In [4], Fejér established the following weighted generalization of the inequality(1.1): Theorem A

FEJÉR INEQUALITIES FOR WRIGHT-CONVEX FUNCTIONS

MING-IN HO

CHINAINSTITUTE OFTECHNOLOGY

NANKANG, TAIPEI

TAIWAN11522 mingin@cc.chit.edu.tw

Received 07 September, 2006; accepted 18 November, 2006 Communicated by P.S. Bullen

ABSTRACT. In this paper, we establish several inequalities of Fejér type for Wright-convex functions.

Key words and phrases: Hermite-Hadamard inequality, Fejér inequality, Wright-convex functions.

2000 Mathematics Subject Classification. 26D15.

1. INTRODUCTION

Iff : [a, b]→Ris a convex function, then

(1.1) f

a+b 2

≤ 1 b−a

Z b

a

f(x)dx≤ f(a) +f(b) 2 is known as the Hermite-Hadamard inequality ([5]).

In [4], Fejér established the following weighted generalization of the inequality(1.1):

Theorem A. Iff : [a, b]→Ris a convex function, then the inequality

(1.2) f

a+b 2

Z b

a

p(x)dx≤ Z b

a

f(x)p(x)dx≤ f(a) +f(b) 2

Z b

a

p(x)dx

holds, wherep: [a, b]→Ris nonnegative, integrable, and symmetric aboutx= ^a+b₂ .

In recent years there have been many extensions, generalizations, applications and similar results of the inequalities(1.1)and(1.2)see [1] – [8], [10] – [16].

In [2], Dragomir established the following theorem which is a refinement of the first inequal- ity of(1.1).

Theorem B. Iff : [a, b]→Ris a convex function, andH is defined on[0,1]by H(t) = 1

b−a Z b

a

f

tx+ (1−t)a+b 2

dx,

233-06

thenH is convex, increasing on[0,1],and for allt∈[0,1], we have

(1.3) f

a+b 2

=H(0)≤H(t)≤H(1) = 1 b−a

Z b

a

f(x)dx.

In [11], Yang and Hong established the following theorem which is a refinement of the second inequality of(1.1):

Theorem C. Iff : [a, b]→Ris a convex function, andF is defined on[0,1]by F (t) = 1

2 (b−a) Z b

a

f

1 +t 2

a+

1−t 2

x

+f

1 +t 2

b+

1−t 2

x

dx,

thenF is convex, increasing on[0,1],and for allt ∈[0,1], we have

(1.4) 1

b−a Z b

a

f(x)dx=F (0)≤F (t)≤F(1) = f(a) +f(b)

2 .

We recall the definition of a Wright-convex function:

Definition 1.1 ([9, p. 223]). We say thatf : [a, b]→ Ris a Wright-convex function, if, for all x,y+δ∈[a, b]withx < y andδ≥0, we have

(1.5) f(x+δ) +f(y)≤f(y+δ) +f(x).

LetC([a, b])be the set of all convex functions on[a, b]andW([a, b])be the set of all Wright- convex functions on[a, b]. ThenC([a, b]) $ W([a, b]). That is, a convex function must be a Wright-convex function but the converse is not true. (see [9, p. 224]).

In [10], Tseng, Yang and Dragomir established the following theorems for Wright-convex functions related to the inequality(1.1), Theorem A and Theorem B:

Theorem D. Letf ∈W([a, b])∩L₁[a, b].Then the inequality(1.1)holds.

Theorem E. Letf ∈ W([a, b])∩L₁[a, b]and letH be defined as in Theorem B. ThenH ∈ W([0,1])is increasing on[0,1], and the inequality(1.3)holds for allt∈[0,1].

Theorem F. Let f ∈ W([a, b])∩L₁[a, b] and letF be defined as in Theorem C. Then F ∈ W([0,1])is increasing on[0,1], and the inequality(1.4)holds for allt∈[0,1].

In [12], Yang and Tseng established the following theorem which refines the inequality(1.2):

Theorem G ([12, Remark 6]). Letf andpbe defined as in Theorem A. IfP,Qare defined on [0,1]by

(1.6) P (t) =

Z b

a

f

tx+ (1−t)a+b 2

p(x)dx (t∈(0,1))

and

(1.7) Q(t) = Z b

a

1 2

f

1 +t

2 a+1−t 2 x

p

x+a 2

+ f

1 +t

2 b+ 1−t 2 x

p

x+b 2

dx (t ∈(0,1)),

thenP,Qare convex and increasing on[0,1]and, for allt∈[0,1],

(1.8) f

a+b 2

Z b

a

p(x)dx=P (0)≤P (t)≤P (1) = Z b

a

f(x)p(x)dx

and (1.9)

Z b

a

f(x)p(x)dx=Q(0) ≤Q(t)≤Q(1) = f(a) +f(b) 2

Z b

a

p(x)dx.

In this paper, we establish some results about Theorem A and Theorem G for Wright-convex functions which are weighted generalizations of Theorem D, E and F.

2. MAINRESULTS

In order to prove our main theorems, we need the following lemma [10]:

Lemma 2.1. Iff : [a, b]→R, then the following statements are equivalent:

(1) f ∈W([a, b]) ;

(2) for alls, t, u, v∈[a, b]withs≤t≤u≤vandt+u=s+v, we have

(2.1) f(t) +f(u)≤f(s) +f(v).

Theorem 2.2. Letf ∈ W([a, b])∩L₁[a, b]and letp: [a, b]→ Rbe nonnegative, integrable, and symmetric aboutx= ^a+b₂ .Then the inequality(1.2)holds.

Proof. For the inequality (2.1) and the assumptions thatpis nonnegative, integrable, and sym- metric aboutx= ^a+b₂ , we have

f

a+b 2

Z b

a

p(x)dx

= Z ^a+b₂

a

f

a+b 2

p(x)dx+ Z ^a+b₂

a

f

a+b 2

p(a+b−x)dx

= Z ^a+b₂

a

f

a+b 2

+f

a+b 2

p(x)dx

≤ Z ^a+b₂

a

[f(x) +f(a+b−x)]p(x)dx

x≤ a+b

2 ≤ a+b

2 ≤a+b−x

= Z ^a+b₂

a

f(x)p(x)dx+ Z b

a+b 2

f(x)p(x)dx

= Z b

a

f(x)p(x)dx,

and

f(a) +f(b) 2

Z b

a

p(x)dx

= Z ^a+b₂

a

f(a) +f(b) 2

p(x)dx+ Z ^a+b₂

a

f(a) +f(b) 2

p(a+b−x)dx

= Z ^a+b₂

a

[f(a) +f(b)]p(x)dx

≥ Z ^a+b₂

a

[f(x) +f(a+b−x)]p(x)dx (a≤x≤a+b−x≤b)

= Z ^a+b₂

a

f(x)p(x)dx+ Z b

a+b 2

f(x)p(x)dx= Z b

a

f(x)p(x)dx.

This completes the proof.

Remark 2.3. If we set p(x) ≡ 1 (x∈[a, b]) in Theorem 2.2, then Theorem 2.2 generalizes Theorem D.

Remark 2.4. FromC([a, b])$W([a, b]), Theorem 2.2 generalizes Theorem A.

Theorem 2.5. Letf andpbe defined as in Theorem 2.2 and letP be defined as in(1.6). Then P ∈W([0,1])is increasing on[0,1], and the inequality(1.8)holds for allt ∈[0,1].

Proof. Ifs, t, u, v∈[0,1]ands ≤t≤u≤v,t+u=s+v, then forx∈ a,^a+b₂

we have b≥sx+ (1−s)a+b

2 ≥tx+ (1−t)a+b 2

≥ux+ (1−u)a+b

2 ≥vx+ (1−v)a+b 2 ≥a and ifx∈_a+b

2 , b , then

a≤sx+ (1−s)a+b

2 ≤tx+ (1−t)a+b 2

≤ux+ (1−u)a+b

2 ≤vx+ (1−v)a+b 2 ≤b, where

tx+ (1−t)a+b 2

+

ux+ (1−u)a+b 2

=

sx+ (1−s)a+b 2

+

vx+ (1−v)a+b 2

.

By the inequality(2.1), we have

(2.2) f

tx+ (1−t)a+b 2

+f

ux+ (1−u)a+b 2

≤f

sx+ (1−s)a+b 2

+f

vx+ (1−v)a+b 2

for allx∈[a, b]. Now, using the inequality(2.2)andpis nonnegative on[a, b], we have (2.3)

f

tx+ (1−t)a+b 2

+f

ux+ (1−u)a+b 2

p(x)

≤

f

sx+ (1−s)a+b 2

+f

vx+ (1−v)a+b 2

p(x) for allx∈[a, b]. Integrating the inequality(2.3)overxon[a, b], we have

P (t) +P(u)≤P (s) +P (v).

HenceP ∈W([0,1]).

Next, if0≤s≤t≤1andx∈ a,^a+b₂

, then tx+ (1−t)a+b

2 ≤sx+ (1−s)a+b 2

≤s(a+b−x) + (1−s)a+b 2

≤t(a+b−x) + (1−t)a+b 2 , where

sx+ (1−s)a+b 2

+

s(a+b−x) + (1−s)a+b 2

=

tx+ (1−t)a+b 2

+

t(a+b−x) + (1−t)a+b 2

. By the inequality (2.1)and the assumptions that p is nonnegative, integrable, and symmetric aboutx= ^a+b₂ , we have

P (s) = Z b

a

f

sx+ (1−s)a+b 2

p(x)dx

= Z ^a+b₂

a

f

sx+ (1−s)a+b 2

p(x)dx

+ Z ^a+b₂

a

f

s(a+b−x) + (1−s)a+b 2

p(a+b−x)dx

= Z ^a+b₂

a

f

sx+ (1−s)a+b 2

+f

s(a+b−x) + (1−s)a+b 2

p(x)dx

≤ Z ^a+b₂

a

f

tx+ (1−t)a+b 2

+f

t(a+b−x) + (1−t)a+b 2

p(x)dx

= Z ^a+b₂

a

f

tx+ (1−t)a+b 2

p(x)dx

+ Z ^a+b₂

a

f

t(a+b−x) + (1−t)a+b 2

p(a+b−x)dx

= Z b

a

f

tx+ (1−t)a+b 2

p(x)dx=P (t).

Thus,P is increasing on[0,1], and the inequality(1.8)holds for allt∈[0,1].

This completes the proof.

Remark 2.6. If we set p(x) ≡ 1 (x∈[a, b]) in Theorem 2.5, then Theorem 2.2 generalizes Theorem E.

Theorem 2.7. Letf andpbe defined as in Theorem 2.2 and letQbe defined as in(1.7). Then Q∈W([0,1])is increasing on[0,1], and the inequality(1.9)holds for allt ∈[0,1].

Proof. Ifs, t, u, v∈[0,1]ands ≤t≤u≤v,t+u=s+v, then for allx∈[a, b]we have a ≤

1 +v 2

a+

1−v 2

x≤

1 +u 2

a+

1−u 2

x

≤

1 +t 2

a+

1−t 2

x≤

1 +s 2

a+

1−s 2

x≤b

and

a ≤

1 +s 2

b+

1−s 2

x≤

1 +t 2

b+

1−t 2

x

≤

1 +u 2

b+

1−u 2

x≤

1 +v 2

b+

1−v 2

x≤b, where

1 +u 2

a+

1−u 2

x

+

1 +t 2

a+

1−t 2

x

=

1 +v 2

a+

1−v 2

x

+

1 +s 2

a+

1−s 2

x

and

1 +t 2

b+

1−t 2

x

+

1 +u 2

b+

1−u 2

x

=

1 +s 2

b+

1−s 2

x

+

1 +v 2

b+

1−v 2

x

. By the inequality(2.1), we have

(2.4) f

1 +u 2

a+

1−u 2

x

+f

1 +t 2

a+

1−t 2

x

≤f

1 +v 2

a+

1−v 2

x

+f

1 +s 2

a+

1−s 2

x

and (2.5) f

1 +t 2

b+

1−t 2

x

+f

1 +u 2

b+

1−u 2

x

≤f

1 +s 2

b+

1−s 2

x

+f

1 +v 2

b+

1−v 2

x

for allx ∈[a, b]. Now, using the inequality(2.4), (2.5)and the assumptions thatpis nonnega- tive on[a, b], we have

1 2f

1 +u 2

a+

1−u 2

x

p

x+a 2

(2.6)

+ 1 2f

1 +t 2

a+

1−t 2

x

p

x+a 2

+ 1 2f

1 +t 2

b+

1−t 2

x

p

x+b 2

+ 1 2f

1 +u 2

b+

1−u 2

x

p

x+b 2

≤ 1 2f

1 +v 2

a+

1−v 2

x

p

x+a 2

+1 2f

1 +s 2

a+

1−s 2

x

p

x+a 2

+1 2f

1 +s 2

b+

1−s 2

x

p

x+b 2

+1 2f

1 +v 2

b+

1−v 2

x

p

x+b 2

Integrating the inequality(2.6)overxon[a, b], we have

Q(t) +Q(u)≤Q(s) +Q(v). HenceQ∈W([0,1]).

Next, if0≤s≤t≤1andx∈[a, b], then 1 +t

2

a+

1−t 2

x≤

1 +s 2

a+

1−s 2

x

≤

1 +s 2

b+

1−s 2

(a+b−x)

≤

1 +t 2

b+

1−t 2

(a+b−x)

and

1 +t 2

a+

1−t 2

(a+b−x)≤

1 +s 2

a+

1−s 2

(a+b−x)

≤

1 +s 2

b+

1−s 2

x

≤

1 +t 2

b+

1−t 2

x,

where 1 +s

2 a

+

1−s 2

x

+

1 +s 2

b+

1−s 2

(a+b−x)

=

1 +t 2

a+

1−t 2

x

+

1 +t 2

b+

1−t 2

(a+b−x)

,

and

1 +s 2

a+

1−s 2

(a+b−x)

+

1 +s 2

b+

1−s 2

x

=

1 +t 2

a+

1−t 2

(a+b−x)

+

1 +t 2

b+

1−t 2

x

.

By the inequality(2.1)and the assumptions thatpis nonnegative and symmetric aboutx= ^a+b₂ , we have

f

1 +s 2

a+

1−s 2

x

p

x+a 2

(2.7)

+f

1 +s 2

b+

1−s 2

(a+b−x)

p

2a+b−x 2

+f

1 +s 2

a+

1−s 2

(a+b−x)

p

a+ 2b−x 2

+f

1 +s 2

b+

1−s 2

x

p

x+b 2

=

f

1 +s 2

a+

1−s 2

x

+ f

1 +s 2

a+

1−s 2

(a+b−x)

p

x+a 2

+

f

1 +s 2

b+

1−s 2

(a+b−x)

+ f

1 +s 2

b+

1−s 2

x

p

x+b 2

≤

f

1 +t 2

a+

1−t 2

x

+ f

1 +t 2

a+

1−t 2

(a+b−x)

p

x+a 2

+

f

1 +t 2

b+

1−t 2

(a+b−x)

+ f

1 +t 2

b+

1−t 2

x

p

x+b 2

=f

1 +t 2

a+

1−t 2

x

p

x+a 2

+f

1 +t 2

b+

1−t 2

(a+b−x)

p

2a+b−x 2

+f

1 +t 2

a+

1−t 2

(a+b−x)

p

a+ 2b−x 2

+f

1 +t 2

b+

1−t 2

x

p

x+b 2

. Integrating the inequality(2.7)overxon[a, b], we have

4Q(s)≤4Q(t)

HenceQis increasing on[0,1], and the inequality(1.9)holds for allt∈[0,1].

This completes the proof.

Remark 2.8. If we set p(x) ≡ 1 (x∈[a, b]) in Theorem 2.7, then Theorem 2.2 generalizes Theorem F.

Remark 2.9. FromC([a, b])$ W([a, b]), Theorem 2.5 and Theorem 2.7 generalize Theorem C.

REFERENCES

[1] J.L. BRENNERAND H. ALZER, Integral inequalities for concave functions with applications to special functions, Proc. Roy. Soc. Edinburgh A, 118 (1991), 173–192.

[2] S.S. DRAGOMIR, Two mappings in connection to Hadamard’s inequalities, J. Math. Anal. Appl., 167 (1992), 49–56.

[3] S.S. DRAGOMIR, Y.J. CHO AND S.S. KIM, Inequalities of Hadamard’s type for Lipschitzian mappings and their applications, J. Math. Anal. Appl., 245 (2000), 489–501.

[4] L. FEJÉR, Über die Fourierreihen, II, Math. Naturwiss. Anz Ungar. Akad. Wiss., 24 (1906), 369–

390. (Hungarian).

[5] J. HADAMARD, Étude sur les propriétés des fonctions entières en particulier d’une fonction con- sidérée par Riemann, J. Math. Pures Appl., 58 (1893), 171–215.

[6] K.C. LEEANDK.L. TSENG, On a weighted generalization of Hadamard’s inequality for G-convex functions, Tamsui-Oxford J. Math. Sci., 16(1) (2000), 91–104.

[7] M. MATI ´C AND J. PE ˇCARI ´C, On inequalities of Hadamard’s type for Lipschitzian mappings, Tamkang J. Math., 32(2) (2001), 127–130.

[8] C.E.M. PEARCEANDJ. PE ˇCARI ´C, On some inequalities of Brenner and Alzer for concave Func- tions, J. Math. Anal. Appl., 198 (1996), 282–288.

[9] A.W. ROBERTSANDD.E. VARBERG, Convex Functions, Academic Press, New York, 1973.

[10] K.L. TSENG, G.S. YANGANDS.S. DRAGOMIR, Hadamard inequalities for Wright-convex func- tions, Demonstratio Math., 37(3) (2004), 525–532.

[11] G.S. YANGANDM.C. HONG, A note on Hadamard’s inequality, Tamkang. J. Math., 28(1) (1997), 33–37.

[12] G.S. YANGANDK.L. TSENG, On certain integral inequalities related to Hermite-Hadamard in- equalities, J. Math. Anal. Appl., 239 (1999), 180–187.

[13] G.S. YANG AND K.L. TSENG, Inequalities of Hadamard’s Type for Lipschitzian mappings, J.

Math. Anal. Appl., 260 (2001), 230–238.

[14] G.S. YANG AND K.L. TSENG, On certain multiple integral inequalities related to Hermite- Hadamard inequality, Utilitas Mathematica, 62 (2002), 131–142.

[15] G.S. YANGANDK.L. TSENG, Inequalities of Hermite-Hadamard-Fejér type for convex functions and Lipschitizian functions, Taiwanese J. Math., 7(3) (2003), 433–440.

[16] G.S. YANG ANDC.S. WANG, Some refinements of Hadamard’s inequalities, Tamkang J. Math., 28(2) (1997), 87–92.