Open Access

Some companions of perturbed Ostrowski-type inequalities based on the quadratic kernel function with three sections and applications

Journal of Inequalities and Applications20132013:226

https://doi.org/10.1186/1029-242X-2013-226

Received: 1 August 2012

Accepted: 18 April 2013

Published: 6 May 2013

Abstract

In this paper, based on the quadratic kernel function with three sections, which was defined by Liu in 2009, we establish some companions of perturbed Ostrowski-type inequalities for the case when f L 1 [ a , b ] , f L 2 [ a , b ] and f L 2 [ a , b ] , respectively. The special cases of these results offer better estimation than the conventional trapezoidal formula and the midpoint formula. The results we get can apply to composite quadrature rules in numerical integration and probability density functions. The effectiveness of these applications is also illustrated through several specific examples due to better error estimates.

MSC:26D15, 41A55, 41A80, 65C50.

Keywords

perturbed Ostrowski type inequalitydifferentiable mappingcomposite quadrature ruleprobability density function

1 Introduction

In 1938, Ostrowski [1] established the following interesting integral inequality for differentiable mappings with bounded derivatives.

Theorem 1.1 Let f : [ a , b ] R be a differentiable mapping on ( a , b ) whose derivative is bounded on ( a , b ) and denote f = sup t ( a , b ) | f ( t ) | < . Then for all x [ a , b ] we have
| f ( x ) 1 b a a b f ( t ) d t | [ 1 4 + ( x a + b 2 ) 2 ( b a ) 2 ] ( b a ) f .
(1.1)

The constant 1 4 is sharp in the sense that it cannot be replaced by a smaller one.

In [2], Guessab and Schmeisser proved the following companion of Ostrowski’s inequality.

Theorem 1.2 Let f : [ a , b ] R be satisfying the Lipschitz condition, i.e., | f ( t ) f ( s ) | M | t s | . Then for all x [ a , a + b 2 ] we have
| f ( x ) + f ( a + b x ) 2 1 b a a b f ( t ) d t | [ 1 8 + 2 ( x 3 a + b 4 b a ) 2 ] ( b a ) M .
(1.2)

The constant 1 8 is sharp in the sense that it cannot be replaced by a smaller one. In (1.2), the point x = 3 a + b 4 gives the best estimator.

Motivated by [2], Dragomir [3] proved some companions of Ostrowski’s inequality for absolutely continuous functions. Recently, Alomari [4] studied the companion of Ostrowski inequality (1.2) for differentiable bounded mappings. In [5], Liu established some companions of an Ostrowski-type integral inequality for functions whose first derivatives are absolutely continuous and second derivatives belong to L p ( 1 p ) spaces.

Theorem 1.3 Let f : [ a , b ] R be such that f is absolutely continuous on [ a , b ] and f L [ a , b ] . Then for all x [ a , a + b 2 ] we have
| f ( x ) + f ( a + b x ) 2 ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 1 b a a b f ( t ) d t | [ 1 96 + 1 2 ( x 3 a + b 4 ) 2 ( b a ) 2 ] ( b a ) 2 f .
(1.3)

The constant 1 96 is sharp in the sense that it cannot be replaced by a smaller one.

For other related results, the reader may refer to [626] and the references therein.

The main aim of this paper is to establish some companions of perturbed Ostrowski-type inequalities for the case when f L 1 [ a , b ] , f L 2 [ a , b ] and f L 2 [ a , b ] , respectively. For our purpose, we will use the quadratic kernel function with three sections (see (2.1) below) which was defined by Liu in [5]. The special cases of the results we get offer better estimation than the conventional trapezoidal formula and the midpoint formula. These results can apply to composite quadrature rules in numerical integration and probability density functions. The effectiveness of these applications is also illustrated through several specific examples due to better error estimates.

2 Main results

To prove our main results, we need the following lemmas.

Lemma 2.1 [5]

Let f : [ a , b ] R be such that f is absolutely continuous on [ a , b ] . Denote by K ( x , t ) : [ a , b ] R the kernel given by
K ( x , t ) = { 1 2 ( t a ) 2 , t [ a , x ] , 1 2 ( t a + b 2 ) 2 , t ( x , a + b x ] , 1 2 ( t b ) 2 , t ( a + b x , b ] ,
(2.1)
then the identity
1 b a a b K ( x , t ) f ( t ) d t = 1 b a a b f ( t ) d t f ( x ) + f ( a + b x ) 2 + ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2
(2.2)

holds.

2.1 The case when f L 1 [ a , b ] and is bounded

Theorem 2.1 Let f : [ a , b ] R be such that f is absolutely continuous on [ a , b ] . If f L 1 [ a , b ] and γ f ( t ) Γ , t [ a , b ] , then for all x [ a , a + b 2 ] we have
| f ( x ) + f ( a + b x ) 2 ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | ( S γ ) [ ( b a ) 2 48 + b a 4 | x 3 a + b 4 | ]
(2.3)
and
| f ( x ) + f ( a + b x ) 2 ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | ( Γ S ) [ ( b a ) 2 48 + b a 4 | x 3 a + b 4 | ] ,
(2.4)

where S = ( f ( b ) f ( a ) ) / ( b a ) .

Proof From (2.2) and the facts
1 b a a b f ( t ) d t = f ( b ) f ( a ) b a
(2.5)
and
1 b a a b K ( x , t ) d t = 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ,
(2.6)
it follows that
1 b a a b K ( x , t ) f ( t ) d t 1 ( b a ) 2 a b K ( x , t ) d t a b f ( t ) d t = 1 b a a b f ( t ) d t f ( x ) + f ( a + b x ) 2 + ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] .
(2.7)
We denote
R n ( x ) = 1 b a a b K ( x , t ) f ( t ) d t 1 ( b a ) 2 a b K ( x , t ) d t a b f ( t ) d t .
(2.8)
If C R is an arbitrary constant, then we have
R n ( x ) = 1 b a a b ( f ( t ) C ) [ K ( x , t ) 1 b a a b K ( x , s ) d s ] d t .
(2.9)
Furthermore, we have
| R n ( x ) | 1 b a max t [ a , b ] | K ( x , t ) 1 b a a b K ( x , s ) d s | a b | f ( t ) C | d t .
(2.10)
To compute
max t [ a , b ] | K ( x , t ) 1 b a a b K ( x , s ) d s | = max { | 1 2 ( x a ) 2 [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] | , | 1 2 ( a + b 2 x ) 2 [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] | , 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 } = max { b a 24 | 6 x 5 a b | , b a 12 | 3 x 2 a b | , 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 } ,
(2.11)
we denote
y 1 = b a 24 | 6 x 5 a b | , y 2 = b a 12 | 3 x 2 a b | , y 3 = 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 .
If we choose y 1 = 0 , then we get x 1 = 5 a + b 6 . If we choose y 2 = 0 , then we get x 2 = 2 a + b 3 . A direct computation gives that
{ y 2 max { y 1 , y 3 } , x [ a , 3 a + b 4 ] , y 1 > max { y 2 , y 3 } , x ( 3 a + b 4 , a + b 2 ] .
(2.12)
Therefore, we get
max t [ a , b ] | K ( x , t ) 1 b a a b K ( x , s ) d s | = max { y 1 , y 2 } = ( b a ) 2 48 + b a 4 | x 3 a + b 4 | .
(2.13)
We also have
a b | f ( t ) γ | d t = ( S γ ) ( b a )
(2.14)
and
a b | f ( t ) Γ | d t = ( Γ S ) ( b a ) .
(2.15)

Therefore, we obtain (2.3) and (2.4) by using (2.7)-(2.10), (2.1)-(2.15) and choosing C = γ and C = Γ in (2.10), respectively. □

Corollary 2.1 Under the assumptions of Theorem  2.1, choose
  1. (1)
    x = 3 a + b 4 , we have
    | f ( 3 a + b 4 ) + f ( a + 3 b 4 ) 2 + f ( b ) f ( a ) b a ( b a ) 2 96 1 b a a b f ( t ) d t | 1 48 ( S γ ) ( b a ) 2 ,
    (2.16)
    | f ( 3 a + b 4 ) + f ( a + 3 b 4 ) 2 + f ( b ) f ( a ) b a ( b a ) 2 96 1 b a a b f ( t ) d t | 1 48 ( Γ S ) ( b a ) 2 .
    (2.17)
     
  2. (2)
    x = a , we have
    | f ( a ) + f ( b ) 2 f ( b ) f ( a ) b a ( b a ) 2 12 1 b a a b f ( t ) d t | 1 12 ( S γ ) ( b a ) 2 , | f ( a ) + f ( b ) 2 f ( b ) f ( a ) b a ( b a ) 2 12 1 b a a b f ( t ) d t | 1 12 ( Γ S ) ( b a ) 2 .
     
  3. (3)
    x = a + b 2 , we have
    | f ( a + b 2 ) + f ( b ) f ( a ) b a ( b a ) 2 24 1 b a a b f ( t ) d t | 1 12 ( S γ ) ( b a ) 2 , | f ( a + b 2 ) + f ( b ) f ( a ) b a ( b a ) 2 24 1 b a a b f ( t ) d t | 1 12 ( Γ S ) ( b a ) 3 .
     
Corollary 2.2 Let f be as in Theorem  2.1. Additionally, if f is symmetric about x = a + b 2 , then for all x [ a , a + b 2 ] we have
| f ( x ) ( x 3 a + b 4 ) f ( x ) + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | ( S γ ) [ ( b a ) 2 48 + b a 4 | x 3 a + b 4 | ]
and
| f ( x ) ( x 3 a + b 4 ) f ( x ) + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | ( Γ S ) [ ( b a ) 2 48 + b a 4 | x 3 a + b 4 | ] .

2.2 The case when f L 2 [ a , b ]

Theorem 2.2 Let f : [ a , b ] R be a thrice continuously differentiable mapping in ( a , b ) with f L 2 [ a , b ] . Then for all x [ a , a + b 2 ] we have
| f ( x ) + f ( a + b x ) 2 ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | 1 π f 2 { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2 .
(2.18)
Proof Let R n ( x ) be defined by (2.8). From (2.7), we get
R n ( x ) = 1 b a a b f ( t ) d t f ( x ) + f ( a + b x ) 2 + ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] .
(2.19)
If we choose C = f ( ( a + b ) / 2 ) in (2.9) and use the Cauchy inequality, then we get
| R n ( x ) | 1 b a a b | f ( t ) f ( a + b 2 ) | | K ( x , t ) 1 b a a b K ( x , s ) d s | d t 1 b a [ a b ( f ( t ) f ( a + b 2 ) ) 2 d t ] 1 / 2 × [ a b ( K ( x , t ) 1 b a a b K ( x , s ) d s ) 2 d t ] 1 / 2 .
(2.20)
We can use the Diaz-Metcalf inequality (see [[19], p.83] or [[25], p.424]) to get
a b ( f ( t ) f ( a + b 2 ) ) 2 d t ( b a ) 2 π 2 f 2 2 .
We also have
a b ( K ( x , t ) 1 b a a b K ( x , s ) d s ) 2 d t = a b K ( x , t ) 2 d t ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 = 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 .
(2.21)

Therefore, using the above relations, we obtain (2.18). □

Corollary 2.3 Under the assumptions of Theorem  2.2, choose
  1. (1)
    x = 3 a + b 4 , we have
    | f ( 3 a + b 4 ) + f ( a + 3 b 4 ) 2 + f ( b ) f ( a ) b a ( b a ) 2 96 1 b a a b f ( t ) d t | ( b a ) 5 / 2 48 π 5 f 2 .
    (2.22)
     
  2. (2)
    x = a , we have
    | f ( a ) + f ( b ) 2 f ( b ) f ( a ) b a ( b a ) 2 12 1 b a a b f ( t ) d t | ( b a ) 5 / 2 12 π 5 f 2 .
     
  3. (3)
    x = a + b 2 , we have
    | f ( a + b 2 ) + f ( b ) f ( a ) b a ( b a ) 2 24 1 b a a b f ( t ) d t | ( b a ) 5 / 2 12 π 5 f 2 .
     
Corollary 2.4 Let f be as in Theorem  2.1. Additionally, if f is symmetric about x = a + b 2 , i.e., f ( a + b x ) = f ( x ) , then for all x [ a , a + b 2 ] we have
| f ( x ) ( x 3 a + b 4 ) f ( x ) + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | 1 π f 2 { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2 .

2.3 The case when f L 2 [ a , b ]

Theorem 2.3 Let f : [ a , b ] R be such that f is absolutely continuous on [ a , b ] with f L 2 [ a , b ] . Then for all x [ a , a + b 2 ] we have
| f ( x ) + f ( a + b x ) 2 ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | σ ( f ) b a { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2 ,
(2.23)
where σ ( f ) is defined by
σ ( f ) = f 2 2 ( f ( b ) f ( a ) ) 2 b a = f 2 2 S 2 ( b a )
(2.24)

and S is defined in Theorem  2.1.

Proof Let R n ( x ) be defined by (2.8). If we choose C = 1 b a a b f ( s ) d s in (2.9) and use the Cauchy inequality and (2.21), then we get
| R n ( x ) | 1 b a a b | f ( t ) 1 b a a b f ( s ) d s | | K ( x , t ) 1 b a a b K ( x , s ) d s | d t 1 b a [ a b ( f ( t ) 1 b a a b f ( s ) d s ) 2 d t ] 1 / 2 × [ a b ( K ( x , t ) 1 b a a b K ( x , s ) d s ) 2 d t ] 1 / 2 σ ( f ) b a { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2 .

 □

Corollary 2.5 Under the assumptions of Theorem  2.3, choose
  1. (1)
    x = 3 a + b 4 , we have
    | f ( 3 a + b 4 ) + f ( a + 3 b 4 ) 2 + f ( b ) f ( a ) b a ( b a ) 2 96 1 b a a b f ( t ) d t | ( b a ) 3 / 2 48 5 σ ( f ) .
    (2.25)
     
  2. (2)
    x = a , we have
    | f ( a ) + f ( b ) 2 f ( b ) f ( a ) b a ( b a ) 2 12 1 b a a b f ( t ) d t | ( b a ) 3 / 2 12 5 σ ( f ) .
     
  3. (3)
    x = a + b 2 , we have
    | f ( a + b 2 ) + f ( b ) f ( a ) b a ( b a ) 2 24 1 b a a b f ( t ) d t | ( b a ) 3 / 2 12 5 σ ( f ) .
     
Corollary 2.6 Let f be as in Theorem  2.1. Additionally, if f is symmetric about x = a + b 2 , then for all x [ a , a + b 2 ] we have
| f ( x ) ( x 3 a + b 4 ) f ( x ) + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 1 b a a b f ( t ) d t | σ ( f ) b a { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2 .

3 Application to composite quadrature rules

Let I n : a = x 0 < x 1 < < x n 1 < x n = b be a partition of the interval [ a , b ] and h i = x i + 1 x i ( i = 0 , 1 , 2 , , n 1 ).

Consider the perturbed composite quadrature rules
Q n 1 ( I n , f ) = 1 2 i = 0 n 1 [ f ( 3 x i + x i + 1 4 ) + f ( x i + 3 x i + 1 4 ) ] h i + i = 0 n 1 f ( x i + 1 ) f ( x i ) 96 h i 2
(3.1)
and
Q n 2 ( I n , f ) = 1 2 i = 0 n 1 [ f ( 3 x i + x i + 1 4 ) + f ( x i + 3 x i + 1 4 ) ] h i + Γ + γ 192 i = 0 n 1 h i 3 .
(3.2)

The following results hold.

Theorem 3.1 Let f : [ a , b ] R be such that f is absolutely continuous on [ a , b ] . If f L 1 [ a , b ] and γ f ( t ) Γ , t [ a , b ] , then for all x [ a , a + b 2 ] we have
a b f ( t ) d t = Q n 1 ( I n , f ) + R n 1 ( I n , f ) ,
where Q n 1 ( I n , f ) is defined by formula (3.1), and the remainder R n 1 ( I n , f ) satisfies the estimate
| R n 1 ( I n , f ) | 1 48 ( S γ ) i = 0 n 1 h i 3
(3.3)
and
| R n 1 ( I n , f ) | 1 48 ( Γ S ) i = 0 n 1 h i 3 .
(3.4)
Proof Applying inequality (2.1) and (2.1) to the intervals [ x i , x i + 1 ] , we get
| x i x i + 1 f ( t ) d t 1 2 [ f ( 3 x i + x i + 1 4 ) + f ( x i + 3 x i + 1 4 ) ] h i f ( x i + 1 ) f ( x i ) 96 h i 2 | 1 48 ( S γ ) h i 3
and
| x i x i + 1 f ( t ) d t 1 2 [ f ( 3 x i + x i + 1 4 ) + f ( x i + 3 x i + 1 4 ) ] h i f ( x i + 1 ) f ( x i ) 96 h i 2 | 1 48 ( Γ S ) h i 3

for i = 0 , 1 , 2 , , n 1 . Now summing over i from 0 to n 1 and using the triangle inequality, we get (3.3) and (3.4). □

Theorem 3.2 Let f : [ a , b ] R be a thrice continuously differentiable mapping in ( a , b ) with f L 2 [ a , b ] . Then for all x [ a , a + b 2 ] we have
a b f ( t ) d t = Q n 1 ( I n , f ) + R n 1 ( I n , f ) ,
where Q n 1 ( I n , f ) is defined by formula (3.1), and the remainder R n 1 ( I n , f ) satisfies the estimate
| R n 1 ( I n , f ) | f 2 48 π 5 i = 0 n 1 h i 7 / 2 .
(3.5)
Proof Applying inequality (2.3) to the intervals [ x i , x i + 1 ] , we get
| x i x i + 1 f ( t ) d t 1 2 [ f ( 3 x i + x i + 1 4 ) + f ( x i + 3 x i + 1 4 ) ] h i f ( x i + 1 ) f ( x i ) 96 h i 2 | h i 7 / 2 48 π 5 f 2

for i = 0 , 1 , 2 , , n 1 . Now summing over i from 0 to n 1 and using the triangle inequality, we get (3.5). □

Theorem 3.3 Let f : [ a , b ] R be such that f is absolutely continuous on [ a , b ] with f L 2 [ a , b ] . Then for all x [ a , a + b 2 ] we have
a b f ( t ) d t = Q n 1 ( I n , f ) + R n 1 ( I n , f ) ,
where Q n 1 ( I n , f ) is defined by formula (3.1), and the remainder R n 1 ( I n , f ) satisfies the estimate
| R n 1 ( I n , f ) | σ ( f ) 48 5 i = 0 n 1 h i 5 / 2 .
(3.6)
Proof Applying inequality (2.5) to the intervals [ x i , x i + 1 ] , we get
| x i x i + 1 f ( t ) d t 1 2 [ f ( 3 x i + x i + 1 4 ) + f ( x i + 3 x i + 1 4 ) ] h i f ( x i + 1 ) f ( x i ) 96 h i 2 | h i 5 / 2 48 5 σ ( f )

for i = 0 , 1 , 2 , , n 1 . Now summing over i from 0 to n 1 and using the triangle inequality, we get (3.6). □

To illustrate the effectiveness of the perturbed composite quadrature rules (3.1) and (3.2), we compute the approximate values of several specific examples using these two rules and the composite trapezoidal formula
T n ( I n , f ) = h 2 [ f ( a ) + 2 i = 1 n 1 f ( x i ) + f ( b ) ] ,
respectively, and then we compare their errors. We get Table 1, from which the power of these two rules in numerical integration is demonstrated due to better error estimates.
Table 1

Numerical results

f(x)

n

[a,b]

a b f ( x ) d x

T n

Error of T n

Q n 1 and Q n 2

Error of Q n 1 and Q n 2

cosx − x

20

[ 0 , π 2 ]

−0.233701

−0.234215

5.14E-4

−0.233636

6.5E-5

e 2 x cos ( e x )

20

[0,1]

−1.176887

−1.181466

4.579E-3

−1.176316

5.71E-4

1 x 4 + 4 x 2 + 3

10

[0,1]

0.241549

0.241393

1.56E-4

0.241569

2E-5

tanx + x

20

[ 0 , π 4 ]

0.654999

0.655127

1.28E-4

0.654983

7E-6

ln ( x 2 + 1 )

20

[−1,1]

0.527887

0.529554

1.667E-3

0.527679

2.08E-4

4 Application to probability density functions

Now, let X be a random variable taking values in the finite interval [ a , b ] , with the probability density function f : [ a , b ] [ 0 , 1 ] and with the cumulative distribution function
F ( x ) = Pr ( X x ) = a x f ( t ) d t .

The following results hold.

Theorem 4.1 With the assumptions of Theorem  2.1, we have
| 1 2 [ F ( x ) + F ( a + b x ) ] ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] b E ( X ) b a | ( S γ ) [ ( b a ) 2 48 + b a 4 | x 3 a + b 4 | ]
(4.1)
and
| 1 2 [ F ( x ) + F ( a + b x ) ] ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] b E ( X ) b a | ( Γ S ) [ ( b a ) 2 48 + b a 4 | x 3 a + b 4 | ]
(4.2)

for all x [ a , a + b 2 ] , where E ( X ) is the expectation of X.

Proof By (2.3) and (2.4) on choosing f = F and taking into account
E ( X ) = a b t d F ( t ) = b a b F ( t ) d t ,

we obtain (4.1) and (4.2). □

Corollary 4.1 Under the assumptions of Theorem  4.1 with x = 3 a + b 4 , we have
| 1 2 [ F ( 3 a + b 4 ) + F ( a + 3 b 4 ) ] + b a 96 [ f ( b ) f ( a ) ] b E ( x ) b a | 1 48 ( S γ ) ( b a ) 2
(4.3)
and
| 1 2 [ F ( 3 a + b 4 ) + F ( a + 3 b 4 ) ] + b a 96 [ f ( b ) f ( a ) ] b E ( x ) b a | 1 48 ( Γ S ) ( b a ) 2 .
(4.4)
Theorem 4.2 With the assumptions of Theorem  2.2, we have
| 1 2 [ F ( x ) + F ( a + b x ) ] ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] b E ( X ) b a | 1 π f 2 { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2
(4.5)

for all x [ a , a + b 2 ] , where E ( X ) is the expectation of X.

Proof By (2.18) on choosing f = F and taking into account
E ( X ) = a b t d F ( t ) = b a b F ( t ) d t ,

we obtain (4.5). □

Corollary 4.2 Under the assumptions of Theorem  4.2 with x = 3 a + b 4 , we have
| 1 2 [ F ( 3 a + b 4 ) + F ( a + 3 b 4 ) ] + b a 96 [ f ( b ) f ( a ) ] b E ( x ) b a | ( b a ) 5 / 2 48 π 5 f 2 .
(4.6)
Theorem 4.3 With the assumptions of Theorem  2.3, we have
| 1 2 [ F ( x ) + F ( a + b x ) ] ( x 3 a + b 4 ) f ( x ) f ( a + b x ) 2 + f ( b ) f ( a ) b a [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] b E ( X ) b a | σ ( f ) b a { 1 320 ( a + b 2 x ) 5 + 1 10 ( x a ) 5 ( b a ) [ 1 2 ( x 3 a + b 4 ) 2 + ( b a ) 2 96 ] 2 } 1 / 2
(4.7)

for all x [ a , a + b 2 ] , where E ( X ) is the expectation of X.

Proof By (2.23) on choosing f = F and taking into account
E ( X ) = a b t d F ( t ) = b a b F ( t ) d t ,

we obtain (4.7). □

Corollary 4.3 Under the assumptions of Theorem  4.3 with x = 3 a + b 4 , we have
| 1 2 [ F ( 3 a + b 4 ) + F ( a + 3 b 4 ) ] + b a 96 [ f ( b ) f ( a ) ] b E ( x ) b a | ( b a ) 3 / 2 48 5 σ ( f ) .
(4.8)

Declarations

Acknowledgements

The authors wish to thank the anonymous referees and the editor for their valuable comments. This work was partly supported by the Qing Lan Project of Jiangsu Province, the National Natural Science Foundation of China (Grant No. 41174165) and the Teaching Research Project of NUIST (Grant No. 12JY052).

Authors’ Affiliations

(1)
College of Mathematics and Statistics, Nanjing University of Information Science and Technology
(2)
Department of Mathematics, Hanseo University

References

  1. Ostrowski A: Über die Absolutabweichung einer differentiierbaren Funktion von ihrem Integralmittelwert. Comment. Math. Helv. 1937, 10(1):226–227. 10.1007/BF01214290MathSciNetView ArticleGoogle Scholar
  2. Guessab A, Schmeisser G: Sharp integral inequalities of the Hermite-Hadamard type. J. Approx. Theory 2002, 115(2):260–288. 10.1006/jath.2001.3658MathSciNetView ArticleGoogle Scholar
  3. Dragomir SS: Some companions of Ostrowski’s inequality for absolutely continuous functions and applications. Bull. Korean Math. Soc. 2005, 42(2):213–230.MathSciNetView ArticleGoogle Scholar
  4. Alomari MW: A companion of Ostrowski’s inequality with applications. Transylv. J. Math. Mech. 2011, 3(1):9–14.MathSciNetGoogle Scholar
  5. Liu Z: Some companions of an Ostrowski type inequality and applications. JIPAM. J. Inequal. Pure Appl. Math. 2009., 10(2): Article ID 52Google Scholar
  6. Alomari MW: A companion of Ostrowski’s inequality for mappings whose first derivatives are bounded and applications in numerical integration. Transylv. J. Math. Mech. 2012, 4(2):103–109.MathSciNetGoogle Scholar
  7. Alomari MW: A generalization of companion inequality of Ostrowski’s type for mappings whose first derivatives are bounded and applications in numerical integration. Kragujev. J. Math. 2012, 36(1):77–82.MathSciNetGoogle Scholar
  8. Barnett NS, Dragomir SS, Gomm I: A companion for the Ostrowski and the generalised trapezoid inequalities. Math. Comput. Model. 2009, 50(1–2):179–187. 10.1016/j.mcm.2009.04.005MathSciNetView ArticleGoogle Scholar
  9. Dragomir SS: A companions of Ostrowski’s inequality for functions of bounded variation and applications. RGMIA Res. Rep. Coll. 2002., 5: Article ID 28Google Scholar
  10. Dragomir SS: Ostrowski type inequalities for functions defined on linear spaces and applications for semi-inner products. J. Concr. Appl. Math. 2005, 3(1):91–103.MathSciNetGoogle Scholar
  11. Dragomir SS: Ostrowski’s type inequalities for continuous functions of selfadjoint operators on Hilbert spaces: a survey of recent results. Ann. Funct. Anal. 2011, 2(1):139–205.MathSciNetView ArticleGoogle Scholar
  12. Dragomir SS, Sofo A: An integral inequality for twice differentiable mappings and applications. Tamkang J. Math. 2000, 31(4):257–266.MathSciNetGoogle Scholar
  13. Duoandikoetxea J: A unified approach to several inequalities involving functions and derivatives. Czechoslov. Math. J. 2001, 51(126)(2):363–376.MathSciNetView ArticleGoogle Scholar
  14. Huy VN, Ngô Q-A: New bounds for the Ostrowski-like type inequalities. Bull. Korean Math. Soc. 2011, 48(1):95–104.MathSciNetView ArticleGoogle Scholar
  15. Liu WJ: Several error inequalities for a quadrature formula with a parameter and applications. Comput. Math. Appl. 2008, 56(7):1766–1772. 10.1016/j.camwa.2008.04.016MathSciNetView ArticleGoogle Scholar
  16. Liu WJ: Some weighted integral inequalities with a parameter and applications. Acta Appl. Math. 2010, 109(2):389–400. 10.1007/s10440-008-9323-2MathSciNetView ArticleGoogle Scholar
  17. Liu WJ, Xue QL, Wang SF: New generalization of perturbed Ostrowski type inequalities and applications. J. Appl. Math. Comput. 2010, 32(1):157–169. 10.1007/s12190-009-0240-yMathSciNetView ArticleGoogle Scholar
  18. Liu Z: Note on a paper by N. Ujević. Appl. Math. Lett. 2007, 20(6):659–663. 10.1016/j.aml.2006.09.001MathSciNetView ArticleGoogle Scholar
  19. Mitrinović DS, Pečarić JE, Fink AM: Inequalities involving functions and their integrals and derivatives. 53. In Mathematics and Its Applications (East European Series). Kluwer Academic, Dordrecht; 1991.Google Scholar
  20. Sarikaya MZ: On the Ostrowski type integral inequality. Acta Math. Univ. Comen. 2010, 79(1):129–134.MathSciNetGoogle Scholar
  21. Sarikaya MZ: New weighted Ostrowski and čebyšev type inequalities on time scales. Comput. Math. Appl. 2010, 60(5):1510–1514. 10.1016/j.camwa.2010.06.033MathSciNetView ArticleGoogle Scholar
  22. Set E, Sarıkaya MZ: On the generalization of Ostrowski and grüss type discrete inequalities. Comput. Math. Appl. 2011, 62(1):455–461. 10.1016/j.camwa.2011.05.026MathSciNetView ArticleGoogle Scholar
  23. Tseng K-L, Hwang S-R, Dragomir SS: Generalizations of weighted Ostrowski type inequalities for mappings of bounded variation and their applications. Comput. Math. Appl. 2008, 55(8):1785–1793. 10.1016/j.camwa.2007.07.004MathSciNetView ArticleGoogle Scholar
  24. Tseng K-L, Hwang S-R, Yang G-S, Chou Y-M: Weighted Ostrowski integral inequality for mappings of bounded variation. Taiwan. J. Math. 2011, 15(2):573–585.MathSciNetGoogle Scholar
  25. Ujević N: New bounds for the first inequality of Ostrowski-grüss type and applications. Comput. Math. Appl. 2003, 46(2–3):421–427. 10.1016/S0898-1221(03)90035-6MathSciNetView ArticleGoogle Scholar
  26. Vong SW: A note on some Ostrowski-like type inequalities. Comput. Math. Appl. 2011, 62(1):532–535. 10.1016/j.camwa.2011.05.037MathSciNetView ArticleGoogle Scholar

Copyright

© Liu et al.; licensee Springer. 2013

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.