• No results found

Schur Convexity for a Class of Symmetric Functions and Its Applications

N/A
N/A
Protected

Academic year: 2020

Share "Schur Convexity for a Class of Symmetric Functions and Its Applications"

Copied!
15
0
0

Loading.... (view fulltext now)

Full text

(1)

Volume 2009, Article ID 493759,15pages doi:10.1155/2009/493759

Research Article

Schur-Convexity for a Class of Symmetric

Functions and Its Applications

Wei-Feng Xia

1

and Yu-Ming Chu

2

1School of Teacher Education, Huzhou Teachers College, Huzhou 313000, China 2Department of Mathematics, Huzhou Teachers College, Huzhou 313000, China

Correspondence should be addressed to Yu-Ming Chu,[email protected]

Received 16 May 2009; Accepted 14 September 2009

Recommended by Jozef Banas

Forx x1, x2, . . . , xnRn, the symmetric functionφnx, ris defined by φnx, r φnx1, x2, . . . , xn;r 1≤i1<i2···<irn

r

j1xij/1xij

1/r, wherer1,2, . . . , nandi

1, i2, . . . , inare positive

integers. In this article, the Schur convexity, Schur multiplicative convexity and Schur harmonic convexity ofφnx, rare discussed. As applications, some inequalities are established by use of the theory of majorization.

Copyrightq2009 W.-F. Xia and Y.-M. Chu. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

Throughout this paper we useRnto denote then-dimensional Euclidean space over the field of real numbers andRn {x1, x2, . . . , xn:xi >0, i 1,2, . . . , n}. In particular, we useRto

denoteR1.

For the sake of convenience, we use the following notation system. Forx x1, x2, . . . , xn, y y1, y2, . . . , ynRn, andα >0, let

xyx1y1, x2y2, . . . , xnyn,

xyx1y1, x2y2, . . . , xnyn,

αx αx1, αx2, . . . , αxn,

xα

1, x2α, . . . , xαn

,

1

x

1

x1,

1

x2, . . . ,

1

xn

(2)

logxlogx1,logx2, . . . ,logxn, ex ex1, ex2, . . . , exn.

1.1

The notion of Schur convexity was first introduced by Schur in 1923 1. It has many important applications in analytic inequalities2–7, combinatorial optimization8, isoperimetric problem for polytopes 9, linear regression 10, graphs and matrices 11, gamma and digamma functions12, reliability and availability13, and other related fields. The following definition for Schur convex or concave can be found in 1, 3, 7 and the references therein.

Definition 1.1. LetERn n ≥ 2be a set, a real-valued functionF onEis called a Schur convex function if

Fx1, x1, . . . , xnFy1, y2, . . . , yn 1.2

for each pair ofn-tuplesx x1, . . . , xnandy y1, . . . , ynonE, such thatxis majorized

byyin symbolsxy, that is,

k

i1

xi≤ k

i1

yi, k1,2, . . . , n−1,

n

i1

xi

n

i1

yi,

1.3

wherexi denotes theith largest component inx.F is called Schur concave if−F is Schur

convex.

The notation of multiplicative convexity was first introduced by Montel 14. The Schur multiplicative convexity was investigated by Niculescu 15, Guan 7, and Chu et al.16.

Definition 1.2see7,16. LetERn n≥2be a set, a real-valued functionF :ERis called a Schur multiplicatively convex function onEif

Fx1, x2, . . . , xnFy1, y2, . . . , yn 1.4

for each pair of n-tuples x x1, x2, . . . , xn and y y1, y2, . . . , ynon E, such that x is

logarithmically majorized byyin symbols logx≺logy, that is,

k

i1

xi≤ k

i1

yi, k1,2, . . . , n−1,

n

i1

xi

n

i1

yi.

1.5

(3)

In paper 17, Anderson et al. discussed an attractive class of inequalities, which arise from the notion of harmonic convex functions. Here, we introduce the notion of Schur harmonic convexity.

Definition 1.3. LetERn n ≥ 2be a set. A real-valued functionF onEis called a Schur harmonic convex function if

Fx1, x2, . . . , xnF

y1, y2, . . . , yn

1.6

for each pair ofn-tuplesx x1, x2, . . . , xnandy y1, y2, . . . , ynonE, such that 1/x≺1/y. Fis called a Schur harmonic concave function onEif1.6is reversed.

The main purpose of this paper is to discuss the Schur convexity, Schur multiplicative convexity, and Schur harmonic convexity of the following symmetric function:

φnx, r φnx1, x2, . . . , xn;r

1≤i1<i2···<irn ⎛ ⎝r

j1

xij

1xij ⎞ ⎠

1/r

, 1.7

wherex x1, x2, . . . , xnRn n≥2,r1,2, . . . , n, andi1, i2, . . . , ir are positive integers. As

applications, some inequalities are established by use of the theory of majorization.

2. Lemmas

In order to establish our main results we need several lemmas, which we present in this section.

The following lemma is so-called Schur’s condition which is very useful for determining whether or not a given function is Schur convex or Schur concave.

Lemma 2.1see6,7,18. Letf : Rn 0,nR 0,be a continuous symmetric

function. Iffis differentiable inRn, thenfis Schur convex if and only if

xixj

∂f ∂xi

∂f ∂xj

≥0 2.1

for alli, j 1,2, . . . , nand x x1, . . . , xnRn. Alsof is Schur concave if and only if 2.1is

reversed for alli, j 1,2, . . . , nandx x1, . . . , xnRn. Here,f is a symmetric function inRn

meaning thatfPx fxfor anyxRnand anyn×npermutation matrixP.

Remark 2.2. Sincef is symmetric, the Schur’s condition in Lemma 2.1, that is,2.1can be reduced to

x1−x2

∂f

∂x1 −

∂f ∂x2

(4)

Lemma 2.3see7,16. Letf:RnRbe a continuous symmtric function. Iffis differentiable

inRn, thenfis Schur multiplicatively convex if and only if

logx1−logx2

x1∂x∂f 1 −x2

∂f ∂x2

≥0 2.3

for allx x1, x2, . . . , xnRn. Also f is Schur multiplicatively concave if and only if 2.3 is

reversed.

Lemma 2.4. Letf:RnRbe a continuous symmetric function. Iffis differentiable inRn, then fis Schur harmonic convex if and only if

x1−x2

x2 1

∂f ∂x1 −x

2 2

∂f ∂x2

≥0 2.4

for allx x1, x2, . . . , xnRn. Alsofis Schur harmonic concave if and only if 2.4is reversed.

Proof. From Definitions1.1and1.3, we clearly see the fact thatf:RnRis Schur harmonic convex if and only ifFx 1/f1/x:RnRis Schur concave.

This fact,Lemma 2.1andRemark 2.2together with elementary calculation imply that

Lemma 2.4is true.

Lemma 2.5see5,6. Letx x1, x2, . . . , xnRnandni1xis. Ifcs, then

cx

nc/s−1

cx

1

nc/s−1,

cx2

nc/s−1, . . . ,

cxn nc/s−1

x1, x2, . . . , xn x. 2.5

Lemma 2.6see6. Letx x1, x2, . . . , xnRnandni1xis. Ifc≥0, then

cx

nc/s1

cx

1

nc/s1,

cx2

nc/s1, . . . ,

cxn

nc/s1

x1, x2, . . . , xn x. 2.6

Lemma 2.7see19. Suppose thatx x1, x2, . . . , xnRnandni1xis.If0≤λ≤1,then

sλx

nλ

sλx

1

nλ , sλx2

nλ , . . . , sλxn

nλ

x1, x2, . . . , xn x. 2.7

3. Main Results

Theorem 3.1. Forr∈ {1,2, . . . , n},the symmetric functionφnx, ris Schur concave inRn.

Proof. ByLemma 2.1andRemark 2.2, we only need to prove that

x1−x2

∂φ

nx, r

∂x1 −

∂φnx, r ∂x2

≤0 3.1

(5)

The proof is divided into four cases.

Case 1. Ifr1, then1.7leads to

φnx,1 φnx1, x2, . . . , xn; 1 n

i1

xi

1xi. 3.2

However3.2and elementary computation lead to

x1−x2

∂φ

nx,1

∂x1 −

∂φnx,1 ∂x2

x1−x221x1x2

x1x21x11x2 φnx,

1≤0. 3.3

Case 2. Ifn≥2 andrn, then1.7yields

φnx, n φnx1, x2, . . . , xn;n

n

i1

xi

1xi

1/n

. 3.4

From3.4and elementary computation, we have

x1−x2

∂φ

nx, n

∂x1 −

∂φnx, n ∂x2

x1−x222x1x2

n1x121x22

n

i1

xi

1xi 1/n−1

≤0. 3.5

Case 3. Ifn≥3 andr2, then by1.7we have

φnx,2 φnx1, x2,· · · , xn; 2

x

1

1x1

x2

1x2

1/2⎡

n j3

x1

1x1

xj

1xj

1/2⎤

φn1x2, x3, . . . , xn; 2

x

2

1x2

x1

1x1

1/2⎡

n j3

x2

1x2

xj

1xj 1/2⎤

φn−1x1, x3, . . . , xn; 2.

(6)

Elementary computation and3.6yield

x1−x2

∂φ

nx,2

∂x1 −

∂φnx,2 ∂x2

x1−x22 1x11x2

φnx,2

2

×

⎣ 2x1x2

x1x22x1x2

n

j3

1x1x2 32x12x2xj1xj x1xj2x1xjx2xj2x2xj

⎤ ⎦0.

3.7

Case 4. Ifn≥4 and 3≤rn−1, then from1.7, we have

φnx, r φnx1, x2, . . . , xn;r

φn−1x2, x3, . . . , xn;r

3≤i1<i2<···<ir−1≤n ⎛ ⎝ x1

1x1

r−1

j1

xij

1xij ⎞ ⎠

1/r

×

3≤i1<i2<···<ir−2≤n ⎛ ⎝ x1

1x1

x2

1x2

r−2

j1

xij

1xij ⎞ ⎠

1/r

φn−1x1, x3, . . . , xn;r

3≤i1<i2<···<ir−1≤n ⎛ ⎝ x2

1x2

r−1

j1

xij

1xij

⎞ ⎠

1/r

×

3≤i1<i2<···<ir−2≤n ⎛ ⎝ x1

1x1

x2

1x2

r−2

j1

xij

1xij ⎞ ⎠

1/r

,

3.8

x1−x2

∂φ

nx, r

∂x1 −

∂φnx, r ∂x2

x1−x22 1x121x22

×

⎡ ⎢

3≤i1<i2<···<ir−2≤n

2x1x2 x1/1x1 x2/1x2

r2 j1

xij/

1xij

3≤i1<i2<···<ir−1≤n

1x1x22x1x2

r1 j1

xij/1xij

x1/1x1 rj−11

xij/

1xij

x2/1x2 rj−11

xij/

1xij

⎤ ⎥ ⎦

×φnx, rr ≤0.

3.9

(7)

For the Schur multiplicative convexity or concavity ofφnx, r, we have the following

theorem

Theorem 3.2. It holds thatφnx, ris Schur multiplicatively concave in1,n.

Proof. According toLemma 2.3we only need to prove that

logx1−logx2

x1

∂φnx, r ∂x1 −x2

∂φnx, r ∂x2

≤0 3.10

for allx x1, x2, . . . , n∈1,nandr1,2, . . . , n.Then proof is divided into four cases.

Case 1. Ifr1, then3.2leads to

logx1−logx2

x1

∂φnx,1 ∂x1 −x2

∂φnx,1 ∂x2

logx1−logx2

x1−x2 1x11x2 φnx,

1≤0.

3.11

Case 2. Ifrn, n≥2, then3.4yields

logx1−logx2

x1∂φn∂xx, n 1 −x2

∂φnx, n ∂x2

logx1−logx2

x1−x2

n1x121x22

1−x1x2

n

i1

xi

1xi

1/n−1

≤0.

3.12

Case 3. Ifn≥3 andr2, then3.6implies

logx1−logx2

x1∂φnx,

2

∂x1 −x2

∂φnx,2 ∂x2

φnx,2

2

logx1−logx2

x1−x2 1x121x22

×

⎣ 1−x1x2

x1/1x1 x2/1x2

n

j3

x1x2 1−x1x2

xj/1xj

x1/1x1 xj/1xjx2/1x2 xj/1xj ⎤ ⎦≤0.

(8)

Case 4. Ifn≥4 and 3≤rn−1, then from3.8we have

logx1−logx2

x1∂φn∂xx, r 1 −x2

∂φnx, r ∂x2

logx1−logx2

x1−x2 1x121x22

×

⎡ ⎢

3≤i1<i2<···<ir−2≤n

1−x1x2

x1/1x1 x2/1x2 rj−12

xij/

1xij

3≤i1<i2<···<ir−1≤n

x1x21−x1x2rj−11

xij/

1xij

x1/1x1

r1 j1

xij/

1xijx2/1x2

r1 j1

xij/

1xij

⎤ ⎥ ⎦

× φnrx, r ≤0.

3.14

Therefore,Theorem 3.2follows from3.10and Cases1–4.

Remark 3.3. From3.11and3.12we know thatφnx,1is Schur multiplicatively concave

in0,nandφnx, nis Schur multiplicatively convex in0,1n.

Theorem 3.4. Forr ∈ {1,2, . . . , n},the symmetric functionφnx, ris Schur harmonic convex in Rn

.

Proof. According toLemma 2.4we only need to prove that

x1−x2

x2 1

∂φnx, r

∂x1 −x

2 2

∂φnx, r ∂x2

≥0 3.15

for allx x1, x2, . . . , xnRnandr1,2, . . . , n.

The proof is divided into four cases.

Case 1. Ifr1, then from3.2we have

x1−x2

x2 1

∂φnx,1

∂x1 −x

2 2

∂φnx,1 ∂x2

x1−x22

1x11x2φnx,

1≥0. 3.16

Case 2. Ifn≥2 andrn, then3.4leads to

x1−x2

x2 1

∂φnx, n

∂x1 −x

2 2

∂φnx, n ∂x2

x1−x22x1x22x1x2

n1x121x22

n

i1

xi

1xi 1/n−1

≥0.

(9)

Case 3. Ifn≥3 andr2, then3.6yields

x1−x2

x2 1

∂φnx,2

∂x1 −x

2 2

∂φnx,2 ∂x2

x1−x22 1x11x2

φnx,2

2

×

⎡ ⎣1

n

j3

x1x2x1xjx2xj3x1x2xj1xj x1xj2x1xjx2xj2x2xj

⎤ ⎦

≥0.

3.18

Case 4. Ifn≥4 and 3≤rn−1, then3.8implies

x1−x2

x2 1

∂φnx, r

∂x1 −x

2 2

∂φnx, r ∂x2

x1−x22 1x121x22

φnx, r r

×

⎡ ⎢

3≤i1<i2<···<ir−2≤n

x1x22x1x2

x1/1x1 x2/1x2

r2 j1

xij/

1xij

3≤i1<i2<···<ir−1≤n

x1x2x1x22x1x2

r2 j1

xij/

1xij

x1/1x1rj−11

xij/

1xij

x2/1x2rj−11

xij/

1xij

⎤ ⎥ ⎦

≥0.

3.19

Therefore,3.15follows from Cases1–4and the proof ofTheorem 3.4is completed.

4. Applications

In this section, we establish some inequalities by use of Theorems3.1,3.2and3.4and the theory of majorization.

Theorem 4.1. If x x1, x2,· · ·, xnRn, s ni1xi,Hnx n/

n

i11/xi, and r

{1,2, . . . , n}, then

1φnx, rφncx/nc/s−1, rforcs;

2φnx, rφncHnx−1/cx−1x, rforcni11/xi; 3φnx, rφncx/nc/s1, rforc≥0;

4φnx, rφncHnx 1/cx1x, rforc≥0;

(10)

6φnx, rφnnλ/ni1 1/xiλ/x, rfor0≤λ≤1; 7φnx, rφnsλx/nλ, rfor0≤λ≤1;

8φnx, rφnnλ/ni11/xiλ/x , rfor0≤λ≤1.

Proof. Theorem 4.1follows fromTheorem 3.1,Theorem 3.4and Lemmas2.5–2.7together with the fact that

sλx

sλx

1

, sλx2

, . . . , sλxn

x1, x2, . . . , xn x. 4.1

Theorem 4.2. Ifx x1, x2,· · ·, xnRn,Anx 1/nni1xi, andr∈ {1,2, . . . , n}, then

i

1≤i1<i2<···<irn ⎛ ⎝r

j1

xij

1xij

⎞ ⎠

1/r

rAAnx n1x

n!/r·r!n−r!

;

ii

1≤i1<i2<···<irn ⎛ ⎝r

j1

1 1xij

⎞ ⎠

1/r

r 1

An1x

n!/r·r!n−r! .

4.2

Proof. We clearly see that

Anx, Anx, . . . , Anxx1, x2, . . . , xn x. 4.3

Therefore,Theorem 4.2ifollows from4.3andTheorem 3.1together with1.7, and

Theorem 4.2iifollows from4.3andTheorem 3.4together with1.7.

If we taker1 inTheorem 4.2iandrninTheorem 4.2, respectively, then we have the following corollary.

Corollary 4.3. Ifx x1, x2, . . . , xnRnandGnx ni1xi1/n, then

i Gnx

Gn1x

Anx An1x;

ii An

x

1x

Anx

An1x;

iii An

1 1x

A 1

n1x.

4.4

Remark 4.4. If we take ni1xi 1 in Corollary 4.3i, then we obtain the Weierstrass inequality:see20, page 260

n

i1

1

xi 1

(11)

Theorem 4.5. Ifx x1, x2, . . . , xnRnandr ∈ {1,2, . . . , n}, then

i

1≤i1<i2<···<irn ⎛ ⎝r

j1

xij

1xij ⎞ ⎠

1/r

r Hnx

1Hnx

n!/r·r!n−r!

;

ii

1≤i1<i2<···<irn ⎛ ⎝r

j1

1 1xij

⎞ ⎠

1/r

r 1

1Hnx

n!/r·r!n−r! .

4.6

Proof. We clearly see that

1

Hnx,

1

Hnx, . . . ,

1

Hnx

1

x1,

1

x2, . . . ,

1

xn

x1. 4.7

Therefore,Theorem 4.5ifollows from4.7andTheorem 3.4together with1.7, and

Theorem 4.5iifollows from4.7andTheorem 3.1together with1.7.

If we take r 1 and r ninTheorem 4.5, respectively, then we get the following corollary.

Corollary 4.6. Ifx x1, x2, . . . , xnRn, then

i Gnx

Gn1x

Hnx

1Hnx;

ii Gn1x≥1Hnx;

iii An

x

1x

Hnx

1Hnx;

iv An

1 1x

≤ 1

1Hnx.

4.8

Theorem 4.7. Ifx x1, x2, . . . , xn∈1,nandr ∈ {1,2, . . . , n}, then

1≤i1<i2<···<irn ⎛ ⎝r

j1

xij

1xij ⎞ ⎠

1/r

r Gnx

1Gnx

n!/r·r!n−r!

. 4.9

Proof. We clearly see that

logGnx, Gnx, . . . , Gnx≺logx1, x2, . . . , xn. 4.10

(12)

If we take r 1 and r ninTheorem 4.7, respectively, then we get the following corollary.

Corollary 4.8. Ifx x1, x2, . . . , xn∈1,n, then

i An

x

1x

Gnx

1Gnx;

ii Gn1x≥1Gnx.

4.11

Remark 4.9. FromRemark 3.3and4.10together with1.7we clearly see that inequality in

Corollary 4.8iis reversed forx∈0,1nand inequality inCorollary 4.8iiis true forxRn.

Theorem 4.10. Ifx x1, x2,· · ·, xnRn,then

i

1≤i1<i2···<irn ⎛ ⎝r

j1

1xij 2xij

⎞ ⎠

1/r

1n i1xi

2ni1xi r−1

2

n−1!/r!n−r!

×r

2

n−1!/r·r!n−r−1!

for1≤rn−1;

ii

n

i1

1xi 2xi

1ni1xi

2ni1xi

n−1 2 ;

iii

1≤i1<i2···<irn ⎛ ⎝r

j1

1 2xij

⎞ ⎠

1/r

≤ 1

2ni1xi r−1

2

n−1!/r−1!n−r!

×r

2

n−1!/r·r!n−r−1!

for1≤rn−1;

iv

n

i1

1 2xi

1 2ni1xi

n−1 2 .

4.12

Proof. Theorem 4.10follows from Theorems3.1,3.4, and1.7together with the fact that

1x1,1x2, . . . ,1xn

1

n

i1

xi,1,1, . . . ,1

. 4.13

Theorem 4.11. LetAA1A2· · ·An1be ann-dimensional simplex inRnand letPbe an arbitrary point in the interior ofA. IfBi is the intersection point of straight lineAiP and hyperplanei

A1A2· · ·Ai−1Ai1· · ·An1, i1,2, . . . , n1, then forr ∈ {1,2, . . . , n1}one has

i

1≤i1<i2···<irn1

⎛ ⎝r

j1

PBij

AijBijPBij ⎞ ⎠

1/r

r

n2

n1!/r·r!n−r1!

(13)

ii

1≤i1<i2···<irn1

⎛ ⎝r

j1

AijBij

AijBijPBij ⎞ ⎠ 1/rr n 1

n2

n1!/r·r!n−r1!

;

iii

1≤i1<i2···<irn1

⎛ ⎝r

j1

PAij

AijBijPAij ⎞ ⎠ 1/rr n

2n1

n1!/r·r!n−r1!

;

iv

1≤i1<i2···<irn1

⎛ ⎝r

j1

AijBij

AijBijPAij ⎞ ⎠ 1/rr n 1 2n1

n1!/r·r!n−r1!

.

4.14

Proof. It is easy to see thatni11PBi/AiBi 1 and ni11PAi/AiBi n, these identical equations imply

1

n1, 1

n1, . . . , 1

n1

PB

1

A1B1,

PB2

A2B2, . . . ,

PBn1

An1Bn1

,

n

n1,

n n1, . . . ,

n n1

PA

1

A1B1,

PA2

A2B2, . . . ,

PAn1

An1Bn1

.

4.15

Therefore,Theorem 4.11follows from4.15, Theorems3.1,3.4, and1.7.

Remark 4.12. Mitrinovic’ et al. 21, pages 473–479 established a series of inequalities for

PAi/AiBi and PBi/AiBi,i 1,2, . . . , n1. Obvious, our inequalities in Theorem 4.11are

different from theirs.

Theorem 4.13. Suppose that AMnC n ≥ 2 is a complex matrix, λ1, λ2, . . . , λn, and σ1, σ2, . . . , σn are the eigenvalues and singular values ofA, respectively. IfA is a positive definite Hermitian matrix andr ∈ {1,2, . . . , n},then

i

1≤i1<i2···<irn ⎛ ⎝r

j1

λij

1λij

⎞ ⎠ 1/rr trA

ntrA

n!/r·r!n−r!

;

ii

1≤i1<i2···<irn ⎛ ⎝r

j1

1 1λij

⎞ ⎠ 1/rr n

ntrA

n!/r·r!n−r!

;

iii

1≤i1<i2···<irn ⎛ ⎝r

j1

1λij 2λij

⎞ ⎠ 1/rr n

detIA 1n detIA

n!/r·r!n−r!

;

iv

1≤i1<i2···<irn ⎛ ⎝r

j1

1 trAλij

⎞ ⎠ 1/rr 1 trAndetA

n!/r·r!n−r!

(14)

v

1≤i1<i2···<irn ⎛ ⎝r

j1

1

λij n

i1λi

n i1σi

⎞ ⎠

1/r

1≤i1<i2···<irn ⎛ ⎝r

j1

1

σij n

i1λi

n i1σi

⎞ ⎠

1/r

.

4.16

Proof. i–iiWe clearly see thatλi>0 i1,2, . . . , nandni1λitrA,then we have

trA

n ,

trA

n , . . . ,

trA

n

λ1, λ2, . . . , λn. 4.17

Therefore,Theorem 4.13iandiifollows from4.17, Theorems3.1,3.4, and1.7.

iiiIt is easy to see that 1λ1,1λ2, . . . ,1λnare the eigenvalues of matrixIAand

n

i11λi detIA, then we get

log

n

detIA, n

detIA, . . . , n

detIA

≺log1λ1,1λ2, . . . ,1λn

1λi ≥1, i1,2, . . . , n.

4.18

Therefore,Theorem 4.13iiifollows from4.18,Theorem 3.2, and1.7.

ivIt is not difficult to verify that

log

trA

n

detA, trA

n

detA, . . . , trA

n

detA

≺log

trA

λ1 ,

trA

λ2 , . . . ,

trA

λn

,

trA

λi ≥1, i1,2, . . . , n.

4.19

Therefore,Theorem 4.13ivfollows from4.19, andTheorem 3.2together with1.7.

vA result due to Weyl22gives

logλ1, λ2, . . . , λn≺logσ1, σ2, . . . , σn. 4.20

From4.20, we clearly see that

log

n

i1λi

n i1σi

λ1 ,

n i1λi

n i1σi

λ2 , . . . ,

n i1λi

n i1σi

λn

≺log

n

i1λi

n i1σi

σ1 ,

n i1λi

n i1σi

σ2 , . . . ,

n i1λi

n i1σi

σn

n i1λi

n i1σi

λi ,

n i1λi

n i1σi

σi ≥1, i1,2, . . . , n.

, 4.21

(15)

Acknowledgment

This work was supported by the National Science Foundation of Chinano. 60850005and the Natural Science Foundation of Zhejiang Provinceno. D7080080, Y607128, Y7080185.

References

1 I. Schur, “ ¨Uber eine klasse von mittelbildungen mit anwendungen auf die determinantentheorie,” Sitzungsberichte der Berliner Mathematischen Gesellschaft, vol. 22, pp. 9–20, 1923.

2 G. H. Hardy, J. E. Littlewood, and G. P ´olya, “Some simple inequalities satisfied by convex functions,” Messenger of Mathematics, vol. 58, pp. 145–152, 1929.

3 C. Stepolhkepniak, “An effective characterization of Schur-convex functions with applications,” Journal of Convex Analysis, vol. 14, no. 1, pp. 103–108, 2007.

4 J. S. Aujla and F. C. Silva, “Weak majorization inequalities and convex functions,”Linear Algebra and Its Applications, vol. 369, pp. 217–233, 2003.

5 K. Guan, “The Hamy symmetric function and its generalization,” Mathematical Inequalities & Applications, vol. 9, no. 4, pp. 797–805, 2006.

6 K.-Z. Guan, “Schur-convexity of the complete symmetric function,” Mathematical Inequalities & Applications, vol. 9, no. 4, pp. 567–576, 2006.

7 K.-Z. Guan, “Some properties of a class of symmetric functions,”Journal of Mathematical Analysis and Applications, vol. 336, no. 1, pp. 70–80, 2007.

8 F. K. Hwang and U. G. Rothblum, “Partition-optimization with Schur convex sum objective functions,”SIAM Journal on Discrete Mathematics, vol. 18, no. 3, pp. 512–524, 2004.

9 X.-M. Zhang, “Schur-convex functions and isoperimetric inequalities,”Proceedings of the American Mathematical Society, vol. 126, no. 2, pp. 461–470, 1998.

10 C. Stepolhkepniak, “Stochastic ordering and Schur-convex functions in comparison of linear experiments,”Metrika, vol. 36, no. 5, pp. 291–298, 1989.

11 G. M. Constantine, “Schur convex functions on the spectra of graphs,”Discrete Mathematics, vol. 45, no. 2-3, pp. 181–188, 1983.

12 M. Merkle, “Convexity, Schur-convexity and bounds for the gamma function involving the digamma function,”The Rocky Mountain Journal of Mathematics, vol. 28, no. 3, pp. 1053–1066, 1998.

13 F. K. Hwang, U. G. Rothblum, and L. Shepp, “Monotone optimal multipartitions using Schur convexity with respect to partial orders,”SIAM Journal on Discrete Mathematics, vol. 6, no. 4, pp. 533– 547, 1993.

14 P. Montel, “Sur les fonctions convexes et les fonctions sousharmoniques,”Journal de Math´ematiques, vol. 7, no. 9, pp. 29–60, 1928.

15 C. P. Niculescu, “Convexity according to the geometric mean,”Mathematical Inequalities & Applications, vol. 3, no. 2, pp. 155–167, 2000.

16 Y. Chu, X. Zhang, and G. Wang, “The Schur geometrical convexity of the extended mean values,” Journal of Convex Analysis, vol. 15, no. 4, pp. 707–718, 2008.

17 G. D. Anderson, M. K. Vamanamurthy, and M. Vuorinen, “Generalized convexity and inequalities,” Journal of Mathematical Analysis and Applications, vol. 335, no. 2, pp. 1294–1308, 2007.

18 A. W. Marshall and I. Olkin, Inequalities: Theory of Majorization and Its Applications, vol. 143 of Mathematics in Science and Engineering, Academic Press, New York, NY, USA, 1979.

19 S.-H. Wu, “Generalization and sharpness of the power means inequality and their applications,” Journal of Mathematical Analysis and Applications, vol. 312, no. 2, pp. 637–652, 2005.

20 P. S. Bullen,A Dictionary of Inequalities, vol. 97, Longman, Harlow, UK, 1998.

21 D. S. Mitrinovi´c, J. E. Peˇcari´c, and V. Volenec,Recent Advances in Geometric Inequalities, vol. 28, Kluwer Academic Publishers, Dordrecht, The Netherlands, 1989.

References

Related documents