ON SOME NEW MEAN VALUE INEQUALITIES
LIANG-CHENG WANG AND CAI-LIANG LI SCHOOL OFMATHEMATICSSCIENCE
CHONGQINGINSTITUTE OFTECHNOLOGY
NO. 4OFXINGSHENGLU
YANGJIAPING400050
CHONGQING, THEPEOPLE’SREPUBLIC OFCHINA. wangliangcheng@163.com
CHENGDUELECTROMECHANICALCOLLEGE
CHENGDU610031, SICHUANPROVINCE
THEPEOPLE’SREPUBLIC OFCHINA. dzlcl@163.com
Received 04 March, 2007; accepted 27 April, 2007 Communicated by P.S. Bullen
ABSTRACT. In this paper, using the arithmetic-geometric mean inequality, we obtain some new mean value inequalities. Finally, some applications are given, they are extension of Hölder’s inequalities.
Key words and phrases: Mean value inequality, Hölder’s inequality, Continuous positive function, Extension.
2000 Mathematics Subject Classification. 26D15.
1. INTRODUCTION ANDMAIN RESULTS
Leta > 0, b > 0and t ∈ (0,1). It is well-known that the following arithmetic-geometric mean inequality holds
(1.1) atb1−t≤ta+ (1−t)b.
The arithmetic-geometric mean inequality is a classical inequality with many applications.
Also, there exist extensive works devoted to generalizing or improving the arithmetic-geometric mean inequality. In this respect, we refer the reader to [1] – [7] and the references cited therein for updated results.
In this paper, by (1.1), we obtain some new mean value inequalities. Finally, some applica- tions are given.
In this paper, we agree
q
X
i=q+1
bi = 0, (bi ∈R, q∈N).
The first author is partially supported by the Key Research Foundation of the Chongqing Institute of Technology under Grant 2004DZ94.
073-07
Theorem 1.1. Letxi >0 (i= 1,2, . . . , n; n ≥2)andt∈(0,1).
(1) For the following B(k) 4 1
n2
"
k
k
X
i=1
xi+
n
X
i=1
xti
! n X
i=k+1
x1−ti
!
+
n
X
i=k+1
xti
! k X
i=1
x1−ti
!#
, (k = 1,2, . . . , n) and
C(j) 4 1 n2
"
(n−j + 1)
n
X
i=j
xi+
n
X
i=1
xti
! j−1 X
i=1
x1−ti
!
+
j−1
X
i=1
xti
! n X
i=j
x1−ti
!#
, (j = 1,2, . . . , n), we have
(1.2) 1 n
n
X
i=1
xti
! 1 n
n
X
i=1
x1−ti
!
=B(1) ≤B(2)≤ · · · ≤B(k)≤B(k+ 1) ≤ · · · ≤B(n) = 1 n
n
X
i=1
xi
and (1.3) 1
n
n
X
i=1
xti
! 1 n
n
X
i=1
x1−ti
!
=C(n)≤C(n−1)≤ · · · ≤C(j)≤C(j−1)≤ · · · ≤C(1) = 1 n
n
X
i=1
xi.
(2) For1≤j < k < l ≤n(n ≥3), we have (1.4) (k−j+ 1)
k
X
i=j
xi+ (l−k+ 1)
l
X
i=k
xi+
l
X
i=j
xti
! l X
i=j
x1−ti
!
≤(l−j+ 1)
l
X
i=j
xi+
k
X
i=j
xti
! k X
i=j
x1−ti
! +
l
X
i=k
xti
! l X
i=k
x1−ti
! .
Corollary 1.2. Letxi >0 (i= 1,2, . . . , n, n≥2)andp, q be any two positive numbers.
(1) For D(k) 4 1
n2
"
k
k
X
i=1
xp+qi +
n
X
i=1
xpi
! n X
i=k+1
xqi
!
+
n
X
i=k+1
xpi
! k X
i=1
xqi
!#
, (k = 1,2, . . . , n)
and
E(j) 4 1 n2
"
(n−j+ 1)
n
X
i=j
xp+qi +
n
X
i=1
xpi
! j−1 X
i=1
xqi
!
+
j−1
X
i=1
xpi
! n X
i=j
xqi
!#
, (j = 1,2, . . . , n), we have
(1.5) 1 n
n
X
i=1
xpi
! 1 n
n
X
i=1
xqi
!
=D(1)≤D(2) ≤ · · · ≤D(k)≤D(k+ 1)≤ · · · ≤D(n) = 1 n
n
X
i=1
xp+qi and
(1.6) 1 n
n
X
i=1
xpi
! 1 n
n
X
i=1
xqi
!
=E(n)≤E(n−1)≤ · · · ≤E(j)≤E(j−1)≤ · · · ≤E(1) = 1 n
n
X
i=1
xp+qi .
(2) For1≤j < k < l ≤n(n ≥3), we have (1.7) (k−j+ 1)
k
X
i=j
xp+qi + (l−k+ 1)
l
X
i=k
xp+qi +
l
X
i=j
xpi
! l X
i=j
xqi
!
≤(l−j+ 1)
l
X
i=j
xp+qi +
k
X
i=j
xpi
! k X
i=j
xqi
! +
l
X
i=k
xpi
! l X
i=k
xqi
! .
2. PROOF OFTHEOREM ANDCOROLLARY
Proof of Theorem 1.1. (1) Two equalities are clear in (1.2). To complete the proof of (1.2), we only need to prove thatB(k)≤B(k+ 1)(1≤k ≤n−1). Indeed, from (1.1) we have
(2.1) xtk+1
k
X
i=1
x1−ti =
k
X
i=1
xtk+1x1−ti ≤
k
X
i=1
(txk+1+ (1−t)xi), and
(2.2) x1−tk+1
k
X
i=1
xti =
k
X
i=1
x1−tk+1xti ≤
k
X
i=1
((1−t)xk+1+txi). Using (2.1) and (2.2), after a simple manipulation we get
(2.3) xtk+1
k
X
i=1
x1−ti +x1−tk+1
k
X
i=1
xti ≤kxk+1+
k
X
i=1
xi.
Fork = 1,2, . . . , n−1, by (2.3) we get
B(k) = 1 n2
"
k
k
X
i=1
xi+
n
X
i=1
xti
! n X
i=k+1
x1−ti
! +
n
X
i=k+1
xti
! k X
i=1
x1−ti
!#
= 1 n2
"
k
k
X
i=1
xi+xk+1 +
n
X
i=1
xti
! n X
i=k+2
x1−ti
!
+
k
X
i=1
xti+
n
X
i=k+2
xti
!
x1−tk+1+
n
X
i=k+2
xti
! k X
i=1
x1−ti
!
+xtk+1
k
X
i=1
x1−ti
#
= 1 n2
"
k
k
X
i=1
xi+xk+1 +xtk+1
k
X
i=1
x1−ti +x1−tk+1
k
X
i=1
xti
+
n
X
i=1
xti
! n X
i=k+2
x1−ti
! +
n
X
i=k+2
xti
! k+1 X
i=1
x1−ti
!#
≤ 1 n2
"
k
k
X
i=1
xi+xk+1 +kxk+1+
k
X
i=1
xi
+
n
X
i=1
xti
! n X
i=k+2
x1−ti
! +
n
X
i=k+2
xti
! k+1 X
i=1
x1−ti
!#
= 1 n2
"
(k+ 1)
k+1
X
i=1
xi +
n
X
i=1
xti
! n X
i=k+2
x1−ti
! +
n
X
i=k+2
xti
! k+1 X
i=1
x1−ti
!#
=B(k+ 1).
By same arguments of proof for (1.2), we can also get inequalities in (1.3).
(2) For1≤j < k < l ≤n, from (1.1) we have
k−1
X
i=j
xti
! l X
i=k+1
x1−ti
!
=
k−1
X
i=j l
X
s=k+1
xtix1−ts (2.4)
≤
k−1
X
i=j l
X
s=k+1
(txi+ (1−t)xs)
= (l−k)
k−1
X
i=j
txi+ (k−j)
l
X
i=k+1
(1−t)xi
and
(2.5)
l
X
i=k+1
xti
! k−1 X
i=j
x1−ti
!
≤(l−k)
k−1
X
i=j
(1−t)xi+ (k−j)
l
X
i=k+1
txi.
Using (2.4) and (2.5), after a simple manipulation we have
(2.6)
k−1
X
i=j
xti
! l X
i=k+1
x1−ti
! +
l
X
i=k+1
xti
! k−1 X
i=j
x1−ti
!
≤(l−k)
k−1
X
i=j
xi+ (k−j)
l
X
i=k+1
xi.
From (2.6) we obtain
(k−j + 1)
k
X
i=j
xi+ (l−k+ 1)
l
X
i=k
xi+
l
X
i=j
xti
! l X
i=j
x1−ti
!
= (k−j+ 1)
k
X
i=j
xi+ (l−k+ 1)
l
X
i=k+1
xi+ (l−k)xk
+
k
X
i=j
xti
! k X
i=j
x1−ti
! +
l
X
i=k+1
xti
! l X
i=k+1
x1−ti
!
+xtk
l
X
i=k+1
x1−ti +x1−tk
l
X
i=k+1
xti +xk
+
k−1
X
i=j
xti
! l X
i=k+1
x1−ti
! +
l
X
i=k+1
xti
! k−1 X
i=j
x1−ti
!
≤(k−j + 1)
k
X
i=j
xi+ (l−k+ 1)
l
X
i=k+1
xi+ (l−k)xk
+
k
X
i=j
xti
! k X
i=j
x1−ti
! +
l
X
i=k
xti
! l X
i=k
x1−ti
!
+ (l−k)
k−1
X
i=j
xi+ (k−j)
l
X
i=k+1
xi
= (l−j+ 1)
l
X
i=j
xi+
k
X
i=j
xti
! k X
i=j
x1−ti
! +
l
X
i=k
xti
! l X
i=k
x1−ti
! ,
which implies (1.4).
This completes the proof of Theorem 1.1.
Proof of Corollary 1.2. Replacet, 1−t andxi in Theorem 1.1 by p+qp , p+qq andxp+qi , respec-
tively. We obtain Corollary 1.2.
3. APPLICATIONS
Proposition 3.1. Letxir >0 (i= 1,2, . . . , n,n ≥ 2;r = 1,2, . . . , m,m ≥2)andt ∈ (0,1).
For
F(k) 4 1 n2
k
k
X
i=1 m
X
r=1
xir
! +
n
X
i=1 m
X
r=1
xir
!t!
n
X
i=k+1 m
X
r=1
xir
!1−t
+
n
X
i=k+1 m
X
r=1
xir
!t!
k
X
i=1 m
X
r=1
xir
!1−t
, (k = 1,2, . . . , n) and
G(h) 4 1 n2
n
X
i=1 n
X
j=1
h
X
r=1
xir
!t h X
r=1
xjr
!1−t +
m
X
r=h+1
xtirx1−tjr
, (h= 1,2, . . . , m), we have
1 n2
n
X
i=1 n
X
j=1 m
X
r=1
xtirx1−tjr
! (3.1)
=G(1)≤G(2)≤ · · · ≤G(h)≤G(h+ 1) ≤ · · · ≤G(m)
= 1
n
n
X
i=1 m
X
r=1
xir
!t!
1 n
n
X
i=1 m
X
r=1
xir
!1−t
=F(1)≤F(2)≤ · · · ≤F(k)≤F(k+ 1) ≤ · · · ≤F(n)
= 1 n
n
X
i=1 m
X
r=1
xir.
Proof. Forxir >0,xjr >0(1≤i, j ≤n, r= 1,2, . . . , m) andt ∈(0,1). We write P(i, j;h) 4
h
X
r=1
xir
!t h X
r=1
xjr
!1−t +
m
X
r=h+1
xtirx1−tjr (h = 1,2, . . . , m).
The first named author of this paper showed in [8] that the following chain of Hölder’s inequal- ities holds
m
X
r=1
xtirx1−tjr =P(i, j; 1) (3.2)
≤P(i, j; 2)≤ · · · ≤P(i, j;h)≤P(i, j;h+ 1)≤ · · · ≤P(i, j;m)
=
m
X
r=1
xir
!t m
X
r=1
xjr
!1−t
. From the properties of inequality and (3.2), we have
1 n2
n
X
i=1 n
X
j=1 m
X
r=1
xtirx1−tjr
!
= 1 n2
n
X
i=1 n
X
j=1
P(i, j; 1) (3.3)
≤ 1 n2
n
X
i=1 n
X
j=1
P(i, j; 2)
≤ · · · ≤ 1 n2
n
X
i=1 n
X
j=1
P(i, j;h)
≤ 1 n2
n
X
i=1 n
X
j=1
P(i, j;h+ 1)≤ · · ·
≤ 1 n2
n
X
i=1 n
X
j=1
P(i, j;m)
= 1 n2
n
X
i=1 n
X
j=1 m
X
r=1
xir
!t m
X
r=1
xjr
!1−t
= 1
n
n
X
i=1 m
X
r=1
xir
!t!
1 n
n
X
i=1 m
X
r=1
xir
!1−t
.
It is easy to see that
(3.4) G(h) = 1
n2
n
X
i=1 n
X
j=1
P(i, j;h), h= 1,2, . . . , m.
(3.3) and (3.4) imply inequalities between the first equality and the second equality in (3.1).
Replacingxi in (1.2) byPm
r=1xir, we obtain inequalities between the third equality and the fourth equality in (3.1).
This completes the proof of Proposition 3.1.
Proposition 3.2. Letfi : [a, b]7→(0,+∞) (a < b)be continuous functions(i= 1,2, . . . , n, n≥ 2)andt ∈(0,1). For
H(k) = 1 n2
"
k
k
X
i=1
Z b a
fi(x)dx
+
n
X
i=1
Z b a
fi(x)dx
t! n X
i=k+1
Z b a
fi(x)dx 1−t!
+
n
X
i=k+1
Z b a
fi(x)dx
t! k X
i=1
Z b a
fi(x)dx
1−t!#
, (k = 1,2, . . . , n) and anyy∈[a, b], we have
1 n2
n
X
i=1 n
X
j=1
Z b a
(fi(x))t(fj(x))1−tdx (3.5)
≤ 1 n2
" n X
i=1 n
X
j=1
Z y a
fi(x)dx
tZ y a
fj(x)dx 1−t
+ Z b
y
(fi(x))t(fj(x))1−tdx
!#
≤ 1 n
n
X
i=1
Z b a
fi(x)dx t!
1 n
n
X
i=1
Z b a
fi(x)dx 1−t!
=H(1)≤H(2)≤ · · · ≤H(k)≤H(k+ 1)≤ · · · ≤H(n)
= 1 n
n
X
i=1
Z b a
fi(x)dx.
Proof. For1 ≤ i, j ≤ n, t ∈ (0,1), y ∈ [a, b]and continuous functionsfi : [a, b] 7→ (0,+∞) (i = 1,2, . . . , n;n ≥ 2), in [8], Wang also obtained the following refinement for the integral form of Hölder’s inequalities:
Z b a
(fi(x))t(fj(x))1−tdx (3.6)
≤ Z y
a
fi(x)dx
tZ y a
fj(x)dx 1−t
+ Z b
y
(fi(x))t(fj(x))1−tdx
≤ Z b
a
fi(x)dx
tZ b a
fj(x)dx 1−t
. Using the properties of inequality and (3.6), we have
1 n2
n
X
i=1 n
X
j=1
Z b a
(fi(x))t(fj(x))1−tdx
≤ 1 n2
n
X
i=1 n
X
j=1
Z y a
fi(x)dx
tZ y a
fj(x)dx 1−t
+ Z b
y
(fi(x))t(fj(x))1−tdx
!
≤ 1 n2
n
X
i=1 n
X
j=1
Z b a
fi(x)dx
tZ b a
fj(x)dx 1−t
= 1
n
n
X
i=1
Z b a
fi(x)dx t!
1 n
n
X
i=1
Z b a
fi(x)dx 1−t!
, which is two inequalities of left hand in (3.2).
Replacing xi in (1.2) by Rb
afi(x)dx, we obtain inequalities between the two equalities in (3.2).
This completes the proof of Proposition 3.2.
Remark 3.3. (3.1) and (3.2) are extensions of Hölder’s inequalities.
REFERENCES
[1] D.S. MITRINOVI ´CANDP.M. VASI ´C, Sredine, Matematiˇcka Biblioteka, Beograd, 40 (1969).
[2] D.S. MITRINOVI ´C, Analytic Inequalities, Springer-Verlag, Berlin/New York, 1970.
[3] J.-C. KUANG, Applied Inequalities, Shandong Science and Technology Press, 2004. (Chinese).
[4] L.-C. WANG, Convex Functions and Their Inequalities, Sichuan University Press, Chengdu, China, 2001. (Chinese).
[5] C.L. WANG, Inequalities of the Rado-Popoviciu type for functions and their applications, J. Math.
Anal. Appl., 100 (1984), 436–446.
[6] G.H. HARDY, J.E. LITTLEWOODANDG PÓLYA, Inequalities, 2nd ed., Cambridge, 1952.
[7] FENG QI, Generalized weighted mean values with two parameters , Proc. R. Soc. Lond. A., 454 (1998), 2723–2732.
[8] L.C. WANG, Two mappings related to Hölder’s inequalities, Univ. Beograd. Publ. Elektrotehn. Fak.
Ser. Mat., 15 (2004), 92–97.