(k, s)-Riemann-Liouville fractional integral and ...

8 downloads 0 Views 197KB Size Report
Keywords: Riemann-Liouville fractional integrals, synchronous function, Cheby- ... 11, 14, 21]. The first is the Riemann-Liouville fractional integral of α ≥ 0 for a.
(k, s)-Riemann-Liouville fractional integral and applications Mehmet Zeki SARIKAYA∗† , Zoubir DAHMANI‡, Mehmet Ey¨ up KIRIS§ and ¶ Farooq AHMAD

Abstract In this paper, we introduce a new approach on fractional integration, which generalizes the Riemann-Liouville fractional integral. We prove some properties for this new approach. We also establish some new integral inequalities using this new fractional integration.

Keywords: Riemann-Liouville fractional integrals, synchronous function, Chebyshev inequality, H¨older inequality. 2000 AMS Classification: 26A33;26A51;26D15.

1. Introduction Fractional calculus and its widely application have recently been paid more and more attentions. For more recent development on fractional calculus, we refer the reader to [7, 12, 15, 16, 19]. There are several known forms of the fractional integrals of which two have been studied extensively for their applications [5, 10, 11, 14, 21]. The first is the Riemann-Liouville fractional integral of α ≥ 0 for a continuous function f on [a, b] which is defined by Z x 1 α−1 α Ja f (x) = (x − t) f (t)dt, α ≥ 0, a < x ≤ b. Γ(α) a This integral is motivated by the well known Cauchy formula: Z x Z x Z t1 Z tn−1 1 n−1 dt1 (x − t) f (t)dt, n ∈ N∗ . dt2 ... f (tn )dtn = Γ(n) a a a a The second is the Hadamard fractional integral introduced by Hadamard [9]. It is given by: Z x³ 1 x ´α−1 dt Jaα f (x) = log f (t) , α > 0, x > a. Γ(α) a t t ∗Department of Mathematics, Faculty of Science and Arts, D¨ uzce University, D¨ uzce, Turkey, Email: [email protected] † Corresponding Author. ‡ Laboratory of Pure and Applied Mathematics, UMAB, University of Mostaganem, Algeria, Email: [email protected] § Department of Mathematics, Faculty of Science and Arts, Afyon Kocatepe University, Afyon-TURKEY, Email: [email protected], [email protected] ¶ Centre for Advanced Studies in Pure and Applied Mathematics, Bahauddin Zakariya University, Multan, 60800, Pakistan, Email: [email protected]

2

The Hadamard integral is based on the generalization of the integral Z tn−1 Z x Z Z x³ dt1 t1 dt2 f (tn ) 1 1 x ´n−1 dt ... dtn = log f (t) t1 a t2 tn Γ(α) Γ(α) a t t a a for n ∈ N∗ . In [10], Katugampola gave a new fractional integration which generalizes both the Riemann-Liouville and Hadamard fractional integrals into a single form. This generalization is based on the observation that, for n ∈ N∗ , Z x Z t1 Z tn−1 Z ¢n−1 s (s + 1)1−n x ¡ s+1 s s s t1 dt1 t f (t)dt, t2 dt2 ... tn f (tn )dtn = x − ts+1 Γ(α) a a a a which gives the following fractional version Z ¢α−1 s (s + 1)1−n x ¡ s+1 s α Ja f (x) = x − ts+1 t f (t)dt, Γ(α) a where α and s 6= −1 are real numbers. Recently, in [6], Diaz and Pariguan have defined new functions called k-gamma and k-beta functions and the Pochhammer k-symbol that is respectively generalization of the classical gamma and beta functions and the classical Pochhammer symbol: x

n!k n (nk) k −1 , k > 0, n→∞ (x)n,k

Γk (x) = lim

where (x)n,k is the Pochhammer k-symbol for factorial function. It has been tk

shown that the Mellin transform of the exponential function e− k is the k-gamma function, explicitly given by Z ∞ tk Γk (x) = tx−1 e− k dt, x > 0. 0 x

Clearly, Γ(x) = lim Γk (x), Γk (x) = k k −1 Γ( xk ) and Γk (x + k) = xΓk (x). Furtherk→1

more, k-beta function is defined as follows Z y 1 1 x −1 Bk (x, y) = t k (1 − t) k −1 dt, k 0 k (y) so that Bk (x, y) = k1 B( xk , ky ) and Bk (x, y) = ΓΓk (x)Γ . k (x+y) Later, under the above definitions, in [13], Mubeen and Habibullah have introduced the k-fractional integral of the Riemann-Liouville type as follows: Z x α 1 −1 α (x − t) k f (t)dt, α > 0, x > 0. k J f (x) = kΓk (α) 0

Note that when k → 1, then it reduces to the classical Riemann-liouville fractional integral.

2. (k, s)-Riemann-Liouville fractional integral In this section, we present the (k, s) fractional integration which generalizes all of the above Riemann-Liouville fractional integrals as follows:

3

2.1. Definition. Let f be a continuous function on on a the finite real interval [a, b]. Then (k, s)-Riemann-Liouville fractional integral of f of order α > 0 is defined by: α Z ¢ α −1 (s + 1)1− k x ¡ s+1 s α (2.1) x − ts+1 k ts f (t)dt, x ∈ [a, b], k Ja f (x) := kΓk (α) a where k > 0, s ∈ R\{−1}. In the following theorem, we prove that the (k, s) fractional integral is well defined: 2.2. Theorem. Let f ∈ L1 [a, b], s ∈ R\{−1} and k > 0. Then sk Jaα f (x) exists for any x ∈ [a, b], α > 0. h¡ ¢ α −1 i Proof. Let ∆ := [a, b] × [a, b] and P : ∆ → R ; P (x, t) = xs+1 − ts+1 k ts . It clear to see that P = P+ + P− , where ½ ¡ ¢ α −1 xs+1 − ts+1 k ts , a ≤ t ≤ x ≤ b P+ (x, t) := 0 ,a ≤ x ≤ t ≤ b and

½ ¡ P− (x, t) :=

ts+1 − xs+1 0

¢ αk −1

xs

,a ≤ t ≤ x ≤ b , a ≤ x ≤ t ≤ b.

Since P is measurable on ∆, then we can write Z b Z x Z x ¡ s+1 ¢ α −1 ¢α k ¡ s+1 P (x, t)dt = P (x, t)dt = x − ts+1 k ts dt = x − as+1 k . α a a a By using the repeated integral, we obtain ! ÃZ ! Z b ÃZ b Z b b P (x, t) |f (x)| dt dx = |f (x)| P (x, t)dt dx a

a

a

k α

=

b

!

Z

P (x, t) |f (x)| dt dx = a

a

Z

b

¡ s+1 ¢α x − as+1 k |f (x)| dx

a

¢α k ¡ s+1 b − as+1 k α

≤ That is Z b ÃZ

a

ÃZ

b

|f (x)| a

b

Z

b

|f (x)| dx. a

! P (x, t)dt dx

a

¢α k ¡ s+1 b − as+1 k kf (x)kL1 [a,b] < ∞. α Therefore, the function Q : ∆ → R; Q(x, t) := P (x, t)f (x) is integrable over ∆ by Rb Tonelli’s theorem. Hence, by Fubini’s theorem a P (x, t)f (x)dx is an integrable function on [a, b], as a function of t ∈ [a, b]. That is, sk Jaα f (x) exists. ¤ ≤

Now, we prove the commutativity and the semigroup properties of the (k, s)Riemann-Liouville fractional integral. We have:

4

2.3. Theorem. Let f be continuous on [a, b], k > 0 and s ∈ R\{−1}. Then, £ ¤ s α s β s α+β f (x) = sk Jaβ [ sk Jaα f (x)] , k Ja k Ja f (x) = k Ja for all α > 0, β > 0, x ∈ [a, b]. Proof. Thanks to Definition 1 and by Dirichlet’s formula, we have Z ¢ α −1 £ ¤ (s + 1)1− αk x ¡ s+1 s α s β x − ts+1 k ts ks Jaβ f (t)dt k Ja k Ja f (x) = kΓk (α) a · ¸ β α 1− ¢α ¢β R t ¡ s+1 (s+1)1− k R x ¡ s+1 s+1 k −1 s (s+1) k s+1 k −1 s = kΓk (α) a x t τ f (τ )dτ dt. −t t −τ kΓk (β) a α

(s+1)1− k kΓk (α)

=

Rx¡ a

s+1

x

−t

s+1

¢ αk −1

· t

β

(s+1)1− k kΓk (β)

s

Rt¡ a

s+1

t

−τ

s+1

¢ βk −1

¸ s

τ f (τ )dτ dt.

That is (2.2)

¸ ·Z x α+β Z x ¡ ¢β ¡ s+1 ¢α (s + 1)2− k s+1 k −1 s s+1 k −1 s s+1 = 2 t t −τ dt dτ. τ f (τ ) x −t k Γk (α)Γk (β) a τ ¡ ¢ ¡ ¢ Using the change of variable y = ts+1 − τ s+1 / xs+1 − τ s+1 , we can write s α k Ja

(2.3)

Z

x

£

¤

s β k Ja f (x)

¢ β −1 ¡ s+1 ¢ α −1 ¡ s+1 (xs+1 −τ s+1 ) t − τ s+1 k ts dt = x − ts+1 k s+1

τ

(xs+1 −τ s+1 )

=

α+β −1 k

s+1

R1 0

(1 − y)

α k −1

y

β k −1

dy

=

α+β −1 k

R1 0

(xs+1 −τ s+1 )

α

α+β −1 k

s+1

kBk (α, β).

According to the k-beta function and by (2.2) and (2.3), we obtain s α k Ja

£

¤

s β k Ja f (x)

α+β

= =

(s + 1)1− k kΓk (α + β)

Z

x

¡

xs+1 − τ s+1

¢ α+β k −1

τ s f (τ )dτ

a

s α+β f (x). k Ja

This completes the proof of the Theorem 2.3. 2.4. Theorem. Let α, β > 0, k > 0 and s ∈ R\{−1}. Then, we have (2.4)

s α k Ja

h¡ ¢ β −1 i = xs+1 − as+1 k

¡ s+1 ¢ α+β −1 Γk (β) , x − as+1 k α (s + 1) k Γk (α + β)

where Γk denotes the k-gamma function.

β

(1 − y) k −1 y k −1 dy

¤

5

¡ ¢ ¡ ¢ Proof. By Definition 1 and using the change of variable y = xs+1 − ts+1 / xs+1 − as+1 ; x ∈ ]a, b], we get α Z h¡ ¢β i ¢ α −1 ¡ ¢ α+β −1 (s + 1)1− k x ¡ s+1 s α s+1 s+1 k −1 J x − a = x − ts+1 k ts ts+1 − as+1 k dt k a kΓk (α) a =

¢ α+β −1 Z 1 α ¡ (s + 1)− k xs+1 − as+1 k β α (1 − y) k −1 y k −1 dy kΓk (α) 0 ¡

=

¢ α+β −1 xs+1 − as+1 k Bk (α, β). α (s + 1) k Γk (α)

The case a = x is trivial. The proof of the Theorem 2.4 is complete.

¤

2.5. Remark. (i :) Taking s = 0, k > 0 in (2.4), we obtain h i β α+β Γk (β) −1 α k −1 (2.5) = (x − a) k . k Ja (x − a) Γk (α + β) (ii :) The formula (2.4) for s = 0, k = 1 becomes h i Γ(β) β−1 α+β−1 Jaα (x − a) = (x − a) . Γ(α + β) 2.6. Corollary. Let k > 0 and s ∈ R\{−1}. Then the formula ¡ s+1 ¢ α −2 1 s α x − as+1 k (2.6) α k Ja (1) = (s + 1) k Γk (α + k) is valid for any α > 0. 2.7. Remark. (a :) For s = 0, k > 0 in (2.6), we get (2.7)

α k Ja (1)

=

α 1 −2 (x − a) k . Γk (α + k)

(b :) For s = 0, k = 1 we have Jaα (1) =

1 α−2 (x − a) . Γ(α + 1)

3. Some new (k, s)-Riemann-Liouville fractional integral inequalities Chebyshev inequalities can be represented in (k, s)-fractional integral forms as follows: 3.1. Theorem. Let f, g be two synchronous on [0, ∞), then for all t > a ≥ 0, α > 0, β > 0, the following inequalities for (k, s)-fractional integrals hold: (3.1) (3.2)

s α k Ja f g(t)



1 s α J f (t) sk Jaα g(t) α Ja (1) k a

s α s β s β s α k Ja f g(t) k Ja (1)+ k Ja f g(t) k Ja (1)



s α s β s α s β k Ja f (t) k Ja g(t)+ k Ja g(t) k Ja f (t).

6

Proof. Since the functions f and g are synchronous on [0, ∞), then for all x, y ≥ 0, we have (f (x) − f (y)) (g (x) − g(y)) ≥ 0. Therefore (3.3)

f (x) g (x) + f (y) g (y) ≥ f (x) g (y) + f (y) g (x)

1− α ¡ ¢ α −1 k Multiplying both sides of (3.3) by (s+1) ts+1 − xs+1 k xs , then integrating kΓk (α) the resulting inequality with respest to x over (a, t), we obtain α

(s + 1)1− k kΓk (α)

Z

t

¡

ts+1 − xs+1

¢ αk −1

xs f (x) g (x) dx

a

α Z ¢ α −1 (s + 1)1− k t ¡ s+1 t − xs+1 k xs f (y) g (y) dx kΓk (α) a Z t 1− α ¡ s+1 ¢ α −1 (s + 1) k ≥ t − xs+1 k xs f (x) g (y) dx kΓk (α) a α Z t ¡ s+1 ¢ α −1 (s + 1)1− k + t − xs+1 k xs f (y) g (x) dx, kΓk (α) a

+

i.e. (3.4)

s α k Ja f g(t)

+ f (y) g (y) sk Jaα (1) ≥ g (y) sk Jaα f (t) + f (y) sk Jaα g(t).

1− α ¡ ¢ α −1 k Multiplying both sides of (3.3) by (s+1) ts+1 − y s+1 k y s , then integrating kΓk (α) the resulting inequality with respest to y over (a, t), we obtain α

+ 1)1− k kΓk (α)

(s s α k Ja f g(t) + ≥

1− α k

(s s α k Ja f (t)

+ 1) kΓk (α)

+

(s s α k Ja g(t)

¡

t

¡ s+1 ¢ α −1 t − y s+1 k y s f (y) g (y) dy

ts+1 − y s+1

t



a

Z

t

¡

ts+1 − y s+1

a

1 s α J f (t) sk Jaα g(t) Jaα (1) k a

and the first inequality is proved.

y s dy

¡ s+1 ¢ α −1 t − y s+1 k y s g(y)dy

that is s α k Ja f g(t)

¢ αk −1

a

Z

1− α k

+ 1) kΓk (α)

t

a

Z

α

+ 1)1− k kΓk (α)

(s s α k Ja (1)

Z

¢ αk −1

y s f (y) dy,

7 β 1− ¢ β −1 k ¡ s+1 Multiplying both sides of (3.3) by (s+1) t − y s+1 k y s , then integratkΓk (α) ing the resulting inequality with respest to y over (a, t), we obtain β

+ 1)1− k kΓk (α)

(s s α k Ja f g(t) + sk Jaα (1)

(s + 1) kΓk (α)

(s s α k Ja f (t)

+ 1) kΓk (α)



+ sk Jaα g(t)

¡

t

¡ s+1 ¢ β −1 t − y s+1 k y s f (y) g (y) dy

ts+1 − y s+1

¢ βk −1

y s dy

a

Z

1− β k

(s + 1) kΓk (α)

t

a

Z

1− β k

1− β k

Z

t

¡ s+1 ¢ β −1 t − y s+1 k y s g(y)dy

a

Z

t

¡

ts+1 − y s+1

¢ βk −1

y s f (y) dy,

a

that is s α s β s β s α k Ja f g(t) k Ja (1)+ k Ja f g(t) k Ja (1)

≥ sk Jaα f (t) sk Jaβ g(t)+ sk Jaα g(t) sk Jaβ f (t)

and the second inequality is proved. The proof is completed.

¤

3.2. Theorem. Let f, g be two synchronous on [0, ∞), h ≥ 0, then for all t > a ≥ 0, α > 0, β > 0, the following inequalities for (k, s)-fractional integrals hold: 1 β k

(s + 1) Γk (β + k) + ≥

¡ s+1 ¢ β −2 t − as+1 k

s α k Ja f gh(t)

¡ s+1 ¢ α −2 1 t − as+1 k (s + 1) Γk (α + k) α k

s α s β k Ja f h(t) k Ja g(t)

+

s α s β k Ja gh(t) k Ja f (t)

+ sk Jaα f (t) sk Jaβ gh(t) +

s β k Ja f gh(t)



s α s β k Ja h(t) k Ja f g(t)



s α s β k Ja f g(t) k Ja h(t)

s α s β k Ja g(t) k Ja f h(t).

Proof. Since the functions f and g are synchronous on [0, ∞) and h ≥ 0, then for all x, y ≥ 0, we have (f (x) − f (y)) (g (x) − g(y)) (h (x) + h (y)) ≥ 0. By opening the above, we get (3.5)

f (x) g (x) h (x) + f (y) g (y) h (y) ≥

f (x) g (y) h (x) + f (y) g (x) h (x) − f (y) g (y) h (x) −f (x) g (x) h (y) + f (x) g (y) h (y) + f (y) g (x) h (y) .

8 1− α ¡ ¢ α −1 k Multiplying both sides of (3.5) by (s+1) ts+1 − xs+1 k xs , then integrating kΓk (α) the resulting inequality with respest to x over (a, t), we obtain

α

(s + 1)1− k kΓk (α)

Z

t

¢ α −1 ¡ s+1 t − xs+1 k xs f (x) g (x) h (x) dx

a

α Z ¢ α −1 (s + 1)1− k t ¡ s+1 +f (y) g (y) h (y) t − xs+1 k xs dx kΓk (α) a α Z t ¡ s+1 ¢ α −1 (s + 1)1− k ≥ g (y) t − xs+1 k xs f (x) h (x) dx kΓk (α) a Z 1− α ¢ α −1 (s + 1) k t ¡ s+1 +f (y) t − xs+1 k xs g (x) h (x) dx kΓk (α) a α Z ¢ α −1 (s + 1)1− k t ¡ s+1 t − xs+1 k xs h (x) dx −f (y) g (y) kΓk (α) a α Z t ¡ s+1 ¢ α −1 (s + 1)1− k −h (y) t − xs+1 k xs f (x) g (x) dx kΓk (α) a α Z ¢ α −1 (s + 1)1− k t ¡ s+1 +g (y) h (y) t − xs+1 k xs f (x) dx kΓk (α) a α Z t ¡ s+1 ¢ α −1 (s + 1)1− k +f (y) h (y) t − xs+1 k xs g (x) dx. kΓk (α) a

i.e,

(3.6) ≥

s α k Ja f gh(t)

+ f (y) g (y) h(y) sk Jaα (1)

g (y) sk Jaα f h(t) + f (y) sk Jaα gh(t) − f (y) g (y) sk Jaα h(t) − h(y) sk Jaα f g(t) +g (y) h (y) sk Jaα f (t) + f (y) h (y) sk Jaα g(t).

9 β 1− ¢ β −1 k ¡ s+1 Multiplying both sides of (3.6) by (s+1) t − y s+1 k y s , then integrating kΓk (β) the resulting inequality with respest to y over (a, t), we obtain β Z ¢ β −1 (s + 1)1− k t ¡ s+1 s α J f gh(t) t − y s+1 k y s dy k a kΓk (β) a

Z

β

+sk Jaα (1) ≥

(s + 1)1− k kΓk (β)

¡

Z

t

¡

β

Z

t

(s + 1)1− k kΓk (β) β

− sk Jaα h(t)

(s + 1)1− k kΓk (β)

y s f (y) g (y) h(y)dy

β

(s + 1)1− k kΓk (β)

+ sk Jaα g(t)

(s + 1)1− k kΓk (β)

β

¢ βk −1

y s g (y) dy

¡ s+1 ¢ β −1 t − y s+1 k y s f (y) dy

t

¡ s+1 ¢ β −1 t − y s+1 k y s f (y) g (y) dy

a

(s + 1)1− k kΓk (β)

+ sk Jaα f (t)

ts+1 − y s+1

a

Z

β

− sk Jaα f g(t)

¢ βk −1

a β

+ sk Jaα gh(t)

ts+1 − y s+1

a

+ 1)1− k kΓk (β)

(s s α k Ja f h(t)

t

Z

t

¡ s+1 ¢ β −1 t − y s+1 k y s h(y)dy

a

Z

t

¡ s+1 ¢ β −1 t − y s+1 k y s g (y) h (y) dy

t

¡ s+1 ¢ β −1 t − y s+1 k y s f (y) h (y) dy,

a

Z a

that is s α s β k Ja f gh(t)k Ja (1)

+sk Jaα (1)sk Jaβ f gh(t) ≥

s α s β k Ja f h(t) k Ja g(t)

+

s α s β k Ja gh(t) k Ja f (t)

− sk Jaα h(t) sk Jaβ f g(t) −

s α s β k Ja f g(t) k Ja h(t)

+ sk Jaα f (t) sk Jaβ gh(t) +

s α s β k Ja g(t) k Ja f h(t)

which this completes the proof.

¤

3.3. Corollary. Let f, g be two synchronous on [0, ∞), h ≥ 0, then for all t > a ≥ 0, α > 0, the following inequalities for (k, s)-fractional integrals hold: ¢ α −2 s α ¡ s+1 1 t − as+1 k α k Ja f gh(t) k (s + 1) Γk (α + k) ≥

s α s α k Ja f h(t) k Ja g(t)

+

s α s α k Ja gh(t) k Ja f (t)



s α s α k Ja h(t) k Ja f g(t).

3.4. Theorem. Let f, g and h be three monotonic functions defined on [0, ∞) satisfying the following (f (x) − f (y)) (g (x) − g(y)) (h (x) − h (y)) ≥ 0

10

for all x, y ∈ [a, t] , then for all t > a ≥ 0, α > 0, β > 0, the following inequalities for (k, s)-fractional integrals hold: ¡ s+1 ¢ β −2 s α 1 t − as+1 k β k Ja f gh(t) (s + 1) k Γk (β + k) − ≥

¡ s+1 ¢ α −2 1 t − as+1 k (s + 1) Γk (α + k) α k

s α s β k Ja f h(t) k Ja g(t)

s α s β k Ja gh(t) k Ja f (t)

+

−sk Jaα f (t) sk Jaβ gh(t) −

s β k Ja f gh(t)



s α s β k Ja h(t) k Ja f g(t)

+

s α s β k Ja f g(t) k Ja h(t)

s α s β k Ja g(t) k Ja f h(t).

Proof. The proof is similar to that given in Theorem 3.2.

¤

3.5. Theorem. Let f and g be two functions on [0, ∞), then for all t > a ≥ 0, α > 0, β > 0, the following inequalities for (k, s)-fractional integrals hold: ¡ s+1 ¢ β −2 s α 2 1 t − as+1 k (3.7) β k Ja f (t) (s + 1) k Γk (β + k) +

¡ s+1 ¢ α −2 1 t − as+1 k (s + 1) Γk (α + k) α k

s β 2 k Ja g (t)

≥ 2 sk Jaα f (t) sk Jaβ g(t) (3.8)

s α 2 s β 2 k Ja f (t) k Ja g (t)

s β 2 s α 2 k Ja f (t) k Ja g (t)

+

≥ 2 sk Jaα f g(t) sk Jaβ f g(t).

Proof. Since, 2

(f (x) − g(y)) ≥ 0 then we have (3.9)

f 2 (x) + g 2 (y) ≥ 2f (x)g(y).

β 1− α ¡ 1− ¢ α −1 ¢ β −1 k k ¡ s+1 Multiplying both sides of (3.9) by (s+1) ts+1 − xs+1 k xs and (s+1) t − y s+1 k y s , kΓk (α) kΓk (β) then integrating the resulting inequality with respest to x and y over (a, t) respectively, we obtain (3.7). On the other hand, since

2

(f (x)g(y) − f (y)g(x)) ≥ 0 then under procedures similar to the above we obtain (3.8).

¤

3.6. Corollary. Let f and g be two functions on [0, ∞), then for all t > a ≥ 0, α > 0, the following inequalities for (k, s)-fractional integrals hold: ¡ s+1 ¢ α −2 £s α 2 ¤ 1 s β 2 t − as+1 k α k Ja f (t) + k Ja g (t) (s + 1) k Γk (α + k) ≥

2 sk Jaα f (t) sk Jaα g(t)

s α 2 s α 2 k Ja f (t) k Ja g (t)

2

≥ [sk Jaα f g(t)] .

11

3.7. Theorem. Let f : R → R and defined by Z x f¯(x) = ts f (t)dt, x > a ≥ 0, s ∈ R\{−1} a

then for α ≥ k > 0 s α k Ja f (x)

=

1 s α−k ¯ J f (x) k k a

Proof. By definition of the (k, s)-fractional integral and by using Dirichlet’s formula, we have Z t α Z ¢α (s + 1)1− k x ¡ s+1 s α¯ s+1 k −1 s us f (u)dudt J (x) = t f x − t k a kΓk (α) a a Z x Z x 1− α ¡ s+1 ¢ α −1 (s + 1) k = us f (u) x − ts+1 k ts dtdu kΓk (α) a u α Z x ¢α ¡ s+1 (s + 1)− k = x − us+1 k us f (u)du Γk (α + k) a =

k sk Jaα+k f (x).

This completes the proof of Theorem 3.7.

¤

We give the generalized Cauchy-Buniakovsky-Schwarz inequality as follows: 3.8. Lemma. Let f, g, h : [a, b] → (0, ∞) be three functions 0 ≤ a < b. Then (3.10)

ÃZ

! ÃZ

b

m

b

x

g (t)h (t)f (t)dt

! n

ÃZ

y

g (t)h (t)f (t)dt

a



a

!2

b

g

m+n 2

(t)h

x+y 2

(t)f (t)dt

.

a

where m, n, x, y arbitrary real numbers. Proof. Z

b

 s Z p  g m (t)hx (t)f (t)

a

Z

b

Z m

x

n

a

y

n

y

g (t)h (t)f (t)dt + g (t)h (t)f (t) a

(t)h

x+y 2

2 g m (t)hx (t)f (t)dt dt ≥ 0

b

g m (t)hx (t)f (t)dt

a

s m+n 2

b

a

Z

b

g (t)h (t)f (t)

Z

b

(t)f (t) a

≥ 0

g n (t)hy (t)f (t)dt −

s Z g n (t)hy (t)f (t)

p

a

"

−2 g

b

s Z g m (t)hx (t)f (t)dt a

 b

g n (t)hy (t)f (t)dt dt

12

ÃZ

! ÃZ

b

2

m

g (t)h (t)f (t)dt

≥ 2

n

y

g (t)h (t)f (t)dt

a

ÃZ

!

b

x

a

! sZ

b

g

m+n 2

(t)h

x+y 2

b

Z

g m (t)hx (t)f (t)dt

(t)f (t)dt

a

s

a

b

g n (t)hy (t)f (t)dt a

which this give the requair inequality.

¤

3.9. Theorem. Let f ∈ L1 [a, b]. Then ³ ´2 ´³ ´ ³ m+n α α n( α −1)+1 p ( −1)+1 r+p s m( k −1)+1 r (3.11) f (x) sk Ja k f (x) ≥ sk Ja 2 k f 2 (x) k Ja for k, m, n, r, p > 0 and α > 1. 1− α ¡ ¢ α −1 s k Proof. By taking g(t) = xs+1 − ts+1 k , f (t) = t (s+1) and h(t) = f (t) in kΓk (α) (3.10), we obtain µ ¶µ ¶ α Z α Z ¢m( αk −1) s r ¢n( α −1) s p (s + 1)1− k x ¡ s+1 (s + 1)1− k x ¡ s+1 x − ts+1 t f (t)dt x − ts+1 k t f (t)dt kΓk (α) kΓk (α) a a µ ≥

α

(s + 1)1− k kΓk (α)

Z

x

¶2 ¡ s+1 ¢ m+n ( α −1) s r+p t f 2 (t)dt x − ts+1 2 k

a

which can be written as (3.11).

¤

3.10. Remark. For k = 1 in (3.11), we get the following inequalities ³ ´³ ´ ³ m+n ´2 (α−1)+1 r+s Jam(α−1)+1 f r (x) k Jan(α−1)+1 f s (x) ≥ k Ja 2 f 2 (x) .

References [1] Belarbi, S. and Dahmani, Z., On some new fractional integral inequalities, J. Ineq. Pure and Appl. Math., 10(3) (2009), Art. 86. [2] Dahmani, Z., New inequalities in fractional integrals, International Journal of Nonlinear Scinece, 9(4) (2010), 493-497. [3] Dahmani, Z., On Minkowski and Hermite-Hadamard integral inequalities via fractional integration, Ann. Funct. Anal. 1(1) (2010), 51-58. [4] Dahmani, Z., Tabharit, L. and Taf, S., Some fractional integral inequalities, Nonl. Sci. Lett. A, 1(2) (2010), 155-160. [5] Dahmani, Z., Tabharit, L. and Taf, S., New generalizations of Gruss inequality usin Riemann-Liouville fractional integrals, Bull. Math. Anal. Appl., 2(3) (2010), 93-99. [6] Diaz, R. and Pariguan, E., On hypergeometric functions and Pochhammer k−symbol, Divulg.Math, 15.(2007),179-192. [7] Gorenflo, R. and Mainardi, F., Fractional calculus: integral and differential equations of fractional order, Springer Verlag, Wien (1997), 223-276. [8] Kilbas, A. A.,Srivastava, H. M. and Trujillo, J. J., Theory and Applications of Fractional Differential Equations, North-Holland Mathematics Studies, 204, Elsevier Sci. B.V., Amsterdam, 2006. etude des fonctions donn´ ees par leur d´ eveloppement de [9] Hadamard, J., Essai sur l’´ Taylor,Jour.PureandAppl.Math4(8),1892,101-186. [10] Katugompola, U.N., New Approach Generalized Fractional Integral, Applied Math and Comp. 218(2011),860-865.

13

[11] Latif, M. A. and Hussain, S., New inequalities of Ostrowski type for co-ordineted convex functions via fractional integrals, Journal of Fractional Calculus and Applications,Vol. 2., 2012, No. 9, pp. 1-15. [12] Miller, S. and Ross, B., An introduction to the Fractional Calculus and Fractional Differential Equations, John Wiley & Sons, USA, 1993, p.2. [13] Mubeen, S. and Habibullah, G.M., k−fractional integrals and application, Int. J. Contemp. Math. Sciences, 7(2), 2012, 89-94. [14] Romero, L. G., Luque, L. L., Dorrego, G. A. and Cerutti, R. A., On the k−RiemannLiouville Fractional Derivative, Int. J. Contemp. Math. Sciences, 8(1), 2013, 41-51. [15] Sarikaya, M.Z. and Ogunmez, H., On new inequalities via Riemann-Liouville fractional integration, Abstract and Applied Analysis, Volume 2012, Article ID 428983, 10 pages, doi:10.1155/2012/428983. [16] Sarikaya, M. Z., Set, E., Yaldiz, H. and Basak, N., Hermite -Hadamard’s inequalities for fractional integrals and related fractional inequalities, Mathematical and Computer Modelling, 57 (2013) 2403–2407. [17] Sarikaya, M. Z., On the Hermite-Hadamard-type inequalities for co-ordinated convex function via fractional integrals, Integral Transforms and Special Functions, Volume 25, Issue 2, pp:134-147, 2014. [18] Sarikaya, M. Z., Ostrowski type inequalities involving the right Caputo fractional derivatives belong to Lp , Facta Universitatis, Series Mathematics and Informatics , Vol. 27 No 2 (2012), 191-197. [19] Sarikaya, M. Z. and Yaldiz, H., On weighted Montogomery identities for Riemann-Liouville fractional integrals, Konuralp Journal of Mathematics, Volume 1 No. 1 pp. 48-53, 2013. [20] Sarikaya, M. Z., and Karaca, A., On the k-Riemann-Liouville fractional integral and applications, International Journal of Statistics and Mathematics, Vol. 1(1), pp. 022-032, August, 2014. [21] Tunc, M., On new inequalities for h-convex functions via Riemann-Liouville fractional integration, Filomat 27:4 (2013), 559–565 [22] Samko, S. G., Kilbas, A. A and Marichev, O. I., Fractional Integrals and Derivatives Theory and Application, Gordan and Breach Science, New York, 1993.