Recently, İşcan [1] presented the following concept of p-convex mappings, which is a generalization of ordinary convexity and harmonically convexity.

Definition 1.1 [1] The mapping π : I ⊂ ( 0, ∞ ) → ℝ is named a p-convex mapping, where p ∈ ℝ \ { 0 } . If for all s ∈ [ 0,1 ] and x , y ∈ I , we have

π ( [ s x p + ( 1 − s ) y p ] 1 p ) ≤ s π ( x ) + ( 1 − s ) π ( y ) . (1.1)

Many researchers have worked in the properties and inequalities for p-convex functions. For example, Zhang and Wang [2] introduced some properties for p-convexity. Kunt and İşcan [3] established several Hermite-Hadamard-Fejér inequalities involving p-convex mapping. Dragomir et al. [4] gave some integral inequalities for differentiable p-convex mappings. Mehreen and Anwar [5] presented several Hermite-Hadamard type inequalities related to exponentially p-convex functions. For more results associated with p-convex functions see references in [6] [7].

In [8], Katugampola introduced a class of fractional integral operator, which generalizes Riemann-Liouville and Hadamard fractional integrals simultaneously.

Definition 1.2 [8] Let [ a , b ] ⊂ ℝ be a finite interval. The left-side and right-side Katugampola fractional integrals of order μ ≥ 0 of π ∈ χ c σ ( a , b )   ( c ∈ ℝ ,1 ≤ σ ≤ ∞ ) are defined respectively by

I p a + μ π ( x ) = p 1 − μ Γ ( μ ) ∫ a x s p − 1 ( x p − s p ) 1 − μ π ( s ) d s (1.2)

and

I p b − μ π ( x ) = p 1 − μ Γ ( μ ) ∫ x b s p − 1 ( s p − x p ) 1 − μ π ( s ) d s , (1.3)

where a < x < b , p > 0 and Γ ( μ ) is the Gamma function and its definition is Γ ( μ ) = ∫ 0 ∞     e − s s μ − 1 d s , if the integrals exist.

Theorem 1.1 [8] Let μ > 0 and p > 0 . Then, for x < b , we have

l i m p → 1 I p b − μ π ( x ) = J b − μ π ( x )     and     l i m p → 0 + I p b − μ π ( x ) = H b − μ π ( x ) , (1.4)

and for x > a , we have

lim p → 1 I p a + μ π ( x ) = J a + μ π ( x )     and     lim p → 0 + I p a + μ π ( x ) = H a + μ π ( x ) , (1.5)

where the symbol J a + μ π and J b − μ π denote respectively the left-sided and right-sided Riemann-Liouville fractional integrals of the order μ ∈ ℝ + defined by

J a + μ π ( x ) = 1 Γ ( μ ) ∫ a x ( x − s ) μ − 1 π ( s ) d s ,     a < x (1.6)

and

J b − μ π ( x ) = 1 Γ ( μ ) ∫ x b ( s − x ) μ − 1 π ( s ) d s ,     x < b . (1.7)

And the symbol H a + μ and H b − μ π denote respectively the left-sided and right-sided Hadamard fractional integrals of order μ ∈ ℝ + defined as

H a + μ π ( x ) = 1 Γ ( μ ) ∫ a x ( ln ( x s ) ) μ − 1 π ( s ) d s s ,     a < x (1.8)

and

H b − μ π ( x ) = 1 Γ ( μ ) ∫ x b ( ln ( x s ) ) μ − 1 π ( s ) d s s ,     x < b . (1.9)

Theory of Katugampola fractional integral operators attract widely attention for many authors, some new generalizations, extensions and variations of classically integral inequalities via Katugampola fractional integrals have been established in the literature. For example, Chen and Katugampola [9] obtained Hermite-Hadamard and Hermite-Hadamard-Fejér type inequalities in connection with Katugampola fractional integrals and convex mappings. Delavar and Dragomir [10] studied Katugampola fractional integrals Hermite-Hadamard’s mid-point inequalities via Lipschitzian mappings and convex mappings. Toplu et al. [11] established the Hermite-Hadamard inequality for p-convex mappings via Katugampola fractional integrals. Kermausuor et al. [12] introduced some new Katugampola fractional integrals Hermite-Hadamard type inequalities through strongly η -convex mappings. For more information related to Katugampola fractional integral operators, we refer an interested reader to [13] - [17].

In [18], Hu et al. established the following identity for right Katugampola fractional integrals to derive several parameterized integral inequalities.

Theorem 1.2 Let p , μ > 0 and π : [ a , b ] ⊆ ( 0, ∞ ) → ℝ be a differentiable mapping on ( a , b ) such that π ′ ∈ L 1 ( [ a , b ] ) , where 0 < a < b . Then for all m , n ∈ ℝ , the following identity holds:

m π ( a ) + ( n − m ) π ( [ a p + b p 2 ] 1 p ) + ( 1 − n ) π ( b ) − p α Γ ( α + 1 ) b p − a p I p b − μ π ( a ) = b p − a p p ∫ 0 1     z ( s ) ( s b p + ( 1 − s ) a p ) π ′ ( s b p + ( 1 − s ) a p p ) d s , (1.10)

where

z ( s ) = ( s μ − m ,   s ∈ [ 0, 1 2 ) , s μ − n ,         s ∈ [ 1 2 ,1 ] . (1.11)

These studies motivated us to establish some trapezium-type inequalities involving Katugampola fractional integrals for the mappings whose first derivative absolute values are p-convex. We emphasize that our main results generalize the ones obtained by Sarikaya and Budak [19]. Also, we present two examples to support our results.

Before stating the results, we define some notations as follows:

m : = ( λ a p + ( 1 − λ ) b p ) 1 p , (2.1)

n : = ( λ b p + ( 1 − λ ) a p ) 1 p (2.2)

and

Φ π ( μ , p , λ , a , b ) : = − p [ π ( n ) + π ( m ) ] ( 1 − 2 λ ) ( b p − a p ) + Γ ( 1 + μ ) p μ + 1 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 [ I p n + μ π ( m ) + I p m − μ π ( n ) ] . (2.3)

Lemma 2.1 Assume that μ , p > 0 and π : [ a p , b p ] ⊆ ( 0, ∞ ) → ℝ be a differentiable mapping on ( a p , b p ) with 0 < a < b satisfying π ′ ∈ L 1 ( [ a p , b p ] ) . Then the following identity exists:

Φ π ( μ , p , λ , a , b ) = ∫ 0 1 [ ( 1 − s ) μ − s μ ] ( s m p + ( 1 − s ) n p ) 1 p − 1 π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) d s , (2.4)

where λ ∈ ( 0,1 ] \ 1 2 .

Proof. It suffices to note that:

ξ = ∫ 0 1 [ ( 1 − s ) μ − s μ ] ( s m p + ( 1 − s ) n p ) 1 p − 1 π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) d s = [ ∫ 0 1 ( 1 − s ) μ ( s m p + ( 1 − s ) n p ) 1 p − 1 π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) d s ]       + [ − ∫ 0 1     s μ ( s m p + ( 1 − s ) n p ) 1 p − 1 π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) d s ] : = ξ 1 + ξ 2 . (2.5)

Integrating by parts, we obtain

ξ 1 = p ( 1 − 2 λ ) ( b p − a p ) [ ∫ 0 1 ( 1 − s ) μ d ( π ( [ s m p + ( 1 − s ) n p ] 1 p ) ) ] = p ( 1 − 2 λ ) ( b p − a p ) [ ( 1 − s ) μ π ( [ s m p + ( 1 − s ) n p ] 1 p ) | 0 1     + μ ∫ 0 1 ( 1 − s ) μ − 1 π ( [ s m p + ( 1 − s ) n p ] 1 p ) d s ] = − p π ( n ) ( 1 − 2 λ ) ( b p − a p )     + μ p ( 1 − 2 λ ) ( b p − a p ) ∫ 0 1 ( 1 − s ) μ − 1 π ( [ s m p + ( 1 − s ) n p ] 1 p ) d s . (2.6)

Using the change of variable x p = s ( λ a p + ( 1 − λ ) b p ) + ( 1 − s ) ( λ b p + ( 1 − λ ) a p ) for s ∈ [ 0,1 ] , then the Equality (2.6) can be written as

ξ 1 = − p π ( n ) ( 1 − 2 λ ) ( b p − a p ) + μ p 2 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 ∫ n m x p − 1 ( m p − x p ) 1 − μ π ( x ) d x = − p π ( n ) ( 1 − 2 λ ) ( b p − a p ) + Γ ( 1 + μ ) p μ + 1 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 I p n + μ π ( m ) , (2.7)

and similarly, we have

ξ 2 = − p ( 1 − 2 λ ) ( b p − a p ) [ ∫ 0 1     s μ d ( π ( [ s m p + ( 1 − s ) n p ] 1 p ) ) = − p ( 1 − 2 λ ) ( b p − a p ) [ s μ π ( [ s m p + ( 1 − s ) n p ] 1 p ) | 0 1     + μ ∫ 0 1     s μ − 1 π ( [ s m p + ( 1 − s ) n p ] 1 p ) d s ] = − p π ( m ) ( 1 − 2 λ ) ( b p − a p ) + μ p 2 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 ∫ n m x p − 1 ( x p − n p ) 1 − μ π ( x ) d x = − p π ( m ) ( 1 − 2 λ ) ( b p − a p ) + Γ ( 1 + μ ) p μ + 1 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 I p m − μ π ( n ) . (2.8)

Adding the Equality (2.7) and Equality (2.8) together, we get

ξ = ξ 1 + ξ 2 = − p [ π ( n ) + π ( m ) ] ( 1 − 2 λ ) ( b p − a p ) + Γ ( 1 + μ ) p μ + 1 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 [ I p n + μ π ( m ) + I p m − μ π ( n ) ] . (2.9)

This completes the proof.

Remark 2.1 Choosing p = 1 in Lemma 2.1, we have Lemma 2.1 presented by Sarikaya and Budak in [19].

Remark 2.2 Taking λ = 0 in Lemma 2.1, we have

− p [ π ( a ) + π ( b ) ] b p − a p + Γ ( 1 + μ ) p μ + 1 ( b p − a p ) μ + 1 [ I p a + μ π ( b ) + I p b − μ π ( a ) ] = ∫ 0 1 [ ( 1 − s ) μ − s μ ] ( s b p + ( 1 − s ) a p ) 1 p − 1 π ′ ( [ s b p + ( 1 − s ) a p ] 1 p ) d s . (2.10)

Similarly, putting λ = 1 in Lemma 2.1, we obtain

p [ π ( a ) + π ( b ) ] b p − a p + Γ ( 1 + μ ) p μ + 1 ( b p − a p ) μ + 1 [ I p a − μ π ( b ) + I p b + μ π ( a ) ] = ∫ 0 1 [ ( 1 − s ) μ − s μ ] ( s a p + ( 1 − s ) b p ) 1 p − 1 π ′ ( [ s a p + ( 1 − s ) b p ] 1 p ) d s , (2.11)

which is proved by Toplu et al. in [11].

We now present some katugampola fractional integrals inequalities with multiple parameters related to p-convex mappings.

Theorem 3.1 Let π : [ a p , b p ] ⊂ ( 0, ∞ ) → R be a differentiable mapping on ( a p , b p ) with 0 < a < b such that π ′ ∈ L 1 ( [ a p , b p ] ) . If | π ′ | q is p-convex on [ a p , b p ] for p > 0 , q ≥ 1 , then the following inequality for katugampola fractional integrals holds:

| Φ π ( μ , p , λ , a , b ) | ≤ η 1 − p [ 2 μ + 1 ( 1 − 1 2 μ ) ] [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q , (3.1)

where μ > 0 , λ ∈ ( 0,1 ] \ 1 2 and

η = ( a ,           1 ≤ p < ∞ , b ,         0 < p < 1. (3.2)

Proof. Case 1: p ≥ 1 . By means of Lemma 2.1, one has

| Φ π ( μ , p , λ , a , b ) | ≤ ∫ 0 1 | ( 1 − s ) μ − s μ | | ( s m p + ( 1 − s ) n p ) 1 p − 1 | | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | d s . (3.3)

For all 0 ≤ s ≤ 1 , applying the fact that

( s ( λ a p + ( 1 − λ ) b p ) + ( 1 − s ) ( λ b p + ( 1 − λ ) a p ) ) 1 p − 1 ≤ ( a p ) 1 p − 1 = a 1 − p , (3.4)

we obtain

| Φ π ( μ , p , λ , a , b ) | ≤ a 1 − p [ ∫ 0 1 | ( 1 − s ) μ − s μ | | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | d s ] . (3.5)

Using the Hölder inequality for inequality (3.5), we deduce

| Φ π ( μ , p , λ , a , b ) | ≤ a 1 − p [ ∫ 0 1 | ( 1 − s ) μ − s μ | d s ] 1 − 1 q [ ∫ 0 1 | ( 1 − s ) μ − s μ | | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ] 1 q = a p − 1 ( ∫ 0 1 2 [ ( 1 − s ) μ − s μ ] d s + ∫ 1 2 1 [ s μ − ( 1 − s ) μ ] d s ) 1 − 1 q         ⋅ [ ∫ 0 1 | ( 1 − s ) μ − s μ | | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ] 1 q . (3.6)

Since | π ′ | q is p-convex, we get

| Φ π ( μ , p , λ , a , b ) | ≤ a p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] 1 − 1 q [ ∫ 0 1 | ( 1 − s ) μ − s μ | [ s | π ′ ( m ) | q + ( 1 − s ) | π ′ ( n ) | q ] d s ] 1 q = a p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] 1 − 1 q [ 1 μ + 1 ( 1 − 1 2 μ ) ] 1 q [ | π ′ ( m ) | q + | π ′ ( n ) | q ] 1 q = a p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q . (3.7)

Case 2: 0 < p < 1 . For all 0 ≤ s ≤ 1 , we deduce that

( s ( λ a p + ( 1 − λ ) b p ) + ( 1 − s ) ( λ b p + ( 1 − λ ) a p ) ) 1 p − 1 ≤ ( b p ) 1 p − 1 = b 1 − p . (3.8)

Using above process with relation inequality (3.8), we obtain

| Φ π ( μ , p , λ , a , b ) | ≤ b p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] 1 − 1 q [ ∫ 0 1 | ( 1 − s ) μ − s μ | [ s | π ′ ( m ) | q + ( 1 − s ) | π ′ ( n ) | q ] d s ] 1 q = b p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] 1 − 1 q [ 1 μ + 1 ( 1 − 1 2 μ ) ] 1 q [ | π ′ ( m ) | q + | π ′ ( n ) | q ] 1 q = b p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] ⋅ [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q . (3.9)

Thus, the proof is completed.

Corollary 3.1 In Theorem 3.1, if we take q = 1 , then we have

| Φ π ( μ , p , λ , a , b ) | ≤ η 1 − p [ 1 μ + 1 ( 1 − 1 2 μ ) ] [ | π ′ ( m ) | + | π ′ ( n ) | ] , (3.10)

where η is define by (3.2).

Corollary 3.2 In Theorem 3.1, if we choose p = 1 , then we obtain Theorem 2.3 in [19].

Corollary 3.3 In Theorem 3.1, if we put λ = 0 , then we get

− p [ π ( a ) + π ( b ) ] b p − a p + Γ ( 1 + μ ) p μ + 1 ( b p − a p ) μ + 1 [ I p a + μ π ( b ) + I p b − μ π ( a ) ] ≤ η p − 1 [ 2 μ + 1 ( 1 − 1 2 μ ) ] [ | π ′ ( a ) | q + | π ′ ( b ) | q 2 ] 1 q , (3.11)

where η is defined by (3.2).

Now, we prepare to introduce the second theorem as follows.

Theorem 3.2 Let p , μ > 0 and π : [ a p , b p ] ⊂ ( 0, ∞ ) → ℝ be a differentiable mapping on ( a p , b p ) with 0 < a < b such that π ′ ∈ L 1 ( [ a , b ] ) . If | π ′ | q is

p-convex on [ a p , b p ] for q > 1 and give constant r > 1 such that 1 r + 1 q = 1 ,

then the following inequality holds:

| Φ π ( μ , p , λ , a , b ) | ≤ η 1 − p [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q . (3.12)

where η is defined by (3.2) and λ ∈ ( 0,1 ] \ 1 2 .

Proof. Case 1: p ≥ 1 . Continuing from inequality (3.5), and using Hölder inequality, we obtain

| Φ π ( μ , p , λ , a , b ) | ≤ a 1 − p [ ∫ 0 1 | ( 1 − s ) μ − s μ | | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | d s ] ≤ a 1 − p [ ∫ 0 1 | ( 1 − s ) μ − s μ | r d s ] 1 r [ ∫ 0 1 | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ] 1 q = a 1 − p ( ∫ 0 1 2 [ ( 1 − s ) μ − s μ ] r d s + ∫ 1 2 1 [ s μ − ( 1 − s ) μ ] r d s ) 1 r         ⋅ [ ∫ 0 1 | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ] 1 q . (3.13)

Since | π ′ | q is p-convex, we get

∫ 0 1 | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ≤ ∫ 0 1 [ s | π ′ ( m ) | q + ( 1 − s ) | π ′ ( n ) | q ] d s = | π ′ ( m ) | q + | π ′ ( n ) | q 2 . (3.14)

By calculation, we have

( ∫ 0 1 2 [ ( 1 − s ) μ − s μ ] r d s + ∫ 1 2 1 [ s μ − ( 1 − s ) μ ] r d s ) 1 r = [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r . (3.15)

Using inequality (3.14) and (3.15) in inequality (3.13), we deduce

| Φ π ( μ , p , λ , a , b ) | ≤ a 1 − p [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q . (3.16)

Case 2: 0 < p < 1 . By utilizing the above process with relation inequality (3.8), we obtain

| Φ π ( μ , p , λ , a , b ) | ≤ b 1 − p [ ∫ 0 1 | ( 1 − s ) μ − s μ | r d s ] 1 r [ ∫ 0 1 | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ] 1 q ≤ b 1 − p [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q . (3.17)

Thus, the proof is completed.

Corollary 3.4 In Theorem 3.2, if we take p = 1 , then we have Theorem 2.6 in [19].

Next, we will use the well-known Young’s inequality

X Y ≤ X r r + Y q q , ∀ X , Y ≥ 0 , r ⋅ q > 1 , 1 r + 1 q = 1 (3.18)

in Theorem 3.2 to get the following two corollaries.

Corollary 3.5 Under all assumptions of Theorem 3.2, we deduce

| Φ π ( μ , p , λ , a , b ) | ≤ η 1 − p [ 2 r μ + 1 ( 1 − 1 2 μ r ) r + | π ′ ( m ) | q + | π ′ ( n ) | q 2 q ] , (3.19)

where η is defined by (3.2).

Corollary 3.6 Under the conditions of Theorem 3.2, if we take λ = 0 , then we have

− p [ π ( a ) + π ( b ) ] b p − a p + Γ ( 1 + μ ) p μ + 1 ( b p − a p ) μ + 1 [ I p a + μ π ( b ) + I p b − μ π ( a ) ] ≤ η 1 − p [ 2 r μ + 1 ( 1 − 1 2 μ r ) r + | π ′ ( a ) | q + | π ′ ( b ) | q 2 q ] , (3.20)

where η is defined by (3.2).

We will apply the following special functions in the next theorem.

1) The beta function,

β ( x , y ) = ∫ 0 1     s x − 1 ( 1 − s ) y − 1 d s ,   x , y > 0. (3.21)

2) The hypergeometric function,

2 F 1 ( a , b ; c ; z ) = 1 β ( b , c − b ) ∫ 0 1     s b − 1 ( 1 − s ) c − b − 1 ( 1 − z s ) − a d s ,     c > b > 0 ,   | z | < 1. (3.22)

Theorem 3.3 Let p , μ > 0 and π : [ a p , b p ] ⊂ ( 0, ∞ ) → ℝ be a differentiable mapping on ( a p , b p ) with 0 < a < b such that π ′ ∈ L 1 ( [ a , b ] ) . If | π ′ | q is

p-convex on [ a p , b p ] for q > 1 and give constant r > 1 such that 1 r + 1 q = 1 ,

then the following inequality exists:

| Φ π ( μ , p , λ , a , b ) | ≤ [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r { m ( 1 − p ) q 2 ⋅ F 2 1 ( q − q p ,1 : 3 ; m p − n p m p ) | π ′ ( m ) | q     + m ( 1 − p ) q 2 [ 2 ⋅ F 2 1 ( q − q p ,1 : 2 ; m p − n p m p ) − F 2 1 ( q − q p ,1 : 3 ; m p − n p m p ) ] | π ′ ( n ) | q } 1 q . (3.23)

Proof. Applying Lemma 2.1, Hölder inequality and the p-convexity of | π ′ | q , one has

| Φ π ( μ , p , λ , a , b ) | ≤ ∫ 0 1 | ( 1 − s ) μ − s μ | | ( s m p + ( 1 − s ) n p ) 1 p − 1 | | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) d s | ≤ [ ∫ 0 1 | ( 1 − s ) μ − s μ | r d s ] 1 r [ ∫ 0 1 ( s m p + ( 1 − s ) n p ) ( 1 p − 1 ) q | π ′ ( [ s m p + ( 1 − s ) n p ] 1 p ) | q d s ] 1 q ≤ [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r [ ∫ 0 1 ( s m p + ( 1 − s ) n p ) ( 1 p − 1 ) q [ s | π ′ ( m ) | q + ( 1 − s ) | π ′ ( n ) | q ] d s ] 1 q = [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r { | π ′ ( m ) | q ∫ 0 1     s ( s m p + ( 1 − s ) n p ) ( 1 p − 1 ) q d s         + | π ′ ( n ) | q ∫ 0 1 ( 1 − s ) ( s m p + ( 1 − s ) n p ) ( 1 p − 1 ) q d s } 1 q . (3.24)

By calculation, we obtain

∫ 0 1     s ( s m p + ( 1 − s ) n p ) ( 1 p − 1 ) q d s = m ( 1 − p ) q 2 ⋅ F 2 1 ( q − q p ,1 ; 3 ; m p − n p m p ) (3.25)

and

∫ 0 1 ( 1 − s ) ( s m p + ( 1 − s ) n p ) ( 1 p − 1 ) q d s = m ( 1 − p ) q 2 [ 2 ⋅ F 2 1 ( q − q p ,1 ; 2 ; m p − n p m p ) − F 2 1 ( q − q p ,1 ; 3 ; m p − n p m p ) ] . (3.26)

Utilizing inequality (3.25) and (3.26) in inequality (3.24), we can obtain desired inequality (3.23). The proof is completed.

Corollary 3.7 Under the conditions of Theorem 3.3, if we take λ = 0 , then we obtain

− p [ π ( a ) + π ( b ) ] b p − a p + Γ ( 1 + μ ) p μ + 1 ( b p − a p ) μ + 1 [ I p a + μ π ( b ) + I p b − μ π ( a ) ] ≤ [ 2 r μ + 1 ( 1 − 1 2 μ r ) ] 1 r { b ( 1 − p ) q 2 ⋅ F 2 1 ( q − q p ,1 : 3 ; b p − a p b p ) | π ′ ( b ) | q       + b ( 1 − p ) q 2 [ 2 ⋅ F 2 1 ( q − q p ,1 : 2 ; b p − n p a p ) − F 2 1 ( q − q p ,1 : 3 ; b p − a p b p ) ] | π ′ ( a ) | q } 1 q . (3.27)

In this part, we obtain two examples to illustrate our main results.

Example 4.1 Let π ( x ) = 1 2 x 2 , x ∈ ( 0, ∞ ) , then | π ′ ( x ) | = x is p-convex for

p < 1 . Choosing λ = 0 , μ = 1.5 , p = 0.5 , q = 1 , a = 4 and b = 9 , then all the assumptions in Theorem 3.1 are satisfied. We have

| Φ π ( μ , p , λ , a , b ) | = − p [ π ( n ) + π ( m ) ] ( 1 − 2 λ ) ( b p − a p ) + Γ ( 1 + μ ) p μ + 1 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 [ I p n + μ π ( m ) + I p m − μ π ( n ) ] = | − 0.5 [ π ( 4 ) + π ( 9 ) ] + Γ ( 2.5 ) 0.5 2.5 [ I 0.5 4 + 1.5 π ( 9 ) + I 0.5 9 − 1.5 π ( 4 ) ] | ≈ 8.9344 ≤ b 1 − p [ 2 μ + 1 ( 1 − 1 2 μ ) ] [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q = 9 ( 1 − 0.5 ) 2 1.5 + 1 ( 1 − 1 2 1.5 ) ( π ′ ( 4 ) + π ′ ( 9 ) 2 ) ≈ 10.0846. (4.1)

This proves that the described result in Theorem 3.1 is correct.

Example 4.2 Let π ( x ) = 1 1 − p x 1 − p , x ∈ ( 0, ∞ ) , then | π ′ ( x ) | = x − p is

p-convex for p > 1 . Taking λ = 0 , μ = 1.5 , p = 2 , q = 1 , a = 1 and b = 3 , then all the assumptions in Theorem 3.1 are satisfied. We have

| Φ π ( μ , p , λ , a , b ) | = − p [ π ( n ) + π ( m ) ] ( 1 − 2 λ ) ( b p − a p ) + Γ ( 1 + μ ) p μ + 1 ( 1 − 2 λ ) μ + 1 ( b p − a p ) μ + 1 [ I p n + μ π ( m ) + I p m − μ π ( n ) ] = | − 2 [ π ( 1 ) + π ( 3 ) ] 8 + Γ ( 2.5 ) 2 2.5 8 2.5 [ I 2 1 + 1.5 π ( 3 ) + I 2 3 − 1.5 π ( 1 ) ] | ≈ 0.0672 ≤ a 1 − p [ 2 μ + 1 ( 1 − 1 2 μ ) ] [ | π ′ ( m ) | q + | π ′ ( n ) | q 2 ] 1 q = 1 ( 1 − 2 ) 2 1.5 + 1 ( 1 − 1 2 1.5 ) ( π ′ ( 1 ) + π ′ ( 3 ) 2 ) ≈ 0.17146 (4.2)

This proves that the described result in Theorem 3.1 is correct.

In this paper, we assume that the absolute value of the derivative of the considered function π ′ is p-convex to obtain some inequalities for Katugampola fractional integrals. More new results can be derived by taking special parameter values for λ , μ and p. We emphasize that certain results proved in this article generalize the ones obtained by Sarikaya and Budak [19] and Topul et al. [11].

The author was supported in general project of education department of Hunan province (No. 19C0359).

The author declares no conflicts of interest regarding the publication of this paper.

Wang, H. (2020) Generalized Hermite-Hadamard Type Inequalities Related to Katugampola Fractional Integrals. Open Access Library Journal, 7: e6793. https://doi.org/10.4236/oalib.1106793