Previous Article in Journal
Drivers’ Skills and Behavior vs. Traffic at Intersections
Previous Article in Special Issue
Integral Representation for the Solutions of Autonomous Linear Neutral Fractional Systems with Distributed Delay

Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

# Some Fractional Dynamic Inequalities of Hardy’s Type via Conformable Calculus

by
Samir Saker
1,*,
Mohammed Kenawy
2,
3 and
Mohammed Zakarya
4,5
1
Department of Mathematics, Faculty of Science, Mansoura University, Mansoura 35516, Egypt
2
Department of Mathematics, Faculty of Science, Fayoum University, Fayoum 63514, Egypt
3
Department of Mathematical Science, College of Science, Princess Nourah bint Abdulrahman University, P.O. Box 105862, Riyadh, Saudi 11656, Arabia
4
Department of Mathematics, College of Science, King Khalid University, P.O. Box 9004, Abha 61413, Saudi Arabia
5
Department of Mathematics, Faculty of Science, Al-Azhar University, Assiut 71524, Egypt
*
Author to whom correspondence should be addressed.
Mathematics 2020, 8(3), 434; https://doi.org/10.3390/math8030434
Submission received: 12 February 2020 / Revised: 8 March 2020 / Accepted: 10 March 2020 / Published: 16 March 2020

## Abstract

:
In this article, we prove some new fractional dynamic inequalities on time scales via conformable calculus. By using chain rule and Hölder’s inequality on timescales we establish the main results. When $α = 1$ we obtain some well-known time-scale inequalities due to Hardy, Copson, Bennett and Leindler inequalities.
MSC:
26A15; 26D10; 26D15; 39A13; 34A40; 34N05

In 1920, Hardy [1] established the inequality
$∑ n = 1 ∞ ( 1 n ∑ i = 1 n w ( i ) ) k ≤ ( k k − 1 ) k ∑ n = 1 ∞ w k ( n ) , k > 1 .$
where $w ( n )$ is a positive sequence defined for all $n ≥ 1$. After that, Hardy [2], by using the calculus of variations, proved the continuous inequality of (1) which has the form
$∫ 0 ∞ ( 1 x ∫ 0 x g ( s ) d s ) k d x ≤ ( k k − 1 ) k ∫ 0 ∞ g k ( x ) d x ,$
for a given positive function g, which is integrable over $( 0 , x ) ,$ and $g k$ is convergent and integrable over $( 0 , ∞ )$ and $k > 1$. In (1) and (2), $( k / ( k − 1 ) ) k$ is a sharp constant. As a generalization of (2), Hardy [3] showed that when $k > 1$, then
$∫ 0 ∞ x − h ( ∫ 0 x g ( s ) d s ) k d x ≤ ( k h − 1 ) k ∫ 0 ∞ x k − h g k ( x ) d x , for h > 1 ,$
and
$∫ 0 ∞ x − h ( ∫ x ∞ g ( s ) d s ) k d x ≤ ( k 1 − h ) k ∫ 0 ∞ x k − h g k ( x ) d x , for h < 1 .$
The constants $( k / ( h − 1 ) ) k$ and $( k / ( 1 − h ) ) k$ in (3) and (4) are the best possible. Copson [4] demonstrated that if $g ( x ) > 0$, $k > 1$ and $g k ( x )$ is integrable on the interval $( 0 , ∞ )$, then
$∫ x ∞ ( g ( s ) s ) d s ,$
converges for $x > 0$ and
$∫ 0 ∞ ( ∫ x ∞ g ( s ) s d s ) k d x ≤ k k ∫ 0 ∞ g k ( x ) d x ,$
where $k k$ is the best possible constant. Some of the generalizations of the discrete Hardy inequality (1) and the discrete version of (5) and its extensions are due to Leindler, we refer to the papers the papers [5,6,7,8]. For example, Leindler in [5] proved that if $p > 1 ,$ $λ ( n )$, $g ( n ) > 0$, then
$∑ n = 1 ∞ λ ( n ) ( ∑ s = 1 n g ( s ) ) p ≤ p p ∑ n = 1 ∞ λ 1 − p ( n ) ( ∑ s = n ∞ λ ( s ) ) p g p ( n ) ,$
and
$∑ n = 1 ∞ λ ( n ) ( ∑ k = n ∞ g ( k ) ) p ≤ p p ∑ n = 1 ∞ λ 1 − p ( n ) ( ∑ k = 1 n λ ( k ) ) p g p ( n ) .$
The converses of (6) and (7) are proved by Leindler in [6]. He proved that if $0 < p ≤ 1$, then
$∑ n = 1 ∞ λ ( n ) ( ∑ k = 1 n g ( k ) ) p ≥ p p ∑ n = 1 ∞ λ 1 − p ( n ) ( ∑ k = n ∞ λ ( k ) ) p g p ( n ) ,$
and
$∑ n = 1 ∞ λ ( n ) ( ∑ k = n ∞ g ( k ) ) p ≥ p p ∑ n = 1 ∞ λ 1 − p ( n ) ( ∑ k = 1 n λ ( p ) ) p g p ( n ) .$
For more generalization Copson in [9] showed that if $k > 1$, $λ ( j ) ≥ 0$, $w ( j ) ≥ 0$, $∀ j ≥ 1$, $Ω ( m ) = ∑ j = 1 m λ ( j )$, and $h > 1$, then
$∑ m = 1 ∞ λ ( m ) Ω h ( m ) ( ∑ j = 1 m w ( j ) λ ( j ) ) k ≤ ( k h − 1 ) k ∑ m = 1 ∞ λ ( m ) Ω k − h ( m ) w k ( m ) ,$
and if $0 ≤ h < 1$ and $k > 1$, then
$∑ m = 1 ∞ λ ( m ) Ω h ( m ) ( ∑ j = m ∞ w ( j ) λ ( j ) ) k ≤ ( k 1 − h ) k ∑ m = 1 ∞ λ ( m ) Ω k − h ( m ) w k ( m ) .$
The integral versions of the inequalities (10) and (11) was proved by Copson in [10] (Theorems 1 and 3). In particular, he proved that if $k ≥ 1$, $h > 1$, and $Ω ( s ) = ∫ 0 s λ ( t ) d t$, then
$∫ 0 ∞ λ ( s ) Ω h ( s ) Φ k ( s ) d s ≤ ( k h − 1 ) k ∫ 0 ∞ λ ( s ) Ω h − k ( s ) g k ( s ) d s ,$
where $Φ ( s ) = ∫ 0 s λ ( t ) g ( t ) d t$, and if $k > 1$, $0 ≤ h < 1$, then
$∫ 0 ∞ λ ( s ) Ω h ( s ) Φ k ( s ) d s ≤ ( k 1 − h ) k ∫ 0 ∞ λ ( s ) Ω h − k ( s ) g k ( s ) d s ,$
where $Φ ( s ) = ∫ s ∞ λ ( t ) g ( t ) d t .$ Leindler in [5] and Bennett in [11] presented interesting different inequalities. Leindler established that if $k > 1$, $Ω * ( m ) = ∑ j = m ∞ λ ( j ) < ∞$, and $0 ≤ h < 1$, then
$∑ m = 1 ∞ λ ( m ) ( Ω * ( m ) ) h ( ∑ j = 1 m w ( j ) λ ( j ) ) k ≤ ( k 1 − h ) k ∑ m = 1 ∞ λ ( m ) ( Ω * ( m ) ) k − h w k ( m ) ,$
and Bennett in [11] showed that if $1 < h ≤ k$, then
$∑ m = 1 ∞ λ ( m ) ( Ω * ( m ) ) h ( ∑ j = m ∞ w ( j ) λ ( j ) ) k ≤ ( k h − 1 ) k ∑ m = 1 ∞ λ ( m ) ( Ω * ( m ) ) k − h w k ( m ) .$
In last decades, studying the dynamic equations and inequalities on time scales become a main field in applied and pure mathematics, we refer to [12,13,14] and the references they are cited. In fact, the book [13] includes forms of the above inequalities on time-scale and their extensions. The timescales idea is returned to Stefan Hilger [15], who investigated the research of dynamic equations on timescales. The books by Bohner and Peterson in [16,17] summarized and organized most timescales calculus. The three most common timescales calculuses are difference, differential, and quantum calculus (see [18]), i.e., at $T = N$, $T = R ,$ and $T = q N 0 = { q s : s ∈ N 0 }$ where $q > 1$.
In recent years, a lot of work has been published for fractional inequalities and the subject becomes an active field of research and several authors were interested in proving inequalities of fractional type by using the Riemann-Liouville and Caputo derivative (see [19,20,21]).
On the other hand, the authors in [22,23] introduced a new fractional calculus called the conformable calculus and gave a new definition of the derivative with the base properties of the calculus based on the new definition of derivative and integrals. By using conformable calculus, some authors have studied classical inequalities like Chebyshev’s inequality [24], Hermite-Hadamard’s inequality [25,26,27], Opial’s inequality [28,29] and Steffensen’s inequality [30].
The main question that arises now is: Is it possible to prove new fractional inequalities on timescales and give a unified approach of such studies? This in fact needs a new fractional calculus on timescales. Very recently Torres and others, in [31,32], combined a time scale calculus and conformable calculus and obtained the new fractional calculus on timescales. So, it is natural to look on new fractional inequalities on timescales and give an affirmative answer to the above question.
In particular, in this paper, we will prove the fractional forms of the classical Hardy, Bennett, Copson and Leindler inequalities. The paper is divided into two sections. Section 2 is an introduction of basics of fractional calculus on timescales and Section 3 contains the main results.

## 1. Preliminaries and Basic Lemmas

We present the fundamental results about the fractional timescales calculus. The results are adapted from [16,17,31,32]. A time-scale $T$ is non-empty closed subset of $R$ ($R$ is the real numbers). The operators of backward jump and forward jump express of the closest point $t ∈ T$ on the right and left of t is defined by, respectively:
$ρ ( t ) : = sup { s ∈ T : s < t } ,$
$σ ( t ) : = inf { s ∈ T : s > t } ,$
where $sup ϕ = inf T$ and $inf ϕ = sup T$ ($ϕ$ denotes the empty set), for any $t ∈$$T$ the notation $f σ ( t )$ refer to $f ( σ ( t ) ) ,$ i.e., $f σ = f ∘ σ$. The graininess function $μ$$: T → [ 0$, $∞ )$, defined by $μ ( t ) : = σ ( t ) − t$.
Definition 1.
The number$T α Δ ( f ) ( t )$(provided it exists) of the function$f : T → R$, for$t > 0$and$α ∈ ( 0 , 1 ]$is the number which has the property that for any$ϵ > 0 ,$there exists a neighborhood U of t S. T.
$| [ f σ ( t ) − f ( s ) ] t 1 − α − T α Δ ( f ( t ) ) ( σ ( t ) − s ) | ≤ ϵ | σ ( t ) − s ) | , f o r a l l t ∈ U$
$T α Δ ( f ( t ) )$is called the conformable$α —$fractional derivative of function f of order α at t, for conformable fractional derivative on$T$at 0, we define it with$T α Δ ( f ( 0 ) = lim t → 0 + T α Δ ( f ( t ) ) .$
The conformable fractional derivative has the following properties.
Theorem 1.
Let$v ,$$u : T → R$are conformable fractional derivative from order$α ∈ ( 0 , 1 ]$, then the following properties are hold:
(i)
The$v + u : T → R$is conformable fractional derivative and
$T α Δ ( v + u ) = T α Δ ( v ) + T α Δ ( u ) .$
(ii)
Fora all$k ∈ R$, then$k v : T → R$is$α —$fractional differentiable and
$T α Δ ( k v ) = k T α Δ ( v ) .$
(iii)
If v and u are$α —$fractional differentiable, we have$v u : T → R$is$α —$fractional differentiable and
$T α Δ ( v u ) = T α Δ ( v ) u + ( v ∘ σ ) T α Δ ( u ) = T α Δ ( v ) ( u ∘ σ ) + v T α Δ ( u ) .$
(iv)
If v is$α —$fractional differentiable, then$1 / v$is$α —$fractional differentiable with
$T ( 1 v ) = − T α Δ ( v ) v ( v ∘ σ ) .$
(v)
If v and u are$α —$fractional differentiable, then$v / u$ is $α —$fractional differentiable with
$T α Δ ( v / u ) = T α Δ ( v ) u − v T α Δ ( u ) u ( u ∘ σ ) ,$
valid $∀ t ∈ T k$, where $u ( t ) ( u ( σ ( t ) ) ≠ 0 .$
Lemma 1.
Let$v : T → R$is continuous and$α —$fractional differentiable at$t ∈ T$for$α ∈ ( 0 , 1 ]$, and$u : R → R$is continuous and differentiable. Then there exists d$∈ [ t , σ ( t ) ]$with
$T α Δ ( u ∘ v ) ( t ) = u ′ ( v ( d ) ) T α Δ ( v ( t ) ) .$
Lemma 2.
Let$u : R → R$is continuously differentiable,$α ∈ ( 0 , 1 ]$, and$v : T → R$be$α —$fractional differentiable. Then$( u ∘ v ) : T → R$is$α —$fractional differentiable and we have
$T α Δ ( u ∘ v ) ( s ) = ( ∫ 0 1 u ′ ( v ( s ) + h μ ( s ) s α − 1 T α Δ ( v ( s ) ) ) d h ) T α Δ ( v ( s ) ) .$
Definition 2.
Let$0 < α ≤ 1 ,$the$α —$fractional integral of$f$, is defined as
$∫ f ( s ) Δ α s = ∫ f ( s ) s α − 1 Δ s .$
The conformable fractional integral satisfying the next properties.
Theorem 2.
Assume a, b,$c ∈ T$,$λ ∈ R$. Let u,$v : T → R$. Then
(i)
$∫ a b [ v ( s ) + u ( s ) ] Δ α s =$$∫ a b v ( s ) Δ α s + ∫ a b u ( s ) Δ α s .$
(ii)
$∫ a b λ v ( s ) Δ α s$$= λ ∫ a b v ( s ) Δ α s .$
(iii)
$∫ a b v ( s ) Δ α s$$= − ∫ b a v ( s ) Δ α s .$
(iv)
$∫ a b v ( s ) Δ α s$$= ∫ a c v ( s ) Δ α s + ∫ c b v ( s ) Δ α s .$
(v)
$∫ a a v ( s ) Δ α s = 0 .$
Lemma 3.
Assume$T$be a time-scale, a,$b ∈ T$where$b > a$. Let u, v are conformable$α$—fractional differentiable,$α ∈ ( 0 , 1 ]$. Then the formula of integration by parts is given by
$∫ a b v ( s ) T α Δ u ( s ) Δ α s = [ v ( s ) u ( s ) ] a b − ∫ a b u σ ( s ) T α Δ v ( s ) Δ α s .$
Lemma 4.
Assume$T$be a time-scale, a,$b ∈ T$and$α ∈ ( 0$,$1 ]$. Let u,$v : T → R$. Then
$∫ a b | v ( s ) u ( s ) | Δ α s ≤ [ ∫ a b | v ( s ) | k Δ α s ] 1 k [ ∫ a b | u ( s ) | l Δ α s ] 1 l ,$
where$k > 1$and$1 / k + 1 / l = 1 .$

## 2. Main Results

Throughout the paper, we will assume that the functions are nonnegative on $[ a , ∞ ) T$ and its integrals exist and are finite. We start with the fractional time-scale inequality of Copson’s type.
Theorem 3.
Assume$1 < c <$k, define
$Φ ( x ) : = ∫ a x λ ( s ) Δ α s and Ω ( x ) : = ∫ a x λ ( s ) g ( s ) Δ α s .$
If
$Ω ( ∞ ) < ∞ , and ∫ a ∞ λ ( s ) ( Φ σ ( s ) ) c − α + 1 Δ α s < ∞ ,$
then
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x ≤ ( k c − α ) k ∫ a ∞ λ ( x ) Φ k ( α − c ) ( x ) ( Φ σ ( x ) ) ( c − α + 1 ) ( 1 − k ) g k ( x ) Δ α x .$
Proof.
By employing the formula of integration by parts (20) on the term
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x ,$
with $u σ ( x ) = ( Ω σ ( x ) ) k$ and $x α Δ v ( x ) = λ ( x ) ( Φ σ ( x ) ) c − α + 1$, we have that
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x = v ( x ) Ω k ( x ) | a ∞ + ∫ a ∞ − v ( x ) x α Δ ( Ω k ( x ) ) Δ α x ,$
where
$− v ( x ) = ∫ x ∞ λ ( s ) ( Φ σ ( s ) ) c − α + 1 Δ α s = ∫ x ∞ x α Δ Φ ( s ) ( Φ σ ( s ) ) α − c − 1 Δ α s .$
By using the chain rule (18), we obtain that
$− x α Δ ( Φ α − c ( x ) ) = − ( α − c ) Φ α − c − 1 ( d ) x α Δ Φ ( x ) , where d ∈ [ x , σ ( x ) ] = ( c − α ) x α Δ Φ ( x ) Φ c − α + 1 ( d ) ≥ ( c − α ) x α Δ ( Φ ( x ) ) ( Φ σ ( x ) ) c − α + 1 .$
Then we have
$x α Δ ( Φ ( x ) ) ( Φ σ ( x ) ) α − c − 1 ≤ − 1 c − α x α Δ Φ α − c ( x ) ,$
and thus
$− v ( x ) = ∫ x ∞ λ ( s ) ( Φ σ ( s ) ) c − α + 1 Δ α s ≤ − 1 c − α ∫ x ∞ x α Δ Φ α − c − 1 ( s ) Δ α s ≤ Φ α − c ( x ) c − α .$
Again, by using the chain rule (18) to calculate
$x α Δ ( Ω k ( x ) ) = k Ω k − 1 ( d ) x α Δ ( Ω ( x ) ) , where d ∈ [ x , σ ( x ) ] ,$
at $x α Δ ( Ω ( x ) ) = λ ( x ) g ( x ) ≥ 0$ and $d ≤ σ ( x )$, we get that
$x α Δ ( Ω k ( x ) ) ≤ k λ ( x ) g ( x ) ( Ω σ ( x ) ) k − 1 .$
Since $Ω ( a ) = 0$, $v ( ∞ ) = 0$ and from (24), (25) and (23) we have
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x ≤ k c − α ∫ a ∞ Φ α − c ( x ) λ ( x ) g ( x ) ( Ω σ ( x ) ) k − 1 Δ α x ,$
which reformulated as
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x = k c − α ∫ a ∞ λ ( x ) Φ α − c ( x ) g ( x ) ( λ ( x ) ( Φ σ ( x ) ) α − c − 1 ) k − 1 k ( λ ( x ) ( Ω σ ( x ) ) k ( Φ σ ( x ) ) c − α + 1 ) k − 1 k Δ α x .$
Using Hölder’s inequality (21) on
$∫ a ∞ λ ( x ) Φ α − c ( x ) g ( x ) ( λ ( x ) ( Φ σ ( x ) ) α − c − 1 ) k − 1 k ( λ ( x ) ( Ω σ ( x ) ) k ( Φ σ ( x ) ) c − α + 1 ) k − 1 k Δ α x$
with indices k and $k / ( k − 1 )$, we have
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x ≤ k c − α [ ∫ a ∞ [ λ ( x ) Φ α − c ( x ) g ( x ) [ λ ( x ) ( Φ σ ( x ) ) α − c − 1 ] k − 1 k ] k Δ α x ] 1 k × [ ∫ a ∞ [ [ λ ( x ) ( Ω σ ( x ) ) k ( Φ σ ( x ) ) c − α + 1 ] k − 1 k ] k k − 1 Δ α x ] k − 1 k ,$
then
$[ ∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x ] 1 k ≤ k c − α [ ∫ a ∞ λ ( x ) Φ k ( α − c ) ( x ) g k ( x ) ( Φ σ ( x ) ) ( c − α + 1 ) ( 1 − k ) Δ α x ] 1 k .$
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 ( Ω σ ( x ) ) k Δ α x ≤ ( k c − α ) k ∫ a ∞ λ ( x ) Φ k ( α − c ) ( x ) g k ( x ) ( Φ σ ( x ) ) ( c − α + 1 ) ( 1 − k ) Δ α x ,$
that is the desired inequality (22). The proof is complete. □
Corollary 1.
At$α = 1$in Theorem 3, we obtain the inequality
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c ( Ω σ ( x ) ) k Δ x ≤ ( k c − 1 ) k ∫ a ∞ Φ k ( 1 − c ) ( x ) ( Φ σ ( x ) ) c ( 1 − k ) λ ( x ) g k ( x ) Δ x .$
that is the timescales version of inequality (2.8) in [33].
Corollary 2.
At$α = 1$, and$T = R$$( Φ σ ( x ) = Φ ( x ) )$in Theorem 3, we obtain the integral inequality
$∫ a ∞ λ ( x ) Φ c ( x ) ( ∫ a x λ ( s ) g ( s ) d s ) k d x ≤ ( k c − 1 ) k ∫ a ∞ Φ k − c ( x ) λ ( x ) g k ( x ) d x ,$
which is of Copson type.
Corollary 3.
At$α = 1$, T$= R$,$λ ( x ) = 1$and$a = 0$,$( Φ ( x ) = ∫ 0 x λ ( s ) d s = x )$in Theorem 3, we have Hardy-Littlewood integral inequality (3)
$∫ 0 ∞ 1 x c ( ∫ 0 x g ( s ) d s ) k d x ≤ ( k c − 1 ) k ∫ 0 ∞ 1 x c − k g k ( x ) d x .$
Also, if$c = k$, we obtain the standard Hardy inequality (2)
$∫ 0 ∞ 1 x k ( ∫ 0 x g ( s ) d s ) k d x ≤ ( k k − 1 ) k ∫ 0 ∞ g k ( x ) d x .$
Theorem 4.
Let$0 ≤ c < 1$and$k > 1$. Define
$Φ ( x ) : = ∫ a x λ ( s ) Δ α s and Ω ( x ) : = ∫ x ∞ λ ( s ) g ( s ) Δ α s .$
If
$Ω ( a ) < ∞ , and ∫ a ∞ λ ( s ) ( Φ σ ( s ) ) c − α + 1 Δ α s < ∞ ,$
then
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x ≤ ( k α − c ) k ∫ a ∞ ( Φ σ ( x ) ) k − c + α − 1 λ ( x ) g k ( x ) Δ α x .$
Proof.
By using the formula of integration by parts (20) on
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x ,$
with $v ( x ) = Ω k ( x )$ and $x α Δ u ( x ) = λ ( x ) ( Φ σ ( x ) ) c − α + 1$, we have
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x = u ( x ) Ω k ( x ) | a ∞ + ∫ a ∞ u σ ( x ) x α Δ ( − Ω k ( x ) ) Δ α x ,$
where
$u ( x ) = ∫ a x λ ( s ) ( Φ σ ( s ) ) c − α + 1 Δ α s = ∫ a x x α Δ Φ ( s ) ( Φ σ ( s ) ) α − c − 1 Δ α s .$
By using chain rule (18), then for $d ∈ [ x , σ ( x ) ] ,$ we get that
$x α Δ ( Φ α − c ( x ) ) = ( α − c ) Φ α − c − 1 ( d ) x α Δ ( Φ ( x ) ) = ( α − c ) x α Δ ( Φ ( x ) ) Φ c − α + 1 ( d ) ≥ ( α − c ) x α Δ ( Φ ( x ) ) ( Φ σ ( x ) ) c − α + 1 .$
So
$x α Δ ( Φ ( x ) ) ( Φ σ ( x ) ) α − c − 1 ≤ 1 α − c x α Δ ( Φ α − c ( x ) ) ,$
and then,
$u σ ( x ) = ∫ a σ ( x ) x α Δ ( Φ ( s ) ) ( Φ σ ( s ) ) α − c − 1 Δ α s ≤ 1 α − c ∫ a σ ( x ) x α Δ ( Φ α − c ( s ) ) Δ α s ≤ ( Φ σ ( x ) ) α − c α − c .$
Again, by using chain rule (18), we obtain
$− x α Δ ( Ω k ( x ) ) = − k Ω k − 1 ( d ) x α Δ ( Ω ( x ) ) , where d ∈ [ x , σ ( x ) ] ,$
since $x α Δ ( Ω ( x ) ) = − λ ( x ) g ( x ) ≥ 0$ and $d ≥ x ,$ then
$− x α Δ ( Ω k ( x ) ) ≤ k λ ( x ) g ( x ) Ω k − 1 ( x ) .$
Using $Φ ( a ) = 0$, $Ω ( ∞ ) = 0$ and (28), (29) and (27), we have that
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x ≤ k α − c ∫ a ∞ ( Φ σ ( x ) ) α − c λ ( x ) g ( x ) Ω k − 1 ( x ) Δ α x ,$
which reformulated as
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x = k α − c ∫ a ∞ ( Φ σ ( x ) ) α − c λ ( x ) g ( x ) ( λ ( x ) ( Φ σ ( x ) ) α − c − 1 ) k − 1 k ( λ ( x ) Ω k ( x ) ( Φ σ ( x ) ) c − α + 1 ) k − 1 k Δ α x .$
By employing Hölder’s inequality (21) on
$∫ a ∞ ( Φ σ ( x ) ) α − c λ ( x ) g ( x ) ( λ ( x ) ( Φ σ ( x ) ) α − c − 1 ) k − 1 k ( λ ( x ) Ω k ( x ) ( Φ σ ( x ) ) c − α + 1 ) k − 1 k Δ α x$
with indices k and $k / ( k − 1 )$, we have
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x ≤ k α − c [ ∫ a ∞ [ ( Φ σ ( x ) ) 1 − c λ ( x ) g ( x ) [ λ ( x ) ( Φ σ ( x ) ) α − c − 1 ] k − 1 k ] k Δ α x ] 1 k × [ ∫ a ∞ [ [ λ ( x ) Ω k ( x ) ( Φ σ ( x ) ) c − α + 1 ] k − 1 k ] k k − 1 Δ α x ] k − 1 k ,$
then we have
$[ ∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c − α + 1 Ω k ( x ) Δ α x ] 1 k ≤ k α − c [ ∫ a ∞ λ ( x ) g k ( x ) ( Φ σ ( x ) ) c − k − α + 1 Δ α x ] 1 k .$
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c Ω k ( x ) Δ α x ≤ ( k α − c ) k ∫ a ∞ ( Φ σ ( x ) ) k − c + α − 1 λ ( x ) g k ( x ) Δ α x ,$
that is the desired inequality (26). The proof is complete. □
Corollary 4.
At$α = 1$in Theorem 4, then
$∫ a ∞ λ ( x ) ( Φ σ ( x ) ) c Ω k ( x ) Δ x ≤ ( k 1 − c ) k ∫ a ∞ ( Φ σ ( x ) ) k − c λ ( x ) g k ( x ) Δ x .$
which is the timescales version inequality (2.22) in [33].
Corollary 5.
At$α = 1$, and T$= R$in Theorem 4, we obtain the next integral inequality
$∫ a ∞ λ ( x ) Φ c ( x ) ( ∫ x ∞ λ ( s ) g ( s ) d s ) k d x ≤ ( k 1 − c ) k ∫ a ∞ Φ k − c ( x ) λ ( x ) g k ( x ) d x ,$
which considered an extension of Hardy’s inequality (4) as in the following corollary.
Corollary 6.
At$α = 1$, T$= R$,$λ ( x ) = 1$and$a = 0$in Theorem 4, we have Hardy-Littlewood integral inequality (4)
$∫ 0 ∞ 1 x c ( ∫ x ∞ g ( s ) d s ) k d x ≤ ( k 1 − c ) k ∫ 0 ∞ 1 x c − k g k ( x ) d x .$
A generalization of Leindler’s inequality (14) on fractional time scales will be proved in the next theorem.
Theorem 5.
Assume$0 ≤ c <$$1 < k$, define
$Φ ( x ) : = ∫ x ∞ λ ( s ) Δ α s and Ω ( x ) : = ∫ a x λ ( s ) g ( s ) Δ α s .$
If
$Ω ( ∞ ) < ∞ , and ∫ a ∞ λ ( s ) Φ c − α + 1 ( s ) Δ α s < ∞ ,$
then
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ ( k α − c ) k ∫ a ∞ Φ k − c + α − 1 ( x ) λ ( x ) g k ( x ) Δ α x .$
Proof.
Using the formula of integration by parts (20) on
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) Ω k ( x ) Δ α x ,$
with $u σ ( x ) = ( Ω σ ( x ) ) k$ and $x α Δ v ( x ) = λ ( x ) Φ c − α + 1 ( x )$, we have
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x = v ( x ) Ω k ( x ) | a ∞ + ∫ a ∞ − v ( x ) x α Δ ( Ω k ( x ) ) Δ α x ,$
where
$v ( x ) = − ∫ x ∞ λ ( s ) Φ c − α + 1 ( s ) Δ α s .$
By chain rule (18), we see for $d ∈ [ x , σ ( x ) ]$ that
$− x α Δ Φ α − c ( x ) = − ( α − c ) Φ α − c − 1 ( d ) x α Δ ( Φ ( x ) ) = − ( α − c ) ( − λ ( x ) ) Φ c − α + 1 ( d ) ≥ ( α − c ) λ ( x ) Φ c − α + 1 ( x ) .$
Hence
$− v ( x ) = ∫ x ∞ λ ( s ) Φ c − α + 1 ( s ) Δ α s ≤ − 1 α − c ∫ x ∞ x α Δ Φ α − c ( s ) Δ α s ≤ Φ α − c ( x ) α − c ,$
from chain rule (18), we obtain
$x α Δ ( Ω k ( x ) ) = k Ω k − 1 ( d ) x α Δ ( Ω ( x ) ) , where d ∈ [ x , σ ( x ) ] ,$
since
$x α Δ ( Ω ( x ) ) = λ ( x ) g ( x ) ≥ 0 and d ≤ σ ( x ) ,$
we get
$x α Δ ( Ω k ( x ) ) ≤ k λ ( x ) g ( x ) ( Ω σ ( x ) ) k − 1 .$
Using $Ω ( a ) = 0$, $v ( ∞ ) = 0$ and (32), (33) and (31 ), we get that
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ k α − c ∫ a ∞ Φ α − c ( x ) λ ( x ) g ( x ) ( Ω σ ( x ) ) k − 1 Δ α x ,$
which reformulated as
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ k α − c ∫ a ∞ Φ ( c − α + 1 ) ( k − 1 k ) ( x ) λ k − 1 k ( x ) Φ c − α ( x ) λ ( x ) g ( x ) λ k − 1 k ( x ) ( Ω σ ( x ) ) k − 1 Φ ( c − α + 1 ) ( k − 1 k ) ( x ) Δ α x .$
Using Hölder’s inequality (21) on
$∫ a ∞ Φ ( c − α + 1 ) ( k − 1 k ) ( x ) λ k − 1 k ( x ) Φ c − α ( x ) λ ( x ) g ( x ) λ k − 1 k ( x ) ( Ω σ ( x ) ) k − 1 Φ ( c − α + 1 ) ( k − 1 k ) ( x ) Δ α x$
with indices k and $k / ( k − 1 )$, we have
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ k α − c [ ∫ a ∞ [ Φ ( c − α + 1 ) ( k − 1 k ) ( x ) λ k − 1 k ( x ) Φ c − α ( x ) λ ( x ) g ( x ) ] k Δ α x ] 1 k × [ ∫ a ∞ [ λ k − 1 k ( x ) ( Ω σ ( x ) ) k − 1 Φ ( c − α + 1 ) ) ( k − 1 k ) ( x ) ] k k − 1 Δ α x ] k − 1 k ,$
then
$[ ∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ] 1 k ≤ k α − c [ ∫ a ∞ Φ k − c + α − 1 ( x ) λ ( x ) g k ( x ) Δ α x ] 1 k .$
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ ( k α − c ) k ∫ a ∞ Φ k − c + α − 1 ( x ) λ ( x ) g k ( x ) Δ α x ,$
that is the desired inequality (30). The proof is complete. □
Corollary 7.
At$α = 1$in Theorem 5, we get
$∫ a ∞ λ ( x ) Φ c ( x ) ( Ω σ ( x ) ) k Δ x ≤ ( k 1 − c ) k ∫ a ∞ Φ k − c ( x ) λ ( x ) g k ( x ) Δ x ,$
which inequality (2.36) in [33].
A generalization of Bennett’s inequality (15) on fractional timescales will be proved in the next theorem.
Theorem 6.
Assume$0 < α ≤ 1$,$1 < c ≤ k$, and define
$Φ ( x ) : = ∫ x ∞ λ ( s ) Δ α s and Ω ( x ) : = ∫ x ∞ λ ( s ) g ( s ) Δ α s .$
If
$Ω ( a ) < ∞ and ∫ a ∞ λ ( s ) Φ c − α + 1 ( s ) Δ α s < ∞ ,$
then
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) Ω k ( x ) Δ α x ≤ ( k c − α ) k ∫ a ∞ Φ k − c + α − 1 ( x ) λ ( x ) g k ( x ) Δ α x .$
Proof.
Using the formula of integration by parts (20) on
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) Ω k ( x ) Δ α x ,$
with $v ( x ) = Ω k ( x )$ and $x α Δ u ( x ) = λ ( x ) Φ c − α + 1 ( x )$, then
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) Ω k ( x ) Δ α x = u ( x ) Ω k ( x ) | a ∞ + ∫ a ∞ u σ ( x ) x α Δ ( − Ω k ( x ) ) Δ α x ,$
where
$u ( x ) = ∫ a x λ ( s ) Φ c − α + 1 ( s ) Δ α s .$
By using chain rule (18), we see for $d ∈ [ x , σ ( x ) ]$ that
$x α Δ ( Φ α − c ( x ) ) = ( α − c ) Φ α − c − 1 ( d ) x α Δ ( Φ ( x ) ) = ( α − c ) ( − λ ( x ) ) Φ c − α + 1 ( d ) ≥ ( c − α ) λ ( x ) Φ c − α + 1 ( x ) .$
we get,
$u σ ( x ) = ∫ a σ ( x ) λ ( s ) Φ c − α + 1 ( s ) Δ α s ≤ 1 c − α ∫ a σ ( x ) x α Δ Φ α − c ( s ) Δ α s = ( Φ σ ( x ) ) α − c c − α − Φ α − c ( a ) c − α ≤ Φ α − c ( x ) c − α ,$
from chain rule (18), we find that
$x α Δ ( Ω k ( x ) ) = k Ω k − 1 ( d ) x α Δ Ω ( x ) , where d ∈ [ x , σ ( x ) ] ,$
since
$x α Δ ( Ω ( x ) ) = − λ ( x ) g ( x ) ≤ 0 and x ≤ d ,$
then
$− x α Δ ( Ω k ( x ) ) ≤ k λ ( x ) g ( x ) Ω k − 1 ( x ) .$
Using $v ( a ) = 0$, $Ω ( ∞ ) = 0$ and (36), (37) and (35), we get that
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ k c − α ∫ a ∞ Φ α − c ( x ) λ ( x ) g ( x ) Ω k − 1 ( x ) Δ α x ,$
which reformulated as
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) Ω k ( x ) Δ α x ≤ k c − α ∫ a ∞ Φ ( c − α + 1 ) ( k − 1 k ) ( x ) λ k − 1 k ( x ) Φ c − α ( x ) λ ( x ) g ( x ) λ k − 1 k ( x ) Ω k − 1 ( x ) Φ ( c − α + 1 ) ( k − 1 k ) ( x ) Δ α x ,$
Using Hölder’s inequality (21) on
$∫ a ∞ λ ( x ) g ( x ) Φ ( c − α + 1 ) ( k − 1 k ) ( x ) λ k − 1 k ( x ) Φ c − α ( x ) λ k − 1 k ( x ) Ω k − 1 ( x ) Φ ( c − α + 1 ) ( k − 1 k ) ( x ) Δ α x$
with indices k and $k / ( k − 1 )$, we have
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) Ω k ( x ) Δ α x ≤ k c − α [ ∫ a ∞ [ λ ( x ) g ( x ) Φ ( c − α + 1 ) ( k − 1 k ) ( x ) λ k − 1 k ( x ) Φ c − α ( x ) ] k Δ α x ] 1 k × [ ∫ a ∞ [ λ k − 1 k ( x ) Ω k − 1 ( x ) Φ ( c − α + 1 ) ( k − 1 k ) ( x ) ] k k − 1 Δ α x ] k − 1 k ,$
then
$[ ∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω ( x ) ) k Δ α x ] 1 k ≤ k c − α [ ∫ a ∞ Φ k − c + α − 1 ( x ) λ ( x ) g k ( x ) Δ α x ] 1 k .$
$∫ a ∞ λ ( x ) Φ c − α + 1 ( x ) ( Ω σ ( x ) ) k Δ α x ≤ ( k c − α ) k ∫ a ∞ Φ k − c + α − 1 ( x ) λ ( x ) g k ( x ) Δ α x ,$
that is the desired inequality (34). The proof is complete. □
Corollary 8.
At$α = 1$in Theorem 5, we have the inequality
$∫ a ∞ λ ( x ) Φ c ( x ) ( Ω σ ( x ) ) k Δ x ≤ ( k c − 1 ) k ∫ a ∞ Φ k − c ( x ) λ ( x ) g k ( x ) Δ x ,$
which the inequality (2.49) in [33].

## 3. Conclusions

The new fractional calculus on timescales is presented with applications to some new fractional inequalities on timescales like Hardy, Bennett, Copson and Leindler types. Inequalities are considered in rather general forms and contain several special integral and discrete inequalities. The technique is based on the applications of well-known inequalities and new tools from fractional calculus.

## Author Contributions

S.S. contributed in preparing the introduction, preliminaries and formulate theorem 3, its proof and its corollaries (cor.1, cor. 2 and cor.3). M.K. contributed in preparing the introduction, preliminaries and formulate theorem 4, its proof and its corollaries (cor.4, cor. 5 and cor.6). G.A. contributed in preparing the introduction, preliminaries and formulate theorem 5, its proof and its corollaries (cor.7). M.Z. contributed in preparing the introduction, preliminaries and formulate theorem 6, its proof and its corollaries (cor.8). All authors contributed equally to the writing of this manuscript. All authors have read and agreed to the published version of the manuscript.

## Funding

This research was funded by the Deanship of Scientific Research at Princess Nourah bint Abdulrahman University through the Fast-track Research Funding Program.

## Acknowledgments

This research was funded by the Deanship of Scientific Research at Princess Nourah bint Abdulrahman University through the Fast-track Research Funding Program. The authors thank the referees for helpful comments that lead to the improvement of the presentation of the results in this paper.

## Conflicts of Interest

The authors declare no conflict of interest.

## References

1. Hardy, G.H. Notes on a theorem of Hilbert. Math. Z. 1920, 6, 314–317. [Google Scholar] [CrossRef]
2. Hardy, G.H. Notes on some points in the integral calculus, LX. An inequality between integrals. Mess. Math. 1925, 54, 150–156. [Google Scholar]
3. Hardy, G.H. Notes on some points in the integral calculus, LXIV. Further inequalities between integrals. Mess. Math. 1928, 57, 12–16. [Google Scholar]
4. Copson, E.T. Note on series of positive terms. J. Lond. Math. Soc. 1927, 2, 9–12. [Google Scholar] [CrossRef]
5. Leindler, L. Generalization of inequalities of Hardy and Littlewood. Acta Sci. Math. 1970, 31, 279–285. [Google Scholar]
6. Leindler, L. Further sharpining of inequalities of Hardy and Littlewood. Acta Sci. Math. 1990, 54, 285–289. [Google Scholar]
7. Leindler, L. A theorem of Hardy-Bennett-type. Acta Math. Hungar. 1998, 87, 315–325. [Google Scholar] [CrossRef] [Green Version]
8. Leindler, L. Two Hardy-Bennett-type theorems. Acta Math. Hungar. 1999, 85, 265–276. [Google Scholar] [CrossRef]
9. Copson, E.T. Note on series of positive terms. J. Lond. Math. Soc. 1928, 3, 49–51. [Google Scholar] [CrossRef]
10. Copson, E.T. Some integral inequalities. Prof. R. Soc. Edinburg. Sect. A 1976, 75, 157–164. [Google Scholar] [CrossRef]
11. Bennett, G. Some elementary inequalities. Quart. J. Math. Oxford 1987, 38, 401–425. [Google Scholar] [CrossRef]
12. Agarwal, R.P.; Bohner, M.; O’Regan, D.; Saker, S.H. Some dynamic wirtinger-type inequalities and their applications. Pacific J. Math. 2011, 252, 1–18. [Google Scholar] [CrossRef] [Green Version]
13. Agarwal, R.P.; O’Regan, D.; Saker, S.H. Hardy Type Ineqaualities on Time Scales; Springer: Basel, Switzerland, 2016. [Google Scholar]
14. Saker, S.H. Some nonlinear dynamic inequalities on time scales and applications. J. Math. Inequal. 2010, 4, 561–579. [Google Scholar] [CrossRef]
15. Hilger, S. Analysis on measure chains—A unified approach to continuous and discrete calculus. Results Math. 1990, 18, 18–56. [Google Scholar] [CrossRef]
16. Bohner, M.; Peterson, A. Dynamic Equations on Time Scales: An Introduction with Applications; Birkhäuser: Boston, MA, USA, 2001. [Google Scholar]
17. Bohner, M.; Peterson, A. Advances in Dynamic Equations on Time Scales; Birkhäuser: Boston, MA, USA, 2003. [Google Scholar]
18. Kac, V.; Cheung, P. Quantum Calculus; Springer: New York, NY, USA, 2001. [Google Scholar]
19. Bogdan, K.; Dyda, B. The best constant in a fractional Hardy inequality. Math. Nach. 2011, 284, 629–638. [Google Scholar] [CrossRef] [Green Version]
20. Jleli, M.; Samet, B. Lyapunov-type inequalities for a fractional differential equation with mixed boundary conditions. Math. Inequal. Appl. 2015, 18, 443–451. [Google Scholar] [CrossRef]
21. Yildiz, C.; Ozdemir, M.E.; Onelan, H.K. Fractional integral inequalities for different functions. New Trends Math. Sci. 2015, 3, 110–117. [Google Scholar]
22. Abdeljawad, T. On conformable fractional calculus. J. Comp. Appl. Math. 2015, 279, 57–66. [Google Scholar] [CrossRef]
23. Khalil, R.; Horani, M.A.; Yousef, A.; Sababheh, M. A new definition of fractional derivative. J. Comp. Appl. Math. 2014, 264, 65–70. [Google Scholar] [CrossRef]
24. Akkurt, A.; Yildirim, M.E.; Yildirim, H. On some integral inequalities for conformable fractional integrals. RGMIA Res. Rep. Collect. 2016, 19, 107. [Google Scholar]
25. Chu, Y.M.; Khan, M.A.; Ali, T.; Dragomir, S.S. Inequalities for α-fractional differentiable functions. J. Inequal. Appl. 2017, 1, 93. [Google Scholar] [CrossRef] [Green Version]
26. Khan, M.A.; Ali, T.; Dragomir, S.S.; Sarikaya, M.Z. Hermite-Hadamard type inequalities for conformable fractional integrals. Rev. Real Acad. Cienc. Exactas 2018, 112, 1033–1048. [Google Scholar] [CrossRef]
27. Set, E.; Gözpınar, A.; Ekinci, A. Hermite-Hadamard type inequalities via conformable fractional integrals. Acta Math. Univ. Comenian. 2017, 86, 309–320. [Google Scholar] [CrossRef]
28. Sarikaya, M.Z.; Budak, H. New inequalities of Opial type for conformable fractional integrals. Turkish J. Math. 2017, 41, 1164–1173. [Google Scholar] [CrossRef]
29. Sarikaya, M.Z.; Budak, H. Opial type inequalities for conformable fractional integrals. Rgmia Res. Rep. Collect. 2016, 19, 93. [Google Scholar]
30. Sarikaya, M.Z.; Yaldiz, H.; Budak, H. Steffensen’s integral inequality for conformable fractional integrals. Int. J. Anal. Appl. 2017, 15, 23–30. [Google Scholar]
31. Benkhettou, N.; Salima, H.; Torres, D.F.M. A conformable fractional calculus on arbitrary time scales. J. King Saud Univ. Science 2016, 1, 93–98. [Google Scholar] [CrossRef]
32. Nwaeze, E.R.; Torres, D.F.M. Chain rules and inequalities for the BHT fractional calculus on arbitrary timescales. Arab J. Math. 2017, 6, 13–20. [Google Scholar] [CrossRef] [Green Version]
33. Saker, S.H.; O’Regan, D.; Agarwal, R.P. Generalized Hardy, Copson, Leindler and Bennett inequalities on time scales. Math. Nachr. 2014, 287, 686–698. [Google Scholar] [CrossRef]

## Share and Cite

MDPI and ACS Style

Saker, S.; Kenawy, M.; AlNemer, G.; Zakarya, M. Some Fractional Dynamic Inequalities of Hardy’s Type via Conformable Calculus. Mathematics 2020, 8, 434. https://doi.org/10.3390/math8030434

AMA Style

Saker S, Kenawy M, AlNemer G, Zakarya M. Some Fractional Dynamic Inequalities of Hardy’s Type via Conformable Calculus. Mathematics. 2020; 8(3):434. https://doi.org/10.3390/math8030434

Chicago/Turabian Style

Saker, Samir, Mohammed Kenawy, Ghada AlNemer, and Mohammed Zakarya. 2020. "Some Fractional Dynamic Inequalities of Hardy’s Type via Conformable Calculus" Mathematics 8, no. 3: 434. https://doi.org/10.3390/math8030434

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.