let
f (x, y, z) = 2(x + y + z) − xyz.
We have to show that
f (x, y, z) ≤ 10 when x2 + y2 + z2 = 9.
To
intoleave
their the condition
quadratic x2 +Consider
mean. y2 + z2 = 9 unchanged, we will mix two variables
= 2(.(y2 + z2 ) − y − z) − x (y − z)2 /2
√
2( x + y + z) − xyz ≤ 2.3( x2 + y2 + z2 ) − 1 = 6 3 − 1 < 10.
The second one: 0 < x ≤ 1. Then
a3t + at3 + t4 ≤ 3
2 4
⇔ at(3 − t ) ≤ 3 − t
⇔ (3 − 2t2)t2(3 − t2)2 ≤ (3 − t4)2
and this one is true since it is equivalent to
Example 6 (Mircea Lascu and Marian Tetiva). Prove that if a, b, c > 0 have
product 1, then the following inequality holds:
a2 + b2 + c2 + 3 ≥ a + b + c + ab + bc + ca.
Solution. Now it’s obvious that we should consider the function:
f (a, b, c) = a2 + b2 + c2 + 3 − a − b − c − ab − bc − ac.
It is also easy to deduce that
√ √ √ √ √ √
f (a, b, c) − f (a, bc, bc) = ( b + c) (( b + c)2 − 1 − a).
2
So, we would like to have the last term positive. Naturally, we choose
√ −1 +2≥ a+2 a
and it is obviously true since a
√
a + 2 a ≤ 2a + 1 ≤ a2 + 2.
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 + (bc)n + (ca)n
n
1(ab)ab 1 bc 1 ca
− −
when a, b, c are
Thearbitrary nonnegative
investigation reals case (n =a2+) shows
− satisfying
of the easiest b + c =that
1. Analysis.
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 = max a, b, c . When one variable is equal to 0,
the function becomes a simple one-variable
function.
Solution. Let’s assume that a ≤ b ≤ c. Put {
(ab)n (bc)n (ca)n }
f
+ + .
We will show that (a, b, c) = 1 − ab 1 − bc 1 − ca
h ( x ) = x n / ( 1 − x)
bc ≤ (b + c)2/4 = 1/4,
so max
f = h(1/4) = 1/(3.4n−1).
1
when a = b = and c = 0. This is the
answer to the problem, since the value is attained for example
2
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
Analysis. Let
x+1 1
f ( x) = − .
4 x+1
We have to show that if
x1.x2 . . . .xn = 1 then f (x1) + f (x2) + · · · + f (xn) ≥ 0.
√ √
The condition x1.x2 . . . .xn = 1 forces us to mix x1, x2 into x1, x2, x1, x2. But
after that? We can mix some pairs of variables, but if the number of
k
ables brings us to result iff n = 2
is odd? More detailed consideration . What we can do
shows thatin other
mixingcases?
vari-Wecan
variables
work
backward!
If
min{show
we will ) + f ( x2 ) + · · · + f ( xn )} = f ( x1∗ ) + f ( x2∗ ) + · · · + f ( xn∗ ),
f ( x1that
x1∗ = x2∗ = · · · = xn∗ = 1,
√
x ƒ= y then f (x) + f (y) > 2 f ( xy), (7)
which leads to a contradiction. But
√ √
(7) ⇔ (x + y − 2 xy)/4 > 2/(1 + xy) − 1/(1 + x) − 1/(1 + y)
√ √ √ √
⇔( x− y)2 > [2(1 + x)(1 + y) − (1 + xy)(2 + x + y)]/(1 + xy)(1 + x)(1
√ √ √ √ √ √
⇔ ( x − y)2/4 > ( x − y)2(1 − xy)/(1 + xy)(1 + x)(1 + y).
which is true. Finally, we conclude that
x2 + y2 + z2 + x + y + z ≥ 2(xy + yz + zx).
Problem
are 5 (1993
nonnegative IMO
reals Shortlist,
which add up proposed by Vietnam). Prove that if a, b, c, d
to 1, then
1 + 176abcd
abc + bdc + cda + dab ≤ .
27
a2 + b2 + c2 + 2 + that
Prove ≥ all
abc for a +
a, bb, +
c>c 0+ we + bc the
ab have + ca. Problem
following 6.
inequality:
Problem 7. For positive real numbers a, b, c prove that
1 Introduction
1.1 Definition
A parallelogram is a quadrilateral with both pairs of opposite sides parellel.
CK = AH.
Consequently, the two triangles BHD and DKB are congruent, hence
BK = DH.
Thus, we ABCD
quadrilateral must=beAD,
obtain BC combining with the hypothesis AB = CD,
a parallelogram. Q
Unfortunately, the following construction will show a counter-example of the
above problem.
E acrossand
trapzoid) the side
DE BD. = Then
DC. weIt have AB =
shows (diagonals
thatDE AB = CD.of an isoscelec
Given an
isoscelec trapzoid ADBE. Let C be the symmetric point of
On the
∠BED =∠ other
DAB.hand, from
Then = ∠BCD.we have ∠BCD = ∠BED. Moreover,
the symmetry
∠BAD
Finally, the quadrilateral ABCD satisfies the condition of the above problem,
but it is not a parallelogram.
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 ƒ= ∠C. A circle (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.
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.
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.
ment is true
with the weight + 1.
to N of N If+ A1,can be by
then expressed upof
the rules to induction
N and there
theare no elements
Assuming the
statement is true up to N, we need to show that the state-
with
true. Thus,the weight that set(NA has a +few elements
we can assume 1). Let statement is
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.
least up which
be a set to 1. contains
Let us denote P(A
the unit ) as a fully
weight, thus expressible subset atof set Let A
A is expressible
A. Then P(A) is fully expressible and is the largest among the fully
expressible subsets. From this we can conclude:
Corollary
largest fully2.expressible
Let A be subset
a set that
of A contains the of
and the sum unit element.
elements of PLet (Ah.) be
(A)Pis If the
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 ∈ A\ P( A).
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 3 elements with
. If A contains a maximum of
weight less than 4, then
a)a 1 = a2 = a3 = 1, a4 = a5 = . = ak−1 = 4 and ak = 5; or
b)a 1 = a2 = 1, a3 = 2, a4 = a5 = . = ak = 4.
M 1. Since a1 4
− ≤ a2 ≤ a3 ≤ ≤ a k (the weights are in ascending order),
it follows that
a2 = 1,aa1 3==
Either
It follows
2 = a3 = 1, a4 = a5 = . = ak−1 = 4 and ak = 5 Or a1 =
a2,
from 4 = a5 = . =
aTheorem ak = A(M,k)
1 that 4. is fully expressible.
N
Theorem 2. Given A(2N, k), s(ai) = ai ≤ N and k ≥ + 2. Let us denote the
largest fully expressible subset of A as P(A). If P (A) is 2expressible up to at
least 3, then:
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 no
N
less than S + 2, thus no less than (
2 + their
into two subsets of equal weight (meaning
2), and set A can be separated
weight is N).
Proof. 1) LetASisbefully
P(A), then the weight of P (and
expressible A) (obviously S isseparated
so A can be a positiveinto
integer). =
If Aof
two parts
equal weight.
Let us assume that A is not fully expressible.
4
3, elements with weight
P(A) contains less thanwith
every element P( A) no
4. Asweight canlarger
be expressed at least
than 4. Thus, a) up to
First
we need to see that N > 4. Due to Proposition 1, set A has at least
which
statementmeans
thatthey is(A
are kin) P
A( M, ). fully
not So P(expressible.
A) = A, which is a contradiction
Thus, N > 4. if 2Nof the 8,
then the≤ weight of the elements in A can be no larger than 4,
b) If N > 4, let B = A\ P( A) = b1; b2; ; b p . We can identify
= s(bj).2,Then
bCorollary
j
S +=2 Sfor+ every
bj ≥ 2N + j· ≤
b1 + b02 < ·{ ·p.+Ifbpthe
and p > 0.of According
number the elementsto
the elements of B with their weights without } misunderstanding, that is in set
P( A) is q, then k = q + p and S ≥ q. Then:
2N = S + b1 + b2 + .... + bp ≥ S + p(2S + 2) = S + (k − q)(S + 2) Since S
≥
=⇒ S⇒2 −
q 2N(k≥−S 1+)S(k−
− 2k + 2)>=0.−Looking
S)(+S2N S + (k − )S quadratic
at 1the + 2k polynomial:
f (x) = x2 − (t + 1)x + 2t − 4
f (2) = 22 − (t + 1) · 2 + 2t − 4 = −2 < 0
f (t − 1) = (t − 1)2 − (t + 1)(t − 1) + 2t − 4 = −2 < 0
This means that on the [2, t − 1] interval f (x) < 0. Since S > 3 and
N N
f (S) > 0, thus S ≥ t and T = ⇒ S≥ .
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
N fully expressible. It follows that A ƒ= P(A). In this case, due to
A is not
S≥ ,
2
N N N
2N ≥ S + 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 ≥
4p
N · (p + 1) ⇒ 3 − > 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 ƒ= 1. N
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(ai) = ai ≤ N and k ≥ + 2,
then set A can be separated into two subsets the difference between the weights
of which is 0 or 2. 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
2
(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
1)1) Let A = {1, 1, 1, 86, 6}, then the weight of A is 16. The number
of elements is 6 = + 2. A cannot be fully expressed (as 5 is not
2
expressible) and 1 + 1 + 6 = 1 + 1 + 6.
Counter-examples include instances when the criterion of express-
ibility up to 3 is not satisfied.
weight. This
Let way it is enough
us order to prove
the elements in ascending ak ≤ N,
that ak−1 + order which can
according be Proof.
to their
seen through:
N
a1 + a2 + .... + ak−2 ≥ 2(k − 2) ≥ 2 · 2= N
From this it follows that ak−1 + ak ≤ N.
Theorem 3. Given A(2N, k) where N is even, s(ai) = ai ≤ N and k ≥
+ 2. Furthermore, let a1 = a2 = 2. Then, set A can be separated into two subsets2
of
N equal weight.
Proof. Letis usL order
the other = l1the
; l2;elements in two subsets,
; lv , containing with theThis
odd weights. beingAC == Cwhile
firstway L
and C L, u + v = k. Since 2N is an even number, v also c1; c2; ; cu ,
{
containing elements the weights of which are even numbers, needs to be even.
Let
} v = 2t and consider the (u + t) set.
{
B = b1; b2; ; b(u + t) where the } bj elements are composed in the follow-
ing way: ∪
ci
{ b =
2 where i = 1, 2, , u, and
i
}
lj − lv−j
bu+j = 2 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 N to Proposition 2, the weight of elements in B
cannot be larger than , and the total weight of B cannot be larger than
N either. 2
N
It can be proved that u + t ≥ + 2 in the following way:
4
u+v− 2 u− 2
v ≥ 2+ +
u+t=2+u− 2+ 2 2
2
12 N
2 u −2 2 N4 u − 22
≥ 2+ · + =N
2+ +
N u− 2
From this: u + t ≥ 2 + 4
+ 2
≥ 2+ 4.
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
(m is a positive case NGiven
integer). divided
A(by 6 yields
2N, a remainder
k) where s(ai) = aiof 2). andNk =4m
N Let 6m++3,2
then set A can be separated into two subsets of equal weight and the value of k
cannot be reduced. ≤
≥
Proof. Since k = 4m + 3 ≥ 3m + 3 = + 2, due to Proposition
N
1, a3 3. If a3 = 3 a3 + a4 + .... + ak 2 3 (k 2) = 12m + 3
≤ ⇒ ≥ − ⇒
a1 + 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
shows ≤that the value k = of
of subsets + 3 weight.
4k equal cannot Example
be reduced. A can
in fact be separated into two 3
Proposition 4 (Ininteger).
4 (m is a positive case N divided
Given by 6 yields
A(2N, a remainder
k) where s(ai) = of
ai 4).NLet
andN k= 6m
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, due to Proposition 1,
2
a3 ≤ 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
value of k = 4k
be separated + two
into 4 cannot be of
subsets equal weight.QExample 4 shows that the
reduced.
Proposition
integer). Given 5 A
(In(2N,
casek)Nwhere
is divisible
s(ai) = aby 6). = 6m
LetkN 3m
N. Es + 2,(mthen
is set
a positive
A can
i
be separated into two subsets of equal weight and the value of k cannot be
reduced. ≤
6m N
Note: k = 3m + 2 = + 2 = + 2.
Proof.
2 2 ≤ a ≤ a ≤ a ≤ 3. There are the
Due to Proposition 1, a1 2 3 4
following scenarios:
a)IftoA3,begins (1, Theorem
withcase
in which 1) or (1, 22) can
values, then A(2N, k) can be ex- pressed up
be used.
b)If it begins with (2, 2) values, then Theorem 3 can be used.
c)There
cases are
a2 =two
a3 =cases
a4 = left
3 iswhen it beginsofeither
true because with (1,1.3In
Proposition ) or (2,case
this 3). In
weboth
can
classify the elements based on their weight.
C = {e | es weight is divisible by 3} = {c1; c2; ; cu}
Due
than to
2m.the construction, the elements are integers, so their weight is less
Theorem 5. Given
into two subsets A(2N,
of equal k) where
weight if: s(ai) = ai N. Set A can be separated
≤
1)K = N + 1 if N is odd;
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 thatsoif that
integers you their
cannot pick
sum a few numbers
is divisible (n 1) all
fromthese numbers
by n, then positive
−
yield the same remainder when divided by n.
N and k
4)Given A(2N, k), s(ai) = ai ≤ + 2. Denote P(A) as
2
N
≥ 2
the largest fully expressible subset of set A. Prove that if P(A) is
expressible up to 3, then A is fully expressible!
N
5)Given A(2N, k) where N ≥ 2, s(ai) = ai ≤ N and k ≥ + 2.
Furthermore, let a2 2
= 2. Prove that set A can be separated into two
subsets of equal weight.
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.
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 Bull.
Theory. Pl, Abraham
ResearchGinzburg
Council,,Israel,
Abraham10F;Ziv Theorem
41-43; 1961.in additive num- ber
[3] HMS Proceeding Lai Chau Seminar, (2015), 73 88
[4] Komal Forum 3906 problema
−
(2014) jul, Online Forum.
−
[5] Lovsz
beyond, L., Pelikn J., Vesztergombi
Springer-Verlag, 2003. K Discrete mathematics. Elementary and
Ha Noi, 20/3/2019.
N. V. Loi
Email: loiscenter@gmail.com.
ON A CLASS OF FUNCTIONAL EQUATIONS IN PLANE
GEOMETRY
Let Γ be the set of all triples of positive numbers (A, B, C) such that
A + B + C = π,
i.e. every of
all triples triple (A, B,
positive Γ isb,the
C) ∈ (a,
numbers c) angles of a triangle, and let F be the set of
satisfying
|b − c| < a < b + c,
i.e.
Ourevery
main triple
purpose(a, is
b, c) F is the side lengths of a triangle.
∈
- Determine all function f : (0, π) → (0, π) such that ( f (A), f (B), f (C)) ∈ Γ
for all (A, B, C) ∈ Γ.
- Determine all function f : (0, ∞) → (0, ∞) such that ( f (a), f (b), f (c)) ∈ F
|b − c| < a < b + c,
f J ( x ) − f J (π − x − y) = 0, ∀ x, y, x + y ∈ (0, π ). (2)
π 2π π
g ( x ) + g ( y ) = g ( x + y) for x, y ∈ (−3 , 3 ), x + y < 3
Since, g is also bounded from below, it follows that g must be linear on
(− π , 2π ) (see [1]-[3]). Therefore, g(x) = ax for some a and hence,
3 3
π(1 − a)
f (x) = ax + .
3
Theorem
(c)) ∈ F for 11.allAny c) ∈ Ffis: of(0,the
function
(a, b, → (0, ∞) satisfying ( f (a), f (b), f
∞)form
(1 − α)π 1
f ( x ) = u sin .α arcsin θ ( x ) + Σ, − ™ α ™ 1, u > 0, (4)
where θ ( x) is difined as follows; for every given triple (a, b, c) F with
x
the circumscribed radius R then of θ ( x) =3 . Proof.2 Applying the
2R
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.
[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.
ON THE R EARRANGEMENT I NEQUALITY
Nguyen Van Ngoc Thang
Long University, Hanoi
Definition 1. Let’s
bers. The triples (a1, consider
a2, a3) andtwo , b2, b2)(aare
(b1triples 1, a2, a3) and (b1, b2, b2) of real num-
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,and
a2, a3) and c(b)1,beb2any , b2)permutation
be similarly of arranged or bboth be oppo-
sitely a1b1 + a2b2 + a(c3b1,3 cis2, called
(i) S =arranged, 3
the Sorted numbers.1, b2, b3.
sum of thenumbers
(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)arranged.
oppositely and (2, 4, 6) are similarly arranged while (1, 3, 5) and (6, 4, 2) are
2. If 0 < a ≤ b ≤ c, then (a, b, c) and ( 1 , 1 , 1 ) are oppositely arranged, while
Theorem 12.
S ≥ P ≥ R. (5)
Proof.
and let Let’s
(c1, c2take
, c3) two
be atriples (a1, a2of
, a3()band (b1, b2, b3) increasingly
c1 ≥ ar-
c2. ranged
permutation 1, b2, b3). Assume that
(i) a2 + b2 + c2 ≥ ab + bc + ca,
a+b+c 1 1 1
(i ) abc + + ,
a2 b2 c2
≤
a2 b2 c2 b c a
(ii) + + ≥ + + ,
ba22 c2b2 a2 c2 a b c
(iii) + + a ≥ a + b + c.
Solution. We may assume that b ac≤ a ≤ b.
1 1 1 1 1 1
(i) The triplets . , , Σ and . , , Σ are similarly arranged. Thus
a b c abc
11 11 11 11 11 11
+ + ≥ + + ,
that is (i). aa bb cc ab bc ca
a b c a c
(ii) The triplets . , , Σ and . Σ are similarly arranged. Hence
b
, ,
b c a bca
aa bb cc ab bc ca
+ + ≥ + + ,
which is (ii). bb cc a a b c c a a b
.1 1 1 Σ
2 2 2
(iii) The triplets (a , b , c ) and , , are opposetely arranged. Hence
1 1 a1 b c 1 1 1
a +b +c
2 2 2
≤a +b +c , 2 2 2
a b c b c a
that is, just (iii).
Problem 4. Let a,2b, c 2be positive
2 2
real numbers.
2 2
Prove the inequality
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
1≥1
1 b c.
a2 + b2 ≥ c2 and ≥ a
By the Rearrangement inequality we have
a2 b2 c2
21 1 1 21 1 1
+b2 +c2 + b 2 + c 2 = a + b + c, (6)
+ b c ≥ a b c
a
and b2 + =a
a c a
21 1 21 1 1
b 2 c2 +b2 +c 1 + b 2 + c 2 = a + b + c. (7)
2
+ c a ≥ a b c
c + =a b
a
Adding (1)a andb (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 + b1 1 2
Solution. The triplets (a, b, c) and ( b 1 , , ) are similarly arranged
(we may assume a ≤ b ≤ c). Thus
+c c+a a+b
1 1 1
1 1 1 ≥a +b +c ,
a +b +c
and also
b+c c+a a+b c+a a+b b+c
1 1 1
b +1 c c1+ a a1+ b ≥ a a + b + b b + c + c c + a .
a +b +c
Adding these two inequalities we fine
.b +a c c b+ a c b Σ
a+ c +a a a+
bb +b b+c= c3.
2 + + + a +
≥a + b +b + c c++ a c
if a
required inequality =
is oblained. = only c. Therefore, the
b holds if and
Equality
Problem 6. (Neshbit’s generalized Inequality). If a, b, c > 0, and n is a
positive integer, then
an + b n + c n≥ an−1 + bn−1 + cn−1 .
b+c c+a a+b 2
≤
Solution.
1 1Assume 1 that a b c. Then tripltes (an, bn, cn) and
. , , , Σ are similarly arranged. Hence,
b+c c+a a+b
b +1 c c1+ a n a1+ b c 1+ a a1+ b n b1+ c
an + bn +c ≥ an + bn +c
1 1 1
n a1+ b ≥ a + bn + cn
n
nb +1c n c1+ a .
a +b +c a+b b+c c+a
Adding these two inequalities we fine
2. an c +b
bn cn Σ a n + cn bn + an n n. (8)
+ + ≥ + +
b+c c+a a+b c+a a+b
We have b+c
an + bn
1 n−1 n−1
−n−1
a+b + bn−1 ) ⇔ (a − b)(a b ) ≥ 0, ∀a, b > 0, n ≥ 1.
2
≥
(a
Therefore the right side ofn (8) nbecomes
an + cn b +a cn + bn
+ + n−1
. (9)
c+a a+b ≥
b+c +b n−1
+c n−1
a
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
Solution. Suppose that the triplet (a, b, c) is increasing. Then the triplet
1 1 1 is increasing also. By the Rearrangement
, ,
b+c− ac+a− ba+b− c
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
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
Adding these inequalities we get c+a− b a+b− c
b+c c +a a+b
2Q ≥ + +
b+c− a c+a− b a+b− c
b+c− a+a c+a− b+b a+b− c+c
= − + − + −
b+ca a c+a c a+b c
− −
=3+ + bb − + = 3 + Q.
b+c a c+a b a+b c
which is equivalent to the desired inequality.
Problem 8. Let x, y, z > 0 be real numbers. Prove the inequality
x2 z2 y2 x2 z2 y2
− +
y+z − + −
z+x
≥ 0.
x+y
Solution. We need to prove that
x2 + y2 + z2≥ z+2 +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
≥ .
x2 ≥ y2 ≥ z2 and + z+x x+y
y z
≥
By the Rearrangement inequality we have
x2 + y2 + z2≥ z+2 +x2 , y2
y+z z+x x+y y+z z+x x+y
as required.
If we assume z 2≥ y ≥ x, then 1
z y2 x2 and 1 1
≥ ≥ .
≥ x+z z+y
x+y
By the Rearrangement inequality we obtain
x2 y2 z2 2 1 2 1 2 1
+ + =z +x +y
y+z z+x x+y 2 x +
1 y y +
1 z
≥z +x2 +y z+x
2 1
x+y z+x
2 2 2
z+x
= + + . z x y
y+z z+x x+y
+ + ≥ x + y + z.
yz zx xy is symmetric we may assume that
Solution. Since the given inequality
x ≥ y ≥ z. Then
1 1
x3 ≥ y3 ≥ z3 and ≥ 1
y zx xy.
By the Rearrangement inequality we havez ≥
x3 y3 z3 3 1 3 1 1
+ + =x +z3
yz zx xy + y zx xy
yz x2 y2 z
1 31 2. (10)
≥ x yz+ y yz+ z 3 zx
1
3
= x
We will prove that + +
y z
2 2 2
x y z
+ + ≥ x + y + z. (11)
y z x1 1 1 (since the inequality
Let x ≥ y ≥ z. Then x2 ≥ y2 ≥ z2 and ≥ ≥ x
z case
(11) is cyclic we also need to consider the y z ≥ y x). By the
Rearrangement inequality we obtain
x2 y2 z2
x2 y2 z2
+ + ≥
y z x + + = x + y + z.
x y z
The case when z ≥ y x is analogus to the previous case. Now by (10) and
(11) we have
x3 y3 z3
+ + ≥ x + y + z.
yz zx xy
The equality occurs if and only if x = y = z.
≤
(13)
Since
(x + y + z + t )2
xy + xz + yt + zt = (x + z)(y + t) ≤ =4
by (13) we deduce that 4
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≥ . 3
a3(b 1+ c) b3(c + a) c3(a + b) 2
1 1 1
Solution. Set x = , y = x = , z = x = . Since abc = 1 we obtain
xyz = a b c
1. With these new notations the required inequality transforms
into
x2 + y2 + z2≥ . 3
y+z x+z x+y 2
In virtue of the Neshbit’s generalized and Am-GM inequalities, we hve
x2 y2 z2 x+y+z √ xyz 3
3
+ + ≥
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
ilarly arranged, then inequality). If (a1, a2, a3) and (b1, b2, b3) are sim-
≥. Σ. Σ.
c( a + b − c) ≤ b ( c + a − b ) .
Problem 16. Let a, b, c > 0. Prove that
a8 + b8 + c8 1 1 1
≥
a 3b 3c 3 + + .
a b c
Problem 17. Leta ≥ b ≥ c and 0 ≤ x ≤ y ≤ z. Prove that
a b
c
a+b+c
≥ √ xyz .a+b+cΣ
x + y + z 3
≥ 3 x+y+z .
Problem 18. Let a, b, c > 0. Prove that
√ √ √
ab + bc + ca ≥ a bc + b ac + c ab
2 Generality
n
≤
3
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.
bers.R(n) = a1bn + a2bn−1 +
(b)
··
+ anb1 is called the Reversed sum of the num-
bers,P(where
(c) n) = ac1c,1 c+, a. 2.c.2 c+ be any·+permutation
1 2 n ··
ancn is calledofthe
thePermutated
numbers b1sum
, b2, .of. .the
, bnnum-
.
·
2.2 The Rearrangement Inequality
Theorem 13.
S( n) ≥ P( n) ≥ R(n). (14)
cn
· · · = xn.
2.3.4 Cauchy-Bunyakowskii- Schwarz Inequality (CBS-Inequality)
Let a1, a2, . . . , an and b1, b2, . . . , bn be real number sequences. Then
(a1b1 + a2b2 + · · · + anbn)2 ≤ (a2 + a2 + · · · + a2 )(b2 + b2 + · · · + b2 ),
1 2 n 1 2 n
(18)
with
or bi equality if and
= kai, for 1 ≤onlyi if≤forn.some
Proof. 1 2 · =1 a≤
constantIf ak, a=i a= =kb·i, ·for i ≤ n
n = 0 or
2 = · · · =2n band
.ab211 +=a2b2+ n =Q 0, =theb .result is trivial. Otherwise, define P =
· · · a+i a i 21
bfor+ 1b2≤
2+ · · · + b2n. Since both are non-
i ≤ n. By Corollary 7,
zero, we let xi = and xn+i =
P Q
a2 + a2 + · · · + a2 b2 + b2 + · · · + b2
2
2 P2 n 2 Q2 n
= 1 + 1
= x21+ x2 +2 · · · + x2 n
≥ x1 xn+1 + x2 xn+2 + · · · + xnx2n + xn+1 x1 + xn+2 x2 + · · · + x2nxn
2(a1b1 + a2b2 + · · · + anbn)
which=is equivalent toPQ ,
the desired result. Equality holds if and only if
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
Problem 22 (IMO, 1978). Let a1, a2, . . . , an be distinct positive integers. Prove
that
a1 a2 an 1 1 1
12 + 22 + · · · + n2 ≥ 1+ 2+ · · · + n.
Solution. Let (c1, c2, . . . , cn) be the permutation of (a1, a2, . . . , an), such
that c1 ≤ c2 ≤ · · · ≤ cn. Then ci ≥ i for 1 ≤ i ≤ n. By the Rearrangement
inequality,
a1 a2 an cn
+ +···+ c1 c2 +··· +
≥ +
12 22 n2 12 22 n2
c1 2 n
≥ + +···+
112 122 1 n2
= + +··· + .
References 1 2 n
[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.
SOME NEW COMBINATORIAL IDENTITIES
x1 =
( n − 1) !
( 1)n−2 ∏n (2 + αi)
x 2 = − by x = −2
i=1
1!(n 2)!
−
n
−
( 1)n−3 ∏ (3 + αi)
above i=1
by x = −3
x 3 = 2!(n − 3)!
.
..
n
(n −
(− ∏ (n + α i) by x = −n.
1)!
xn i=1
=
Corollary
ties: 3. Set ϕ(x) =
∏ (x + αi). Then, there are the following identi-
i=1
i
n (−1)i(n)ϕ(i) = (−1)nn!.
(i) ∑
i=0 (−1)i(n)i in = (−1)nn!.
(ii) n
∑ i n n+i n
(iii) n (−1) ( )( ) = (−1) .
i=0
∑ i i
Proof.i=0
n(i) We invoke by Example 1 to deduce that:
(−1)n−i ϕ(i) (x − α1)(x − α2) . . . (x − αn)
∑ (x + i)(i − 1)!(n − i)!
−1=−
(x + 1)(x + 2) . . . (x + n)
.
i=1
By x
0 we get n (−1)n−i ϕ(i) (−1)n+1 ϕ(0) n
1 i n
−1 = or ∑( i
= ∑ i!(n − i)! n!
(−1)nn!. i=0
− ) ( )ϕ( ) =
i=1 i
(ii) Substituting α1 n
= · · · = αn = 0 in ∑ (−1)i(n)i ϕ(i) = (−1)nn! we get
i=0
∑
n
i n n
i=0
(−1) ( )i = (−1) n!.
n
(iii) By
α i (−1)i(n)(n+i) = (−1)n.
n
i
= i, i = 0, 1, . . . , n we obtain ∑ i i
i=0
Example 2. Solve the following system of linear eqations:
∑ n sx k + yk 1
k=1
=
s2 + k2 s
s = ±1, . . . , ±n; xk, yk ∈ R, k = 1, . . . , n
n
∑
n
∏ ( s2 + k 2 )
s= 1
(k2 2 2 .
∏ − s )(1 + k )
and evaluate the sum S = k=1
1 sƒ=k
Proof. Consider f (x) = − + n x k x + yk p (x )
∑ = , where p(x)
x k=1 x2 + k2 n
x ∏ ( x2 + k 2 )
is a polynomial of degree ™ 2n. Since f (s) = 0 therefore
k=1 p(s) = 0 when
k=1
∏ (r + k 2 )
2
s=1 (s
2
+ k2)( 2n )
2 ∑ (−
n−k n−k = (2n)!.
2
k=0
1) k
2 2
n (s + k ) 1 + k2
n
∏ n
Proof. Sine ∑ n s=1− = 1 by Example 2 and ∏ (k2 − s 2) =
∏(k2 s2)(1 + k2)
sƒ=k
k=1
s
k
n
∏ n
(n − k ) ! ( n + k ) !
(k − s) ∏ (k + s) = ( −1)n−k we get
2k2
sƒ=
sƒ=
k k
n
2k2 ∏n (s2 + k2)
s =1
∑
= 1.
k=1 (−1)n−k(n − k )!( n + k )!( 1 + k2)
(s2 + k2)( 2n )
n−k 2
(−1) k ∏ n
n s=1 n−k
∑
∑n n yk 1
k=1 xk +∑
s− k=1 =
k s+k s
s = ±i, . . . , ±ni; xk, yk ∈ R
n
∏ (s 2 + k 2 )
and evaluate the sum S = n s= 1
∑
k=1 k2 (1 ∏ (k2 − s 2 ) .
+ k2)
1 sƒ=k
yk p( x)
Proof. Consider f (x) = − + n xk +
n
=
∑ ,
x− ∑ x2
x k=1 k x+k n k2)
k=1
x ∏ (−
k=1
where p(x) is a polynomial of degree ™ 2n. Since f (s) = 0 therefore
a = −(−1)n by x = 0
n
a ∏ (12 + k2)
x1 = k =1 by x = 1
n
2.12 ∏ (1 − k 2 )
k=2
n
a ∏ (22 + k2)
x2 = k=1n
∏ by x = 2
2.22
... k=1,kƒ=2 (2
2− k2)
n
a ∏ ( n2 + k 2 )
k= 1
n∏1
xn = by x = n
− ( n2 −
2n2 k=1 2k
n
)a ∏ (12 + k2)
y1 = k= 1 by x = −1
n
2
2.12 ∏ (1 − k )
k=2
n
a ∏ (22 + k2)
y2 = k= 1
n by x = −2
∏
2.22
... k=1,kƒ=2 (2
2− k2)
n
a ∏ (n 2 + k 2 )
k =1
n∏1
yn = 2n2 − (n2 − 2k by x = −n.
k=1
1 n ) yk
From = a(x2 + 12)(x2 + 22) . . . (x2 + n2)
− +∑ x x +
n k n and
x x −k k + ∑ x ∏ (x2 − k 2)
k=1 k=1 k=1
2 ∑
n
a(x + 1 )(x2 + 22) . . . (x2 + n2)
2 2
xk .
x2 − k2 =
n
xk = yk it follows −1 + 2x k=1 ∏ (x2 k2)
−
k=1
n
∏ ( s2 + k 2 )
By replacing x = i we obtain ∑ n s= 1 = 1.
k2 (1 + k2) ∏ (k2 − s2)
k=1
sƒ=k
Corollary 5. There is the following identity
n ∏
n
s=1 (s
2
+ k2)( 2n )
2 ∑ (− n−k = (2n)!.
1 + k2
k=1
n
n
Proof. Since ∏ (s2 + k 2 )
s=12 = 1, by the Example 3 and
2 2 2
∑ k ( 1 + k ) ∏ (k − s )
k=1
sƒ=k
n
∏ n n
∏
( k 2 − 2s ) k
− (k s) ∏ −
sƒ= sƒ=k
sƒ=k
=
(sk) =
+ ( − k )!( n + k )!
we have
n 2k2
2 2
n 2 ∏ (s + k )
s =1
∑
= 1.
k=1 (−1)n−k(n − k)!(n + k)!(1 + k2)
∏
n
n (s2 + k2)( 2n )
We obtain 2 ∑ (− s=1 n−k = (2n)! by multiplicating with
(2n)!. k=1 1 + k2
Example 4. Solute the following system of lineat equations and evaluate the
following sum:
1
n sxk + yk
∑
(i)
k=1 s2 + k2 =
s(s2 + 12)(s2 + 22) . . . (s2 + n2)
s = ±1, . . . , ±n; xk, yk ∈ R.
n xk
(ii) T = ∑ .
k=1 n2 + k2
x k x + yk 1 p( x)
n − n =
Proof. Consider f (x) = ∑ x 2 + k2 n
k=1
x ∏ (x2 + k2) x ∏ ( x2 + k 2 )
k=1 k=1
x ∏ ( x2 + k 2 )
k=1
Since ∑
a(−
n
x k x + yk
− n
=
x 2 + k2
x
n
xk 1
n2 + k2 = .
By x = n there is ∑ n2 ∏n (n2 + k2)
k=1 k=1
Corollary 6. There is the following identity:
n
2 2 Σ (−1)k−12k2( 2n ) (2n)!
∑n Σ ∏ (k + s ) = n .
s =1 1 + k2 n+k
− ∏ (1 + k 2 )
k=0 (n!)2
k=0
= Σ ∏n
( k2 + s2 ) −
n Σ k22k2
∑ (n!)2(s=n1 − k)!(n + k)! (n − k)!1(n + k)! (−11+
)k−1
k=1n Σ n
s=1(k
∏
2
+ s2)( 2nn+k) (n+k2n) Σ (−1)k−12k2
= ∑ −
(n!)2(2n)! (2n)! .
k=1
n (k2 + s2)
1 + k2
∏
Hence n Σ +k
Σ (−1)k−12k2(n2n
s =1 ) (2n)! .
∑ −1 = n
∏ (1 + k2)
k=1
(n!)2 1 + k2
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.
MENELAUS’S THEOREM AND ITS APPLICATIONS IN
SOME GRADE 8 PROBLEMS
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 KE
KC
= = EC .
KD , KD DG
KG EC BD KE BD
=
Since BD = CE, replacing by , hence KD , which is
AB DG DG DG
equal to , this happens because DG ǁ AC. Consequently, KE : KD
AC
does not depend on how to choose points D and E.
= KE
ED EC = =
KE AB
MC MC
EM EC BD AC .
⇒ =
KD
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.
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).
BR AQ PC
. . = 1. (1)
AR QC BP
Proof.
(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 BR BP
CP.AQ BR
= , = . = 1.
CP ⇔ AL = CQ AR AL
⇔
AL AR
BP
CQ
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:
B R AQJ PC
. . = 1. (2)
AR Q C BP
J
AQ AQ
From (1) and (2) we deduce j
QJ C = , that means Q ≡ Q’, which
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.
Denoted the area of the triangle XYZ by [XYZ], we have
BR [ PBR ] AQ [PAR] PC [ PCR ]
= , = [PCR] and = [PBR] ,
AR ] QC BP
BR [AQPARPC
hence . . = 1.
AR QC BP
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 = 1, and the conclusion is inferred immediately, since
EK . .
AB CE KD
DB = CE.
H
PH PO −HHO = 2. H − 1 = 2. (8)
O = O O
Now, consider triangleNK ONP and the points Q, H and K. By (7), (8)
and using the assumption = 3, we obtain
KP
NK
. PH 12
KP OQ .HO = 3. 2 . 3 = 1.
QN
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.
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
PBcuts straight lines AB, AC, CF at Q, R, S, re- spectively.
D
Prove that B and D is the midpoint of QS.
PC =
DC
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.
From pairs of similar triangles DHB and EHA, DHC and FHA we
DB HB DC HC
have
= , = .
AE HA HA AE HB AE
Hence DB
. = . FB
AF = AF HC AF . (12)
DC EC
Applying Menelaus’s theorem to the triangle ABC with the transver-
sal PEF, we obtain
PB CE PB AE FB
AF . . =1⇔ = . . (13)
(12) and (13) give PC EA FB PC AF EC
PB DB
=
PC . (14)
BD DC CD
DQ
We have QR ǁ EF, so: P
DS = . Combining these with
C
F = , P
(14) we obtain DQ = DS, that means D isPthe midpoint of QS.
B
Problem 6 (6th 30-4 Mathematical P Olympiad,
F April 2000, Vietnam). Let
PB
(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
C
N k2
[BCN] = k2 + 1 .
S0
=
CA
k2
Having S0 = 1, so [ BCN ] = .
k2 + 1
[ 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 obtain
FB CN
PA . . = 1, hence
FN CA PB
1 + k2 1 + k2
= k2k3 =
FB ⇒ BF 1 + k2 + k2k3 ,
F
N BN
therefore k2 k2
. =
1 + k2
[BCF]= k2 + 1 1 + k2 + k2k3 .
1 + k2 + k2k3 k3 k1
Similarly, we have [ ACI ] = , [ ABE ] = .
Therefore, 1 + k3 + k1k3 1 + k1 + k1k2
k2
[ EIF ] = 1 .− k3 k1 Σ
+ + .
1 + k2 + k2k3 1 + k3 + k1k3 1 + k1 + k1k2
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.
References
[1]Vu Huu Binh, Some development topics of Geometry 8, Vietnam Educa- tion
Publishing House, 2000, Vietnamese.
[2]Nguyen Van Nho, The theorems in Plane Geometry through Olympic
exams, Vietnam Education Publishing House, 2004, Vietnamese.
Vu Tien Viet
Hanoi Mathematical Society
≥ (ab + bc + ca) ,
or
b2c2 c2 a2b2 1
+ a +
2 ≥
a ( b + c ) b ( c + a) c ( a + b )
By AM-GM inequality, we have
(ab + bc + ca).
2 3
b2c2 + c2 a2 + a2b2≥ 1 3 √3 2 2 2
a ( b + c) b ( c + a ) c ( a + b) abc
(ab + bc + ca) ≥ = ·
2 2 2
Proof 2. It follows from the condition abc = 1 that 2 2
1 b2c2 , 1 c2 a2 , 1 ab
= = =
a3(b + c) b3(c + a) c3(a + b) ·
a ( b + c) b(c + a) c ( a + b)
Noe, the inequality (1) is equivalent to
b2c2 ca22 a2b2 3
+ + ≥ ·
Let a(b + c) b(c + a) c(a + b) 2
2 2 +
bc c2 a2 + a2b2= b2c2 + c2 a2 + a2b=
2 S.
a ( b + c) b ( c + a ) c ( a + b) ab + ac bc + ba ca + cb
By Cauchy-Schwarz inequality
(a2 + a2 + a2)(b2 + b2 + b2) ≥ (a1b1 + a2b2 + a3b3)2,
we have 1 2 3 1 2 3
Hence S ≥
3 The desired conclusion follows.
2·
1 1
Proof 3. Note that for α > 0 we have α + ≥ 2, or α ≥ 2 − . We thus
α α
have
1 1 2 1. a 2 ( b + c) 1 ab + bc
Σ
=
a3(b + c)
Similarly · ≥ 2−
2a a2(b + c) 2a 2 = − ·
a 4
1 1
1 ca + cb
≥ 1 − bc + ca
b 3 ( c + a) b 4 , c 3 ( a + b) ≥ ·
Adding the above inequalities yields
1 1 1 1 1 1 1
+ + ≥
a 3( b + c) b 3 ( c + a) c 3 ( a + b)
+ + − (ab + bc + ca)
a b c 2
=1
(ab + bc + ca).
Finaly, the AM-GM inequality leads us to the2
1 1 1
1 3 3
≥ ( ab + bc + ca) ≥ a 2b 2c 2 = ·
a3 (b + c) + b3 (c + a) + c3(a + b) 2 2 2
√
3
1
Choosing λ = gives √ √
≥ 6( λ − a2b2c2 = 6( λ − λ).
4
√
λ) 3
b2c2 c2 a2b2 3
+ a +
2 ≥ ·
a(b + c) b(c + a) c(a + b) 2
Proof 5. Lemma 1. Let α, β be real numbers and x, y be positive real num-
bers.
Then, we have
α2 β2 (α + β)2
+ ≥ · (*)
x y x+y
We see that, inequality (∗) equivalent to
α2y(x + y) + β 2x(x + y) ≥ (α + β)2xy ⇔ α2y2 + β2x2
2
≥ 2αβxy ⇔ (αy − βx) ≥ 0.
Let α, β, γ be real numbers and x, y, z be positive real numbers.
√
ab √
+ .√ x − ca + cbΣ2 =
ca + cb
. b2c 2 c2 a 2 + a2b 2 Σ 2
+
= ab + ac bc + ba ca + cb x
− 2(ab + bc + ca)x + 2(ab + bc + ca).
Clearly f ( x ) ≥ 0, ∀ x ∈ R. From there, we have ∆J ≤ 0, or
2 . b2c 2 c2 a 2 + a 2b2 Σ
+
(ab + bc + ca) − ab + ac bc + ba ca + cb
(ab + bc + ca) ≤ 0.
2
HOMC 2019 - Scientific seminar, Hanoi, April 3, 2019
Inferred
3
b2c2 c2 a2b2 1 3 √3 2 2 2
a
ab +
a2 ac (ab b= ·
bc ++ 2 c 2
ba + bc +
ca +≥
cb ca) ≥
2
Proof 9. Since abc = 1, a2b2c2
= 1. The inequality (1)
equivalent to
b2c2 a2b2
c +
a(b + c) +
b(c + a)
c≥(a +·b)
2
On space Oxyz consider
vectors
O˙ A ( . a ( b + c ) ,
.b ( c + a), .c( a +
b)),
b ca
, , . O˙B ( √
a(b √ a)
+
c)
b(
84
HOMC 2019 - Scientific seminar, Hanoi, April 3,
2019
√abc a ) ab + a
( b)+ | ·| 32 ac (ab b= ·
√ bc ++ 2 c 2
We have O O˙ B | ba + bc +
3 ca +≥
O˙ A · ˙ cb ca) ≥
=
O˙B = ab 1
.
B c a 2 1 1 1
2 2
+ bc + ca b2c2 Proof −1Put −1+
−1 10. + =
| b a + b + c = t > 0, we
and 2 have
+ a
O ab + ac +
· Σ b a− b− c
˙ I bc + ba −
= ca + cb 2
n c
A f
c e [2 1
r 1
o + +
· r a3(b + c) + b−1 + + c−1−1 c−1
s e (a b (c + a) = +−1
3 a−1 a +
3
c ( a + b) b
d
O a−2 b−2 c−2
˙ A =
b
^
B
a
O +
B −
=
bc +
| ≤
+ c
O
˙ | Consid x2
er the ( = , ( < x<
O ca t t
A functio
˙ n f
| −0
)] 2t2 x
A
| x
· 2 )
b ca2b2 1
85
HOMC 2019 - Scientific seminar, Hanoi, April 3, 2019
Inferred
). he (
W a J x=
e vf J )
>
(tot 0,
− lead3
f (xx))is
convex
function.
86
Using Jensen’s inequality, we have
1 1 1 1
a−1 + b−1 + c−1
−1
f (a−1) + f (b ) + f (c−1 ) ≥ 3 f ( )= ( + + )
3 2 a b c
≥ .3 1
3
abc
So we have to prove the inequality.2
= ·
2
Proof 11. Put x = bc > 0, y = ca > 0, z = ab > 0, s = ab + bc + ca =
x + y + z.
Clearly
= + + ≥ ·
s− x s− y s− z 2
y+z s− z+x
s− < 1, 0 < =
0<
x < 1,
= y 2(x + y + z)
2s 2(x + y + z)
x+y 2s
s− = s− x s− y s − z = 1.
z0 <
< 1, + +
2s 2(x + y + z) 2s 2s 2s
Consider random variable X with
x s− x y s− y z s− z
P(X = )= , P(X = )= , P(X = )= .
s− x 2s s− y 2s s− z 2s
The, we have
x s− x y s− y z s− z 1
EX = s − x · 2s + s − y · 2s + s − z · 2s = 2,
. x Σ2 s − x . y Σ2 s − y . z Σ2 s − z
2
E(X ) = · + · 2s + · 2s =
s− x s− s−
y 1 . x2 z z
2 Σ
2s = ,
2 s− x+s−
y2 +
y s−
s
) = + + −
)−
s − xz s− y s− z · 4
0 Var X 1 . x2 y2 z2 Σ 1
≤ ( ) = E( X2 X
2 2s
x2 y2 z2 x+y+z 3√ 3
s
≥
+ + ≥ =
s− x s− y s− z 2 3
xyz = ·
3 Generalizations 2 2 2
∑ 1 n(n − 1)
. Σ .a
m−1 ∑ Σ≥ (3)
1≤k<l≤n ∏ i
aiaj
1≤i≤n,iƒ=k, 1≤i<j≤n,iƒ=k, · (n + 1)(n − 2)
l l
Hint. Applying the generalized Radon’s inequality.
a special case when n = 3, a = a, a2 = b, a3 = c and m = 3 the inequality
(3) In
would to the inequality (1). 1
HOMC 2019 - Scientific seminar, Hanoi, April 3, 2019
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 Γ1 and Γ2 in the
internally
ν is the setorconsisting
both externally and intersecting m.Γν =and
Λ touching ∅Γ : Suppose Λ ν, plane and
of all circles 1 2 both touching Γi in Ai
and intersecting m in Bi, i 1, 2 . ƒ
Prove, there exists a set W of four points in∈the { plane, such that for every
Λ ν every∈line AiBj , i, j 1, 2 is incident
∈ { with at least one point in W.
}
Problem 2. Two circles Γ1 and Γ2 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 Γ1
and D on Γ2. Lines AC and BD meet at E; lines AN
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.
sition
(by theofproposer).
P for which is a+ variable
If POP PH is a point
minimum is given
on BC, po-= D. COMMENT
by P
then the
Problem 4.n is
A1 A2 . . . A Letcyclic
A1 Aif2 .and
. . Aonly
n beifato
convex polygon,
each vertex n can ≥
Aj one 4. aProve
assign pair that
, cos C = .
C4OMMENT. 4 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.
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
Problem 6. Let ABCD be a convex quadrilateral with AB not parallel to
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
= DE, EF 7. Prove thatbe a convex hexagon such that AB = BC, CD
Let ABCDEF
= FA.
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.
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 T1, T2, T3, respec- tively.
Consider the symmetric images of the lines H1 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 H1 H2 , T1 T2 , which are 2 α
β and α | β , respectively. It follows that the mirror image of H1 H2 across
| − | −
T1 T2 is |parallel to AB. A similar conclusion holds for the other two mirror
images, so the triangle . formed by the three reflec- tions has sides parallel
to those of ABC. Hence there is a homothety O h taking ABC to . Now,
the claim is that O incircle of
. is inscribed in the ABC. This can be true
only if O h takes theOcircumcircle of ABC to its incircle, which suggests
defining
O h as in theO solution above.
Problem 10. Let ABCDEF be a convex hexagon such that
∠B + ∠D + ∠F = 360◦ and
AB CD EF
· · = 1.
Prove that BC DE FA
BC AE FD
·
CA
· = 1.
EF DB
COMMENT. Considering the arguments of the complex numbers on both sides
of the equality
b− c a− e f− d
a − c · f − e ·b − d = 1,
of EC and DF, and let FJ be the intersection of FA and DE. Find, with
proof, the value of the sum.
DB EC FA
+ + .
DDJ EEJ FFJ
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 A1 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 CA1 B1 .
Problem 18. Let ABC be a triangle with ∠ BAC = 60◦. Let AP bisect
∠
AQBAC andwhat
+ QB, let BQ
are bisect ∠ABC,
the angles of thewith P on ?BC and Q on AC. If AB + BP =
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
^MAB.
∠FHP = ∠BAC.
Problem 43. Let ABC O 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
RA + RC + RE ≥ 2.
Problem 45. Let the sides of two rectangles be a, b{ and} c, d {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
∠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
from O to the lines containing the sides of 2.
F . Prove that D2 − H2 ≥ P4
SELECTED PROBLEMS FROM MATHEMATICS
AND YOUTH MAGAZINE 2018
3 5 4035 >
4036 ·
12 1 3 12018
1+ + +· · · +
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
P= (b + c) 2 + ( y + z) 2
ab + bc + ca xy + yz + zx
is a constant.
E = AF =
α.
Problem 5 (5/488). Show that the following sum is a positive integer
. Σ
2
1 1 1 1 +
S=1+ +· · · + + 1+ +· · · +
2 .2017 2 .
2017 Σ2
1 1 1
+Σ
22
+···+ 2017 ·
2017 +···
Problem 6 (1/489 (Grade 6)). Find all pairs of integers (x, y) satisfying
x2 + x = 32018y + 1.
^
Problem 7 (2/489 (Grade 7)). Given a triangle ABC with B = 450,
0 ^
C^ = 30 . Let BM be one of the medians of ABC. Find the angle AMB.
Problem 8 (3/489).
Find the minimum Given
value of thereal numbers x, y satisfying 0 < x, y < 1. ).
expression
2xy − x 4xy − y+1
F = x2 + y2 +
·
Problem 9 (4/489). Given a circle ( O ) with a diameter
point C (C is different from A and B). Draw CH perpendicular AB. Onto (AB
O) pick a
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
f ( x) 10 (5/489).
≥ Suppose
α(x that
− the polynomial
3
a) , f
∀x( x ) = x ≥+ ax +0.bxc +has
that3
non-negative real solutions. Find the maximal real number α so
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.
integer
(3/491).aFind
is a divisor
positiveofintegers + 2isthat
n then na such also positiveof n + 2. Problem 18
a divisor
if the
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
ˆ
a) T IP = TK J.
b) The circle (O) and the circumcircle of the triangle KPJ are tangent
ˆ
to each other.
Problem 20 (5/491). Find integral solutions of the equation
y2 + 2y = 4x2y + 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 a term of M? Why?
15400
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 B^AC 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 equation
n3 − 5n + 10 = 2m.
Problem 29 (4/493). Given a triangle ABC with BC = a , AC = b , AB =
5
0 a.
^ ^
c , and 3B + 2C = . Prove that b + c ≤ 4
180
Problem 30 (5/493). Solve
2 the √
2 system of equations
2 0
.
x − y + √ x − y√ + = √
x + 8y + 4 x − 8 y − 4 xy = 0
m
primes p satisfying
31 (1/494 each of the
(For 6th grade)). Findfollowing numbers1)m,p n +
equalities
all natural pn = pm+n; Problem
and
2) pm + pn = pmn.
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− 4 x
Problem 33 (3/494). Solve the equation √ 2x
= 2 − 2.
x +1
2x + 1
Problem 34 (4/494). Given a right triangle ABC with the right angle
arbitrary
AH be thepoint D (DOn
altitude. =H ). opposite
the Through ray
D draw
of thethe
rayline
HAperpendicular
pick an BD.toThat
A. line
Let
intersects AC at E. Let ƒK be the perpendicular projection
of E on AH. Show that DK has a fixed length when D varies.
Problem
minimum 35
and(5/494).
maximum Given realofnumbers
values x, y such that x2 + y2 = 1. Find the
the expression
√
√
T = 4 + 5x + 4 + 5y.
Problem 36 (1/495 (For 6th grade)). Show that it is impossible to write
292018 as a sum of n consecutive positive integers for any n ∈ N, n ≥ 2.
Problem 37 (2/495 (For 7th grade)). Find the last twelve digits of the number
51040.
Problem 38 (3/495). Find integral solutions of the following systems of
equations
3a2 + 2ab
2 +2 3b22= 12 ) . (z 3)(x2 + y2) 2xy = 0
a) a +b =c − x+y=z
. ;
Problem 39 (4/495). Given a right triangle ABC with the right angle
− 1
A. In the angle B^AC draw the rays Ax, Ay such that C^Ax = A^BC;
b
1 2
^ ^
BˆAy = ACB .The ray Ax intersects the angle bisector of ACB at Q, the
2 AK
ray Ay intersects BC at K. Compute the ratio ·
AQ
x2 2x + 2
Problem 40 (5/495). Solve the equation 3.
3 −
+ 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.
^ ^
D
B^ = 45 , DC^ = CB 30 .=Outside
150 . Prove
ABC that
wethe triangle
choose D soABD
thatis equilateral.
BC =
that [(x + 43
Problem y)((3/496). + x)]23 ≥
y + z)(zGiven xyz(2x + ynumbers
non-negative + z)(2yx,+ y,z and
+ x)( + x + y) .
z.2zShow
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).
Problem 45 (5/496). Solve the system of equations
√
x2 (y − 2x2y).+ y ( xy + 1)(3 − xyΣ) = y5
x (1 −
3
2 4 = x2y2
1 1
x2 ) + x + . + ( x 2 − y2 )2
p2 (1/497
+ q2
1.2.3 .1! +. 2!. + 3!
n. +
that there exist
+ 5895 prime numbers
. Notice that n!p, q=satisfying + n! =46
Problem
(For 6th grade)). Find all positive integers n such ··
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.
Problem 48 (3/497). Suppose that n is a positive integer so that 3n + 7n is
divisible by 11. Find the remainder in the divison of 2n + 17n + 2018n2 by 11.
√ √
.y − .y 2 + 2Σ . x + 4 + 2x = 2.
Problem 54 (4/498). Given a triangle ABC with the angles satisfying
.
2x5(− 1)3 y√− x2 y + 10x3 +2y2 − 5y = 0
x +2x
y − 5 = y − 3x + x −
2π 8π 10π
Problem 57 (7/488). Prove that x02= cos + cos + cos
21 21 21 is a solu-
(n3 −n
√)
π √
Problem
a cos 65 (7/490). Find all positive integers a, b such that .8 + √32 + 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
x2 + 4 + y2 + 4 +
Problem 69 (7/491). Show that z2 + 4 = 2(x + y + z)
n→∞ , √
lim ‚. 1 + 2.1 + 3.1 + · · · + .1 + (n − 1) 1 + n = 3.
Problem 70 (8/491). Letma, mb, mc be the lengths of the medians of the triangle
with the perimeter 2. Show that
max{1 ; 3
√3 √ ,
3 r } ≤ m a + mb + m c < 2
where r is the inradius of the triangle.
Problem 71 (9/491). Assume that a, b, c are three non-negative numbers
a b3 + 1 + b c3 + 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
x3 + mx2 + n = 0. (∗)
Find m, n so the the equation (*) has three distinct non-zero real roots
u, v, t satisfying
u4 + v4 + t4 = 3.
u3 − 2n v3 − 2n t3 − 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
(a + b) + c (b + c) + a (c + a) + b2 2 5
. the system
Problem 75 (7/493). Solve √ of equations
x − 1 = 4 9 + 12y − 6y2
√4
y− 1= 9 + 12x − 6x2
Problem 76 (8/493). Given a right prism with equilateral bases ABC.AJ BJ CJ .
Let α be t√he angle between the line BC’ and the plane (A’BC). Prove that
sin α ≤ 2 3 − 3.
Problem 77 (9/493). Given positive numbers a, b. Show that
1 min a, b 2
Σ1 −
2 maxΣ(a, b) b− a
ln b + ln a
( ) a −
≤
≤ Σ − 1 Σ2 .
2 min ( a,
1 max(a, b) b )
Problem 78 (6/494). Find conditions on a, b besides a > b ≥ 1 so that
the system −
. x2 = (a − y)(a + y + 2)
x3 + y3 = x2 + 72xy + y2.
Problem 82 (6/495).xGiven2 real numbers
y x, z that 2xyz =
2 y, z such 3 1, show that
+ +
1 + xy 1 + zy ≥ .
. . .
1 + xz 4
Σ Σ Σ
Problem 83 (7/495). Suppose that the polynomial P(x) = x2018 − ax2016 +
a (a is a real parameter) has 2018 real solu√tions. Show that there exists at
least one solution x0 of P(x) with |x0| ≤ 2.
Problem 84 (8/495). Given a pyramid S.ABCD with the base ABCD is a
1
a2 + b2 + c2 ≥ (ab + bc + ca)+
2
+
(a + b)(b + c)(c + a)
.2( a + b + c)( a3 b3 + b3 c3 + c3 a3 )
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
(g(x)) = g93
( f ((9/497).
x)) x [Given continuous
a, b]; where functions
a, b are f, g: [a,Show
real numbers. b] [a, ]the
b→
that such that f
equation
f (x) = g(x) has at least one solution.
∀
Problem 94 (6/498). Solve the system of equations
√ √
2( x + 4y − 8) = . y2 + x − 3 + 1 − yΣ ×
2 2
√
× . y2 + x − 3 + 1 + y Σ
2√ y4 + 5 = x
Problem 95 (7/498). Solve the equation
4
√ 3
1 + 2log16 x2 + .4 − log x8 4
4
− = 0.
log2 x2 3
+ log2 x − log x3 + 2
2 ABC2 ( AB < points
AC), B,
Problem 96 (8/498). Given a triangle
given circle (O), with the point A can be varied 3 and2 the inscribed
C are in a
by BC.
fixed, andAtwocircle (OA
points
J
) is
andinternally tangent
O are always to same
on the (O) atsideT determined
(T is outside the
triangle
ABC) and is tangent to the sides AB, AC respectively at P, Q. The
line The lines TB, TC meet again (OJ ) respectively at E,
F ( EPQ intersects
F = T ).BC at R.that
= T, Prove
ƒ EF is parallel to the line BC.
a) The line
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
+1
( −
a1 = 12
; an = n , n = 1, 2, . . . .
a)Find lim an . 2− a √
a1 + a2 + · · · + an 2
b) Show that
n ≥ 1− for n = 1, 2, . . . .
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.
b) Prove that the Euler cirle of AMN always goes through a fixed point.
a b c
(10/490). Find 2all triples of+positive 5integers (a, =
that b, c) such7 . Problem 104
Problem 105 (11/490). The sequence (un) is determined as follows
u1 = 14, u2 = 20, u3 = 32, un+2 = 4un+1 − 8un + 8un−1
√
x3 + 6x2 + 9x ≤ x + 4 − 2.
Problem
is equal to111 (7/492).
1, find For a, b, value
the minimum c are positive numbers with the product
of the expression
1 1 1
P= + + .
2
+ bpositive
holds true fora all
2
k(a + b +a,c)b,≥c 3satisfying
+ c2 + numbers + k(ab + abc
bc +=ca1.)
Problem 116 (11/493). Given the sequence (xn)
x2 = x3 = 1
an = √ √ + √ √
3 +10 5 .7 + 3 2 5 Σn 10 5 .7 − 32 5 Σn + 2
3−
5
is a perfect square.
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.
on
120the(12/494). Let (Ia1),square
side AB.Given (I2)respectively
ABCD and be Ptheisincircles of ADP,
an arbitrary CBP.
point Problem
Assume that
DI 1, 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). √3
Problem 121 (10/495). Consider the sequence ( an ): an = n + [ n] , n
√ √
positive integers ( [ 3 n] is the integral part of 3 n ). Suppose there ex-
ists a positive
consecutive integer
natural k such
numbers withthat
p =the terms
6015 2006a+k; 1.ak+1 . ; ak+p are p + 1
; . . that
Show
×
k > 8.10 + 6.10 .
9 7
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 XA = XK.
.
Σxn
Problem 124 (10/496).
x1 = 0, xn+1 = 1The, ∀ n ∈ N ∗ .( xn ) (n ∈ N ∗ ) is defined as follows
sequence
4
Show that the limit of the sequence exists and find it.
Problem 125 (11/496). Find all functions f : R → R such that
4
f (x) f (y) − 9 xy = f (x + y), ∀x, y ∈ 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)
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 both
b) 3m − n! = 2.
les triangle
16 (Tran Quang AB = CFMJ
with Hung, and Q are
AC. P 4(190)). inside
Let ABCtriangle ABC such that Problem
be an isosce-
∠APB = ∠AQC and AP = AQ. Prove that PQ ǁ BC.
Problem 17 (Nguyen NGoc Hung, CFMJ 5(190)). Given the sum
1 1 1 1
A= + + +···+ .
23 + 3 33 + 4 43 + 5 20183 + 2019
1
Compare A
and .
6 Nam, CFMJ 6(190)). Solve the following equa- tion
Problem 18 (Mai Van
√
3
2x + 2 = x3 + 9x2 + 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.
lateral
Problem inscribed
20 (Nguyen Tan(O
in circle Ngoc, that ∠
) suchCFMJ BCD <
8(190)). Let90ABCD
o
, the diagonal AC
be a quadri-
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
M = 11 + 22 + 33 + · · · + 9999 + 100100.
Show that M has 201 digits and find the first 2 digits of M.
Problem 22 (Cao Ngoc Toan, CFMJ 2(191)). Given that
1 2 3 4 2019
S =2018 .1 + 2 + 3+ · · · + 2018Σ .
Prove that S is not a natural number.
having
Problemthe23form ab such
(Le Son Tung,that ba, (a +Find
ab, 3(191).
CFMJ 1)b,all(bthe
+ 2-digit
1)a arenumbers
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.
Problem 25 (Nguyen Ngoc Hung, CFMJ 5(191). Given real numbers x
and y satisfying
. x3 + y3 = 1948
. √
x y+y + zy + x z= = 3xyz + 2.
xz +
√ √ √
Problem 28 (Doan Van Truc, CFMJ 8(191). Given an acute triangle ABC
(AB =< CE.
AC).Prove that
E the
are centroid
BD G of triangle ADE would move alonga
D and on side AB and side AC respectively so that
fixed line when points D and E move along the segments AB and AC,
respectively.
Problem
integer 29n (Truong
so Quangthat An, CFMJ
S(n).S1(192+193).
(n + Let 1) S(=n) be87.
theofsum
the
digits of positive integer n. Find the smallest value of positive
2
doublets of natural
30 (Nguyen numbers
Ngoc Hung, (a; b)2(192+193).
CFMJ satisfying 1000
Finda all b = 1001a.b2 + Problem
+ possible
×
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.
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 ∆ABC such that ∠CEA = 120 o . Then, we construct
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) with
o
∠EDF = 120
that PF = CE.and F and B are on the same side of DE. Show
Problem 33 (Lai Quang Tho, CFMJ 5(192+193). Solve the following system of
equations
. 2x2 − y3 + 2xy + 2xy2 = 3
(CA = CB) with ∠C = 15o. Suppose that there exists point D inside
∆ABC such th√at ∠ ADB = 105o 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
− + +
= − + 6 10 15 21 28 +··· + .
1 3 − − 36 4950
Compare A with 814.
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:
∠ ABC = 36o, ∠CBD = 30o, ∠ BAD = 81o, ∠ CAD = 27o.
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 xpositive
2
+ y2 integers x, y, z, t satisfying:
=
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
√
(x + 3) −x2 − x + 48 = x − 24.
Problem 43 (Luu Ly Tuong, CFMJ 7(194). n is a positive integer (n
≥
3).
a) Find
positive the greatest
integers positive integer
not exceeding k(n) so
k(n), there that among
always exists n3 arbitrary
num- bersdistinct
(not
necessarily
to the other distinct)
one. in which the sum of two of them is equal
b) Find
positive the greatest
integers positivehinteger
not exceeding h(nalways
(n), there ) so that among
exists n arbitrary
3 dis- distinct
tinct numbers in
which the sum of two of them is equal to the other one.
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
proofs of P(3)
inequality are provided
are given in Section
at the end 3, and
of Section the relativesome
2. Moreover, inequalities
differentare cyclic
given in
Sections 4, 5.
1 Introduction
In 1954 Harold Seymour Shapiro proposed the inequality for a cyclic sum in
n variables as follows
x1
xn−1 n
x2 · · xn
+ + xn + x1 + ≥ ,
x2 + x3 + · 2
(1)
x3 + x4 x1 + x2
where xi 0,≥xi + xi+1 > 0 and xi+n = xi for i 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
x1
P(n, p, q) := x2 xn
+ xn−1 +
+ · · + pxn + qx1
px2 + qx3 px3 + qx4 · 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 Cauchyxinequality, we have x2
1
( x + x + x )2 = x ( px + qx ) +
..1 2 3 . .1
px2 + qx3 px3 + qx1 .x2 ( px3 + qx1 )
. .
+ x3 2 3
px1 + qx2
Σ
2 ≤ P(3, p, q)(p + q)(x1x2 + x2x3 + x3x1).
x3(px1 + qx2)
It follows that (x1 + x2 + x3)2 3
.
≥
( p + q)(x1x2 + x2x3 + x3x1) p+q
P(3, p, q)
≥
≥
Obviously, (2) is true for every n 4 if p = 0 or q = 0.
it isIntrue
thiswhen
note, pby studyingq,the inequality
and (2) pin <
false when theq.case n = 4, we
Moreover, we show
give athat
case n = 5. It is worth ≥ if p <the
sufficient condition of p,saying thatwhich
q under then the inequality
q, inequality (2) is true(2)
in is
thefalse
the
next section. For short, for any n 3 and p = q = 1, let P(n) stand for every
even n 4. Two open questions are discussed at the end of
≥
≥
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-
(x1+x2 + x3 + x4)2 ≤
Σ
P(4, p, q) x1(px2 + qx3) + x2(px3 + qx4) + x3(px4 + qx1) + x4(px1 + qx2) .
Σ
Hence
(x1 + x2 + x3 + x4)2
.
P(4, p, q) px1x2 + 2qx1x3 + px1x4 + px2x3 + 2qx2x4 + px3x4
≥
It is an equality if and only if
t3 = x3 − x4.
≥ We thus≥have
This implies that ω is positive. P(4, p, ≥ q) 4. It is
easily seen that if p q, i.e. p ∈ 12, then ω 0 for all t1, t2, t3 R.
Now let us consider the cases when ω vanishes. This ≥
depends con-
siderably at t1 = x1 of p with q. If p x=3 q,= i.e.
on the0 comparison
form ω attains p =t 1 ,=then
x2 the quadratic
0 and 2 2x4 = 0. By
(4) we
−
assert that P(4, p, q) = 4 whenever x1 = x3 and x2 = x4. Also, if
p > 12, then ω vanishes if and only if
t3 = x3 − x4 = 0.
Combining these facts with (4) we conclude that P(4, p, q) = 4 when
x1 = x2 = x3 = x4.
Now we give a counter-example to the inequality (3) in the case p < q,
(6) ⇔ p(2q − 1)(a2 + b2) + 2(p2 + q2 − q)ab > 0 ⇔ p(1 − 2p)( a − b)2 > 0.
+ (2 − 5q)x3x5 + (2 − 5p)x4x5 ≥ 0.
The matrix of ϕ in an appropriate system of basic vectors is of the form
2 2 − 5p 2 − 5q 2 − 5q 2 − 5p
2 − 5p 2 2 − 5p 2 − 5q 2 − 5q
1
B = 2 2− 5q 2 − 5p 2 2 − 5p 2 − 5q
2 − 5q 2 − 5q 2 − 5p 2 2 − 5p
− 5p
which has the principal2minors
2 − 5q 2 − 5q 2 − 5p 2
D 1, D
5p(4 − 5p) 25q(5pq − 1) 125(1 − 5pq )2
,D ,D , D 0.
1 = 2 = 3 = 4 = 5 =
4 4 16
This implies that the necessary and sufficient condition for the positivity
of the quadratic form ϕ is
5− √
√ 5+ 5
5 ≤p≤ .
10 10
We thus obtain a sufficient condition under which the inequality (2)
holds for n = 5. √
5− 5 5+ 5
Theorem 16. If √
10 ≤p≤ , then (2) is true for n = 5.
10
Remark
According3.toConsider
the proofthe inequality
of the (2) in
second part of the case n≥
Theorem 154,this
n isinequality < q.
even, andispfalse.
andWe observe
Theorem 16that p = 0 oranswer
if detailed
is the q = 0, for
thenn it=is5.the case for every n ≥ 4,
3 Proof of P(3)
a b c
S= + + ,
b+c a+c a+b
b c a
M = b + c+ a + c+a + b,
c a b
N= + + .
b+c a+c a+b
Hence,
This implies
M + N + 2S ≥ 6.
2S ≥ 3.
The equality occurs if and only if a = b = c.
Proof 2. By calculating and applying the AM-GM inequality, we have
a b c
a b c
=. + 1Σ + . + 1Σ + . + 1Σ − 3
+ +
b+c a+c a+b b+c c+a a+b
1 1 1
= ( a + b + c) . + + Σ− 3
b+c c+a a+b
1 1 1 1
= .(b + c) + (c + a) + ( a + b)Σ . + + Σ− 3
2 b+c c+a a+b
3
≥ 3.3 (b + c)(c + a)( a + b) · 3. 1 1
9 3 +
= − 3= , c+a a+b
2 2
We have .
1− x
f ( x) =
−2
f J (x) =
> 0,
(x − 1)3
for all x ∈ (0, 1). We deduce that f is a convex function defined on (0, 1).
By the Jensen’s inequality, we have a+b+c
f ( a ) + f ( b) + f ( c) ≥ 3 f . Σ= 3 f
1 3
3 ( )= .
3 2
The inequality is proved. The equality occurs if and only if a = b = c.
Proof 5. Put
xy = + a,
= bc + c,
z = a + b.
y+z− x
It follows a= 2 ,
z+x− y
b= 2 ,
x+y− z
c= 2 .
The inequality is now equivalent to the following ones:
y+z− x z+x− y x+y− z 3
+ + ≥ ,
y + 2z − x z + x2 − y x + y2− z 2
x + y + z ≥ 3.
x y z
Put
x
u=
,
y
y
v=z,
w= .
The last inequality becomes z x
1 1 1
( + u) + ( + v) + ( + w) ≥ 6,
u numbers.
where u, v, w are positive v This inequality
w holds as
1
t + ≥ 2,
t
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
k
+b+ c
1
k−b k−c
3 ≥ .
T= −a 2
Indeed, consider the function
t
f (t) = .
k− t
As Proof 4 we
inequality, havethat f is a convex function on interval (0, k). By the Jensen’s
shows
a+b+c
k
3 3 1
T 2 = .
≥ a+3b+c
= 3
k 2
k−
The inequality is proved. The equality occurs if and only if a = b = c.
Proof 7. By the Cauchy-Schwarz inequality, we have
( a + b + c) 2
Σ √a Σ2
. √ . √
. b c
= √ a ( b + c) + √ . b(c + a) + √ . c ( a + b)
b+c c+a a+b
a b c
≤ 2. + + Σ (ab + bc + ca).
Hence,
b+c a+c a+b ( a + b + c) 2 3
a + b + c≥ ≥ .
b + c a + c a + b 2(ab + bc + ca) 2
b+c
a+b a+c a+b
t= 2 .
We see that
a2 + b2 + c(a + b) c
f
(a, b, c) = 2t2 + 2tc c
≥ + = f (t, t, c).
t2 + c2 + 2ct 2t
This implies
3
f ( a, b, c) ≥ f (tJ , tJ , tJ ) = ,
2
where
a+b+c
tJ = 3 .
Similarly, ≥ −1 4−(ac+−b a+ c) 8b
c+a
8b
c+a c+a = .
b 4b
+ 1
− 4−(aa+−b b+ c) 8c
≥ a+b
8c
= .
a+b a+b
4c +11
c
We deduce
a + b + c≥ 6(a + b +=c) . 3
b + c a + c a + b 4(a + b + c) 2
The inequality is proved. The equality occurs if and only if a = b = c.
HOMC 2019 - Scientific seminar, Hanoi, April 3,
2019
Proof 11. We shall prove that
a ≥ 3
b+c 3a 2
3 3 . (8)
3
2(a 2 + b 2 + c 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
a 32 + b 32 + b 23≥ 3a 21b
a 2 + c 2 + c 2 ≥ 3a 2 c.
Hence, the above inequality holds. Similarly, we have two more inequal-
ities:
b 3b 32
≥
c+a 3 3 3
(9)
2(a 2 + b 2 + c 2 )
From the inequalities (8), (9), (10) we derive
a b c 3(a 32 + b 32 + c 23 ) 3
+ + ≥ 3 3 = .
b+c a+c a+b 3
2
2(a 2 + b 2 + c 2 )
143
HOMC 2019 - Scientific seminar, Hanoi, April 3, 2019
a3 + a3 + b3 ≥ 3a2b.
144
b3 + b3 + c3 ≥ 3b2c.
It follows c3 + c3 + a3 ≥ 3c2a.
a + b + c = 1.
By the AM-GM inequality, we have
a 9 a ( b + c)
+ . a 9a(b + c) = 3a,
b+c 4 ≥2 b + c. 4
b 9b(c + a)
+ . b 9b(c + a) = 3b,
c+a 4 ≥2 c+ a. 4
c 9 c ( a + b)
+ . c 9c( a + b) = 3c.
≥2 .
Equivalently,
a b c
a+b 4
9
+ + ≥ 3− (ab + bc + ca). (11)
+ c prove
In other side, web can a + cthat a + b 2
1 1
ab + bc + ca ≤ 3(a + b + c)2 = . 3 (12)
HOMC 2019 - Scientific seminar, Hanoi, April 3,
2019
From (11) and (12) it follows
a + b + c≥ . 3
b+c a+c a+b 2
Equivalently, +c 2
c
+1
+1 c
c
x + 3
y ≥ − 1.
y+1 x+1 2 x+y
Applying the AM-GM inequality, we have
x+1 y+1
+ ≥ 2.
Or, y+1 x+1
x + y ≥ 2− 1− 1.
y+1 x+1 x+1 y+1
We shall prove that
1 1 3 1
2− − .
Equivalently,
≥ −
x+1 x+y y+1 2
1 1 1 1
− ≥ − ,
2 y+1 x+1 x+y
y− 1
145
HOMC 2019 - Scientific seminar, Hanoi, April 3, 2019
y− 1
.
≥
2(y + 1) (x + 1)(x + y)
146
only if since xa
inequality is true, = ≥ y b 1. The equality
= c. The
occurs last
if and
Proof
assume15.
thatAsa the
≥ babove
≥ 1. proof, oneprove
We shall that that c = 1. Moreover, we can
can assume
a b 1 3
+ + ≥ .
Put b+1 a+1 a+b 2
A = a + b, B = ab.
The inequality becomes
a2 + b2 + a + b 1 3
+ ≥ .
Equivalently, (a + 21)(b + 1) a + b 2
A 2B + A 1 3
− + ≥ ,
A+B+1 A 2
Hence,
2A3 − A2 − A + 2 ≥ B(7A − 2).
and
7A − 2 > 2(a + b − 1) > 0,
( A − 2)2(A + 2) ≥ 0.
occurs if
inequality holds, and
as A only ≥ 2.if The proof
a is=completed.
b =The equality
c. The last
Proof 16. Put
a b c
M( 3) : = + + .
b+c a+c a+b
By the Cauchy-Schwarz inequality, we have
( a + b + c) 2
Σ. . . Σ2
. . .
a
. a(b + c) + . b b ( c + a) + c.
= c ( a + b)
b+c c+a a+b
≤ M(3)Σa(b + c) + b(c + a) + c(a + b)Σ. It
follows
( a + b + c )2 a2 + b2 + c2
M =1+ .
2(ab + bc + ca) 2(ab + bc + ca)
(3) ≥
We can prove that
a2 + b2 + c2 1
≥ .
This follows 2(ab + bc + ca) 2
3
M( 3) ≥ .
2
1b+c 1 1
Σ(b + c) + ( a + c ) + ( a + b ) Σ Σ + a + c + a + b Σ ≥ 9.
The last inequality holds by the Cauchy-Schwarz inequality. In particu-
lar, we have
1 1 1
Σ(b + c ) + ( a + c) + ( a + b ) Σ Σ + + Σ
b+c a+c a+b
= + Σ + + Σ.
32 21 (a +(ac)(
−b + c) (a + b()(a b−+ c) (a + b)((ab+−c)c)2
b) 2 c )2
and
we deduce only if
that the inequality a
is proved. = equality
The b occurs= if c. By this
Proof 19. (Nesbitt, 1903). Put
xy =
= ab +
+ b,c,
z = c + a.
a + b + c+ d≥ 2. (13)
b+c c+d d+a a+b
a b c d
S= + + + ,
b+c c+d d+a a+b
b c d a
M = b + c+ c + d+ d + a+ a + b,
c d a b
N= + + + .
Evedently,
b+c c+d d+a a+b
M + N = 4.
By the AM-GM inequality we have
Thus,
M + N + 2S ≥ 8.
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
proof
applying of M(3) as
the Cauchy-Schwarz presented above,
inequality, which we to obtain
is similarly the By
M ( a + b + c + d )2
. (14)
We then have
(4) ≥
ab + ac + bc + bd + cd + ca + da + db
(a + b + c + d + e )2
≥ .
that+ bc + cd + de + ea + ac + ce + eb + bd + da
We shall prove ab
( a + b + c + d + e) 2 5
≥ .
ab + bc + cd + cd + de + ea + ac + ce + eb + bd + da 2
Indeed, this inequality is equivalent to the followings:
2(a2 + b2 + c2 + d2)
≥ (ab + bc + cd + cd + de + ea + ac + ce + eb + bd + da),
(a − b)2 + (b − c)2 + (c − d)2 + (d − e)2 + (e − a)2 + (a − c)2+
+ (c − e)2 + (e − b)2 + (b − d)2 + (d − a)2 ≥ 0.
The equality occurs if and only if a = b = c = d = e.
S := ab + bc + cd + de + e f + f a + ac + ce + ea + bd + d f + f b.
We then have
2S = (a + b + c + d + e + f )2
− (a2 + b2we
By the Cauchy-Schwarz inequality, + chave
2
+ d2 + e2 + f 2 + 2ad + 2bd + 2c f ).
a2 + b2 + c2 + d2 + e2 + f 2 + 2ad + 2bd + 2c f
= (a + d)2 + (b + e)2 + (d + f )2
1 . (19)
Hence, ≥ (a + b + c + d + e + f ) 2
3
2
2S ≤ ( a + b + c + d + e + f ) 2.
3
Combining
and only if a = b = c = d = e = receive
(18) and (19) we f . M( 6) ≥ 3. The equality occurs if
5 Generalizations of P(3)
Problem 2. Let a, b, cr be arbitrarilyr real numbers. 3
Prove that
a + b + c r , (20)
pbr + qcr pcr + qar
≥
where p, q > par + qbr p+q
0.
Proof. By the Cauchy-Schwarz inequality, we have
ar + b r
c r
c2r
pbr + qcr + b2r
pcr + qar +
par + qbr pbrcr + qbrar pcrar + qbrcr
= r r a2r rr r +r r 2
pa b + (qa
a c+ b + c ) 3
= ≥ .
(p + q)(arbr + brcr + crar )
p+q
F(0)
O x
+ c a (c
It follows F (t) ≥ 0 for every t ≥ 0. This implies that F is monotone on the
J
interval (0, ∞). We deduce that if α ≥ β, then
aα + bα + cα≥ aβ + bβ + cβ .
b α + cα cα + aα aα + b α
The inequality is proved.
bβ + cβ cβ + aβ a β + bβ
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
a2 + b2 + 2
c≥ +a + b . c
2 2 2
b +c c +a a +b 2 2 2 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
3 a3 3 + 3 b33 + 3 c ≥
3 +a2 2 + b22 . c2
b +c c +a a + b3 b 2 + c2 c + a2 a + b2
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
ak + b k + ck ≥ ak−1 + bk−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:
k
k k
+ 2 akn−1 n
a1
pa2 + qa3 a a
+··· pan + qa1 + pa1 + qa2
pa3 + qa4
+
a1k−1 + a2k−1 + k−1
· · · + an−1 + ak−1
n
.
≥
Proof. Put
p+q
ak ak
ak + ak2 n−1
pa2 +1qa3 n
+ .
M :=
pa3 + qa4
+··· pa1 + qa2
We denote pan + qa1
+
an+1 := a1, an+2 := a2.
For each one of ak−1i , i = 1, 2, . . . , n we have
. .
k−1
ak−1 a i
qa i+2) .
= ak−1 i+1 +
i pi+1 + q
a a i+2 · pa i (
By the Cauchy-Schwarz inequality, we have
. Σ2
a1k−1 + ak−1
2 +···+ ak−1
n−1 + ak−1
n
≤ MΣp(a1k−2 a2 + ak−2 k−2 k−2
2 a3 + · · · + a n−1 an + a n a1)
1 2 n−1 n
k−2 k−2 k−2 k−2
+ q (a a3 + a a4 + · · · + a a1 + a a2).Σ (25)
By proving similar to the homogeneous inequalities as above, we obtain the
following inequalities:
k−2 k−2 k−2 k−2
a1 a2 + a 2 a3 + · · · + a n−1 an + a n a1
1 2 n−1 n
k−1 k−1 k−1 k−1
≤a +a +···+a +a , (26)
k−2 k−2 k−2 k−2
a1 a3 + a 2 a4 + · · · + a n−1 a1 + a n a2
≤ a1k−1 + ak−1
2
+ · · · + ak−1
n−1
+ ak−1
n
. (27)
Combining (25), (26), and (27) we receive
. Σ2
a1k−1 + ak−1
2 +···+ ak−1
n−1 + ak−1
n
≤ M(p + q)(ak−1 + ak−1 + · · · + ak−1 + ak−1).
1 2 n−1 n
Remark
Shapiro’s 4.
cycle one.k =Thus
When 1 and p = qnot
it does = 1,true
thefor
inequalityin
every n Problem 6 turns out
3. However, the
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 k k
1 2 n
a + a akn−1 + a
a2 + a3 a3 + a4 + · · · + n + a1 a1 + a2
ak−1
1
+ aak−1
2
+ · · · + ak−1 + ak−1n
2 n−1
≥ .
.b + c
. 2b c+a
Σ
Σ
2c + α − 1 ≥ α 2b ,
α b+c
+α− 1≥ α ,
c 2+c a
a+b
+α− 1≥ α ,
b + c2a c +2ba 2c
a+b a + b
( α − 1) . + + Σ ≥ (α − 1)3.
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 α
be given. Then 1
≥
aα + bα + cα≥ . 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
a
2 = .b +α c .a ( b + c) ,
1+α
b
c+a
b 2 = . .b ( c + a ),
1+α α
c
c
2 = a+b
.c ( a + b.
Therefore, we apply the Cauchy-Schwarz inequality to receive
2 2 2
. Σa b 1+ α Σ 2
1+ α 1+ α Σ . aα cα
Σ.
a +b +( c+
c ) + b( c + ) + c( a +
bα
a b ) + +
b+c c+a a+b
The proof would be completed by the following inequality
2 2 2
.a 1+α + b 1+α + c 1+α Σ2 ≥ 3( ab + bc + ca ). (28)
For the proof of (28), we 1have
+α 1+α
1+α
a
≥ a + b + c. b c (29)
2 + 2 + 2
a 2 + b1 + α+ c
2
3 + 3α
≥ .a + b + cΣ + 3 − 2
2
α−1 3 + 3α
≥ ( a 2+ b + c) + α − 1.3 + 3 − 2 =3a ++ 3α
b + c.
= ( a + b + c) + Σ .a + b + c ΣΣ + 3 −
2 2
Thus, (29) is proved. It follows that
2 2 2
.a 1+α + b 1+α + c 1+α Σ2 ≥ ( a + b + c)2 . (34)
In other side, we have