Skip to main content

New Ostrowski-Grüss type inequalities with the derivatives bounded by functions

Abstract

In this paper, we establish some new Ostrowski-Grüss type inequalities involving multiple interior points with the first-order derivative bounded by functions instead of constants, some of which provide sharp bounds. Then we establish a new 2D Ostrowski-Grüss type inequality involving multiple interior points with the second mixed partial derivative bounded by functions. For illustrating the applications of the Ostrowski-Grüss type inequalities established, we apply them to derive error bounds for some numerical integration formulae.

MSC:26D10, 26D20.

1 Introduction

As is known, the Ostrowski-type inequality [1] can be used to estimate the absolute deviation of a function from its integral mean, while the Grüss inequality [2] can be used to estimate the absolute deviation of the integral of the product of two functions from the product of their respective integral. Recently, various generalizations of the Ostrowski inequality and the Grüss inequality have been established (for example, see [321] and the references therein). These inequalities can be used to provide explicit error bounds for numerical quadrature formulae such as the Simpson quadrature formula, trapezoid quadrature formula and so on. Among the generalizations, many Ostrowski-Grüss type inequalities have been established [1321]. Now we list some important results in the literature.

In [13], Dragomir et al. presented an Ostrowski-Grüss type inequality for the first time as follows:

| f ( x ) 1 b a a b f ( t ) d t f ( b ) f ( a ) b a ( x a + b 2 ) | 1 4 (ba)(Γγ)

for all x[a,b], where γ, Γ are two constants such that γ f (t)Γ, t[a,b].

In [14], under the same conditions as above, Matić et al. improved the above result to the following form:

| f ( x ) 1 b a a b f ( t ) d t f ( b ) f ( a ) b a ( x a + b 2 ) | 1 4 3 (ba)(Γγ).

Then in [15], Cheng presented a sharp inequality as follows:

| f ( x ) 1 b a a b f ( t ) d t f ( b ) f ( a ) b a ( x a + b 2 ) | 1 8 (ba)(Γγ).
(1)

This inequality is sharp in the sense that the constant 1 8 cannot be replaced by a smaller one.

In [16], Feng et al. further generalized the inequalities above, and presented a sharp Ostrowski-Grüss type inequality involving multiple interior points as follows:

| 1 b a i = 0 k ( m i + 1 m i ) f ( x i ) 1 b a a b f ( t ) d t f ( b ) f ( a ) ( b a ) 2 [ b 2 a 2 2 i = 0 k 1 m i + 1 ( x i + 1 x i ) ] | 1 8 ( b a ) ( Γ γ ) ,
(2)

where x i [a,b], i=1,,k1, are interior points, m i [ x i 1 , x i ], i=1,2,,k, x 0 =a, x k =b, m 0 =a, m k + 1 =b, and γ, Γ are defined as before.

We notice that little attention is paid to the Ostrowski-Grüss type inequalities involving multiple interior points with f (t) bounded by functions instead of constants so far in the literature. So, in this paper, motivated by the above works, we extend the Ostrowski-Grüss type inequalities to the case involving multiple interior points with the bounds of f (t) shown as γ(t) f (t)Γ(t), t[a,b]. Some bounds will be derived based on the inequalities, and some of the bounds are sharp. A new 2D Ostrowski-Grüss type inequality will also be derived with the bounds of 2 f ( x , y ) x y shown as γ(x,y) 2 f ( x , y ) x y Γ(x,y), x[a,b], y[c,d]. We also present some applications for the Ostrowski-Grüss type inequalities established, in which new error bounds for some numerical integration formulae are derived.

2 Main results

Lemma 2.1 [[16], Lemma 2.1]

Let IR be an open interval, a,bI, a<b. f:IR is a differential function. Furthermore, suppose that t i [a,b], i=0,1,,k, I k :a= t 0 < t 1 << t k 1 < t k =b is a division of the interval [a,b], and δ i [ t i 1 , t i ], i=1,2,,k, δ 0 =a, δ k + 1 =b. Then we have

i = 0 k ( δ i + 1 δ i )f( t i )= a b f(t)dt+ a b ζ(t, I k ) f (t)dt,
(3)

where

ζ(t, I k )={ t δ 1 , t [ t 0 , t 1 ) , t δ 2 , t [ t 1 , t 2 ) , t δ k 1 , t [ t k 2 , t k 1 ) , t δ k , t [ t k 1 , t k ] .
(4)

Lemma 2.2 Let

φ(t, I k )={ t δ 1 C , t [ t 0 , t 1 ) , t δ 2 C , t [ t 1 , t 2 ) , t δ k 1 C , t [ t k 2 , t k 1 ) , t δ k C , t [ t k 1 , t k ] ,
(5)

where C= 1 2 (b+a) 1 b a i = 0 k 1 δ i + 1 ( t i + 1 t i ). Then a b φ(t, I k )dt=0.

Proof Based on the definition of φ(t, I k ), we have

a b φ ( t , I k ) d t = i = 0 k 1 [ ( t i + 1 δ i + 1 C ) 2 2 ( t i δ i + 1 C ) 2 2 ] = 1 2 i = 0 k 1 [ ( t i + 1 t i ) ( t i + 1 + t i 2 δ i + 1 2 C ) ] = 1 2 { i = 0 k 1 ( t i + 1 2 t i 2 ) 2 i = 0 k 1 ( t i + 1 t i ) δ i + 1 2 C i = 0 k 1 ( t i + 1 t i ) } = 1 2 ( b 2 a 2 ) i = 0 k 1 ( t i + 1 t i ) δ i + 1 C ( b a ) = 0 .
(6)

 □

Theorem 2.1 Under the conditions of Lemma  2.1, if there exist two functions γ(t), Γ(t) with γ(t) f (t)Γ(t), t[a,b], then the following inequality holds:

| i = 0 k ( δ i + 1 δ i ) f ( t i ) a b f ( t ) d t a b ζ ( t , I k ) [ γ ( t ) + Γ ( t ) 2 ] d t | 1 2 Γ γ { 1 2 i = 0 k 1 [ t i + 1 2 + t i 2 ] i = 0 k 1 δ i + 1 ( t i + 1 + t i ) + i = 0 k 1 δ i + 1 2 } .
(7)

Proof Consider

a b ζ ( t , I k ) [ f ( t ) γ ( t ) + Γ ( t ) 2 ] d t = a b ζ ( t , I k ) f ( t ) d t a b ζ ( t , I k ) [ γ ( t ) + Γ ( t ) 2 ] d t ,
(8)
a b | ζ ( t , I k ) | d t = i = 0 k 1 t i t i + 1 | t δ i + 1 | d t a b | ζ ( t , I k ) | d t = i = 0 k 1 [ ( t i + 1 δ i + 1 ) 2 2 + ( t i δ i + 1 ) 2 2 ] a b | ζ ( t , I k ) | d t = 1 2 i = 0 k 1 [ t i + 1 2 + t i 2 ] i = 0 k 1 δ i + 1 ( t i + 1 + t i ) + i = 0 k 1 δ i + 1 2
(9)

and

| a b ζ ( t , I k ) [ f ( t ) γ ( t ) + Γ ( t ) 2 ] d t | a b | ζ ( t , I k ) | | [ f ( t ) γ ( t ) + Γ ( t ) 2 ] | d t 1 2 Γ γ a b | ζ ( t , I k ) | d t .
(10)

Combining (8)-(10) and Lemma 2.1, we obtain the desired result. □

Corollary 2.1 Under the conditions of Lemma  2.1, if W 1 f(t) W 2 , t[a,b], where W 1 , W 2 are two constants, then we have the following inequality:

| i = 0 k ( δ i + 1 δ i ) f ( t i ) a b f ( t ) d t ( W 2 + W 1 2 ) { 1 2 ( b 2 a 2 ) i = 0 k 1 δ i + 1 ( t i + 1 t i ) } | ( W 2 W 1 2 ) { 1 2 i = 0 k 1 [ t i + 1 2 + t i 2 ] i = 0 k 1 δ i + 1 ( t i + 1 + t i ) + i = 0 k 1 δ i + 1 2 } .
(11)

Proof In fact, in (7) we have γ(t)= W 1 , Γ(t)= W 2 . On the other hand,

a b ζ ( t , I k ) d t = i = 0 k 1 t i t i + 1 ( t δ i + 1 ) d t = i = 0 k 1 [ ( t i + 1 δ i + 1 ) 2 2 ( t i δ i + 1 ) 2 2 ] = 1 2 i = 0 k 1 [ t i + 1 2 t i 2 ] i = 0 k 1 δ i + 1 ( t i + 1 t i ) = 1 2 ( b 2 a 2 ) i = 0 k 1 δ i + 1 ( t i + 1 t i ) .
(12)

Then, combining (7) and (12), we get the desired inequality (11). □

Next we present a sharp Ostrowski-Grüss type inequality containing multiple interior points with f (t) bounded by functions as follows.

Theorem 2.2 Under the conditions of Lemma  2.1, if there exist two functions γ(t), Γ(t) with γ(t) f (t)Γ(t), t[a,b], then we have the following inequality:

| i = 0 k ( δ i + 1 δ i ) f ( t i ) a b f ( t ) d t f ( b ) f ( a ) b a [ b 2 a 2 2 i = 0 k 1 δ i + 1 ( t i + 1 t i ) ] a b φ ( t , I k ) [ Γ ( t ) + γ ( t ) 2 ] d t | 1 2 Γ γ a b | φ ( t , I k ) | d t ,
(13)

where φ(t, I k ) is defined as in Lemma  2.2.

The inequality (13) is sharp in the sense that the constant 1 2 on the right-hand side cannot be replaced by a smaller one.

Proof We have the following observations:

a b φ ( t , I k ) [ f ( t ) Γ ( t ) + γ ( t ) 2 ] d t = a b φ ( t , I k ) f ( t ) d t a b φ ( t , I k ) [ Γ ( t ) + γ ( t ) 2 ] d t = i = 0 k 1 t i t i + 1 ( t δ i + 1 C ) f ( t ) d t a b φ ( t , I k ) [ Γ ( t ) + γ ( t ) 2 ] d t = i = 0 k 1 [ ( t i + 1 δ i + 1 C ) f ( t i + 1 ) ( t i δ i + 1 C ) f ( t i ) ] i = 0 k 1 t i t i + 1 f ( t ) d t a b φ ( t , I k ) [ Γ ( t ) + γ ( t ) 2 ] d t = i = 0 k ( δ i + 1 δ i ) f ( t i ) a b f ( t ) d t f ( b ) f ( a ) b a [ b 2 a 2 2 i = 0 k 1 δ i + 1 ( t i + 1 t i ) ] a b φ ( t , I k ) [ Γ ( t ) + γ ( t ) 2 ] d t ,
(14)

and

| a b φ ( t , I k ) [ f ( t ) Γ ( t ) + γ ( t ) 2 ] d t | 1 2 Γ γ a b | φ ( t , I k ) | dt.
(15)

Combining (14) and (15), we obtain the desired inequality (13).

To prove the sharpness of (13), we take k=2, δ 0 =a, δ 1 = b + 5 a 6 , δ 2 = 5 b + a 6 , δ 3 =b, t 1 = b + a 2 , and

f(t)={ σ 1 t , a t < b + 5 a 6 , σ 2 t + b + 5 a 6 σ 1 b + 5 a 6 σ 2 , b + 5 a 6 t < b + a 2 , σ 1 t + b + a 2 σ 2 + b + 5 a 6 σ 1 b + 5 a 6 σ 2 b + a 2 σ 1 , b + a 2 t < 5 b + a 6 , σ 2 t + 5 b + a 6 σ 1 + b + a 2 σ 2 + b + 5 a 6 σ 1 b + 5 a 6 σ 2 b + a 2 σ 1 5 b + a 6 σ 2 , 5 b + a 6 t b ,
(16)

where σ 1 , σ 2 are two constants with σ 1 < σ 2 . Then one can see

f (t)={ σ 2 , t [ b + 5 a 6 , b + a 2 ) [ 5 b + a 6 , b ] , σ 1 , t [ a , b + 5 a 6 ) [ b + a 2 , 5 b + a 6 ) ,
(17)

and

φ(t, I k )={ t b + 5 a 6 , t [ a , b + a 2 ) , t 5 b + a 6 , t [ b + a 2 , b ] .
(18)

So, in fact, γ(t) σ 1 , Γ(t) σ 2 . According to the left-side hand of (15), we have

| a b φ ( t , I k ) [ f ( t ) γ ( t ) + Γ ( t ) 2 ] d t | = ( σ 2 σ 1 2 ) a b | φ ( t , I k ) | dt,

while according to the right-side hand of (15), we also have

a b | φ ( t , I k ) | | [ f ( t ) γ ( t ) + Γ ( t ) 2 ] | d t = a b + 5 a 6 φ ( t , I k ) | σ 1 σ 2 2 | d t + b + 5 a 6 b + a 2 φ ( t , I k ) | σ 2 σ 1 2 | d t b + a 2 5 b + a 6 φ ( t , I k ) | σ 1 σ 2 2 | d t + 5 b + a 6 b φ ( t , I k ) | σ 1 σ 2 2 | d t = ( σ 2 σ 1 2 ) a b | φ ( t , I k ) | d t .

So, (15) holds in the equality form, which confirms the proof. □

Corollary 2.2 Under the conditions of Lemma  2.1, if W 1 f(t) W 2 , t[a,b], where W 1 , W 2 are two constants, then we have the following inequality:

| i = 0 k ( δ i + 1 δ i ) f ( t i ) a b f ( t ) d t f ( b ) f ( a ) b a [ b 2 a 2 2 i = 0 k 1 δ i + 1 ( t i + 1 t i ) ] | ( W 2 W 1 2 ) a b | φ ( t , I k ) | d t .
(19)

Remark 2.1 In Corollary 2.2, if we take k=1, δ 0 =a, δ 1 = a + b 2 , δ 2 =b, t 0 =a, t 1 =b, then Corollary 2.2 becomes the trapezoid-type inequality, which provides error bound for the trapezoid quadrature formula in calculating a b f(t)dt.

| f ( b ) + f ( a ) 2 ( b a ) a b f ( t ) d t | ( W 2 W 1 2 ) a b | φ ( t , I k ) | dt,

where φ(t, I k )=t a + b 2 , t[a,b].

If we take k=2, δ 0 =a, δ 1 =a, δ 2 =b, δ 3 =b, t 0 =a, t 1 =x, t 2 =b, then Corollary 2.2 becomes the following Ostroski-Grüss type inequality as denoted in (1) [[15], Th. 1.5]:

| ( b a ) f ( x ) a b f ( t ) d t [ f ( b ) f ( a ) ] ( x a + b 2 ) | ( W 2 W 1 2 ) a b | φ ( t , I k ) | d t = ( W 2 W 1 8 ) ( b a ) 2 ,

where

φ(t, I k )={ t a ( x a + b 2 ) , t [ a , x ) , t b ( x a + b 2 ) , t [ x , b ] .

If we take k=2, δ 0 =a, δ 1 = 5 a + b 6 , δ 2 = a + 5 b 6 , δ 3 =b, t 0 =a, t 1 = a + b 2 , t 2 =b, then Corollary 2.2 becomes the following Simpson-type inequality, which provides better bound than the corresponding result in [21] (the constant on the right-side hand of the inequality is 1 12 therein).

| 1 3 [ f ( b ) + f ( a ) 2 + 2 f ( ( b + a ) 2 ) ] ( b a ) a b f ( t ) d t | ( W 2 W 1 2 ) a b | φ ( t , I k ) | d t = 5 ( W 2 W 1 ) 72 ( b a ) 2 ,

where

φ(t, I k )={ t 5 a + b 6 , t [ a , a + b 2 ) , t a + 5 b 6 , t [ a + b 2 , b ] .

Remark 2.2 Under the conditions of Corollary 2.2, furthermore, assume that the conditions of [[16], Th. 2.4] hold. Then, proceeding in the same manner as the proof in [[16], Eqs. (13)-(14)], we obtain a b |φ(t, I k )|dt ( b a ) 2 4 . So, Corollary 2.2 provides better bound than the inequality in (2) [[16], Th. 2.4].

In the following, we extend the result in Theorem 2.1 to 2D case, in which 2 f ( x , y ) x y is bounded by two functions.

Theorem 2.3 Let f:[a,b]×[c,d]R be an absolutely continuous function such that the second-order mixed partial derivative exists and there exist two functions γ(x,y), Γ(x,y) with γ(x,y) 2 f ( x , y ) x y Γ(x,y), x[a,b], y[c,d]. Suppose that x i [a,b], y i [c,d], i=0,1,,k. I k :a= x 0 < x 1 << x k 1 < x k =b is a division of the interval [a,b], while J k :a= y 0 < y 1 << y k 1 < y k =d is a division of the interval [c,d]. α i [ x i 1 , x i ], β i [ y i 1 , y i ], i=1,2,,k, α 0 =a, α k + 1 =b, β 0 =c, β k + 1 =d. Then

| { i = 1 k 1 j = 1 k 1 ( α i + 1 α i ) ( β j + 1 β j ) f ( x i , y j ) + i = 1 k 1 ( α i + 1 α i ) ( y k β k ) f ( x i , y k ) i = 1 k 1 ( α i + 1 α i ) ( y 0 β 1 ) f ( x i , y 0 ) + j = 1 k 1 ( x k α k ) ( β j + 1 β j ) f ( x k , y j ) + ( x k α k ) ( y k β k ) f ( x k , y k ) ( x k α k ) ( y 0 β 1 ) f ( x k , y 0 ) j = 1 k 1 ( x 0 α 1 ) ( β j + 1 β j ) f ( x 0 , y j ) ( x 0 α 1 ) ( y k β k ) f ( x 0 , y k ) + ( x 0 α 1 ) ( y 0 β 1 ) f ( x 0 , y 0 ) j = 1 k 1 a b ( β j + 1 β j ) f ( x , y j ) d x a b [ ( y k β k ) f ( x , y k ) ( y 0 β 1 ) f ( x , y 0 ) ] d x i = 1 k 1 c d ( α i + 1 α i ) f ( x i , y ) d y c d [ ( x k α k ) f ( x k , y ) ( x 0 α 1 ) f ( x 0 , y ) ] d y + a b c d f ( x , y ) d y d x } [ f ( b , d ) f ( b , c ) f ( a , d ) + f ( a , c ) ] ( b a ) ( d c ) × [ 1 2 ( b 2 a 2 ) i = 0 k 1 α i + 1 ( x i + 1 x i ) ] [ 1 2 ( d 2 c 2 ) j = 0 k 1 β j + 1 ( y j + 1 y j ) ] a b c d ϕ ( x , y , I k , J k ) [ γ ( x , y ) + Γ ( x , y ) 2 ] d y d x | 1 2 Γ γ { 1 2 i = 0 k 1 [ x i + 1 2 + x i 2 ] i = 0 k 1 α i + 1 ( x i + 1 + x i ) + i = 0 k 1 α i + 1 2 } × { 1 2 j = 0 k 1 [ y j + 1 2 + y j 2 ] j = 0 k 1 β j + 1 ( y j + 1 + y j ) + j = 0 k 1 β j + 1 2 } ,
(20)

where

ϕ(x,y, I k , J k )=(x α i + 1 )(y β j + 1 ),(x,y)[ x i , x i + 1 )×[ y j , y j + 1 ),i,j=0,1,,k1.
(21)

Proof Similar to [[16], Eq. (21)], we obtain that

a b c d ϕ ( x , y , I k , J k ) 2 f ( x , y ) x y d y d x = i = 1 k 1 j = 1 k 1 ( α i + 1 α i ) ( β j + 1 β j ) f ( x i , y j ) + i = 1 k 1 ( α i + 1 α i ) ( y k β k ) f ( x i , y k ) i = 1 k 1 ( α i + 1 α i ) ( y 0 β 1 ) f ( x i , y 0 ) + j = 1 k 1 ( x k α k ) ( β j + 1 β j ) f ( x k , y j ) + ( x k α k ) ( y k β k ) f ( x k , y k ) ( x k α k ) ( y 0 β 1 ) f ( x k , y 0 ) j = 1 k 1 ( x 0 α 1 ) ( β j + 1 β j ) f ( x 0 , y j ) ( x 0 α 1 ) ( y k β k ) f ( x 0 , y k ) + ( x 0 α 1 ) ( y 0 β 1 ) f ( x 0 , y 0 ) j = 1 k 1 a b ( β j + 1 β j ) f ( x , y j ) d x a b [ ( y k β k ) f ( x , y k ) ( y 0 β 1 ) f ( x , y 0 ) ] d x i = 1 k 1 c d ( α i + 1 α i ) f ( x i , y ) d y c d [ ( x k α k ) f ( x k , y ) ( x 0 α 1 ) f ( x 0 , y ) ] d y + a b c d f ( x , y ) d y d x .
(22)

On the other hand, we have

| a b c d ϕ ( x , y , I k , J k ) [ 2 f ( x , y ) x y γ ( x , y ) + Γ ( x , y ) 2 ] d y d x | 1 2 a b c d | ϕ ( x , y , I k , J k ) | | Γ ( x , y ) γ ( x , y ) | d y d x ,
(23)

and

a b c d | ϕ ( x , y , I k , J k ) | d y d x = i = 0 k 1 j = 0 k 1 x i x i + 1 y j y j + 1 | ( x α i + 1 ) ( t β j + 1 ) | d y d x = { 1 2 i = 0 k 1 [ x i + 1 2 + x i 2 ] i = 0 k 1 α i + 1 ( x i + 1 + x i ) + i = 0 k 1 α i + 1 2 } × { 1 2 j = 0 k 1 [ y j + 1 2 + y j 2 ] j = 0 k 1 β j + 1 ( y j + 1 + y j ) + j = 0 k 1 β j + 1 2 } .
(24)

Combining (22)-(24), we get the desired inequality (20). □

Corollary 2.3 Under the conditions of Theorem  2.3, if there exist two constants W 1 , W 2 such that W 1 f(x,y) W 2 , then we have the following inequality:

| { i = 1 k 1 j = 1 k 1 ( α i + 1 α i ) ( β j + 1 β j ) f ( x i , y j ) + i = 1 k 1 ( α i + 1 α i ) ( y k β k ) f ( x i , y k ) i = 1 k 1 ( α i + 1 α i ) ( y 0 β 1 ) f ( x i , y 0 ) + j = 1 k 1 ( x k α k ) ( β j + 1 β j ) f ( x k , y j ) + ( x k α k ) ( y k β k ) f ( x k , y k ) ( x k α k ) ( y 0 β 1 ) f ( x k , y 0 ) j = 1 k 1 ( x 0 α 1 ) ( β j + 1 β j ) f ( x 0 , y j ) ( x 0 α 1 ) ( y k β k ) f ( x 0 , y k ) + ( x 0 α 1 ) ( y 0 β 1 ) f ( x 0 , y 0 ) j = 1 k 1 a b ( β j + 1 β j ) f ( x , y j ) d x a b [ ( y k β k ) f ( x , y k ) ( y 0 β 1 ) f ( x , y 0 ) ] d x i = 1 k 1 c d ( α i + 1 α i ) f ( x i , y ) d y c d [ ( x k α k ) f ( x k , y ) ( x 0 α 1 ) f ( x 0 , y ) ] d y + a b c d f ( x , y ) d y d x } [ f ( b , d ) f ( b , c ) f ( a , d ) + f ( a , c ) ] ( b a ) ( d c ) × [ 1 2 ( b 2 a 2 ) i = 0 k 1 α i + 1 ( x i + 1 x i ) ] [ 1 2 ( d 2 c 2 ) j = 0 k 1 β j + 1 ( y j + 1 y j ) ] ( W 2 + W 1 2 ) [ 1 2 ( b 2 a 2 ) i = 0 k 1 α i + 1 ( x i + 1 x i ) ] × [ 1 2 ( d 2 c 2 ) j = 0 k 1 β j + 1 ( y j + 1 y j ) ] | ( W 2 W 1 2 ) { 1 2 i = 0 k 1 [ x i + 1 2 + x i 2 ] i = 0 k 1 α i + 1 ( x i + 1 + x i ) + i = 0 k 1 α i + 1 2 } × { 1 2 j = 0 k 1 [ y j + 1 2 + y j 2 ] j = 0 k 1 β j + 1 ( y j + 1 + y j ) + j = 0 k 1 β j + 1 2 } .
(25)

Proof In fact, in (20) we have γ(x,y)= W 1 , Γ(x,y)= W 2 . Furthermore, we have the following observation:

a b c d ϕ ( x , y , I k , J k ) d y d x = i = 0 k 1 j = 0 k 1 x i x i + 1 y j y j + 1 ( x α i + 1 ) ( y β j + 1 ) d y d x = [ 1 2 ( b 2 a 2 ) i = 0 k 1 α i + 1 ( x i + 1 x i ) ] [ 1 2 ( d 2 c 2 ) j = 0 k 1 β j + 1 ( y j + 1 y j ) ] .
(26)

Combining (20) and (26), we obtain the desired result. □

Remark 2.3 In Corollary 2.3, if we take k=2, α 0 =a, α 1 = 5 a + b 6 , α 2 = a + 5 b 6 , α 3 =b, x 0 =a, x 1 = a + b 2 , x 2 =b, β 0 =c, β 1 = 5 c + d 6 , β 2 = c + 5 d 6 , β 3 =d, y 0 =c, y 1 = c + d 2 , y 2 =d, then Corollary 2.3 becomes the 2D Simpson-type inequality.

3 Applications to numerical quadrature formulae

In this section, we present some applications of the results established above, and derive error bounds for some numerical quadrature formulae.

Example 1 Let G(f, I h ) be a numerical quadrature formula for a b f(t)dt, and denote

E(t)= | a b f ( t ) d t G ( f , I h ) | .
(27)

Theorem 3.1 For (27), if there exist two functions γ(t), Γ(t) with γ(t) f (t)Γ(t), t[a,b], and

G ( f , I h ) = i = 0 n 1 { 1 3 [ f ( T i + 1 ) + f ( T i ) 2 + 2 f ( ( T i + 1 + T i ) 2 ) ] h i T i T i + 1 φ ( t , I i k ) [ Γ ( t ) + γ ( t ) 2 ] d t } ,

where I h :a= T 0 < T 1 << T n 1 < T n =b is a division of [a,b], h i = T i + 1 T i , i=0,1,,n1, and

φ(t, I i , k )={ t 5 T i + T i + 1 6 , t [ T i , T i + T i + 1 2 ) , t T i + 5 T i + 1 6 , t [ T i + T i + 1 2 , T i + 1 ] ,
(28)

then we have the following estimate:

| E ( t ) | 5 72 Γ γ i = 0 n 1 h i 2 .
(29)

Proof Applying Theorem 2.2 (with [a,b] replaced by [ T i , T i + 1 ], k=2, δ 0 = T i , δ 1 = 5 T i + T i + 1 6 , δ 2 = T i + 5 T i + 1 6 , δ 3 = T i + 1 , t 0 = T i , t 1 = T i + T i + 1 2 , t 2 = T i + 1 , we obtain that

| 1 3 [ f ( T i + 1 ) + f ( T i ) 2 + 2 f ( ( T i + 1 + T i ) 2 ) ] h i T i T i + 1 f ( t ) d t T i T i + 1 φ ( t , I i k ) [ Γ ( t ) + γ ( t ) 2 ] d t | 1 2 Γ γ T i T i + 1 | φ ( t , I i , k ) | d t ,
(30)

where

φ(t, I i , k )={ t 5 T i + T i + 1 6 C i , t [ T i , T i + T i + 1 2 ) , t T i + 5 T i + 1 6 C i , t [ T i + T i + 1 2 , T i + 1 ] .
(31)

Furthermore,

C i = 1 2 ( T i + T i + 1 ) 1 T i + 1 T i i = 0 1 δ i + 1 ( t i + 1 t i )=0,
(32)

and

T i T i + 1 | φ ( t , I i , k ) | dt= 5 36 h i 2 .
(33)

Combining (30), (32) and (33) and a summation with respect to i from 0 to n1 yield (29). □

Corollary 3.1 Under the conditions of Theorem  3.1, if there exist two constants W 1 , W 2 with W 1 f (t) W 2 , t[a,b], h i h, i=0,1,,n1, and G(f, I h ) denote the Simpson complex formula, that is,

G(f, I h )= i = 0 n 1 { 1 3 [ f ( T i + 1 ) + f ( T i ) 2 + 2 f ( ( T i + 1 + T i ) 2 ) ] h i } ,

then we have

| E ( t ) | 5 72 ( W 2 W 1 )(ba)h.
(34)

Proof Considering T i T i + 1 φ(t, I i , k )dt=0 and ba=nh, from (29) we can obtain (33). □

Example 2 Let G(f, I h , J h ) be a numerical quadrature formula for a b c d f(x,y)dxdy, and denote

E(x,y)= | a b c d f ( x , y ) d x d y G ( f , I h , J h ) | .
(35)

Theorem 3.2 For (35), if there exist two functions γ(x,y), Γ(x,y) with γ(x,y) 2 f ( x , y ) x y Γ(x,y), x[a,b], y[c,d], and

G ( f , I h , J h ) = i = 0 n 1 j = 0 n 1 { ( x i + 1 x i ) ( y j + 1 y j ) 9 [ f ( x i , y j + y j + 1 2 ) + f ( x i + 1 , y j + y j + 1 2 ) + f ( x i + x i + 1 2 , y j ) + f ( x i + x i + 1 2 , y j + 1 ) + 4 f ( x i + x i + 1 2 , y j + y j + 1 2 ) ] ( x i + 1 x i ) ( y j + 1 y j ) 36 [ f ( x i , y j ) + f ( x i , y j + 1 ) + f ( x i + 1 , y j ) + f ( x i + 1 , y j + 1 ) ] + ( y j + 1 y j ) 6 x i x i + 1 [ f ( x , y j ) + f ( x , y j + y j + 1 2 ) + f ( x , y j + 1 ) ] d x + ( x i + 1 x i ) 6 y j y j + 1 [ f ( x i , y ) + f ( x i + x i + 1 2 , y ) + f ( x i + 1 , y ) ] d y + x i x i + 1 y j y j + 1 ϕ ( s , t , I i , h , J j , h ) [ γ ( x , y ) + Γ ( x , y ) 2 ] d y d x } ,

where I h :a= X 0 < X 1 << X n 1 < X n =b is a division of the interval [a,b], while J h :a= Y 0 < Y 1 << Y n 1 < Y n =d is a division of the interval [c,d], h 1 , i = x i + 1 x i , h 2 , j = y j + 1 y j , i,j=0,1,,n1, and

ϕ(x,y, I i , h , J j , h )={ ( x 5 X i + X i + 1 6 ) ( y 5 Y j + Y j + 1 6 ) , x [ X i , X i + X i + 1 2 ) , y [ Y j , Y j + Y j + 1 2 ) , ( x 5 X i + X i + 1 6 ) ( y Y j + 5 Y j + 1 6 ) , x [ X i , X i + X i + 1 2 ) , y [ Y j + Y j + 1 2 , Y j + 1 ] , ( x X i + 5 X i + 1 6 ) ( y 5 Y j + Y j + 1 6 ) , x [ X i + X i + 1 2 , X i + 1 ] , y [ Y j , Y j + Y j + 1 2 ) , ( x X i + 5 X i + 1 6 ) ( y Y j + 5 Y j + 1 6 ) , x [ X i + X i + 1 2 , X i + 1 ] , y [ Y j + Y j + 1 2 , Y j + 1 ] ,
(36)

then we have the following estimate:

| E ( x , y ) | 25 2592 Γ γ ( i = 0 n 1 h 1 , i 2 ) ( i = 0 n 1 h 2 , i 2 ) .
(37)

The desired result can be obtained by a suitable application of Theorem 2.3 (with [a,b] replaced by [ x i , x i + 1 ], [c,d] replaced by [ y j , y j + 1 ], k=2, α 0 = x i , α 1 = 5 x i + x i + 1 6 , α 2 = x i + 5 x i + 1 6 , α 3 = x i + 1 , x 0 = x i , x 1 = x i + x i + 1 2 , x 2 = x i + 1 , β 0 = y j , β 1 = 5 y j + y j + 1 6 , β 2 = y j + 5 y j + 1 6 , β 3 = y j + 1 , y 0 = y j , y 1 = y j + y j + 1 2 , y 2 = y j + 1 ), and a summation with respect to i, j from 0 to n1.

In Theorem 3.2, after a simple computation, one can see x i x i + 1 y j y j + 1 ϕ(x,y, I i , h , J j , h )dydx=0. So, we obtain the following corollary.

Corollary 3.2 Under the conditions of Theorem  3.2, if there exist two constants W 1 , W 2 with W 1 2 f ( x , y ) x y W 2 , x[a,b], y[c,d], h 1 , i = h 2 , j h, i,j=0,1,,n1, and G(f, I h , J h ) denote the 2D Simpson complex formula, that is,

G ( f , I h , J h ) = i = 0 n 1 j = 0 n 1 { ( x i + 1 x i ) ( y j + 1 y j ) 9 [ f ( x i , y j + y j + 1 2 ) + f ( x i + 1 , y j + y j + 1 2 ) + f ( x i + x i + 1 2 , y j ) + f ( x i + x i + 1 2 , y j + 1 ) + 4 f ( x i + x i + 1 2 , y j + y j + 1 2 ) ] ( x i + 1 x i ) ( y j + 1 y j ) 36 [ f ( x i , y j ) + f ( x i , y j + 1 ) + f ( x i + 1 , y j ) + f ( x i + 1 , y j + 1 ) ] + ( y j + 1 y j ) 6 x i x i + 1 [ f ( x , y j ) + f ( x , y j + y j + 1 2 ) + f ( x , y j + 1 ) ] d x + ( x i + 1 x i ) 6 y j y j + 1 [ f ( x i , y ) + f ( x i + x i + 1 2 , y ) + f ( x i + 1 , y ) ] d y } ,

then we have the following estimate:

| E ( x , y ) | 25 2592 ( W 2 W 1 )(ba)(dc) h 2 .
(38)

4 Conclusions

We have established some new Ostrowski-Grüss type inequalities involving multiple interior points with the derivatives bounded by functions in both 1D and 2D cases and derived some sharp bounds related to them. As one can see, the inequalities are of new forms and provide better bounds than some existing results in the literature. As for applications, new error bounds for some numerical quadrature formulae are derived based on the inequalities established.

References

  1. Ostrowski A: Über die Absolutabweichung einer differentiierbaren Funktion von ihren Integralmittelwert. Comment. Math. Helv. 1938, 10: 226–227.

    MathSciNet  Article  Google Scholar 

  2. Grüss G: Über das Maximum des absoluten Betrages von [1/(ba)] a b f(x)g(x)dx[1/ ( b a ) 2 ] a b f(x)dx a b g(x)dx. Math. Z. 1935, 39: 215–226. 10.1007/BF01201355

    MathSciNet  Article  Google Scholar 

  3. Alomari M, Darus M, Dragomir SS, Cerone P: Ostrowski type inequalities for functions whose derivatives are s -convex in the second sense. Appl. Math. Lett. 2010, 23: 1071–1076. 10.1016/j.aml.2010.04.038

    MATH  MathSciNet  Article  Google Scholar 

  4. Anastassiou GA: High order Ostrowski type inequalities. Appl. Math. Lett. 2007, 20: 616–621. 10.1016/j.aml.2006.07.005

    MATH  MathSciNet  Article  Google Scholar 

  5. Tseng KL, Hwang SR, Yang GS, Chou YM: Improvements of the Ostrowski integral inequality for mappings of bounded variation I. Appl. Math. Comput. 2010, 217: 2348–2355. 10.1016/j.amc.2010.07.034

    MATH  MathSciNet  Article  Google Scholar 

  6. Dragomir SS: The discrete version of Ostrowski inequalities in normed linear spaces. J. Inequal. Pure Appl. Math. 2002., 3(1): Article ID 2

  7. Dragomir SS, Sofo A: An inequality for monotonic functions generalizing Ostrowski and related results. Comput. Math. Appl. 2006, 51: 497–506. 10.1016/j.camwa.2005.08.029

    MATH  MathSciNet  Article  Google Scholar 

  8. Cerone P, Cheung WS, Dragomir SS: On Ostrowski type inequalities for Stieltjes integrals with absolutely continuous integrands and integrators of bounded variation. Comput. Math. Appl. 2007, 54: 183–191. 10.1016/j.camwa.2006.12.023

    MATH  MathSciNet  Article  Google Scholar 

  9. Liu Z: Some Ostrowski type inequalities. Math. Comput. Model. 2008, 48: 949–960. 10.1016/j.mcm.2007.12.004

    MATH  Article  Google Scholar 

  10. Pachpatte BG: On an inequality of Ostrowski type in three independent variables. J. Math. Anal. Appl. 2000, 249: 583–591. 10.1006/jmaa.2000.6913

    MATH  MathSciNet  Article  Google Scholar 

  11. Xue QL, Zhu J, Liu WJ: A new generalization of Ostrowski-type inequality involving functions of two independent variables. Comput. Math. Appl. 2010, 60: 2219–2224. 10.1016/j.camwa.2010.08.011

    MATH  MathSciNet  Article  Google Scholar 

  12. Aljinović AA, Pečarić J: Discrete weighted Montgomery identity and discrete Ostrowski type inequalities. Comput. Math. Appl. 2004, 48: 731–745. 10.1016/j.camwa.2004.03.004

    MATH  MathSciNet  Article  Google Scholar 

  13. Dragomir SS, Wang S: An inequality of Ostrowski-Grüss’ type and its applications to the estimation of error bounds for some special means and for some numerical quadrature rules. Comput. Math. Appl. 1997, 33(11):15–20. 10.1016/S0898-1221(97)00084-9

    MATH  MathSciNet  Article  Google Scholar 

  14. Ujević N: New bounds for the first inequality of Ostrowski-Grüss type and applications. Comput. Math. Appl. 2003, 46: 421–427. 10.1016/S0898-1221(03)90035-6

    MATH  MathSciNet  Article  Google Scholar 

  15. Cheng XL: Improvement of some Ostrowski-Grüss type inequalities. Comput. Math. Appl. 2001, 42: 109–114. 10.1016/S0898-1221(01)00135-3

    MATH  MathSciNet  Article  Google Scholar 

  16. Feng QH, Meng FW: Some generalized Ostrowski-Grüss type integral inequalities. Comput. Math. Appl. 2012, 63: 652–659. 10.1016/j.camwa.2011.11.017

    MATH  MathSciNet  Article  Google Scholar 

  17. Niezgoda M: A new inequality of Ostrowski-Grüss type and applications to some numerical quadrature rules. Comput. Math. Appl. 2009, 58: 589–596. 10.1016/j.camwa.2009.03.089

    MATH  MathSciNet  Article  Google Scholar 

  18. Matić M, Pečarić J, Ujević N: Improvement and further generalization of inequalities of Ostrowski-Grüss type. Comput. Math. Appl. 2000, 39: 161–175.

    MATH  Google Scholar 

  19. Liu Z: Some Ostrowski-Grüss type inequalities and applications. Comput. Math. Appl. 2007, 53: 73–79. 10.1016/j.camwa.2006.12.021

    MATH  MathSciNet  Article  Google Scholar 

  20. Yang S: A unified approach to some inequalities of Ostrowski-Grüss type. Comput. Math. Appl. 2006, 51: 1047–1056. 10.1016/j.camwa.2005.09.005

    MATH  MathSciNet  Article  Google Scholar 

  21. Pearce CEM, Pečarić J, Ujević N, Varos̆anec S: Generalizations of some inequalities of Ostrowski-Grüss type. Math. Inequal. Appl. 2000, 3(1):25–34.

    MATH  MathSciNet  Google Scholar 

Download references

Acknowledgements

The authors would like to thank the referees very much for their valuable suggestions on improving this paper.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Qinghua Feng.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

QF carried out the main part of this article. All authors read and approved the final manuscript.

Rights and permissions

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

Reprints and Permissions

About this article

Cite this article

Feng, Q., Meng, F. New Ostrowski-Grüss type inequalities with the derivatives bounded by functions. J Inequal Appl 2013, 456 (2013). https://doi.org/10.1186/1029-242X-2013-456

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1029-242X-2013-456

Keywords

  • Ostrowski-Grüss type inequality
  • sharp
  • numerical integration
  • error bound