## OPIALâTYPE INEQUALITIES FOR FRACTIONAL ...

Calculus. Volume 5, Number 1 (2015), 93â106 doi:10.7153/fdc-05- ... fractional integral operator involving generalized MittagâLeffler function. We deduce some.

F ractional Differential Calculus

Volume 5, Number 1 (2015), 93–106

doi:10.7153/fdc-05-09

OPIAL–TYPE INEQUALITIES FOR FRACTIONAL INTEGRAL OPERATOR INVOLVING MITTAG––LEFFLER FUNCTION

G HULAM FARID , J OSIP P E Cˇ ARI C´

AND

Abstract. In this paper we give generalization of Opial-type inequalities by using generalized fractional integral operator involving generalized Mittag–Leffler function. We deduce some results which already have been proved. Also we consider n -exponential convexity of some non-negative differences of inequalities involving Mittag-Leffler function and deduce their exponential convexity and log-convexity.

1. Introduction and preliminaries Fractional calculus refers to integration or differentiation of fractional order. Several mathematicians contributed to this subject over the years. People like Liouville, Riemann, and Weyl made major contributions to the theory of fractional calculus. The story on the fractional calculus continued with contributions from Fourier, Abel, Lacroix, Leibniz, Grunwald and Letnikov. For a historical survey the reader may see [15, 16, 18]. Fractional integral inequalities are useful in establishing the uniqueness of solutions for certain fractional partial differential equations. They also provide upper and lower bounds for the solutions of fractional boundary value problems. These considerations have led various researchers in the field of integral inequalities to explore certain extensions and generalizations by involving fractional calculus operators (see, [1, 2, 8, 6, 14, 25, 26]). In [3, 4, 9, 10, 11] we have established Opial-type integral inequalities for different kinds of fractional derivatives and fractional integral operators for example Riemann– Liouville, Caputo, Canvati etc. In this paper we give Opial-type integral inequalities for fractional integral operator containing a generalized Mittag–Leffler function in the kernel . Definition of this generalized fractional integral operator containing Mittag– Leffler function is as follows. D EFINITION 1.1. Let α , β , k, l, γ be positive real numbers and ω ∈ R. Then the γ ,δ ,k generalized fractional integral operator containing Mittag–Leffler function εα ,β ,l,ω ,a+ for a real-valued continuous function f is defined by: γ ,δ ,k

(εα ,β ,l,ω ,a+ f )(x) =

Z x a

γ ,δ ,k

(x − t)β −1Eα ,β ,l (ω (x − t)α ) f (t)dt,

(1.1)

Mathematics subject classification (2010): 26A33, 26D15, 33E12. Keywords and phrases: Opial-type inequality, fractional integral, fractional derivative, Mittag–Leffler function. c D l , Zagreb

Paper FDC-05-09

93

94

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C γ ,δ ,k

where the function Eα ,β ,l is generalized Mittag–Leffler function defined as γ ,δ ,k

Eα ,β ,l (t) =

tn

(γ )kn

∑ Γ(α n + β ) (δ )ln ,

(1.2)

n=0

and (a)n is the Pochhammer symbol: (a)n = a(a + 1) . . .(a + n − 1), (a)0 = 1 . γ ,δ ,k

If δ = l = 1 in (1.1), then integral operator εα ,β ,l,ω ,a+ reduces to an integral opγ ,1,k

erator containing generalized Mittag–Leffler function Eα ,β ,1 introduced by Srivastava and Tomovski in . Along δ = l = 1 in addition if k = 1 (1.1) reduces to an integral γ operator defined by Prabhakar in  containing Mittag–Leffler function Eα ,β . Let γ

γ

γ ,δ ,k

eα ,β (t) = t β −1 Eα ,β (ω t α ). For ω = 0 in (1.1), integral operator εα ,β ,l,ω ,a+ would correspond essentially to the right-handed Riemann–Liouville fractional integral operator (see, ), Z x 1 I β f (x) = (x − t)β −1 f (t)dt, β > 0. Γ(β ) a We define a variant of Sobolev space:   dm m,1 1 1 W [a, b] = f ∈ L [a, b] : m f ∈ L [a, b] . dt D EFINITION 1.2. (Prabhakar derivative ) Let f ∈ L1 [0, b], 0 < t < b 6 ∞, −γ and f ∗ eα ,m−β ,ω ∈ W m,1 [0, b] , m = [β ] then the Prabhakar derivative is defined by following relation   d m −γ γ Dα ,β ,ω ,0+ f (t) = m εα ,m−β ,ω ,0+ f (t) . dt D EFINITION 1.3. (Caputo-Prabhakar derivative ) Let f ∈ L1 [0, b] , 0 < t < −γ b 6 ∞, and f ∗ eα ,m−β ,ω ∈ W m,1 [0, b], m = [β ] then the Caputo-Prabhakar derivative for f ∈ ACm [0, b] is defined by following relation 

C

 dm γ −γ Dα ,β ,ω ,0+ f (t) = εα ,m−β ,ω ,0+ m f (t) dt   m−1 −γ γ = Dα ,β ,ω ,0+ f (t) − ∑ t k−α Eα ,k−β +1 (ω t α ) f (k) (0+). k=0

R EMARK 1.4. Let β > 0 and f ∈ ACm [0, b], 0 < t < b 6 ∞, then !   m−1 k t γ γ C f (k) (0+) . Dα ,β ,ω ,0+ f (t) = Dα ,β ,ω ,0+ f (t) − ∑ k=0 k! In 1960. Opial established the following integral inequality .

95

O PIAL - TYPE INEQUALITIES FOR FRACTIONAL . . .

Let x(t) ∈ C(1) [0, h] be such that x(0) = x(h) = 0 , and x(t) > 0 in (0, h). Then Z h Z h h ′ 2 |x(t)x′ (t)|dt 6 (1.3) x (t) dt, 4 0 0 where constant h4 is the best possible. Opial’s inequality (1.3) is studied extensively by many researchers. It recognizes as fundamental result in the theory of differential equations (see the monograph ). Following theorems include generalizations of Opial’s inequality and for it we need next characterization: for detail see [24, page 234–238]. We say that a function u : [a, b] −→ R belongs to the class U(v, K) if it admits the representation u(x) =

Z x

K(x,t)v(t) dt

a

where v is a continuous function and K is an arbitrary non-negative kernel such that v(x) > 0 implies u(x) > 0 for every x ∈ [a, b]. We also assume that all integrals under consideration exist and are finite. The following theorem is given in  (also see [1, p. 89] and [24, p. 236]). T HEOREM 1.5. Let u1 ∈ U(v1 , K), u2 ∈ U(v2 , K) and v2 (x) > 0 for every x ∈ [a, b]. Further, let φ (u) be convex, non-negative and increasing for u > 0 , f (u) be convex for u > 0 and f (0) = 0. If f is differentiable function and M = max K(x,t), then       Z b    Z b v1 (t) ′  v1 (t) f u2 (t)φ u1 (t) dt 6 f M v2 (t)φ M φ v (t) 2 u2 (t) v2 (t) dt . v2 (t) a a Extension of above result stated in the following theorem .

T HEOREM 1.6. With same assumptions as in Theorem 1.5 we have     Z b v1 (t) ′  u1 (t)  dt v2 (t)φ M f u2 (t)φ v2 (t) u2 (t) a     Zb v1 (t) dt v2 (t)φ 6 f M v2 (t) a   Z b  v1 (t) 1 f M(b − a)v2 (t)φ dt. 6 b−a a v2 (t)

In [24, p. 236]), also the following result is proved.

T HEOREM 1.7. Let φ : [0, ∞) −→ R be a differentiable function such that for q > 1

1

1 the function φ (x q ) is convex and φ (0) = 0 . Let u ∈ U(v, K) where ( ax (K(x,t)) p dt) p 6 M and 1p + 1q = 1 . Then Z b a

R

1−q

|u(x)|

q   φ (|u(x)|)|v(x)| dx 6 q φ M M ′

q

1

Z b

The reverse of above inequality holds if φ (x q ) is concave.

a

|v(x)|q dx

 1q  .

(1.4)

96

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C

Properties of non-negative difference of inequality (1.4) are studied in . In the following result we have extension of inequality (1.4) (see, ). T HEOREM 1.8. Let φ : [0, ∞) −→ R be a differentiable function such that for q > 1

1

1 the function φ (x q ) is convex and φ (0) = 0 . Let u ∈ U(v, K) where ( ax (K(x,t)) p dt) p 6 M and 1p + 1q = 1 . Then R

Z b

|u(x)|1−q φ ′ (|u(x)|)|v(x)|q dx

a

Z 1  q   b q |v(x)|q dx M φ q M a Z b   1 q φ (b − a) q M|v(x)| dx . 6 q M (b − a) a

6

(1.5)

1

The reverse of above inequality holds if φ (x q ) is concave. In  we denote the non-negative difference of extreme terms in the above inequality by Ψφ (u, v), as follows: q q M (b − a)

Ψφ (u, v) =

Z b

Z b a

  1 φ (b − a) q M|v(x)| dx

|u(x)|1−q φ ′ (|u(x)|)|v(x)|q dx ,

(1.6)

a

and among other properties of above functional we have proved the following results. T HEOREM 1.9. Let φ : [0, ∞) −→ R be a differentiable function such that for q > 1

1

1 the function φ (x q ) is convex and φ (0) = 0 . Let u ∈ U(v, K) where ( ax (K(x,t)) p dt) p 6 M and 1p + 1q = 1 . If φ ∈ C2 (I), where I ⊆ (0, ∞) is compact interval, then there exists ξ ∈ I such that the following equality holds Ψφ (u, v) =

R

ξ φ ′′ (ξ ) − (q − 1) φ ′(ξ ) 2 q ξ 2q−1   Z b Z b |u(x)|q |v(x)|q dx . × (b − a)M q |v(x)|2q dx − 2

(1.7)

a

a

T HEOREM 1.10. Let φ1 , φ2 : [0, ∞) −→ R be differentiable functions such that for 1

q > 1 the function φi (x q ) is convex and φi (0) = 0 , i = 1, 2 . Let u ∈ U(v, K) where 1

( ax (K(x,t)) p dt) p 6 M and interval and R

(b − a)M q

1 p

Z b a

+ 1q = 1 . If φ1 , φ2 ∈ C2 (I), where I ⊆ (0, ∞) is compact |v(x)|2q dx − 2

Z b a

|u(x)|q |v(x)|q dx 6= 0 ,

O PIAL - TYPE INEQUALITIES FOR FRACTIONAL . . .

97

then there exists an ξ ∈ I such that we have Ψφ1 (u, v) ξ φ1′′ (ξ ) − (q − 1) φ1′ (ξ ) = , Ψφ2 (u, v) ξ φ2′′ (ξ ) − (q − 1) φ2′ (ξ )

(1.8)

provided the denominators are not equal to zero. In Section 2 we give Opial-type integral inequalities and related results using generalized fractional integral operator involving generalized Mittag–Leffler function. As special cases some results of [3, 4, 5] are deduced. In Section 3 exponential convexity of non-negative differences of last term in (1.5) with other two terms is given. 2. Inequalities for generalized fractional integral involving generalized Mittag–Leffler function We need the following lemma . L EMMA 2.1. Series given in (1.2) is absolutely convergent for all values of t provided k < l + α . The following results appear as generalizations of Opial-type integral inequalities for Riemann–Liouville fractional integral. T HEOREM 2.2. Let u1 , u2 and φ , f be same as in Theorem 1.5, also let α , β , k, l, γ , ω > 0 such that k < l + α and β > 1 , then we have γ ,δ ,k

Eα ,β ,l (ω (b − a)α )(b − a)β −1     γ ,δ ,k  Z b ( ε v )(x) v1 (x) ′ + 1 α , β ,l, ω ,a γ ,δ ,k  dx   f (εα ,β ,l,ω ,a+ v2 )(x)φ γ ,δ ,k × v2 (x)φ v (x) a 2 (εα ,β ,l,ω ,a+ v2 )(x)     Z b v1 (x) γ ,δ ,k dx . v2 (x)φ 6 f Eα ,β ,l (ω (b − a)α )(b − a)β −1 (2.1) v2 (x) a Proof. Let us define the followings ( γ ,δ ,k (x − t)β −1Eα ,β ,l (ω (x − t)α ), a 6 t 6 x ; K(x,t) := 0, x < t 6 b, γ ,δ ,k

u1 (x) := (εα ,β ,l,ω ,a+ v1 )(x) = γ ,δ ,k

u2 (x) := (εα ,β ,l,ω ,a+ v2 )(x) = One can observe that ∞

Z x a

Z x a

γ ,δ ,k

(x − t)β −1Eα ,β ,l (ω (x − t)α )v1 (t) dt , γ ,δ ,k

(x − t)β −1Eα ,β ,l (ω (x − t)α )v2 (t) dt .

∞ (γ )kn ω n (x − t)nα (γ )kn ω n (b − a)nα γ ,δ ,k 6∑ = Eα ,β ,l (ω (b − a)α ) Γ( α n + β )( δ ) Γ( α n + β )( δ ) nl nl n=0 n=0

(2.2) (2.3)

98

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C γ ,δ ,k

and Lemma 2.1 ensure the existence of Eα ,β ,l (ω (b − a)α ) for k < l + α , further (x − t)β −1 6 (x − a)β −1 6 (b − a)β −1 for β > 1 . Then γ ,δ ,k

K(x,t) 6 Eα ,β ,l (ω (b − a)α )(b − a)β −1, β > 1, k < l + α .

(2.4)

γ ,δ ,k

Therefore we can take M = Eα ,β ,l (ω (b − a)α )(b − a)β −1 , functions u1 , u2 as in (2.2) and (2.3), in Theorem 1.5 to get (2.1).  In the following we give the extension of Theorem 2.2. T HEOREM 2.3. With same assumptions as in Theorem 2.2 we have γ ,δ ,k

Eα ,β ,l (ω (b − a)α )(b − a)β −1   γ ,δ ,k   (εα ,β ,l,ω ,a+ v1 )(x)  v1 (x) ′  γ ,δ ,k  dx  v2 (x)φ × f (εα ,β ,l,ω ,a+ v2 )(x)φ γ ,δ ,k v2 (x) a (εα ,β ,l,ω ,a+ v2 )(x)     Z b v1 (x) γ ,δ ,k α β −1 dx v2 (x)φ 6 f Eα ,β ,l (ω (b − a) )(b − a) v2 (x) a   Z b  v1 (x) 1 γ ,δ ,k f Eα ,β ,l (ω (b − a)α )(b − a)β −1v2 (x)φ dx. (2.5) 6 b−a a v2 (x) Z b

Proof. Proof is similar to the proof of Theorem 2.2, here we use Theorem 1.6 instead of Theorem 1.5. 

R EMARK 2.4. If k = l = δ = 1 , then we get Opial-type inequalities for integral operator introduced by Prabharkar in . If k = l = 1 and ω = 0 in (2.1), then we get result for Reimann–Liouville fractional integral and using it in (2.5) we get [5, Corollary 3.2]. T HEOREM 2.5. Let u and φ be same as in Theorem 1.7. Also, let α , β , k, l, γ , ω > 0 such that k < l + α and β > 1 , with q > 1 . Then we have Z b a

6

γ ,δ ,k

q(b − a)1−qβ

γ ,δ ,k

(Eα ,β ,l (ω (b − a)α ))q ×φ

6

γ ,δ ,k

|(εα ,β ,l,ω ,a+ v)(x)|1−q φ ′ (|(εα ,β ,l,ω ,a+ v)(x)|)|v(x)|q dx

1 γ ,δ ,k Eα ,β ,l (ω (b − a)α )(b − a)β − q

q(b − a)−qβ γ ,δ ,k

(Eα ,β ,l (ω (b − a)α ))q

Z b a

Z

b

q

|v(x)| dx

a

 q1 !

  φ Eαγ ,,δβ,k,l (ω (b − a)α )(b − a)β |v(x)| dx, 1

the reverse of above inequality holds if φ (x q ) is concave.

99

O PIAL - TYPE INEQUALITIES FOR FRACTIONAL . . .

Proof. Let us define the followings K(x,t) :=

(

γ ,δ ,k

(x − t)β −1Eα ,β ,l (ω (x − t)α ), a 6 t 6 x ; 0, x < t 6 b,

γ ,δ ,k

u(x) := (εα ,β ,l,ω ,a+ v)(x) =

Z x a

γ ,δ ,k

(x − t)β −1Eα ,β ,l (ω (x − t)α )v(t) dt .

(2.6)

From (2.4), for α > 0 and β > 1 , k < l + α we have γ ,δ ,k

K(x,t) 6 Eα ,β ,l (ω (b − a)α )(b − a)β −1, from which we can have Z

x

(K(x,t)) p dt

a

1

p

1

γ ,δ ,k

6 Eα ,β ,l (ω (b − a)α )(b − a)β − q , 1

γ ,δ ,k

here we use 1p + 1q = 1 . Therefore we can take M = Eα ,β ,l (ω (b − a)α )(b − a)β − q and function u defined by (2.6) in Theorem 1.8 and get required inequality.  R EMARK 2.6. If k = l = δ = 1 , then we get Opial-type inequalities for integral operator introduced by Prabharkar in . If k = l = 1 and ω = 0 in Theorem 2.5, then we get extension of (1.4) for Riemann–Liouville fractional integral given in [4, Theorem 3.1]. By using φ (x) = x p+q we have the following result.

1 p

C OROLLARY 2.7. Let α , β , k, l, γ , ω > 0 such that k < l + α and β > 1 , with + q1 = 1 and v ∈ L1 [a, b]. Then following inequalities hold Z b a

γ ,δ ,k

|(εα ,β ,l,ω ,a+ v)(x)| p |v(x)|q dx

6 q(b − a)

  p β − q1

γ ,δ ,k (Eα ,β ,l (ω (b − a)α )) p

γ ,δ ,k

6 q(b − a) pβ (Eα ,β ,l (ω (b − a)α )) p

Z b

Z

b

q

|v(x)| dx

a

 p+q q

|v(x)| p+q dx.

a

R EMARK 2.8. If k = l = δ = 1 , then we get result for integral operator introduced by Prabharkar in . If k = l = 1 and ω = 0 in above inequality, then we get [4, Corollary 3.2].

100

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C

C OROLLARY 2.9. Let v ∈ ACm [0, x] , x ∈ [a, b], u and φ be same as in Theorem 1.7. Let α > 0 , m − β > 1, m = [β ], γ ∈ R, ω > 0, with q > 1. Then we have Zb  a

6

C

 1−q     γ γ Dα ,β ,ω ,0+ v (x) φ ′ C Dα ,β ,ω ,0+ v (x) |v (x)|q dx q

q(m−β )−1

(b − a) 

6

−γ

Eα ,m−β ω (b − a)α



 1  −γ ×φ (b − a)(m−β )− q Eα ,m−β ω (b − a)α 

a

q

−γ

(b − a)q(m−β ) Eα ,m−β ω (b − a)α ×

Zb a

1/q  q (m)  v (x) dx 

Zb



  γ α  (m) φ (b − a)(m−β ) Eα−,m− ω (b − a) (x) v dx, β 1

the reverse of above inequality holds if φ (x q ) is concave. Following Remark 1.4 we obtain the following inequalities for Prabhakar derivative. C OROLLARY 2.10. Let v ∈ ACm [0, x] , x ∈ [a, b], u and φ be same as in Theorem 1.7. Let α > 0 , m − β > 1, m = [β ], γ ∈ R, ω > 0, with q > 1. If v(k) (0+) = 0, k = 0, 1, 2, . . . m − 1, then we have Zb  a

6

 1−q     γ γ q Dα ,β ,ω ,0+ v (x) φ ′ Dα ,β ,ω ,0+ v (x) |v (x)| dx q

q(m−β )−1

(b − a) 

6

−γ

Eα ,m−β ω (b − a)α

(m−β )− 1q

 ×φ (b − a)

−γ

Eα ,m−β

 1/q  Zb q   ω (b − a)α  v(m) (x) dx  a

q −γ

(b − a)q(m−β ) Eα ,m−β ω (b − a)α ×

Zb a





  γ α  (m) (x) dx. φ (b − a)(m−β ) Eα−,m− β ω (b − a) v

 1 The reverse of the above inequalities hold when φ x q is concave.

101

O PIAL - TYPE INEQUALITIES FOR FRACTIONAL . . .

Now we give generalizations of Theorem 1.9 and Theorem 1.10.

T HEOREM 2.11. With same assumptions as in Theorem 1.9, let α , β , k, l, γ , ω > 0 such that k < l + α and β > 1 , with 1p + 1q = 1 and v ∈ L1 [a, b]. Then there exists ξ ∈ I such that the following equality holds γ ,δ ,k

Ψφ ((εα ,β ,l,ω ,a+ v)(x), v(x)) =

ξ φ ′′ (ξ ) − (q − 1) φ ′(ξ ) γ ,δ ,k (Eα ,β ,l (ω (b − a)α ))q 2 q ξ 2q−1 ×(b − a)qβ

Z b

|v(x)|2q dx − 2

Z b a

a

γ ,δ ,k

|(εα ,β ,l,ω ,a+ v)(x)|q |v(x)|q dx.

Proof. From the proof of Theorem 2.5 we have for β > 1, k < l + α Z

x

p

(K(x,t)) dt

a

1

p

1

γ ,δ ,k

6 Eα ,β ,l (ω (b − a)α )(b − a)β − q .

γ ,δ ,k

γ ,δ ,k

1

Using the function u = (εα ,β ,l,ω ,a+ v)(x) and M = Eα ,β ,l (ω (b − a)α )(b − a)β − q in Theorem 1.9 we get required equality. 

T HEOREM 2.12. With same assumptions as in Theorem 1.10, let α , β , k, l, γ , ω > 0 such that k < l + α and β > 1 , with 1p + 1q = 1 and v ∈ L1 [a, b]. Then there exists ξ ∈ I such that the following equality holds γ ,δ ,k

Ψφ1 ((εα ,β ,l,ω ,a+ v)(x), v(x)) γ ,δ ,k

Ψφ2 ((εα ,β ,l,ω ,a+ v)(x), v(x))

=

ξ φ1′′ (ξ ) − (q − 1) φ1′ (ξ ) , ξ φ2′′ (ξ ) − (q − 1) φ2′ (ξ )

(2.7)

provided the denominators are not equal to zero.

Proof. Proof is similar to the poof of Theorem 2.11, here we use Theorem 1.10 instead of Theorem 1.9. 

R EMARK 2.13. If k = l = δ = 1 , then we get results for integral operator introduced by Prabharkar in . If k = l = 1 and ω = 0 , then we get results for Riemann– Liouville fractional integral given in [3, Theorem 3.1, Theorem 3.2].

102

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C

3. Method of exponential convexity Following definitions and properties of exponentially convex functions comes from , also [13, 19]. Throughout we consider I is an interval in R. D EFINITION 3.1. A function ψ : I → R is n -exponentially convex in the Jensen sense on I if   n xi + x j ∑ ξi ξ j ψ 2 > 0 i, j=1 holds for all choices ξi ∈ R and xi ∈ I , i = 1, . . . , n . A function ψ : I → R is n -exponentially convex if it is n -exponentially convex in the Jensen sense and continuous on I . R EMARK 3.2. It is clear from the definition that 1 -exponentially convex functions in the Jensen sense are in fact nonnegative functions. Also, n -exponentially convex functions in the Jensen sense are k -exponentially convex in the Jensen sense for every k ∈ N, k 6 n . By definition of positive semi-definite matrices and some basic linear algebra we have the following proposition. P ROPOSITION 3.3. If ψ is an n -exponentially convex in the Jensen sense, then    xi + x j k the matrix ψ is a positive semi-definite matrix for all k ∈ N, k 6 n . 2 i, j=1 k   xi + x j > 0 for all k ∈ N, k 6 n . Particularly, det ψ 2 i, j=1 D EFINITION 3.4. A function ψ : I → R is exponentially convex in the Jensen sense on I if it is n -exponentially convex in the Jensen sense for all n ∈ N. A function ψ : I → R is exponentially convex if it is exponentially convex in the Jensen sense and continuous. R EMARK 3.5. It is known (and easy to show) that ψ : I → (0, ∞) is a log-convex in the Jensen sense if and only if   x+y + β 2 ψ (y) > 0 α 2 ψ (x) + 2αβ ψ 2 holds for every α , β ∈ R and x, y ∈ I . It follows that a function is log-convex in the Jensen-sense if and only if it is 2 -exponentially convex in the Jensen sense. Also, using basic convexity theory it follows that a function is log-convex if and only if it is 2 -exponentially convex. Next we need divided differences, commonly used when dealing with functions that have different degree of smoothness.

O PIAL - TYPE INEQUALITIES FOR FRACTIONAL . . .

103

D EFINITION 3.6. The second order divided difference of a function f : I → R at mutually different points y0 , y1 , y2 ∈ I is defined recursively by [yi ; f ] = f (yi ) , i = 0, 1, 2 f (yi+1 ) − f (yi ) , i = 0, 1 [yi , yi+1 ; f ] = yi+1 − yi [y1 , y2 ; f ] − [y0 , y1 ; f ] [y0 , y1 , y2 ; f ] = . y2 − y0

(3.1)

R EMARK 3.7. The value [y0 , y1 , y2 ; f ] is independent of the order of the points y0 , y1 and y2 . This definition may be extended to include the case in which some or all the points coincide. Namely, taking the limit y1 → y0 in (3.1), we get lim [y0 , y1 , y2 ; f ] = [y0 , y0 , y2 ; f ] =

y1 →y0

f (y2 ) − f (y0 ) − f ′ (y0 )(y2 − y0 ) , y2 6= y0 (y2 − y0)2

provided that f ′ exists, and furthermore, taking the limits yi → y0 , i = 1, 2 in (3.1), we get f ′′ (y0 ) lim lim [y0 , y1 , y2 ; f ] = [y0 , y0 , y0 ; f ] = y2 →y0 y1 →y0 2 provided that f ′′ exists. Motivated by inequalities in (1.5) we define the following functionals, non-negative differences of last term with other two terms as follows: Ψ1φs (u, v) = Ψ2φs (u, v) =

Z  1   1 q   b q |v(x)|q dx φ (b − a) q M|v(x)| dx − q φ M M a a (3.2) Z b Z b   1 φ (b − a) q M|v(x)| dx− |u(x)|1−q φ ′ (|u(x)|)|v(x)|q dx.

q q M (b − a)

q M q (b − a)

Z b

a

a

(3.3) We use a method of producing n -exponentially convex and exponentially convex functions given in , to prove the n -exponential convexity for the functionals Ψiφ (u, v), i = 1, 2 defined in (3.2), (3.3). T HEOREM 3.8. Let J be an interval in R and ϒ = {φs : s ∈ J} be a family of functions defined on an interval I in R, such that the function s 7→ [y0 , y1 , y2 ; Fφs ] is n exponentially convex in the Jensen sense on J for every three mutually different points 1 y0 , y1 , y2 ∈ I , where Fφs (y) = φs (y q ). Let Ψiφ (u, v), i = 1, 2 be functionals defined in (3.2), (3.3). Then s 7→ Ψiφs (u, v), i = 1, 2 are n -exponentially convex functions in the Jensen sense on J . If the functions s 7→ Ψiφs (u, v), i = 1, 2 are also continuous on J , then are n -exponentially convex on J . Proof. See the proof of Theorem 5.11 in .



104

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C

C OROLLARY 3.9. Let J be an interval in R and ϒ = {φs : s ∈ J} be a family of functions defined on an interval I in R, such that the function s 7→ [y0 , y1 , y2 ; Fφs ] is exponentially convex in the Jensen sense on J for every three mutually different points 1 y0 , y1 , y2 ∈ I , where Fφs (y) = φs (y q ). Let Ψiφ (u, v), i = 1, 2 be functionals defined in (3.2), (3.3). Then functions s 7→ Ψiφs (u, v), i = 1, 2 are exponentially convex function in the Jensen sense on J . If the functions s 7→ Ψiφs (u, v), i = 1, 2 are continuous on J , then are exponentially convex on J . Let us denote means for φs , φ p ∈ Ω by   1 Ψ (u,v) s−p   , s 6= p ,  Ψiiφφs (u,v)  p µs,p (Ψi , Ω) = d Ψ (u,v)  s   exp dsΨ iφ(u,v) , s = p,

(3.4)

i φs

for i = 1, 2 .

T HEOREM 3.10. Let J be an interval in R and Ω = {φs : s ∈ J} be a family of functions defined on an interval I in R, such that the function s 7→ [y0 , y1 , y2 ; Fφs ] is 2 exponentially convex in the Jensen sense on J for every three mutually different points 1 y0 , y1 , y2 ∈ I , where Fφs (y) = φs (y q ). Let Ψiφ (u, v), i = 1, 2 be functionals defined in (3.2), (3.3). Then the following statements hold: (i) If the functions s 7→ Ψiφ (u, v), i = 1, 2 are continuous on J , then are 2 -exponentially convex functions on J . If the functions s 7→ Ψiφs (u, v), i = 1, 2 are additionally positive, then are also log-convex on J , and for r, s,t ∈ J such that r < s < t , we have t−r t−s s−r Ψiφs (u, v) 6 Ψiφr (u, v) Ψiφt (u, v) i = 1, 2. (3.5)

(ii) If the functions s 7→ Ψiφs (u, v), i = 1, 2 are positive and differentiable on J , then for every s, p, r,t ∈ J , such that s 6 r and p 6 t , we have

µs,p (Ψi , Ω) 6 µr,t (Ψi , Ω) , i = 1, 2. Proof. See the proof of Theorem 5.13 in .

(3.6)



R EMARK 3.11. The results from Theorem 3.8, Corollary 3.9 and Theorem 3.10 still hold when two of the points y0 , y1 , y2 ∈ I coincide, for a family of differentiable functions φs such that the function s 7→ [y0 , y1 , y2 ; Fφs ] is n -exponentially convex in the Jensen sense (exponentially convex in the Jensen sense, log-convex in the Jensen sense). Furthermore, they still hold when all three points coincide for a family of twice differentiable functions with the same property. The proofs can be obtained by recalling Remark 3.7 and suitable characterization of convexity.

O PIAL - TYPE INEQUALITIES FOR FRACTIONAL . . .

105

4. Concluding remarks In this last section we are interested to give remarks on proved results in the sense that they are extended results to fractional calculus, and also similar results can be obtained taking other non negative differences as functionals. R EMARK 4.1. Opial and Opial-type inequalities have many applications in differential calculus (see references), of course have equal importance in fractional differential calculus. Here in this paper given results are generalizations of Opial-type inequalities for fractional differential calculus. R EMARK 4.2. As we prove the n -exponential convexity of the functionals Ψiφs (u, v), i = 1, 2 obtained from the inequalities given in (1.5), similarly we can define and prove the n -exponential convexity of functionals obtained from the inequalities given for fractional integral operators involving ML-functions but here we omit the details. Some of the estimates can be applied for proving existence and uniqueness of some linear and nonlinear fractional differential equations containing Caputo, Prabhakar, Caputo-Prabhakar derivative operators  which is a focus of our next research. Acknowledgement. Research of first author is supported by COMSATS Institute of Information Technology, Islamabad Pakistan. The research of Josip Peˇcari´c was fully supported by the Croatian Science Foundation under the project 5435. The author Zivorad Tomovski is supported under the European Commission and the Croatian Ministry of Science, Education and Sports Co-Financing Agreement No. 291823. In particular, ZT acknowledges project financing from the Maria Curie FP7-PEOPLE2011-COFUND program NEWFELPRO Grant Agreement No. 37 – Anomalous diffusion. REFERENCES  R. P. A GARWAL AND P. Y. H. PANG, Opial Inequalities with Applications in Differential and Difference Equations, Kluwer Academic Publishers, Dordrecht, Boston, London 1995.  G. A. A NASTASSIOU, Advanced inequalities, 11, World Scientific, 2011. ˇ ´ , More on certain Opial-type inequality for  M. A NDRI C´ , A. BARBIR , G. FARID AND J. P E CARI C fractional derivatives, Nonlinear Functional Analysis and Applications, 19, No. 4 (2014), 565–583. ˇ ´ , Opial-type inequality due to Agarwal–Pang  M. A NDRI C´ , A. BARBIR , G. FARID AND J. P E CARI C and fractional differential inequalities, Integral Transforms and Special Functions, 25, No. 4 (2014). ˇ ´ , An Opial-type inequality and exponentially  M. A NDRI C´ , A. BARBIR , S. I QBAL , AND J. P E CARI C convex functions, to appear. ˇ ´ AND I. P ERI C´ , Improvements of composition rule for the Canavati fractional  M. A NDRI C´ , J. P E CARI C derivatives and applications to Opial-type inequalities, Dynam. Systems. Appl., 20 (2011), 383–394. ˇ ´ AND ATIQ UR R EHMAN, Exponential convexity, positive  M. A NWAR , J. JAK Sˇ ETI C´ , J. P E CARI C semi-definite matrices and fundamental inequalities, J. Math. Inequal., 4 (2) (2010), 171–189.  L. C URIEL , L. G ALU E´ , A generalization of the integral operators involving the Gauss hypergeometric function, Revista T´ecnica de la Facultad de Ingenieria Universidad del Zulia, 19 (1) (1996), 17–22. ˇ ´ , Opial type integral inequalities for fractional derivatives, Fractional  G. FARID AND J. P E CARI C Differential Calculus, 2, No. 1 (2012), 31–54.

106

ˇ ´ AND Z. T OMOVSKI G. FARID , J. P E CARI C

ˇ ´ , Opial type integral inequalities for fractional derivatives II, Fractional  G. FARID AND J. P E CARI C Differential Calculus, 2, No. 2 (2012), 139–155. ˇ ´ , Opial type integral Inequalities for Widder derivatives and linear differen G. FARID AND P E CARI C tial operators, Int. J. Analysis Appl., Int. J. Ana. App., Vol. 7, No. 1 (2015), 38–49.  R. G ARRA , R. G ORENFLO , F. P OLITO , Z. T OMOVSKI , Hilfer-Prabhakar Derivatives and Some Applications, Applied Mathematics and Computation, Vol. 242, 576–589, 2014. ˇ ´ , Exponential Convexity Method, J. Convex Anal., 20 (1) (2013), 181–  J. JAK Sˇ ETI C´ AND J. P E CARI C 197.  A. A. K ILBAS , M. S AIGO , R. K. S AXENA, Generalized Mittag–Leffler function and generalized fractional calculus operators, Integral Transform. Spec. Funct. 15 (2004) 31–49.  A. A. K ILBAS , H. M. S RIVASTAVA AND J. J. T RUJILLO, Theory and Applications of fractional derivatial Equations, North-Holland Mathematics Studies, 204, Elsevier, New York-London, 2006.  K. M ILLER AND B. ROSS , An introduction to the fractional calculus and fractional differential Equations, John Wiley and Sons Inc., New York, 1993. ˇ ´ , Generalizations of two inequalities of Godunova and Levin,  D. S. M ITRINOVI C´ AND J. E. P E CARI C Bull. Polish Acad. Sci. Math., 36 (1988), 645–648.  K. O LDHAM AND J. S PANIER, The fractional calculus, Academic Press, New York-London, 1974. ˇ ´ AND J. P ERI C´ , Improvements of the Giaccardi and Petrovi´c inequality and related Sto J. P E CARI C larsky type means, An. Univ. Craiova Ser. Mat. Inform., 39 (1) (2012), 65–75. ˇ ´ , F. P ROSCHAN AND Y. L. T ONG, Convex Functions, Partial Orderings and Statistical  J. E. P E CARI C Applications, Academic Press, Inc. (1992).  T. O. S ALIM , AND A. W. FARAJ , A Generalization of Mittag–Leffler function and integral operator associated with fractional calculus, J. Fract. Calc. Appl. Vol. 3. July 2012, No. 5, pp. 1–13. ˇ T OMOVSKI , Fractional calculus with an integral operator containing  H. M. S RIVASTAVA , AND Z. generalized Mittag–Leffler function in the kernal, Appl. Math. Comput. (2009), doi:10.1016j.amc. 2009.01.055.  Z. O PIAL, Sur une in´egalit´e, Ann. Polon. Math., 8 (1960), 29–32. ˇ ´ , F RANK P ROSCHAN , Y. L. T ONG, Convex Functions, Partial Orderings, and Statistical  J. P E CARI C Applications, Academic Press, Inc., 1992.  T. R. P RABHAKAR, A singular integral equation with a generalized Mittag–Leffler function in the kernel, Yokohama Math. J. 19 (1971) 7–15.  D. V. W IDDER, The Laplace transform, Princeton Uni. Press, New Jersey, 1941.