Está en la página 1de 23

Inequalities of Olympiad

Caliber
Jos e Luis Daz-Barrero
RSME Olympiad Committee
BARCELONA TECH
Jos e Luis Daz-Barrero
RSME Olympic Committee
UPCBARCELONA TECH
jose.luis.diaz@upc.edu

Basic facts to prove


inequalities
Hereafter, some useful facts for proving inequalities are presented:
1. If x y and y z then holds x z for any x, y, z R.
2. If x y and a b then x + a y + b for any x, y, a, b R.
3. If x y then x + z y + z for any x, y, z R.
4. If a b and c > 0 then ac bc for any a, b R.
5. If a b and c < 0 then bc ac for any a, b R.
6. If x y and a b then xa yb for any x, y R
+
or a, b R
+
.
7. Let a, b, c R such that a b c then a + b a + c b + c.
8. Let x, y, z, t R such that x y z t then x+y+z x+y+t x+z+t
y + z + t.
9. If x R then x
2
0 with equality if and only if x = 0.
10. If A
i
R
+
and x
i
R, (1 i n) then holds
A
1
x
2
1
+ A
2
x
2
2
+ A
3
x
2
3
+ . . . + A
n
x
2
n
0,
with equality if and only if x
1
= x
2
= x
3
= . . . = x
n
= 0.
1
Jos e Luis Daz-Barrero
RSME Olympic Committee
UPCBARCELONA TECH
jose.luis.diaz@upc.edu

Inequalities Warm-up
1. Prove the following statements:
1. If ab > 0, then holds
a
b
+
b
a
2. If ab < 0, then holds
a
b
+
b
a
2.
2. Let a > 1, b < 1. Then, a + b > 1 + ab holds.
3. Let a, b be positive numbers, then holds (a + b)
_
a + b
2
a

b + b

a.
4. Let a, b be positive numbers, then holds
1
2
(a + b) +
1
4

_
a + b
2
.
5. If a > 1, then
1
a 1
+
1
a
+
1
a + 1
>
3
a
holds. (Pietro Mengoli 16251686)
6. Let a, b be positive numbers such that a + b = 1, then holds
_
a +
1
a
_
2
+
_
b +
1
b
_
2

25
2
2. Prove the following statements:
1. Let a, b, c be nonnegative real numbers, then holds
6a + 4b + 5c 5

ab + 3

bc + 7

ca.
2. Let a, b, c be positive numbers such that a b c, then holds
a
3
1/a + 1/b + 1/c

abc
a + b + c
3

a
2
+ b
2
+ c
2
3
c
3. Let a, b, c be the length of the sides of a triangle ABC. Then, holds

a(a + c b) +

b(a + b c) +

c(b + c a)
_
(a
2
+ b
2
+ c
2
)(a + b + c).
4. Let a, b, c be positive numbers such that a + b + c = 1, then holds
_
1 +
1
a
__
1 +
1
b
__
1 +
1
c
_
64.
5. Let a, b, c be positive numbers, then holds
a
b + c
+
b
c + a
+
c
a + b

3
2
(Nesbitt 1903)
6. Let a, b, c be positive numbers lying in the interval (0, 1]. Then holds
a
1 + b + ca
+
b
1 + c + ab
+
c
1 + a + bc
1
I hope you enjoy solving the preceding proposals and be sure that I will be very
pleased checking and discussing your nice solutions.
2
Jos e Luis Daz-Barrero
RSME Olympic Committee
UPCBARCELONA TECH
jose.luis.diaz@upc.edu

SET 1
1. Let x, y, z be strictly positive real numbers. Prove that
_
x
y
+
z
3

xyz
_
2
+
_
y
z
+
x
3

xyz
_
2
+
_
z
x
+
y
3

xyz
_
2
12
2. Let x, y, and z be three distinct positive real numbers such that
x +
_
y +

z = z +
_
y +

x
Prove that 40xz < 1.
3. Let a, b, and c be positive real numbers. Prove that
_
5a b
b + c
_
2
+
_
5b c
c + a
_
2
+
_
5c a
a + b
_
2
12
4. Let a, b, c be positive real numbers such that a + b + c = 2. Prove that
bc
4

3a
2
+ 4
+
ca
4

3b
2
+ 4
+
ab
4

3c
2
+ 4

2
3
4

3
5. Let a, b, c be three positive numbers such that ab + bc + ca = 1. Prove that
_
a
2
+ b
2
+ c
2
_
4
_
a
2 3
_
a
b + c
+ b
2 3

b
c + a
+ c
2 3
_
c
a + b
_3

1
2
6. Let a, b, c be three positive numbers such that a
2
+ b
2
+ c
2
= 1. Prove that
_
1
a
3
(b + c)
5
+
1
b
3
(c + a)
5
+
1
c
3
(a + b)
5
_
1/5

3
2
I hope you enjoy solving the preceding proposals and be sure that I will be very
pleased checking and discussing your nice solutions.
3
Jos e Luis Daz-Barrero
RSME Olympic Committee
UPCBARCELONA TECH
jose.luis.diaz@upc.edu

Solutions Warm-up
1. Prove the following statements:
1. If ab > 0, then holds
a
b
+
b
a
2. If ab < 0, then holds
a
b
+
b
a
2.
2. Let a > 1, b < 1. Then, a + b > 1 + ab holds.
3. Let a, b be positive numbers, then holds (a + b)
_
a + b
2
a

b + b

a.
4. Let a, b be positive numbers, then holds
1
2
(a + b) +
1
4

_
a + b
2
.
5. If a > 1, then
1
a 1
+
1
a
+
1
a + 1
>
3
a
holds. (Pietro Mengoli 16251686)
6. Let a, b be positive numbers such that a + b = 1, then holds
_
a +
1
a
_
2
+
_
b +
1
b
_
2

25
2
Solution 1.1 From the inequality (x 1)
2
0 or x
2
+ 1 2x immediately follows
x + 1/x 2 after division by x > 0. Equality holds when x = 1. Likewise, from
(x + 1)
2
0 or x
2
+ 1 2x we get x + 1/x 2 after division by x < 0 (fact
5). Equality holds when x = 1. An alternative proof can be obtained nding the
maximum and minimum of the function f(x) = x +1/x. Finally, putting x = a/b in
the preceding the inequalities claimed are proven.
Solution 1.2 From a > 1 and b < 1 or a 1 > 0 and 1 b > 0, we have that
a + b 1 ab = a(1 b) + b 1 = (a 1)(1 b) > 0
and we are done.
Solution 1.3 We have a + b 2

ab and
_
a + b
2
=

a)
2
+ (

b)
2
2

a +

b
2
on account of mean inequalities. Multiplying up the preceding we get the inequality
claimed. Equality holds when a = b.
Solution 1.4 We observe that
1
2
(a + b) +
1
4

_
a + b
2
or equivalently
a + b
2

_
a + b
2
+
1
4
=
_
1
2

_
a + b
2
_
2
0
4
which trivially holds with equality when a + b = 1/2.
Solution 1.5 Note that the inequality given is equivalent to
1
a 1
+
1
a + 1
>
2
a
. Now
applying HM-AM inequality to the positive numbers a 1 and a + 1 yields
2
1
a1
+
1
a+1

(a 1) + (a + 1)
2
= a
with equality if and only if a1 = a+1 which is impossible. So, the inequality given
is strict.
Solution 1.6 On account that x
2
+ y
2
2
_
x+y
2
_
2
, we have
_
a +
1
a
_
2
+
_
b +
1
b
_
2

1
2
__
a +
1
a
_
+
_
b +
1
b
__
2
=
1
2
_
a + b +
1
a
+
1
b
_
2
=
1
2
_
1 +
1
ab
_
2

25
2
because from a + b = 1 immediately follows
1
2
=
a + b
2

ab and
1
ab
4. Equality
holds when a = b = 1/2 and we are done.
Notice that this inequality is a generalization of the well-known inequality
_
sin
2
x +
1
sin
2
x
_
2
+
_
cos
2
x +
1
cos
2
x
_
2

25
2
2. Prove the following statements:
1. Let a, b, c be nonnegative real numbers, then holds
6a + 4b + 5c 5

ab + 3

bc + 7

ca.
2. Let a, b, c be positive numbers such that a b c, then holds
a
3
1/a + 1/b + 1/c

abc
a + b + c
3

a
2
+ b
2
+ c
2
3
c
3. Let a, b, c be the length of the sides of a triangle ABC. Then, holds

a(a + c b) +

b(a + b c) +

c(b + c a)
_
(a
2
+ b
2
+ c
2
)(a + b + c).
4. Let a, b, c be positive numbers such that a + b + c = 1, then holds
_
1 +
1
a
__
1 +
1
b
__
1 +
1
c
_
64.
5. Let a, b, c be positive numbers, then holds
a
b + c
+
b
c + a
+
c
a + b

3
2
(Nesbitt 1903)
6. Let a, b, c be positive numbers lying in the interval (0, 1]. Then holds
a
1 + b + ca
+
b
1 + c + ab
+
c
1 + a + bc
1
5
Solution 2.1 Applying AM-GM inequality, we have
5

ab + 3

bc + 7

ca 5
_
a + b
2
_
+ 3
_
b + c
2
_
+ 7
_
c + a
2
_
= 6a + 4b + 5c
Equality holds when a = b = c and we are done.
Solution 2.2 We begin proving that a
3
1
a
+
1
b
+
1
c
. Indeed, this inequality is equi-
valent to
a
b
+
a
c
2 which trivially holds on account of the fact that 0 < a b c.
Next we prove that
3

abc
a + b + c
3
To do it, we need the inequality a
2
+b
2
+c
2
ab+bc+ca which follows immediately
adding up the trivial inequalities a
2
+ b
2
2ab, b
2
+ c
2
2bc, c
2
+ a
2
2ca, and
the identity a
3
+b
3
+c
3
3abc = (a +b +c)(a
2
+b
2
+c
2
ab bc ca) (that can
be easily checked). Combining the preceding results, we get a
3
+b
3
+c
3
3abc 0.
Now putting a = x
3
, b = y
3
, c = z
3
, we have
x
3
+ y
3
+ z
3
3xyz
x
3
+ y
3
+ z
3
3
xyz
a + b + c
3

abc
Applying this inequality to the positive numbers 1/a, 1/b and 1/c yields
1
a
+
1
b
+
1
c
3

3
_
1
abc
Inverting terms immediately follows
3
1
a
+
1
b
+
1
c

abc. To prove that


a + b + c
3

_
a
2
+b
2
+c
2
3
we square both sides and we get 3(a
2
+ b
2
+ c
2
) (a + b + c)
2
or
equivalently, a
2
+ b
2
+ c
2
ab + bc + ca. It holds as was proven before. Finally, the
inequality
_
a
2
+b
2
+c
2
3
c trivially holds after squaring its both sides and rearranging
terms.
Solution 2.3 Since a, b, c are the lengths of the sides of a triangle, then a + b >
c, b + c > a and a + c > b. Then, we can write

a(a + c b) =
_
a(a + c b)
_
a + c b,

b(a + b c) =
_
b(a + b c)
_
a + b c,

c(b + c a) =
_
c(b + c a)
_
b + c a.
Applying CBS inequality to the vectors u =
_
_
a(a + c b),
_
b(a + b c),
_
c(b + c a)
_
and v =
_
a + c b,

a + b c,

b + c a
_
yields
_

a(a + c b) +

b(a + b c) +

c(b + c a)
_
2

_
a(a + c b) + b(a + b c) + c(b + c a)
_ _
a + c b + a + b c + b + c a
_
= (a
2
+ b
2
+ c
2
)(a + b + c)
from which the statement follows. Equality holds when a = b = c and we are done.
6
Solution 2.4 We have
_
1 +
1
a
__
1 +
1
b
__
1 +
1
c
_
= 1 +
1
a
+
1
b
+
1
c
+
1
ab
+
1
bc
+
1
ca
+
1
abc
1 + 9 + 27 + 27 = 64
on account that from a + b + c = 1, and applying mean inequalities, immediately
follows
1
abc
27,
1
a
+
1
b
+
1
c
9 and
1
ab
+
1
bc
+
1
ca
=
c
abc
+
a
abc
+
b
abc
=
1
abc
27.
Equality holds when a = b = c = 1/3.
Solution 2.5 We have
a + b
b + c
+
b + c
a + b
+
b + c
c + a
+
c + a
b + c
+
c + a
a + b
+
a + b
c + a
6
or
1 +
2a
b + c
+ 1 +
2b
c + a
+ 1 +
2c
a + b
6
from which immediately follows
a
b + c
+
b
c + a
+
c
a + b

3
2
Equality holds when a = b = c.
Solution 2.6 Since 1 + ab = (1 a)(1 b) + a + b, then
1 + c + ab = (1 a)(1 b) + a + b + c a + b + c
Likewise, we have
1 + a + bc = (1 b)(1 c) + a + b + c a + b + c,
1 + b + ca = (1 c)(1 a) + a + b + c a + b + c.
Therefore,
a
1 + b + ca
+
b
1 + c + ab
+
c
1 + a + bc

a
a + b + c
+
b
a + b + c
+
c
a + b + c
= 1
Equality holds when at least two of the a, b, c are equal to one, and we are done.
7
Jos e Luis Daz-Barrero
RSME Olympic Committee
UPCBARCELONA TECH
jose.luis.diaz@upc.edu

Solutions
1. Let x, y, z be strictly positive real numbers. Prove that
_
x
y
+
z
3

xyz
_
2
+
_
y
z
+
x
3

xyz
_
2
+
_
z
x
+
y
3

xyz
_
2
12
Solution 1. Applying AM-QM inequality, yields
1
3
_
_
x
y
+
z
3

xyz
_
2
+
_
y
z
+
x
3

xyz
_
2
+
_
z
x
+
y
3

xyz
_
2
_

1
9
_
x
y
+
y
z
+
z
x
+
x + y + z
3

xyz
_
2
from which follows
_
x
y
+
z
3

xyz
_
2
+
_
y
z
+
x
3

xyz
_
2
+
_
z
x
+
y
3

xyz
_
2

1
3
_
x
y
+
y
z
+
z
x
+
x + y + z
3

xyz
_
2

1
3
_
3
3
_
x
y

y
z

z
x
+
x + y + z
3

xyz
_
2
=
1
3
_
3 +
x + y + z
3

xyz
_
2
12
because
x + y + z
3

xyz
3 on account of AM-GM inequality. Equality holds when x =
y = z and we are done.
Solution 2. Setting a =
x
3

xyz
, b =
y
3

xyz
and c =
z
3

xyz
into the statement yields
_
a +
b
c
_
2
+
_
b +
c
a
_
2
+
_
c +
a
b
_
2
12
To prove the preceding inequality we set u =
_
a +
b
c
, b +
c
a
, c +
a
b
_
, v = (1, 1, 1)
into the CBS inequality and we obtain
_
a +
b
c
_
2
+
_
b +
c
a
_
2
+
_
c +
a
b
_
2
8
=
1
3
(1
2
+ 1
2
+ 1
2
)
_
_
a +
b
c
_
2
+
_
b +
c
a
_
2
+
_
c +
a
b
_
2
_

1
3
__
a +
b
c
_
+
_
b +
c
a
_
+
_
c +
a
b
__
2
Now, taking into account that abc = 1 and applying the AM-GM inequality twice, we
get
_
a +
b
c
_
+
_
b +
c
a
_
+
_
c +
a
b
_
3
3

_
a +
b
c
__
b +
c
a
__
c +
a
b
_
3
3

2
3
_
ab
c
bc
a
ca
b
= 6
Therefore, on account of the preceding, we have
1
3
__
a +
b
c
_
+
_
b +
c
a
_
+
_
c +
a
b
__
2
12
and we are done. Note that equality holds when x = y = z.
2. Let x, y, and z be three distinct positive real numbers such that
x +
_
y +

z = z +
_
y +

x
Prove that 40xz < 1.
Solution. From x +
_
y +

z = z +
_
y +

x we obtain
x z =
_
y +

x
_
y +

z
and
(xz)
_
_
y +

x +
_
y +

z
_
=
_
_
y +

x +
_
y +

z
__
_
y +

x
_
y +

z
_
=

z
Since x = z, dividing by both sides of the preceding expression by

z yields
_
x +

z
_
_
_
y +

x +
_
y +

z
_
= 1
On the other hand, since x, y, and z are positive, then
4

x <
_
y +

x and
4

z <
_
y +

z. Therefore,
_
x +

z
_ _
4

x +
4

z
_
<
_
x +

z
_
_
_
y +

x +
_
y +

z
_
= 1
Applying AM-GM inequality, we obtain 2
4

xz

x +

z, and 2
8

xz
4

x +
4

z.
Multiplying up the last two inequalities, we get
4
8
_
(xz)
3

_
x +

z
_ _
4

x +
4

z
_
< 1
9
Thus, (xz)
3/8
<
1
4
and
xz <
_
1
4
_
8/3
=
1
3

65536
<
1
3

64000
=
1
40
from which follows 40xz < 1.

3. Let a, b, and c be positive real numbers. Prove that


_
5a b
b + c
_
2
+
_
5b c
c + a
_
2
+
_
5c a
a + b
_
2
12
Solution. WLOG we can assume that a b c from which immediately follows
that a + b a + c b + c and
1
b + c

1
c + a

1
a + b
. Since the rst and the last
sequences are sorted in the same way, by applying rearrangement inequality, we get
2
_
a
b + c
+
b
c + a
+
c
a + b
_
2
_
a
a + b
+
b
b + c
+
c
c + a
_
2
_
a
b + c
+
b
c + a
+
c
a + b
_
2
_
b
a + b
+
c
b + c
+
a
c + a
_
a
b + c
+
b
c + a
+
c
a + b

a
a + b
+
b
b + c
+
c
c + a
Adding up the preceding inequalities, yields
5
_
a
b + c
+
b
c + a
+
c
a + b
_
6 +
a
a + b
+
b
b + c
+
c
c + a
from which we obtain
1
3
_
5a b
b + c
+
5b c
c + a
+
5c a
a + b
_
2
Taking into account AM-QM inequality, we have

_
1
3
_
_
5a b
b + c
_
2
+
_
5b c
c + a
_
2
+
_
5c a
a + b
_
2
_

1
3
_
5a b
b + c
+
5b c
c + a
+
5c a
a + b
_
2
from which the statement follows. Equality holds when a = b = c and we are done.

4. Let a, b, c be positive real numbers such that a + b + c = 2. Prove that


bc
4

3a
2
+ 4
+
ca
4

3b
2
+ 4
+
ab
4

3c
2
+ 4

2
3
4

3
Solution. Squaring both terms, we get
_
bc
4

3a
2
+ 4
+
ca
4

3b
2
+ 4
+
ab
4

3c
2
+ 4
_
2

4
9

3
10
Applying CBS to the vectors u =
_

bc,

ca,

ab
_
and
v =
_
bc

3a
2
+ 4
,
_
ca

3b
2
+ 4
,

ab

3c
2
+ 4
_
we obtain
_
bc
4

3a
2
+ 4
+
ca
4

3b
2
+ 4
+
ab
4

3c
2
+ 4
_
2
(ab + bc + ca)
_
bc

3a
2
+ 4
+
ca

3b
2
+ 4
+
ab

3c
2
+ 4
_
From a + b + c = 2 and the well-known inequality (a + b + c)
2
3(ab + bc + ca)
immediately follows ab + bc + ca 4/3. Equality holds when a = b = c = 2/3.
On the other hand,

3
_
bc

3a
2
+ 4
_
=
bc
_
a
2
+ 4/3

bc

a
2
+ ab + bc + ca
=
bc
_
(a + b)(a + c)

1
2
_
bc
a + b
+
bc
a + c
_
The last inequality holds in account of HM-GM inequality. Likewise,

3
_
ca

3b
2
+ 4
_

1
2
_
ca
b + c
+
ca
b + a
_
and

3
_
ab

3c
2
+ 4
_

1
2
_
ab
c + a
+
ab
c + b
_
Adding the preceding inequalities, we obtain

3
_
bc

3a
2
+ 4
+
ca

3b
2
+ 4
+
ab

3c
2
+ 4
_

1
2
_
ca
b + c
+
ca
b + a
+
bc
a + b
+
bc
a + c
+
ab
c + a
+
ab
c + b
_
=
1
2
(a + b + c) = 1
Therefore,
bc

3a
2
+ 4
+
ca

3b
2
+ 4
+
ab

3c
2
+ 4

3
3
Equality holds when a = b = c = 2/3. From the preceding, we have
_
bc
4

3a
2
+ 4
+
ca
4

3b
2
+ 4
+
ab
4

3c
2
+ 4
_
2

4
3

3
3
and the statement follows. Equality holds when a = b = c = 2/3, and we are done.

5. Let a, b, c be three positive numbers such that ab + bc + ca = 1. Prove that


_
a
2
+ b
2
+ c
2
_
4
_
a
2 3
_
a
b + c
+ b
2 3

b
c + a
+ c
2 3
_
c
a + b
_3

1
2
11
Solution 1. First we observe that from a
2
+b
2
+c
2
ab +bc +ca and the constrain,
immediately follows that a
2
+ b
2
+ c
2
1. Multiplying and dividing the LHS of the
inequality claimed by (a
2
+ b
2
+ c
2
)
3
we have
(a
2
+ b
2
+ c
2
)
7
_
_
_
a
2
3
_
a
b+c
+ b
2 3
_
b
c+a
+ c
2
3
_
c
a+b
a
2
+ b
2
+ c
2
_
_
_
3
(a
2
+ b
2
+ c
2
)
7
_
a
2
+ b
2
+ c
2
a
2
b+c
a
+ b
2
c+a
b
+ c
2
a+b
c
_
=
(a
2
+ b
2
+ c
2
)
8
2(ab + bc + ca)
=
(a
2
+ b
2
+ c
2
)
8
2

1
2
In the preceding we have used the inequality f(1/3) f(1), where
f() =
_
w
1
a

+ w
2
b

+ w
3
c

w
1
+ w
2
+ w
3
_
1/
is the mean powered inequality with weights w
1
=
a
2
a
2
+ b
2
+ c
2
, w
2
=
b
2
a
2
+ b
2
+ c
2
,
and w
3
=
c
2
a
2
+ b
2
+ c
2
respectively. Equality holds when a = b = c = 1/

3, and we
are done.
Solution 2. First, using the constrain, we write the inequality claimed in the most
convenient form
a
2 3
_
a
b + c
+ b
2 3

b
c + a
+ c
2 3
_
c
a + b

1
3

2
_
ab + bc + ca
a
2
+ b
2
+ c
2
_
4/3
=
1
3

2
_
ab + bc + ca
a
2
+ b
2
+ c
2
_
4/3
(ab + bc + ca)
To prove the preceding inequality, we consider the function f : [0, +) R dened
by f(t) = t
7/3
which is convex, as can be easily checked. Applying Jensens inequali-
ty, with q
1
=
(b + c)
2
A
, q
2
=
(c + a)
2
A
, q
3
=
(a + b)
2
A
, where A = (a+b)
2
+(b +c)
2
+
(c + a)
2
; and x
1
=
a
b + c
, x
2
=
b
c + a
, x
3
=
c
a + b
, we have
q
1
f
_
a
b + c
_
+ q
2
f
_
b
c + a
_
+ q
3
f
_
c
a + b
_
f
_
q
1
a
b + c
+ q
2
b
c + a
+ q
3
c
a + b
_
,
or equivalently,
(b + c)
2
A
_
a
b + c
_
7/3
+
(c + a)
2
A
_
b
c + a
_
7/3
+
(a + b)
2
A
_
c
a + b
_
7/3

_
(a(b + c) + b(c + a) + c(a + b)
A
_
7/3
12
Rearranging terms, and after simplication, we obtain
a
2 3
_
a
b + c
+ b
2 3

b
c + a
+ c
2 3
_
c
a + b

_
2(ab + bc + ca)
_
7/3
_
2(a
2
+ b
2
+ c
2
+ ab + bc + ca)
_
4/3

_
2(ab + bc + ca)
_
7/3
_
4(a
2
+ b
2
+ c
2
)
_
4/3
=
1
3

2
_
ab + bc + ca
a
2
+ b
2
+ c
2
_
4/3
(ab + bc + ca)
on account of the well-known inequality a
2
+ b
2
+ c
2
ab + bc + ca. Equality holds
when a = b = c = 1/

3, and we are done.

6. Let a, b, c be three positive numbers such that a


2
+ b
2
+ c
2
= 1. Prove that
_
1
a
3
(b + c)
5
+
1
b
3
(c + a)
5
+
1
c
3
(a + b)
5
_
1/5

3
2
Solution 1. To prove the inequality claimed we consider the function f : R R
dened by f(t) = t
5
. This function is convex in [0, +) as can be easily checked.
Applying Jensens inequality, namely
3

k=1
q
k
f(x
k
) f
_
3

k=1
q
k
x
k
_
with
q
1
=
a
2
a
2
+ b
2
+ c
2
, q
2
=
b
2
a
2
+ b
2
+ c
2
, q
3
=
c
2
a
2
+ b
2
+ c
2
and
x
1
=
1
a(b + c)
, x
2
=
1
b(c + a)
, x
3
=
1
c(a + b)
,
yields
1
a
2
+ b
2
+ c
2
_
a
2
_
1
a(b + c)
_
5
+ b
2
_
1
b(c + a)
_
5
+ c
2
_
1
c(a + b)
_
5
_

_
a
b+c
+
b
c+a
+
c
a+b
_
5
(a
2
+ b
2
+ c
2
)
5
or equivalently,
1
a
3
(b + c)
5
+
1
b
3
(c + a)
5
+
1
c
3
(a + b)
5

_
a
b + c
+
b
c + a
+
c
a + b
_
5
where we have used the constrain a
2
+ b
2
+ c
2
= 1. Now, it is easy to see that
a
b + c
+
b
c + a
+
c
a + b

3
2
13
Indeed, WLOG we can assume that a b c from which follows a+b a+c b+c
and
1
b+c

1
c+a

1
a+b
. Now, applying rearrangement inequality, we get
a
b + c
+
b
c + a
+
c
a + b

a
a + b
+
b
b + c
+
c
c + a
a
b + c
+
b
c + a
+
c
a + b

b
a + b
+
c
b + c
+
a
c + a
Adding the preceding expressions, we obtain
a
b + c
+
b
c + a
+
c
a + b

3
2
(Nesbitts Inequality)
Equality holds when a = b = c.
Finally, substituting this result in the preceding the statement follows. Equality
holds when a = b = c =

3/3, and we are done.


Solution 2. We recall that H olders inequality can be stated as follows: Let a
1
, a
2
, ..., a
n
and b
1
, b
2
, ..., b
n
, (n 2), be sequences of positive real numbers and let p, q R

+
such
that
1
p
+
1
q
= 1, then the following inequality holds
_
n

i=1
a
p
i
_1
p
_
n

i=1
b
q
i
_1
q

i=1
a
i
b
i
Particularizing the preceding result for n = 3 with p = 5/4, q = 5 for which (1/p +
1/q = 1), we get
_
3

i=1
a
5/4
i
_
4
5
_
3

i=1
b
5
i
_
1
5

i=1
a
i
b
i
Putting in the above result a
1
= a
8/5
, a
2
= b
8/5
, a
3
= c
8/5
, b
1
=
1
a
3/5
(b + c)
, b
2
=
1
b
3/5
(c + a)
, b
3
=
1
c
3/5
(a + b)
yields

a
8/5

5/4
+

b
8/5

5/4
+

c
8/5

5/4

4/5

1
a
3/5
(b + c)

5
+

1
b
3/5
(c + a)

5
+

1
c
3/5
(a + b)

1/5

a
b + c
+
b
c + a
+
c
a + b
or equivalently,
(a
2
+b
2
+c
2
)
4/5
_
1
a
3
(b + c)
5
+
1
b
3
(c + a)
5
+
1
c
3
(a + b)
5
_
1/5

a
b + c
+
b
c + a
+
c
a + b
from which the statement follows on account of the constrain and Nesbitts inequal-
ity. Equality holds when when a = b = c =

3/3, and we are done.

Solution 3 by Emilio Fern andez Moral. On account of mean inequalities we have


_
1
a
3
(b + c)
5
+
1
b
3
(c + a)
5
+
1
c
3
(a + b)
5
_
1/5

_
3
3
_
a
3
b
3
c
3
(a + b)
5
(b + c)
5
(c + a)
5
_
1/5
14
=
5
_
3
abc
1
3
_
(a + b)(b + c)(c + a)

3
1/5+1/2
2 (abc)
1/5

3
2
because
3
_
(a + b)(b + c)(c + a)
(a + b) + (b + c) + (c + a)
3
=
2(a + b + c)
3

2
_
1
3
and (abc)
1/5
= (a
2
b
2
c
2
)
1/10
=
_
3

a
2
b
2
c
2
_
3/10

_
a
2
+ b
2
+ c
2
3
_3/10
=
_
1
3
_
3/10
and 3
1/5+1/2+3/10
= 3. Equality holds when a = b = c =

3/3.
15
Jos e Luis Daz-Barrero
RSME Olympic Committee
UPCBARCELONA TECH
jose.luis.diaz@upc.edu

Some Results
Hereafter, some classical discrete inequalities are stated and proven. We begin with
Theorem 1 (General Mean Inequalities) Let a
1
, a
2
, . . . , a
n
be positive real numbers.
Then, the function f : R R dened by
f() =
_
a

1
+ a

2
+ . . . + a

n
n
_
1/
is nondecreasing and the following limits hold
lim

f() = min
1in
_
a
i
_
, lim
0
f() =
n

a
1
a
2
. . . a
n
, lim
+
f() = max
1in
_
a
i
_
Proof. Let 0 < a < b. Consider the function g : [0, +) R dened by g() =
b/a
.
Since
g

() =
b
a
_
b
a
1
_

b/a2
0, ( 0),
then g is convex in [0, +). Applying Jensens inequality, we get
g
_
_
1
n
n

j=1

a
j
_
_

1
n
n

j=1
g(
a
j
)
or
_
_
1
n
n

j=1

a
j
_
_
b/a

1
n
n

j=1

b
j
from which follows f(a) f(b). For negative values of consider the function
h() = ()
b/a
which is convex in (, 0).
Let L
1
= lim

f(). Then, ln L
1
= lim

ln
_
a

1
+ a

2
+ . . . + a

n
n
_
1/
. That is,
ln L
1
= lim

_
1

ln
_
a

1
+ a

2
+ . . . + a

n
n
__
= lim

1
ln a
1
+ a

2
ln a
2
+ . . . + a

n
ln a
n
a

1
+ a

2
+ . . . + a

n
= ln
_
min
1in
_
a
i
_
_
16
from which follows L
1
= min
1in
_
a
i
_
. Likewise, we obtain that L
2
= lim
+
f() =
max
1in
_
a
i
_
. Finally, denoting by L = lim
0
f(), we have
ln L = lim
0
_
1

ln
_
a

1
+ a

2
+ . . . + a

n
n
__
= lim
0
a

1
ln a
1
+ a

2
ln a
2
+ . . . + a

n
ln a
n
a

1
+ a

2
+ . . . + a

n
=
ln a
1
+ ln a
2
+ . . . + ln a
n
n
= ln
n

a
1
a
2
. . . a
n
from which immediately follows lim
0
f() =
n

a
1
a
2
. . . a
n
and the proof is complete.

In 1821 Cauchy published his famous inequality as the second of the two notes
on the theory of inequalities that formed the nal part of his book Cours dAnalyse
Alg ebrique. Namely,
Theorem 2 Let a
1
, a
2
, ..., a
n
, b
1
, b
2
, ..., b
n
be any real numbers. Then, the following
inequality holds:
_
n

k=1
a
k
b
k
_
2

_
n

k=1
a
2
k
__
n

k=1
b
2
k
_
Proof. Consider the quadratic polynomial A(x) =
n

k=1
(a
k
x b
k
)
2
. Then,
A(x) = x
2
n

k=1
a
2
k
2x
n

k=1
a
k
b
k
+
n

k=1
b
2
k
0
Since the preceding equation is nonnegative its discriminant must be less or equal
than zero. That is,
_
n

k=1
a
k
b
k
_
2

_
n

k=1
a
2
k
__
n

k=1
b
2
k
_
0
from which the statement follows. Equality holds when the n-tuples (a
1
, a
2
, . . . , a
n
)
and (b
1
, b
2
, . . . , b
n
) are proportional. This completes the proof.

A generalization of CBS inequality is the well known inequality of H older. It is stated


and proved in the following theorem.
Theorem 3 Let a
1
, a
2
, ..., a
n
and b
1
, b
2
, ..., b
n
, (n 2) be sequences of positive real
numbers and let p, q R

+
such that
1
p
+
1
q
= 1, then the following inequality holds
_
n

i=1
a
p
i
_1
p
_
n

i=1
b
q
i
_1
q

i=1
a
i
b
i
.
Proof 1. We will argue by induction. Assume that the inequality holds for n and we
have to prove it for n + 1. In fact,
n+1

i=1
a
i
b
i
=
n

i=1
a
i
b
i
+ a
n+1
b
n+1

_
n

i=1
a
p
i
_1
p
_
n

i=1
b
q
i
_1
q
+ a
n+1
b
n+1
17

_
n

i=1
a
p
i
+ a
p
n+1
_1
p
_
n

i=1
b
q
i
+ b
q
n+1
_1
q
=
_
n+1

i=1
a
p
i
_
1
p
_
n+1

i=1
b
q
i
_
1
q
.
To complete the inductive process, we must verify that the case when n = 2 also
holds. Indeed, the inequality
a
1
b
1
+ a
2
b
2
(a
p
1
+ a
p
2
)
1
p
(b
q
1
+ b
q
2
)
1
q
,
immediate follows from the fact that the function
f(x) = (a
p
1
+ a
p
2
)
1
p
(b
q
1
+ x
q
)
1
q
a
1
b
1
a
2
x,
is strictly positive for all x > 0. This completes the proof.
Proof 2. First, we write the inequality in the most convenient form

n
i=1
a
i
b
i
_
n
i=1
a
p
i
_ 1
p
_
n
i=1
b
q
i
_1
q
1
or equivalently,
n

i=1
_
a
p
i

n
i=1
a
p
i
_
1
p
_
b
q
i

n
i=1
b
q
i
_
1
q
1.
Now, using the power mean inequality,
x
1

y
1

x +
1

y,
we have
n

i=1
_
a
p
i

n
i=1
a
p
i
_
1
p
_
b
q
i

n
i=1
b
q
i
_
1
q

i=1
_
1
p
a
p
i

n
i=1
a
p
i
+
1
q
b
q
i

n
i=1
b
q
i
_
=
1
p

n
i=1
a
p
i

n
i=1
a
p
i
+
1
q

n
i=1
b
q
i

n
i=1
b
q
i
=
1
p
+
1
q
= 1,
as desired. Equality holds if and only if the n-tuples (a
p
1
, a
p
2
, ..., a
p
n
) and (b
q
1
, b
q
2
, ..., b
q
n
)
are proportional.
Notice that for p = q = 2, we get the inequality
(a
1
b
1
+ a
2
b
2
+ ... + a
n
b
n
)
2
(a
2
1
+ a
2
2
+ ... + a
2
n
)(b
2
1
+ b
2
2
+ ... + b
2
n
)
This is the Cauchy-Bunyakowski-Schwarz inequality.
Using H olders inequality can be easily proven the following inequality of Minkowski.
Theorem 4 Let a
1
, a
2
, ..., a
n
and b
1
, b
2
, ..., b
n
, (n 2) be sequences of positive real
numbers and let p > 1, be a real number, then
_
n

i=1
(a
i
+ b
i
)
p
_1
p

_
n

i=1
a
p
i
_1
p
+
_
n

i=1
b
p
i
_1
p
When p < 1 inequality reverses.
18
Proof. Let be q =
p
p1
, then
1
p
+
1
q
= 1, and applying H olders inequality, we obtain
n

i=1
(a
i
+ b
i
)
p
=
n

i=1
a
i
(a
i
+ b
i
)
p1
+
n

i=1
b
i
(a
i
+ b
i
)
p1

_
n

i=1
a
p
i
_1
p
_
n

i=1
(a
i
+ b
i
)
(p1)q
_1
q
+
_
n

i=1
b
p
i
_1
p
_
n

i=1
(a
i
+ b
i
)
(p1)q
_1
q
=
_
_
_
n

i=1
a
p
i
_1
p
+
_
n

i=1
b
p
i
_1
p
_
_
_
n

i=1
(a
i
+ b
i
)
p
_
p1
p
Multiplying the preceding inequality by
_
n
i=1
(a
i
+ b
i
)
p
_

p1
p
, we get
_
n

i=1
(a
i
+ b
i
)
p
_1
p

_
n

i=1
a
p
i
_1
p
+
_
n

i=1
b
p
i
_1
p
and the proof is complete. Equality holds if and only if (a
1
, a
2
, ..., a
n
) and (b
1
, b
2
, ..., b
n
)
are proportional.

A most useful result is presented in the following


Theorem 5 (Rearrangement) Let a
1
, a
2
, , a
n
and b
1
, b
2
, , b
n
be sequences of
positive real numbers and let c
1
, c
2
, , c
n
be a permutation of b
1
, b
2
, , b
n
. The
sum S = a
1
b
1
+a
2
b
2
+ +a
n
b
n
is maximal if the two sequences a
1
, a
2
, , a
n
and
b
1
, b
2
, , b
n
are sorted in the same way and minimal if the two sequences are sorted
oppositely, one increasing and the other decreasing.
Proof. Let a
i
> a
j
. Consider the sums
S
1
= a
1
c
1
+ + a
i
c
i
+ + a
j
c
j
+ + a
n
c
n
S
2
= a
1
c
1
+ + a
i
c
j
+ + a
j
c
i
+ + a
n
c
n
We have obtained S
2
from S
1
by switching the positions of c
i
and c
j
. Then
S
1
S
2
= a
i
c
i
+ a
j
c
j
a
i
c
j
a
j
c
i
= (a
i
a
j
)(c
i
c
j
)
Therefore,
c
i
> c
j
S
1
> S
2
and c
i
< c
j
S
1
< S
2
.
This completes the proof.
In the sequel, after dening the concepts of convex and concave function, the ine-
quality of Jensen is presented.
Denition. Let f : I R R be a real function. We say that f is convex (concave)
on I if for all a
1
, a
2
I and for all t [0, 1], we have
f(ta
1
+ (1 t)a
2
) f(a
1
) + (1 t)f(a
2
)
When f is concave the inequality reverses.
19
Theorem 6 Let f : I R R be a convex function. Let q
1
, q
2
, . . . , q
n
be nonnegative
real numbers such that q
1
+ q
2
+ . . . + q
n
= 1. Then for all a
k
I (1 k n) holds
f
_
n

k=1
q
k
a
k
_

k=1
q
k
f(a
k
)
Equality holds when a
1
= a
2
= . . . = a
n
. If f is concave the preceding inequality
reverses.
Proof. We will argue by mathematical induction. For n = 2, we have
f(q
1
a
1
+ q
2
a
2
) q
1
f(a
1
) + q
2
f(a
2
)
The preceding inequality holds because f is convex and q
1
+ q
2
= 1. Assume that
f
_
n

k=1
q
k
a
k
_

k=1
q
k
f(a
k
)
and we have to see that
f
_
n+1

k=1
q
k
a
k
_

n+1

k=1
q
k
f(a
k
)
when q
1
+ q
2
+ . . . + q
n
+ q
n+1
= 1. Let q =

n
k=1
q
k
. Then, we have
n

k=1
q
k
q
= 1,
n

k=1
_
q
k
q
_
a
k
I and q + q
n+1
= 1. Now, taking into account the case when n = 2
and the inductive hypothesis, yields
f
_
n+1

k=1
q
k
a
k
_
= f
_
q
_
n

k=1
_
q
k
q
_
a
k
_
+ q
n+1
a
n+1
_
qf
_
n

k=1
_
q
k
q
_
a
k
_
+ q
n+1
f(a
n+1
)
q
n

k=1
_
q
k
q
_
f(a
k
) + q
n+1
f(a
n+1
)
=
n+1

k=1
q
k
f(a
k
)
and by the principle of mathematical induction (PMI) the statement is proven. Equa-
lity holds when a
1
= a
2
= . . . = a
n
.

20
Think about
Pulchra dicuntur quae visa placent beauty is that which, be-
ing seen, pleases. This denition, applies well to Mathematical
beauty in which lack of understanding is so often responsible of
lack of pleasure.
Thomas Aquinas denition of Beauty

También podría gustarte