1 Introduction

Throughout this paper, we adopt the following notations:

R=(,), R 0 =[0,),and R + =(0,).
(1.1)

We recall some definitions of several convex functions.

Definition 1.1 A function f:IRR is said to be convex if

f ( λ x + ( 1 λ ) y ) λf(x)+(1λ)f(y)
(1.2)

holds for all x,yI and λ[0,1].

Definition 1.2 ([1])

For f:[0,b]R and m(0,1], if

f ( λ x + m ( 1 λ ) y ) λf(x)+m(1λ)f(y)
(1.3)

is valid for all x,y[0,b] and λ[0,1], then we say that f(x) is an m-convex function on [0,b].

Definition 1.3 ([2])

For f:[0,b]R and α,m(0,1], if

f ( λ x + m ( 1 λ ) y ) λ α f(x)+m ( 1 λ α ) f(y)
(1.4)

is valid for all x,y[0,b] and λ[0,1], then we say that f(x) is an (α,m)-convex function on [0,b].

In recent decades, plenty of inequalities of Hermite-Hadamard type for various kinds of convex functions have been established. Some of them may be reformulated as follows.

Theorem 1.1 ([[3], Theorem 2.2])

Let f: I RR be a differentiable mapping and a,b I with a<b. If | f (x)| is convex on [a,b], then

| f ( a ) + f ( b ) 2 1 b a a b f ( x ) d x | ( b a ) ( | f ( a ) | + | f ( b ) | ) 8 .
(1.5)

Theorem 1.2 ([[4], Theorem 2])

Let f: R 0 R be m-convex and m(0,1]. If fL[a,b] for 0a<b<, then

1 b a a b f(x)dxmin { f ( a ) + m f ( b / m ) 2 , m f ( a / m ) + f ( b ) 2 } .
(1.6)

Theorem 1.3 ([[2], Theorem 2.2])

Let I R 0 be an open interval and let f:IR be a differentiable function such that f L[a,b] for 0a<b<. If | f ( x ) | q is m-convex on [a,b] for some m(0,1] and q1, then

(1.7)

Theorem 1.4 ([[2], Theorem 3.1])

Let I R 0 be an open interval and let f:IR be a differentiable function such that f L[a,b] for 0a<b<. If [ f ( x ) ] q is (α,m)-convex on [a,b] for some α,m(0,1] and q1, then

where

v 1 = 1 ( α + 1 ) ( α + 2 ) ( α + 1 2 α )
(1.8)

and

v 2 = 1 ( α + 1 ) ( α + 2 ) ( α 2 + α + 2 2 1 2 α ) .
(1.9)

For more and detailed information on this topic, please refer to the monograph [5] and newly published papers [616].

In this paper, we establish some Hermite-Hadamard type integral inequalities for n-time differentiable functions which are (α,m)-convex.

2 A lemma

In order to find inequalities of Hermite-Hadamard type for (α,m)-convex functions, we need the following lemma.

Lemma 2.1 ([[17], Lemma 2.1] or [[18], Lemma 2.1])

Let f:[a,b]RR be an n-time differentiable function such that f ( n 1 ) (x) for nN is absolutely continuous on [a,b]. Then the identity

a b f ( x ) d x = k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) + ( 1 ) n a b K n ( t , x ) f ( n ) ( x ) d x
(2.1)

holds for all t[a,b], where the kernel K n :[a,b]×[a,b]R is defined by

K n (t,x)={ ( x a ) n n ! , x [ a , t ] , ( x b ) n n ! , x [ t , b ] .
(2.2)

3 Hermite-Hadamard type inequalities for (α,m)-convex functions

We now set off to establish some new integral inequalities of Hermite-Hadamard type for n-time differentiable (α,m)-convex functions.

Theorem 3.1 Let f: R 0 R be an n-time differentiable function for nN and let 0a<b< and α,m(0,1]. If f ( n ) (x)L[a, b m ] and | f ( n ) ( x ) | q for q1 is (α,m)-convex on [0, b m ], then

(3.1)

where t[a,b] and B(α,β) is the beta function

B(α,β)= 0 1 t α 1 ( 1 t ) β 1 dt,α,β>0.
(3.2)

Proof If a<t<b, by Lemma 2.1, Hölder’s integral inequality, and the (α,m)-convexity of | f ( n ) ( x ) | q , we have

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! [ a t ( x a ) n | f ( n ) ( x ) | d x + t b ( b x ) n | f ( n ) ( x ) | d x ] 1 ( b a ) n ! { [ a t ( x a ) n d x ] 1 1 / q [ a t ( x a ) n | f ( n ) ( x ) | q d x ] 1 / q + [ t b ( b x ) n d x ] 1 1 / q [ t b ( b x ) n | f ( n ) ( x ) | q d x ] 1 / q } = 1 ( b a ) n ! { [ ( t a ) n + 1 n + 1 ] 1 1 / q [ a t ( x a ) n | f ( n ) ( t x t a a + m x a t a × t m ) | q d x ] 1 / q + [ ( b t ) n + 1 n + 1 ] 1 1 / q × [ t b ( b x ) n | f ( n ) ( b x b t t + m x t b t × b m ) | q d x ] 1 / q } 1 ( b a ) n ! { [ ( t a ) n + 1 n + 1 ] 1 1 / q ( a t ( x a ) n [ ( t x t a ) α | f ( n ) ( a ) | q + m ( 1 ( t x t a ) α ) | f ( n ) ( t m ) | q ] d x ) 1 / q + [ ( b t ) n + 1 n + 1 ] 1 1 / q × ( t b ( b x ) n [ ( b x b t ) α | f ( n ) ( t ) | q + m ( 1 ( b x b t ) α ) | f ( n ) ( b m ) | q ] d x ) 1 / q } .

Substituting

a t ( x a ) n { ( t x t a ) α | f ( n ) ( a ) | q + m [ 1 ( t x t a ) α ] | f ( n ) ( t m ) | q } d x = ( t a ) n + 1 n + 1 [ α B ( n + 2 , α ) | f ( n ) ( a ) | q + m ( 1 α B ( n + 2 , α ) ) | f ( n ) ( t m ) | q ]

and

t b ( b x ) n { ( b x b t ) α | f ( n ) ( t ) | q + m [ 1 ( b x b t ) α ] | f ( n ) ( b m ) | q } d x = ( b t ) n + 1 ( n + 1 ) ( n + α + 1 ) [ ( n + 1 ) | f ( n ) ( t ) | q + α m | f ( n ) ( b m ) | q ]

into the above inequality leads to the inequality (3.1) for t(a,b).

If t=a or t=b, by virtue of Lemma 2.1 and the property that | f ( n ) ( x ) | q is (α,m)-convex on [0, b m ], we have

and

The inequality (3.1) for t=a or t=b follows. Theorem 3.1 is thus proved. □

Corollary 3.1 Under the conditions of Theorem  3.1,

(1) when q=1, we have

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) ( n + 1 ) ! { ( t a ) n + 1 [ α B ( n + 2 , α ) | f ( n ) ( a ) | + m ( 1 α B ( n + 2 , α ) ) | f ( n ) ( t m ) | ] + ( b t ) n + 1 [ 1 n + α + 1 ( ( n + 1 ) | f ( n ) ( t ) | + α m | f ( n ) ( b m ) | ) ] } ;

(2) when α=1, we have

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) ( n + 1 ) ! ( 1 n + 2 ) 1 / q { ( t a ) n + 1 [ | f ( n ) ( a ) | q + m ( n + 1 ) | f ( n ) ( t m ) | q ] 1 / q + ( b t ) n + 1 [ ( ( n + 1 ) | f ( n ) ( t ) | q + m | f ( n ) ( b m ) | q ) ] 1 / q } ;

(3) when m=1, we have

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) ( n + 1 ) ! { ( t a ) n + 1 [ α B ( n + 2 , α ) | f ( n ) ( a ) | q + ( 1 α B ( n + 2 , α ) ) | f ( n ) ( t ) | q ] 1 / q + ( b t ) n + 1 [ 1 n + α + 1 ( ( n + 1 ) | f ( n ) ( t ) | q + α | f ( n ) ( b ) | q ) ] 1 / q } ;

(4) when m=α=q=1, we have

Corollary 3.2 Under the conditions of Theorem  3.1,

(1) when t=a, we have

(3.3)

(2) when t= a + b 2 , we have

(3) when t=b, we have

Theorem 3.2 Let t[a,b] and f: R 0 R be an n-time differentiable function for nN, and let 0a<b< and α,m(0,1]. If f ( n ) (x)L[a, b m ], | f ( n ) ( x ) | q for q>1 is (α,m)-convex on [0, b m ], and nqp0, then

(3.4)

Proof When a<t<b, by Lemma 2.1 and Hölder’s integral inequality, we have

(3.5)

where

a t ( x a ) ( n q p ) / ( q 1 ) dx= q 1 n q + q p 1 ( t a ) ( n q + q p 1 ) / ( q 1 )
(3.6)

and

t b ( b x ) ( n q p ) / ( q 1 ) dx= q 1 n q + q p 1 ( b t ) ( n q + q p 1 ) / ( q 1 ) .
(3.7)

Since | f ( n ) ( x ) | q is (α,m)-convex on [0, b m ], we have

a t ( x a ) p | f ( n ) ( x ) | q d x a t ( x a ) p { ( t x t a ) α | f ( n ) ( a ) | q + m [ 1 ( t x t a ) α ] | f ( n ) ( t m ) | q } d x = ( t a ) p + 1 p + 1 [ α B ( p + 2 , α ) | f ( n ) ( a ) | q + m ( 1 α B ( p + 2 , α ) ) | f ( n ) ( t m ) | q ]

and

t b ( b x ) p | f ( n ) ( x ) | q d x t b ( b x ) p { ( b x b t ) α | f ( n ) ( t ) | q + m [ 1 ( b x b t ) α ] | f ( n ) ( b m ) | q } d x = ( b t ) p + 1 ( p + 1 ) ( p + α + 1 ) [ ( p + 1 ) | f ( n ) ( t ) | q + α m | f ( n ) ( b m ) | q ] .

Hence, the inequality (3.4) follows.

When t=a or t=b, the proof of the inequality (3.4) is similar to the above argument. The proof of Theorem 3.2 is complete. □

Corollary 3.3 Under the conditions of Theorem  3.2,

(1) if α=1, then

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! ( q 1 n q + q p 1 ) 1 1 / q [ 1 ( p + 1 ) ( p + 2 ) ] 1 / q × { ( t a ) n + 1 [ | f ( n ) ( a ) | q + m ( p + 1 ) | f ( n ) ( t m ) | q ] 1 / q + ( b t ) n + 1 [ ( ( p + 1 ) | f ( n ) ( t ) | q + m | f ( n ) ( b m ) | q ) ] 1 / q } ;

(2) if m=1, then

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! ( q 1 n q + q p 1 ) 1 1 / q ( 1 p + 1 ) 1 / q { ( t a ) n + 1 × [ α B ( p + 2 , α ) | f ( n ) ( a ) | q + ( 1 α B ( p + 2 , α ) ) | f ( n ) ( t ) | q ] 1 / q + ( b t ) n + 1 [ 1 p + α + 1 ( ( p + 1 ) | f ( n ) ( t ) | q + α | f ( n ) ( b ) | q ) ] 1 / q } ;

(3) if m=α=1, we have

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! ( q 1 n q + q p 1 ) 1 1 / q [ 1 ( p + 1 ) ( p + 2 ) ] 1 / q { ( t a ) n + 1 [ | f ( n ) ( a ) | q + ( p + 1 ) | f ( n ) ( t ) | q ] 1 / q + ( b t ) n + 1 [ ( p + 1 ) | f ( n ) ( t ) | q + | f ( n ) ( b ) | q ] 1 / q } .

Corollary 3.4 Under the conditions of Theorem  3.2,

(1) if t=a, then

(3.8)

(2) if t= a + b 2 , then

| 1 b a a b f ( x ) d x k = 0 n 1 [ 1 + ( 1 ) k ] ( b a ) k 2 k + 1 ( k + 1 ) ! f ( k ) ( a + b 2 ) | ( b a ) n 2 n + 1 n ! ( q 1 n q + q p 1 ) 1 1 / q ( 1 p + 1 ) 1 / q × { [ α B ( p + 2 , α ) | f ( n ) ( a ) | q + m ( 1 α B ( p + 2 , α ) ) | f ( n ) ( a + b 2 m ) | q ] 1 / q + [ 1 p + α + 1 ( ( p + 1 ) | f ( n ) ( a + b 2 ) | q + α m | f ( n ) ( b m ) | q ) ] 1 / q } ;

(3) if t=b, then

Corollary 3.5 Under the conditions of Theorem  3.2,

(1) if p=0, then

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! ( q 1 n q + q 1 ) 1 1 / q [ 1 ( α + 1 ) ( α + 2 ) ] 1 / q { ( t a ) n + 1 [ | f ( n ) ( a ) | q + α m | f ( n ) ( t m ) | q ] 1 / q + ( b t ) n + 1 [ | f ( n ) ( t ) | q + α m | f ( n ) ( b m ) | q ] 1 / q } ;

(2) if p=q, then

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! ( q 1 n q 1 ) 1 1 / q ( 1 q + 1 ) 1 / q { ( t a ) n + 1 × [ α B ( q + 2 , α ) | f ( n ) ( a ) | q + m ( 1 α B ( q + 2 , α ) ) | f ( n ) ( t m ) | q ] 1 / q + ( b t ) n + 1 [ 1 q + α + 1 ( ( q + 1 ) | f ( n ) ( t ) | q + α m | f ( n ) ( b m ) | q ) ] 1 / q } ;

(3) if p=nq, then

| 1 b a a b f ( x ) d x 1 b a k = 0 n 1 ( b t ) k + 1 + ( 1 ) k ( t a ) k + 1 ( k + 1 ) ! f ( k ) ( t ) | 1 ( b a ) n ! ( 1 n q + 1 ) 1 / q { ( t a ) n + 1 × [ α B ( n q + 2 , α ) | f ( n ) ( a ) | q + m ( 1 α B ( n q + 2 , α ) ) | f ( n ) ( t m ) | q ] 1 / q + ( b t ) n + 1 [ 1 n q + α + 1 ( ( n q + 1 ) | f ( n ) ( t ) | q + α m | f ( n ) ( b m ) | q ) ] 1 / q } .