FOUR APPLICATIONS OF RCF AND LCF THEOREMS

Vasile Crtoaje
Abstract
In this paper are presented four new and difficult symmetric inequalities with right convex and left concave functions, as applications of RCF-Theorem and LCF-Theorem from [1] and [2]. Note that all the functions involved in the proposed inequalities have more than one inflexion point.

1. Introduction
In [1] and [2], we have proved the following theorems: Right Convex Function Theorem (RCF-Theorem). Let f (u) be a function defined on an interval I R and convex for u s , s I . If the inequality

f ( x1) + f ( x2 ) +L+ f ( xn ) nf ( holds for all x1 , x2 ,L, xn I such that

x1 + x2 +L+ xn ) n

x1 + x2 +L+ xn =s, n x + x +L+ xn s. then it also holds for all x1 , x2 ,L, xn I such that 1 2 n

x2 = x3 = L = xn s and

Left Concave Function Theorem (LCF-Theorem). Let f (u) be a function

defined on an interval I R and concave for u s , s I . If the inequality f ( x1) + f ( x2 ) +L+ f ( xn ) nf ( holds for all x1 , x2 ,L, xn I such that x1 + x2 +L+ xn =s, n x + x +L+ xn s. then it also holds for all x1 , x2 ,L, xn I such that 1 2 n x1 = x2 = L = xn1 s and x1 + x2 +L+ xn ) n

Remark 1.1. The hypothesis in RCF-Theorem is equivalent to the condition that f ( x) + (n 1) f ( y) nf (s) for all x , y I such that x s y and

x + (n 1) y = ns .
Remark 1.2. Let g (u) =
and x + (n 1) y = ns .

f (u) f (s) . The hypothesis in RCF-Theorem is us equivalent to the condition that g ( x) g ( y) for all x , y I such that x s y

Remark 1.3. The hypothesis in LCF-Theorem is equivalent to the condition that (n 1) f ( x) + f ( y) nf (s) for all x , y I such that x s y and

(n 1) x + y = ns .
Remark 1.4. Let f (u) f (s) . The hypothesis in LCF-Theorem is us equivalent to the condition that g ( x) g ( y) for all x , y I such that x s y g (u) =
and (n 1) x + y = ns . In this paper, following closely theorems above, we will prove the following four statements.

Proposition 1.1. If a1 , a2 ,L, an are nonnegative real numbers such that a1 + a2 +L+ an = n , then
2 2 2 a1 + a2 +L+ an n 2 2 2 (a1 a1)2 + (a2 a2 )2 +L+ (an an )2

n2+

n 1

(1)

Proposition 1.2. If a , b , c are positive real numbers such that abc = 1 , then

a 2 + b2 + c2 3 18(a + b + c ab bc ca ) .

(2)

Proposition 1.3. If a1 , a2 , L, a8 are nonnegative real numbers such that

a1a2 La8 1 , then

1 a1 (1+ a1)
2

1 a2 (1+ a2 )
2

+L+

1 a8 (1+ a8 )2

0.

(3)

Proposition 1.4. If a1 , a2 , a3 , a4, a5 are positive real numbers such that

a1a2a3a4a5 1 , then 1+ a1 1+ a2 1+ a3 1+ a4 1+ a5 + + + + 5. 2 2 2 2 2 1+ a1 1+ a2 1+ a3 1+ a4 1+ a5 (4)

2. Proofs of the proposed inequalities

Proof of Proposition 1.1. Let A = n 2 + 1 , A 1 . Since a1 + a2 +L+ an = n , n 1 a1 + a2 +L+ an ), n

we may write (1) as

f (a1) + f (a2 ) +L+ f (an ) nf (

(5)

where f (u) = A(u 2 u)2 u 2 +1 , u 0 . The second derivative,

f (u) =12 A(u 2 u) + 2( A 1) , shows that f (u) has two inflexion points for u 0 . Since f (u) 0 for u 1 , the function f is right convex for u s , where a1 + a2 +L+ an =1 . n By RCF-Theorem, it suffices to prove (5) for a1 1 a2 = a3 = L = an and a1 + a2 +L+ an = n . According to Remark 1.2, this means to show that s= g ( x) g ( y) for 0 x 1 y and x + (n 1) y = n , where g (u) = We have g ( x) g ( y) = ( x y)[ A( x2 + xy + y 2 ) A( x + y) 1] = = n(n 1)(1 y)( Ay n +1)2 0 , and the proof is complete. Equality occurs only if a1 = a2 = a3 = L = an =1+ n2 n 3n + 3
2

f (u) f (1) = A(u3 u 2 ) u 1 . u 1

1 n 3n + 3
2

and

, or any cyclic permutation thereof.

Proof of Proposition 1.2. We will show that for any real numbers x , y , z x+ y+ z with = 0 , the inequality holds 3 x+ y+ z f ( x) + f ( y ) + f ( z ) 3 f ( ), (6) 3 where f (u) = 18(eu eu ) e2u ,
u R .

Replacing

then

x, y, z

by

ln a , ln b , ln c , respectively, the desired inequality (2) follows. In order to prove (6), we will apply LCF-Theorem to the function f defined on R, with s = 0 . From the second derivative

f (u) =18(eu eu ) 4 e2u , it follows that f (u) has two inflexion points in R. Since f (u) < 0 for u 0 , the function f (u ) is left concave for u s = 0 . According to LCF-Theorem, it suffices to consider only the case x = y 0 . This means to prove the initial inequality for a = b 1 and a2c =1 . Then, the inequality successively becomes 2a 2 + c 2 3 18(2a + c a 2 2ac) , 2a6 3a 4 +1+18a 2 (a 4 2a3 + 2a 1) 0 , (a 2 1)2 (2a 2 +1) +18a 2 (a 1)3(a +1) 0 , (a 1)2 (2a 1)2 (a +1)(5a +1) 0 . The last inequality is clearly true, and the proof is completed. Equality occurs 1 1 when (a , b , c) = (1,1,1) , and also when (a , b , c) = ( , , 4 ) or any cyclic 2 2 permutation thereof.

Proof of Proposition 1.3. According to Lemma 2.1 and Lemma 2.2 below, it suffices to consider the case where all ai 3 and a1a2 La8 = 1 . We will show that for all xi ln 3 such that x1 + x 2 +L + x8 = 0 , the inequality holds f ( x1) + f ( x2 ) +L+ f ( x8 ) 5 f ( where f (u) = x1 + x2 +L+ x8 ), 8 (7)

1 eu . Replacing then each xi by ln ai , the required (1+ eu )2

inequality (3) follows. We will prove (7) by applying RCF-Theorem to the function f defined on I = ( , ln 3] , with s = 0 . Taking derivatives, we get

f (u) =

eu (8eu e2u 3) , (1+ eu )4

which shows that f has two inflexion point in R. We first have to show that

f (u) is convex for s u ln 3 ; this means that f (u) 0 for 0 u ln 3 or,

equivalently, 8t t 2 3 0 for 1 t 3 . This is true since
8t t 2 3 8t 3t 3 = 5t 3 > 0 .

According to RCF-Theorem, it suffices to prove the inequality (7) for 0 x2 = x3 = L = x8 ln 3 and x1 + x2 +L+ x8 = 0 ; that is, to prove the initial

inequality (3) for 1 a2 = a3 = L = a8 3 show that

and

a1a2 La8 = 1 . Thus, we must

1 a 7(1 b) + 0 (1+ a)2 (1+ b)2

for 1 b 3 and ab7 =1 . Taking into account that b7 (b7 1) 1 a = 7 2 , (1+ a)2 (b +1) we have to show that b7 (b6 + b5 + b4 + b3 + b2 + b +1) 7 0 . (b6 b5 + b4 b3 + b2 b +1)2 Since
b6 b5 + b4 b3 + b2 b +1 = b4 (b2 b +1) (b 1)(b2 +1) b4 (b2 b +1) ,

it suffices to prove the inequality b6 + b5 + b4 + b3 + b2 + b +1 7 0 . b(b2 b +1)2 This inequality is equivalent to (b 1)6 0 , which is clearly true. Equality in the given inequality occurs if and only if a1 = a2 = L = a8 =1 .
Lemma 2.1. If the inequality (3) holds for any 0 < ai 3 such that

a1a2 La8 = 1 , then it holds for any 0 ai 3 such that a1a2 La8 1 . Proof. Assume that 0 ai 3 and a1a2 La8 1 . Always there are eight positive numbers bi such that b1b2 Lb8 = 1 and ai bi 3 for i =1,2,L,8 . According to the hypothesis, the inequality holds function f ( x) =

(1+ b i)2 0 . Since the

i=1 i

1 b

1 x x 3 has the derivative f ( x) = < 0 for x [0 , 3) , 2 (1+ x) (1+ x)3 1 ai 1 bi f (x) is strictly decreasing on [0 , 3] . Therefore, for all (1+ ai )2 (1+ bi )2
8

subscripts i , and hence

(1+ a )2 (1+ b i)2 0 .

i =1 i i =1 i

1 ai

1 b

Lemma 2.2. If the inequality (3) holds for any 0 ai 3 such that

a1a2 La8 1 , then it holds for any ai 0 such that a1a2 La8 1 . Proof. Assume that all ai 0 , and a1a2 La8 1 . Define the numbers x1 , x2 ,L , x8 as , for ai 3 ai xi = ai + 3 . a 1 , for ai > 3 i 1 xi (1+ xi ) = 2 1 ai (1+ ai )2 for all i .

It is easy to show that 0 xi 3 , xi ai and

Since 0 xi 3 and x1x2 L x8 a1a2 La8 1 , from the hypothesis we have

(1+ x )2
i =1 i

1 xi

0 , and hence

(1+ a i)2 0 .
i =1 i

1 a

Remark 2.1. If n = 9 , then the inequality

(1+ a i)2 0 is not true for any

i =1 i

1 a

positive numbers ai with

inequality becomes 1 a1 (1+ a1)2

ai =1 .
i =1

Indeed, for a2 = a3 = L = a9 = 3 , the

1 0 , which is false.

Proof of Proposition 1.4. According to Lemma 2.3 and Lemma 2.4 below (which can be proved in the same way as the preceding Lemma 2.1 and Lemma 2.2), it suffices to consider the case where all ai 2 1 and a1a2a3a4a5 =1 . In this case, the inequality can be proved by applying LCF-Theorem to the function f (u) = with s = 0 . The second derivative
f (u) = eu (1 4 eu 6 e2u + 4 e3u + e4u ) (1+ e2u )3
1+ eu 1+ e2u

defined on I = [ ln( 2 1), ) ,

shows that f has four inflexion point in R. Finally, we have to prove the inequality for 2 1 a1 = a2 = a3 = a4 1 and a1a2a3a4a5 =1 ; that is 4(1+ a) 1+ b + 5 1+ a2 1+ b2 for
2 1 a 1 and a 4b =1 . Since

1+ b 1+ b

= 2

a 4 (1+ a 4 ) 1+ a8

and

1+ a 4

2 4 , 1+ a8 1+ a 4 (1+ a 2 )2

we get
4(1+ a) 1+ b 4(1+ a) 4a 4 5 5 = 1+ a 2 1+ b2 1+ a 2 (1+ a 2 )2 = 1 4a + 6a 2 4a3 + a 4 (1+ a 2 )(1+ a 4 ) = (1 a)4 (1+ a 2 )(1+ a 4 ) 0,

which completes the proof. Equality holds only if a1 = a2 = a3 = a4 = a5 =1 .

Lemma 2.3. If the inequality (4) holds for any ai 2 1 such that

a1a2a3a4a5 =1 , then it holds for any ai 2 1 such that a1a2a3a4a5 1 .

Lemma 2.4. If the inequality (4) holds for any ai 2 1 such that

a1a2a3a4a5 1 , then it holds for any ai > 0 such that a1a2a3a4a5 1 .

Remark 2.2. If n = 6 , then the inequality
n i 1+ a2 6 i =1 i n

1+ a

is not true for any 1

positive numbers ai with inequality becomes 1+ a1

2 1+ a1

ai =1 . Indeed, for a2 = a3 = a4 = a5 = a6 = 2 , the

i =1

0 , which is false.

References
[1] V. Crtoaje, A generalization of Jensens Inequality, Gazeta Matematica Seria A, 2 (2005), 124-138. [2] V. Crtoaje, Algebraic Inequalities. Old and New Method, Gil Publishing House, 2006. [3] M. Tetiva, A new proof for the Right Convex Function Theorem, Gazeta Matematica Seria A, 2 (2006), 126-133. Vasile Crtoaje Department of Automatics and Computers Petroleum-Gas University Ploiesti City, Bucuresti 39, Romania email: vcirtoaje@upg-ploiesti.ro