Sei sulla pagina 1di 5

About Surányi’s Inequality

Mihály Bencze
Str. Harmanului 6, 505600 Sacele, Jud Brasov, Romania

Abstract. In the Miklós Schweitzer Mathematical Competition (Hungary)


Professor János Surányi proposed the following problem, which is interesting and
presents an aspect of a theorem. In this paper we present a new demonstration,
some interesting applications and a generalization.

Theorem 1. (János Surányi). If xk > 0 (k = 1, 2, ..., n) then the following


inequality holds:
n n n
! n !
X Y X X
n n−1
(n − 1) xk + n xk ≥ xk xk .
k=1 k=1 k=1 k=1
.

Proof. Using mathematical induction, for n = 2 we obtain x21 + x22 + 2x1 x2 ≥


2
(x1 + x2 ) , which is true.
We suppose that is true for n and we prove for n + 1.
Because the inequality is symmetric and homogeneous we can suppose that x1 ≥
x2 ≥ ... ≥ xn+1 and x1 +x2 +...+xn = 1, so we must prove the following inequality:
n+1 n+1 n+1
! n+1 !
X Y X X
n+1
n xk + (n + 1) xk ≥ xk xnk
k=1 k=1 k=1 k=1
which can be written in the form
n n n n
!
X Y Y X
n+1 n+1 n n
n xk +nxn+1 +nxn+1 xk +xn+1 xk −(1 + xn+1 ) xk + xn+1 ≥ 0
k=1 k=1 k=1 k=1
From the inductive condition holds
Yn n
X n
X
nxn+1 xk ≥ xn+1 xn−1
k − (n − 1) xn+1 xnk
k=1 k=1 k=1
It remains to prove that:
n n
! n n
!
X X X X
n+1
n xk − n
xk − xn+1 n xnk − xn−1
k
k=1 k=1 k=1 k=1
n
!
Y
+ xn+1 xk + (n − 1) xnn+1 − xn−1
n+1 ≥ 0,
k=1
but this inequality can be decomposed in two inequalities in the following manner:
First, from the Chebyshev inequality we have:
Xn Xn
n n
xk − xn−1
k ≥ 0.
k=1 k=1
Second, because
1 n−1
nxn+1
k + x ≥ 2xnk (k = 1, 2, ..., n) ,
n k
1
2

then after addition we have:


n
Y
xk + (n − 1) xnn+1 − xn−1
n+1
k=1
n
Y
= (xk − xn+1 + xn+1 ) + (n − 1) xnn+1 − xn−1
n+1
k=1
n
X
≥ xnn+1 + xn−1
n+1 (xk − xn+1 ) + (n − 1) xnn+1 − xn−1
n+1 = 0
k=1
or !
n n n n
X X 1 X X
n xn+1
k − xnk ≥ n xnk − xn−1
k ,
n
k=1 k=1 k=1 k=1
but from xn+1 ≤ n1 holds the desired inequality.
If in Theorem 1 we take n = 3, then we obtain:.

Application 1. If x1 , x2 , x3 ≥ 0, then
x31 + x32 + x33 + 3x1 x2 x3 ≥ x21 (x2 + x3 ) + x22 (x3 + x1 ) + x23 (x1 + x2 )
which is the well known Schur’s inequality. Therefore, the inequality of Surányi has
generalized the Schur inequality.

Application 2. If a, b, c denote the sides of triangle ABC, s the semiperimeter,


R the radius of the circumcircle, r the radius of the incircle, then:.
1). R ≥ 2r (the inequality of Euler)
2). s2 ≥ r2 + 16Rr
3
3). (4R + r) ≥ s2 (16R − 5r).

Proof. In Application 1 we take:


1). x1 = a, x2 = b, x3 = c
2). x1 = s − a, x2 = s − b, x3 = s − c
3). x1 = ra , x2 = rb , x3 = rc
where ra , rb , rc are the radii of exinscribed circles.
If In Theorem 1 we take n = 4, then we obtain the following:.

Application 3. If x1 , x2 , x3 , x4 ≥ 0, then
4 4
!
X Y X
x4k + 2 xi xj x2i + x2j

2 xk ≥
k=1 k=1 1≤i<j≤4
.

Remark. Because x2i + x2j ≥ 2xi xj , then


4
X 4
Y X
x4k + 2 xk ≥ x2i x2j ,
k=1 k=1 1≤i<j≤4

but this is the Turkevici inequality. Therefore the inequality of Surányi gives a
refinement and a generalization of Turkevici‘s inequality..

Application 4. Denote ra , rb , rc , rd and ha , hb , hc , hd the radii of exinscribed


spheres and the altitudes in tetrahedron ABCD, then
3

1).
X 1 4 1X 1
3 4
+Q ≥
ha ha r h3a
2).
X 1 4 2X 1
3 + Q ≥
ra4 ra r ra3
where r is the radius of inscribed sphere..

Proof. In Application 3 we take:


1). x1 = h1a , x2 = h1b , x3 = h1c , x4 = h1d and
P 1 1
ha = r
1 1 1 1 1 2
P
2). x1 = ra , x2 = rb , x3 = rc , x4 = rd and ra = r
The inequality of Turkevici can be generalized in following way:.

Theorem 2. If xk > 0 (k = 1, 2, ..., n) , then


v
u n n
X 2
uY X
(xi − xj ) + n t n
x2k ≥ x2k
1≤i<j≤n k=1 k=1

Finally, we generalize the inequality of Surányi in following way:

Theorem 3. If ak ∈ I (I ⊆ R) (k = 1, 2, ..., n) , f : I → R and f and f 0 are


convex functions, then:
n n
! n  
X 1X X (n − 1) ai + aj
(n − 1) f (ak ) + nf ak ≥ f
n i,j=1
n
k=1 k=1
.

Proof. We suppose that a1 ≥ a2 ≥ ... ≥ an , so the desired inequality can be


discomposed in the following two inequalities:
(1).
n−1 n−1
X  kak + ak+1 + ... + an   
X X (n − 1) ai + aj
(n − 1 − k) f (ak )+ f ≥ f
n n
k=1 k=1 1≤i<j≤n

and
(2).
n−1  
X a1 + a2 + ... + an
(k − 1) f (ak ) + (n − 2) f (an ) + nf
n
k=1
n−1
X  kak + ak+1 + ... + an   
X (n − 1) ai + aj
≥ f + f
n n
k=1 1≤i<j≤n

The inequality (1) is the consequence of inequalities


  n  
kak + ak+1 + ... + an X (n − 1) ak + aj
(n − 1 − k) f (ak ) + f ≥ f ,
n n
j=k+1

where k ∈ {1, 2, ..., n − 1} but this holds from Karamata‘s inequality using for
 
kak + ak+1 + ... + an
ak , ak , ..., ak ,
n
4

and  
(n − 1) ak + ak+1 (n − 1) ak + ak+2 (n − 1) ak + an
, , ..., .
n n n
The inequality of Karamata says that: If f : I → R is convex x1 ≥ x2 ≥ ... ≥ xn
and y1 ≥ y2 ≥ ... ≥ yn , x1 ≥ y1 , x1 + x2 ≥ y1 + y2 , ..., x1 + x2 + ... + xn−1 ≥
y1 + y2 + ... + yn−1 , x1 + x2 + ... + xn = y1 + y2 + ... + yn , then
f (x1 ) + f (x2 ) + ... + f (xn ) ≥ f (y1 ) + f (y2 ) + ... + f (yn ) .
In our case
 
kak + ak+1 + ... + an
(x1 , x2 , ..., xn−k ) = ak , ak , ..., ak ,
n
and
 
(n − 1) ak + ak+1 (n − 1) ak + ak+2 (n − 1) ak + an
(y1 , y2 , ..., yn−k ) = , , ..., .
n n n
Now we prove the inequality (2).
Denote
n−1  
X a1 + a2 + ... + an
F (a1 , a2 , ..., an ) = (i − 1) f (ai ) + (n − 2) f (an ) + nf
i=1
n
n−1    
X iai + ai+1 + ... + an X (n − 1) ai + aj
− f − f ,
i=1
n n
1≤i<j≤n

for which we prove that:


F (a1 , a2 , ..., an ) ≥ F (a2 , a2 , a3 , ..., an ) ≥ ...
≥ F (an−1 , an−1 , ..., an−1 , an ) ≥ F (an , an , ..., an ) = 0.
In F (ak , ak , ..., ak , ak+1 , ak+2 , ..., an ) , contain ak the following expression
n
X
(i − 1) f (ak )
i=1
  X k  
kak + ak+1 + ... + an kak + ak+1 + ... + an
+ nf − f
n i=1
n
  X k n  
X (n − 1) ak + ak X (n − 1) ai + ak
− f − f
n j=1 i=k+1
n
1≤i<j≤k
  n  
kak + ak+1 + ... + an X (n − 1) ai + ak
= (n − k) f −k f .
n n
i=k+1

Denote Gk (a) = F (a, a, ..., a, ak+1 , ak+2 , ..., an ), where a ∈ [ak+1 , ak ] , then
  
0 k (n − k) 0 ka + ak+1 + ... + an
Gk (a) = f
n n
n  !
1 X
0 (n − 1) ai + a
− f ≥ 0,
n−k n
i=k+1
5

because
n
ka + ak+1 + ... + an 1 X (n − 1) ai + a

n n−k n
i=k+1
or
n
X
(n − k) a ≥ ai ,
i=k+1
which is true.
Sincee f is convex, then f 0 is increasing but f 0 is convex, so
  n
!
0 ka + ak+1 + ... + an 0 1 X (n − 1) ai + a
f ≥f
n n−k n
i=k+1
n  
1 X
0 (n − 1) ai + a
≥ f ,
n−k n
i=k+1
which follows from Jensen‘s inequality.
Therefore G is increasing and
F (ak , ak , ..., ak , ak+1 , ak+2 , ..., an ) ≥ F (ak+1 , ak+1 , ..., ak+1 , ak+2 , ..., an )
which proves the affirmation..

Remark. If in Theorem 3 we take f (a) = ena and eak = xk (k = 1, 2, ..., n) ,


then we obtain the inequality of Surányi..

Application 5. If ak > 0 (k = 1, 2, ..., n) and α ≥ 2, then


n n
!α n  α
X
α 1X X (n − 1) ai + aj
(n − 1) ak + n ak ≥ .
n i,j=1
n
k=1 k=1
.

Proof. In Theorem 3 we take f (a) = aα .


References.
[1]. Mihály Bencze: Inequalities (manuscript), 1982.
[2]. D.S. Mitrinović, J.E. Pečarić, A.M. Fink: Classical and New Inequalities in
Analysis, Kluwer Academic Publishers, 1993.
[3]. Octogon Mathematical Magazine (1993-2005).

Potrebbero piacerti anche