Advances in Difference Equations
Volume 2009 (2009), Article ID 981728, 47 pages
doi:10.1155/2009/981728
Research Article

A Survey on Semilinear Differential Equations and Inclusions Involving Riemann-Liouville Fractional Derivative

1Department of Mathematical Sciences, Florida Institute of Technology, Melboune, FL 32901-6975, USA
2Laboratoire de Mathématiques, Université de Sidi Bel Abbès, BP 89, 22000 Sidi Bel Abbès, Algeria

Received 16 July 2008; Revised 4 December 2008; Accepted 5 February 2009

Academic Editor: Alberto Cabada

Copyright © 2009 Ravi P. Agarwal et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

We establish sufficient conditions for the existence of mild solutions for some densely defined semilinear functional differential equations and inclusions involving the Riemann-Liouville fractional derivative. Our approach is based on the 𝒞 0 -semigroups theory combined with some suitable fixed point theorems.

1. Introduction

Differential equations and inclusions of fractional order have recently proved to be valuable tools in the modeling of many phenomena in various fields of science and engineering. Indeed we can find numerous applications in viscoelasticity, electrochemistry, electromagnetism, and so forth. For details, including some applications and recent results, see the monographs of Kilbas et al. [1], Kiryakova [2], Miller and Ross [3], Podlubny [4] and Samko et al. [5], and the papers of Agarwal et al. [6], Diethelm et al. [7, 8], El-Sayed [911], Gaul et al. [12], Glockle and Nonnenmacher [13], Lakshmikantham and Devi [14], Mainardi [15], Metzler et al. [16], Momani et al. [17, 18], Podlubny et al. [19], Yu and Gao [20] and the references therein. Some classes of evolution equations have been considered by El-Borai [21, 22], Jaradat et al. [23] studied the existence and uniqueness of mild solutions for a class of initial value problem for a semilinear integrodifferential equation involving the Caputo's fractional derivative.

In this survey paper, we give existence results for various classes of initial value problems for fractional semilinear functional differential equations and inclusions, both cases of finite and infinite delay are considered. More precisely the paper is organized as follows. In the second section we introduce notations, definitions, and preliminary facts that will be used in the remainder of this paper. In the third section we will be concerned with semilinear functional differential equations with finite as well infinite delay. In the forth section, we consider semilinear functional differential equation of neutral type for the both cases of finite and infinite delay. Section 5 is devoted to the study of functional differential inclusions, we examine the case when the right-hand side is convex valued as well as nonconvex valued. In Section 6, we will be concerned with perturbed functional differential equations and inclusions. In the last section, we give some existence results of extremal solutions in ordered Banach spaces.

2. Preliminaries

In this section, we introduce notations, definitions, and preliminary facts which are used throughout this paper. Let ( 𝐸 , ) be a Banach space and 𝐽 a compact real interval. 𝐶 ( 𝐽 , 𝐸 ) is the Banach space of all continuous functions from 𝐽 into 𝐸 with the normFor image/svg+xml𝜓∈𝐶([−𝑟,0],𝐸) the norm of image/svg+xml𝜓 is defined byFor 𝜙 𝐶 ( [ 𝑟 , 𝑏 ] , 𝐸 ) the norm of 𝜙 is defined by 𝐵 ( 𝐸 ) denotes the Banach space of bounded linear operators from 𝐸 into 𝐸 , with norm 𝐿 1 ( 𝐽 , 𝐸 ) denotes the Banach space of measurable functions 𝑦 𝐽 𝐸 which are Bochner integrable normed by

Definition 2.1. A semigroup of class ( 𝐶 0 ) is a one parameter family 𝑇 ( 𝑡 ) 𝑡 0 𝐵 ( 𝐸 ) satisfying the conditions
(i) 𝑇 ( 0 ) = 𝐼 ; (ii) 𝑇 ( 𝑡 ) 𝑇 ( 𝑠 ) = 𝑇 ( 𝑡 + 𝑠 ) , for all 𝑡 , 𝑠 0 ; (iii)the map 𝑡 𝑇 ( 𝑡 ) ( 𝑥 ) is strongly continuous, for each 𝑥 𝐸 , that is,

It is well known that the operator 𝐴 generates a ( 𝐶 0 ) semigroup if 𝐴 satisfies

(i) 𝐷 ( 𝐴 ) = 𝐸 ; (ii)the Hille-Yosida condition, that is, there exists 𝑀 0 and 𝜔 such that ( 𝜔 , ) 𝜌 ( 𝐴 ) , s u p { ( 𝜆 𝜔 ) 𝑛 | ( 𝜆 𝐼 𝐴 ) 𝑛 | 𝜆 > 𝜔 , 𝑛 } 𝑀 , where 𝜌 ( 𝐴 ) is the resolvent set of 𝐴 and 𝐼 is the identity operator in 𝐸 . For more details on strongly continuous operators, we refer the reader to the books of Goldstein [24], Fattorini [25], and the papers of Travis and Webb [26, 27], and for properties on semigroup theory we refer the interested reader to the books of Ahmed [28], Goldstein [24], and Pazy [29].

In all our paper we adopt the following definitions of fractional primitive and fractional derivative.

Definition 2.2 (see [4, 5]). The Riemann-Liouville fractional primitive of order 𝛼 of a function ( 0 , 𝑏 ] of order 𝛼 + is defined byprovided the right side is pointwise defined on ( 0 , 𝑏 ] , and where Γ is the gamma function.

For instance, 𝐼 𝛼 0 exists for all 𝛼 > 0 , when 𝐶 ( ( 0 , 𝑏 ] , ) 𝐿 1 ( ( 0 , 𝑏 ] , ) ; note also that when 𝐶 ( [ 0 , 𝑏 ] , ) , then 𝐼 𝛼 0 𝐶 ( [ 0 , 𝑏 ] , ) and moreover 𝐼 𝛼 0 ( 0 ) = 0 .

Definition 2.3 (see [4, 5]). The Riemann-Liouville fractional derivative of order 𝛼 ( 0 , 1 ) of a continuous function ( 0 , 𝑏 ] is defined by

Let ( 𝑋 , 𝑑 ) be a metric space. We use the notations Consider 𝐻 𝑑 𝒫 ( 𝑋 ) × 𝒫 ( 𝑋 ) + { } given bywhere 𝑑 ( 𝐴 , 𝑏 ) = i n f 𝑎 𝐴 𝑑 ( 𝑎 , 𝑏 ) , 𝑑 ( 𝑎 , 𝐵 ) = i n f 𝑏 𝐵 𝑑 ( 𝑎 , 𝑏 ) . Then ( 𝒫 b d , c l ( 𝑋 ) , 𝐻 𝑑 ) is a metric space and ( 𝒫 c l ( 𝑋 ) , 𝐻 𝑑 ) is a generalized metric space (see [30]).

A multivalued map 𝐹 𝐽 𝒫 c l ( 𝑋 ) is said to be measurable if, for each 𝑥 𝑋 , the function 𝑌 𝐽 defined byis measurable.

Definition 2.4. A measurable multivalued function 𝐹 𝐽 𝑃 b d , c l ( 𝑋 ) is said to be integrably bounded if there exists a function 𝑤 𝐿 1 ( 𝐽 , 𝑅 + ) such that 𝑣 𝑤 ( 𝑡 ) a.e. 𝑡 𝐽 for all 𝑣 𝐹 ( 𝑡 ) .

A multivalued map 𝐹 𝑋 𝒫 ( 𝑋 ) is convex (closed) valued if 𝐹 ( 𝑥 ) is convex (closed) for all 𝑥 𝑋 . image/svg+xml𝐹 is bounded on bounded sets if image/svg+xml𝐹(𝐵)= 𝑥∈𝐵 𝐹(𝑥) is bounded in image/svg+xml𝑋 for all image/svg+xml𝐵∈𝒫 bd (𝑋) , that is, image/svg+xmlsup 𝑥∈𝐵 {sup{|𝑦|∶𝑦∈𝐹(𝑥)}}<∞ .

image/svg+xml𝐹 is called upper semicontinuous (u.s.c. for short) on image/svg+xml𝑋 if for each image/svg+xml𝑥 0 ∈𝑋 the set image/svg+xml𝐹(𝑥 0 ) is nonempty, closed subset of image/svg+xml𝑋 , and for each open set image/svg+xml𝒰 of image/svg+xml𝑋 containing image/svg+xml𝐹(𝑥 0 ) , there exists an open neighborhood image/svg+xml𝒱 of image/svg+xml𝑥 0 such that image/svg+xml𝐹(𝒱)⊆𝒰.𝐹 is said to be completely continuous if image/svg+xml𝐹(𝐵) is relatively compact for every image/svg+xml𝐵∈𝑃 bd (𝑋). If the multivalued map image/svg+xml𝐹 is completely continuous with nonempty compact valued, then image/svg+xml𝐹 is u.s.c. if and only if image/svg+xml𝐹 has closed graph, that is, image/svg+xml𝑥 𝑛 →𝑥 ,𝑦 𝑛 →𝑦 ,𝑦 𝑛 ∈𝐹(𝑥 ) imply image/svg+xml𝑦 ∈𝐹(𝑥 ).

Definition 2.5. A multivalued map image/svg+xml𝛽∶𝐽×𝐶([−𝑟,0],𝐸)→𝑃(𝐸) is said to be Carathéodory if
(i)image/svg+xml𝑡↦𝛽(𝑡,𝑥) is measurable for each 𝑥 𝐶 ( [ 𝑟 , 0 ] , 𝐸 ) , (ii) 𝑥 𝛽 ( 𝑡 , 𝑥 ) is u.s.c. for almost all 𝑡 𝐽 . Furthermore, a Carathéodory map 𝛽 is said to be 𝐿 1 -Carathéodory if
(iii)for each real number 𝜌 > 0 , there exists a function 𝜌 𝐿 1 ( 𝐽 , 𝑅 + ) such thatfor a.e. 𝑡 𝐽 , and for all | 𝑥 | 𝜌 .

Definition 2.6. A multivalued operator 𝐹 𝑋 𝒫 c l ( 𝑋 ) is called
(a) 𝛾 -Lipschitz if and only if there exists 𝛾 > 0 such that(b)contraction if and only if it is 𝛾 -Lipschitz with 𝛾 < 1 , (c) 𝐹 has a fixed point if there exists 𝑥 𝑋 such that 𝑥 𝐹 ( 𝑥 ) . The fixed point set of the multivalued operator 𝐹 will be denoted by F i x 𝐹 .

For more details on multivalued maps and the proof of the known results cited in this section we refer interested reader to the books of Deimling [31], Gorniewicz [32], and Hu and Papageorgiou [33].

Essential for the main results of this paper, we state a generalization of Gronwall's lemma for singular kernels [34, Lemma 7.1.1].

Lemma 2.7. Let 𝑣 , 𝑤 [ 0 , 𝑏 ] [ 0 , ) be continuous functions. If 𝑤 ( ) is nondecreasing and there are constants 𝑎 > 0 and 0 < 𝛼 < 1 such that then there exists a constant 𝑘 = 𝑘 ( 𝛼 ) such that for every 𝑡 [ 0 , 𝑏 ] .

In the sequel, the following fixed point theorems will be used. The following fixed point theorem for contraction multivalued maps is due to Covitz and Nadler [35].

Theorem 2.8. Let ( 𝑋 , 𝑑 ) be a complete metric space. If 𝐹 𝑋 𝒫 c l ( 𝑋 ) is a contraction, then F i x 𝐹 .

The nonlinear alternative of Leray-Schauder applied to completely continuous operators [36].

Theorem 2.9. Let 𝑋 be a Banach space, and 𝐶 𝑋 convex with 0 𝐶 . Let 𝐹 𝐶 𝐶 be a completely continuous operator. Then either
(a) 𝐹 has a fixed point, or (b) the set = { 𝑥 𝐶 𝑥 = 𝜆 𝐹 ( 𝑥 ) , 0 < 𝜆 < 1 } is unbounded.

The following is the multivalued version of the previous theorem due to Martelli [37].

Theorem 2.10. Let 𝑇 𝑋 𝒫 c p , c v ( 𝑋 ) be an upper semicontinuous and completely continuous multivalued map. If the set is bounded, then 𝑇 has a fixed point.

To state existence results for perturbed differential equations and inclusions we will use the following fixed point theorem of Krasnoselskii-Scheafer type of the sum of a completely continuous operator and a contraction one due to Burton and Kirk [38].

Theorem 2.11. Let 𝑋 be a Banach space, and 𝒜 , 𝑋 𝑋 two operators satisfying
(i) 𝒜 is a contraction; (ii) is completely continuous. Then either
(a) the operator equation 𝑦 = 𝒜 ( 𝑦 ) + ( 𝑦 ) has a solution, or (b) the set = { 𝑢 𝑋 𝑢 = 𝜆 𝒜 ( 𝑢 / 𝜆 ) + 𝜆 ( 𝑢 ) } is unbounded for 𝜆 ( 0 , 1 ) .

Recently Dhage states the multivalued version of the previous theorem.

Theorem 2.12 (see [39, 40]). Let 𝑋 be a Banach space, 𝒜 𝑋 𝑃 c l , c v , b d ( 𝑋 ) and 𝑋 𝑃 c p , c v ( 𝑋 ) , two multivalued operators satisfying
(a) 𝒜 is a contraction; (b) is completely continuous. Then either
(i) The operator inclusion 𝜆 𝑥 𝒜 𝑥 + 𝑥 has a solution for 𝜆 = 1 , or (ii) the set = { 𝑢 𝑋 𝑢 𝜆 𝒜 𝑢 + 𝜆 𝑢 , 0 < 𝜆 < 1 } is unbounded.

In the literature devoted to equations with finite delay, the phase space is much of time the space of all continuous functions on [ 𝑟 , 0 ] , 𝑟 > 0 , endowed with the uniform norm topology. When the delay is infinite, the notion of the phase space plays an important role in the study of both qualitative and quantitative theory, a usual choice is a seminormed space introduced by Hale and Kato [41] and satisfying the following axioms.

(A1)There exist a positive constant 𝐻 and functions 𝐾 ( ) , 𝑀 ( ) + + with 𝐾 continuous and 𝑀 locally bounded, such that for any 𝑎 > 0 , if 𝑥 ( , 𝑎 ] 𝐸 , 𝑥 , and 𝑥 ( ) is continuous on [ 0 , 𝑎 ] , then for every 𝑡 [ 0 , 𝑎 ] the following conditions hold:(i) 𝑥 𝑡 is in ; (ii) | 𝑥 ( 𝑡 ) | 𝐻 𝑥 𝑡 ; (iii) 𝑥 𝑡 𝐾 ( 𝑡 ) s u p { | 𝑥 ( 𝑠 ) | 0 𝑠 𝑡 } + 𝑀 ( 𝑡 ) 𝑥 0 , and 𝐻 , 𝐾 , and 𝑀 are independent of 𝑥 ( ) .(A2)For the function 𝑥 ( ) in ( 𝐴 1 ) , 𝑥 𝑡 is a -valued continuous function on [ 0 , 𝑎 ] . (A3)The space is complete. Denote byHereafter are some examples of phase spaces. For other details we refer, for instance, to the book by Hino et al. [42].

Example 2.13. The spaces B C , B U C , 𝐶 , and 𝐶 0 .
BC is the space of bounded continuous functions defined from ( , 0 ] to 𝐸 ;
BUC is the space of bounded uniformly continuous functions defined from ( , 0 ] to 𝐸 ; We have that the spaces 𝐶 0 , and ( 𝐴 1 ) - ( 𝐴 3 ) satisfy conditions B C . ( 𝐴 1 ) , ( 𝐴 3 ) satisfies ( 𝐴 2 ) but 𝐶 𝑔 , 𝑈 𝐶 𝑔 , 𝐶 𝑔 is not satisfied.

Example 2.14. The spaces 𝐶 0 𝑔 , and 𝑔 .
Let ( , 0 ] be a positive continuous function on 𝑔 . We define We consider the following condition on the function 𝑎 > 0 , s u p 0 𝑡 𝑎 s u p { ( 𝑔 ( 𝑡 + 𝜃 ) / 𝑔 ( 𝜃 ) ) < 𝜃 𝑡 } < . .
𝐶 𝑔 For all 𝐶 0 𝑔 Then we have that the spaces ( 𝐴 3 ) and ( 𝐴 1 ) satisfy conditions ( 𝐴 3 ) . They satisfy conditions 𝑔 1 and 𝐶 𝛾 if 𝛾 holds.

Example 2.15. The space 𝐶 𝛾 .
For any real constant 𝐶 𝛾 = 𝜙 𝐶 ( ( , 0 ] , 𝐸 ) l i m 𝜃 𝑒 𝛾 𝜃 𝜙 ( 𝜃 ) e x i s t i n 𝐸 ( 2 . 2 1 ) , we define the functional space 𝜙 = s u p { 𝑒 𝛾 𝜃 | 𝜙 ( 𝜃 ) | 𝜃 0 } . ( 2 . 2 2 ) bysendowed with the following normThen in the space 0 < 𝛼 1 . the axioms 𝐷 𝛼 𝑦 ( 𝑡 ) = 𝐴 𝑦 ( 𝑡 ) + 𝑓 𝑡 , 𝑦 𝑡 , 𝑡 𝐽 = [ 0 , 𝑏 ] , ( 3 . 1 ) 𝑦 ( 𝑡 ) = 𝜙 ( 𝑡 ) , 𝑡 [ 𝑟 , 0 ] , ( 3 . 2 ) are satisfied.

3. Semilinear Functional Differential Equations

3.1. Introduction

Functional differential and partial differential equations arise in many areas of applied mathematics and such equations have received much attention in recent years. A good guide to the literature for functional differential equations is the books by Hale [43] and Hale and Verduyn Lunel [44], Kolmanovskii and Myshkis [45], and Wu [46] and the references therein.

In a series of papers (see [4750]), the authors considered some classes of initial value problems for functional differential equations involving the Riemann-Liouville and Caputo fractional derivatives of order 𝐷 𝛼 In [51, 52] some classes of semilinear functional differential equations involving the Riemann-Liouville have been considered. For more details on the geometric and physical interpretation for fractional derivatives of both the Riemann-Liouville and Caputo types see [53, 54].

In the following, we consider the semilinear functional differential equation of fractional order of the formwhere 𝜙 [ 𝑟 , 0 ] 𝐸 is the standard Riemann-Liouville fractional derivative, 𝜙 ( 0 ) = 0 is a continuous function, ( 𝐸 , | | ) is a closed linear operator (possibly unbounded), 𝑦 a given continuous function with [ 𝑟 , 𝑏 ] , and 𝑡 𝐽 a real Banach space. For any function 𝑦 𝑡 defined on 𝐶 ( [ 𝑟 , 0 ] , 𝐸 ) and any 𝑦 𝑡 ( 𝜃 ) = 𝑦 ( 𝑡 + 𝜃 ) , 𝜃 [ 𝑟 , 0 ] . ( 3 . 3 ) we denote by 𝑦 𝑡 ( ) the element of 𝑡 𝑟 defined byHere 𝛼 = 1 represents the history of the state from time 𝐴 𝐷 ( 𝐴 ) 𝐸 𝐸 , up to the present time ( 𝐶 0 ) .

The reason for studying (3.1) is that it appears in mathematical models of viscoelasticity [55], and in other fields of science [54, 56]. Equation (3.1) is equivalent to solve an integral equation of convolution type. It is also of interest to explore the neighborhood of the diffusion ( { 𝑇 ( 𝑡 ) } 𝑡 0 ). In this survey paper, we use the fractional derivative in the Riemann-Liouville sense. The problems considered in the survey are subject to zero data, which in this case, the Riemann-Liouville and Caputo fractional derivatives coincide. From a practical point of view, in some mathematical models it is more appropriate to consider traditional initial or boundary data. This is what we are considering in this survey.

In all our paper we suppose that the operator 𝑀 = s u p 𝑇 ( 𝑡 ) 𝐵 ( 𝐸 ) 𝑡 𝐽 . ( 3 . 4 ) is the infinitesimal generator of a 𝑦 [ 𝑟 , 𝑏 ] 𝐸 -semigroup 𝑦 ( 𝑡 ) = 𝜙 ( 𝑡 ) , 𝑡 [ 𝑟 , 0 ] , . Denote byBefore stating our main results in this section for problem (3.1) and (3.2) we give the definition of the mild solution.

Definition 3.1 (see [23]). One says that a continuous function 𝑓 𝐽 × 𝐶 ( [ 𝑟 , 0 ] , 𝐸 ) 𝐸 is a mild solution of problem (3.1) and (3.2) if ( 𝐻 1 ) and

3.2. Existence Results for Finite Delay

By using the Banach's contraction principle, we get the following existence result for problem (3.1) and (3.2).

Theorem 3.2. Let | 𝑓 ( 𝑡 , 𝑢 ) 𝑓 ( 𝑡 , 𝑣 ) | 𝑘 𝑢 𝑣 𝐶 , f o r 𝑡 𝐽 a n d e v e r y 𝑢 , 𝑣 𝐶 ( [ 𝑟 , 0 ] , 𝐸 ) . ( 3 . 6 ) continuous. Assume the following.
[ 𝑟 , 𝑏 ] . There exists a nonnegative constant 𝐶 ( [ 𝑟 , 𝑏 ] , 𝐸 ) 𝐶 ( [ 𝑟 , 𝑏 ] , 𝐸 ) such that Then there exists a unique mild solution of problem (3.1) and (3.2) on

Proof. Transform the IVP (3.1) and (3.2) into a fixed point problem. Consider the operator 1 = , 𝑛 + 1 = 𝑛 . ( 3 . 8 ) defined byLet us define the iterates of operator 𝑛 byIt will be sufficient to prove that | | 𝑛 ( 𝑦 ) ( 𝑡 ) 𝑛 | | ( 𝑧 ) ( 𝑡 ) ( 𝑘 𝑀 ) 𝑛 𝑡 Γ ( 𝑛 𝛼 + 1 ) 𝑛 𝛼 𝑦 𝑧 . ( 3 . 9 ) is a contraction operator for | | | | 𝑀 ( 𝑦 ) ( 𝑡 ) ( 𝑧 ) ( 𝑡 ) Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 | | 𝑓 𝑠 , 𝑦 𝑠 𝑓 𝑠 , 𝑧 𝑠 | | 𝑑 𝑠 𝑘 𝑀 Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑦 𝑠 𝑧 𝑠 𝐶 𝑑 𝑠 𝑘 𝑀 Γ ( 𝛼 ) 𝑡 0 𝑦 𝑧 ( 𝑡 𝑠 ) 𝛼 1 = 𝑑 𝑠 𝑘 𝑀 𝑡 Γ ( 𝛼 + 1 ) 𝛼 𝑦 𝑧 . ( 3 . 1 0 ) sufficiently large. For every 𝑛 = 1 we haveIndeed,Therefore (3.9) is proved for 𝑛 + 1 . Assuming by induction that (3.9) is valid for 𝑛 , thenand then (3.9) follows for .
Now, taking ( 𝐻 2 ) sufficiently large in (3.9) yield the contraction of operator { 𝑇 ( 𝑡 ) } 𝑡 𝐽 .
Consequently 𝑡 > 0 has a unique fixed point by the Banach's contraction principle, which gives rise to a unique mild solution to the problem (3.1) and (3.2).

The following existence result is based upon Theorem 2.9.

Theorem 3.3. Assume that the following hypotheses hold.
( 𝐻 3 ) The semigroup 𝑝 , 𝑞 𝐶 ( 𝐽 , + ) is compact for | 𝑓 ( 𝑡 , 𝑢 ) | 𝑝 ( 𝑡 ) + 𝑞 ( 𝑡 ) 𝑢 𝐶 , f o r e a c h 𝑡 𝐽 , a n d e a c h 𝑢 𝐶 ( [ 𝑟 , 0 ] , 𝐸 ) . ( 3 . 1 2 ) . [ 𝑟 , 𝑏 ] . There exist functions such that Then the problem (3.1) and (3.2) has at least one mild solution on { 𝑦 𝑛 }

Proof. Transform the IVP (3.1) and (3.2) into a fixed point problem. Consider the operator 𝑦 𝑛 𝑦 as defined in Theorem 3.2. To show that 𝐶 ( [ 𝑟 , 𝑏 ] , 𝐸 ) is continuous, let us consider a sequence | | 𝑦 𝑛 | | | | | | 1 ( 𝑡 ) ( 𝑦 ) ( 𝑡 ) Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑓 𝑇 ( 𝑡 𝑠 ) 𝑠 , 𝑦 𝑛 𝑠 ) 𝑓 𝑠 , 𝑦 𝑠 | | | | 𝑑 𝑠 𝑀 𝑏 𝛼 𝑓 𝛼 Γ ( 𝛼 ) , 𝑦 𝑛 𝑓 ( , 𝑦 ) . ( 3 . 1 3 ) such that 𝑓 in 𝑦 𝑛 ( 𝑦 ) 𝑀 𝑏 𝛼 𝑓 Γ ( 𝛼 + 1 ) , 𝑦 𝑛 𝑓 , 𝑦 0 a s 𝑛 . ( 3 . 1 4 ) . ThenSince 𝜌 > 0 is a continuous function, then we haveThus 𝑡 𝐽 is continuous. Now for any | | | | 1 | ( 𝑦 ) ( 𝑡 ) | = Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑇 ( 𝑡 𝑠 ) 𝑓 𝑠 , 𝑦 𝑠 | | | | 𝑑 𝑠 𝑀 𝑝 Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑑 𝑠 + 𝑀 𝑝 + 𝜌 𝑞 Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑑 𝑠 𝑀 𝑏 𝛼 Γ ( 𝛼 + 1 ) 𝑝 + 𝜌 𝑞 = 𝛿 < . ( 3 . 1 5 ) , and each we have for each 𝐶 ( [ 𝑟 , 𝑏 ] , 𝐸 ) Thus 𝜏 2 > 𝜏 1 . maps bounded sets into bounded sets in 𝜖 > 0 .
Now, let 𝜖 𝜏 1 𝜏 2 , 𝑦 𝐵 𝜌 Thus if | | 𝜏 ( 𝑦 ) 2 𝜏 ( 𝑦 ) 1 | | 1 | | | | Γ ( 𝛼 ) 𝜏 1 0 𝜖 𝜏 2 𝑠 𝛼 1 𝑇 𝜏 2 𝜏 𝑠 1 𝑠 𝛼 1 𝑇 𝜏 1 𝑓 𝑠 𝑠 , 𝑦 𝑠 | | | | + | | | | 𝑑 𝑠 𝜏 1 𝜏 1 𝜖 𝜏 2 𝑠 𝛼 1 𝑇 𝜏 2 𝜏 𝑠 1 𝑠 𝛼 1 𝑇 𝜏 1 𝑓 𝑠 𝑠 , 𝑦 𝑠 | | | | + | | | | 𝑑 𝑠 𝜏 2 𝜏 1 𝜏 2 𝑠 𝛼 1 𝑇 𝜏 2 𝑓 𝑠 𝑠 , 𝑦 𝑠 | | | | 𝑀 𝑑 𝑠 𝑝 + 𝜌 𝑞 | | | | Γ ( 𝛼 ) 𝜏 1 0 𝜖 𝜏 2 𝑠 𝛼 1 𝜏 1 𝑠 𝛼 1 𝑇 𝜏 1 | | | | + | | | | 𝑠 𝑑 𝑠 𝜏 1 0 𝜖 𝜏 2 𝑠 𝛼 1 𝑇 𝜏 1 𝑇 𝜏 𝜖 𝑠 2 𝜏 1 | | | | + 𝜖 𝑇 ( 𝜖 ) 𝑑 𝑠 𝜏 1 𝜏 1 𝜖 𝜏 2 𝑠 𝛼 1 𝜏 1 𝑠 𝛼 1 𝑑 𝑠 + 𝜏 2 𝜏 1 𝜏 2 𝑠 𝛼 1 𝑀 𝑑 𝑠 𝑝 + 𝜌 𝑞 Γ ( 𝛼 ) 𝜏 1 0 𝜖 𝜏 2 𝑠 𝛼 1 𝜏 1 𝑠 𝛼 1 𝜏 𝑑 𝑠 + 𝑇 2 𝜏 1 𝜖 𝑇 ( 𝜖 ) 𝐵 ( 𝐸 ) 𝜏 1 0 𝜖 𝜏 2 𝑠 𝛼 1 + 𝑑 𝑠 𝜏 1 𝜏 1 𝜖 𝜏 2 𝑠 𝛼 1 𝜏 1 𝑠 𝛼 1 𝑑 𝑠 + 𝜏 2 𝜏 1 𝜏 2 𝑠 𝛼 1 . 𝑑 𝑠 ( 3 . 1 6 ) and 𝜏 1 𝜏 2 we have for any 𝜖 As 𝑇 ( 𝑡 ) and 𝑡 > 0 sufficiently small, the right-hand side of the above inequality tends to zero, since is a strongly continuous operator and the compactness of 𝐵 𝜌 for 𝐸 implies the continuity in the uniform operator topology [29]. By the Arzelá-Ascoli theorem it suffices to show that 0 < 𝑡 < 𝑏 maps 𝜖 into a precompact set in 0 < 𝜖 < 𝑡 .
Let 𝑦 𝐵 𝜌 be fixed and let 𝜖 ( 𝑦 ) ( 𝑡 ) = 𝑇 ( 𝜖 ) Γ ( 𝛼 ) 0 𝑡 𝜖 ( 𝑡 𝑠 𝜖 ) 𝛼 1 𝑇 ( 𝑡 𝑠 𝜖 ) 𝑓 𝑠 , 𝑦 𝑠 𝑑 𝑠 . ( 3 . 1 7 ) be a real number satisfying 𝑇 ( 𝑡 ) . For 𝑡 > 0 we defineSince 𝐸 is a compact operator for 𝜖 , 0 < 𝜖 < 𝑡 . , the setis precompact in 𝑌 ( 𝑡 ) = { ( 𝑦 ) ( 𝑡 ) 𝑦 𝐵 𝜌 } for every 𝐸 MoreoverTherefore, the set = { 𝑦 𝐶 ( [ 𝑟 , 𝑏 ] , 𝐸 ) 𝑦 = 𝜆 ( 𝑦 ) f o r s o m e 0 < 𝜆 < 1 } ( 3 . 2 0 ) is precompact in 𝑦 . Hence the operator 𝑡 𝐽 is completely continuous. Now, it remains to show that the setis bounded.
Let | 𝑦 ( 𝑡 ) | 𝑀 𝑏 𝛼 𝑝 + Γ ( 𝛼 + 1 ) 𝑀 𝑞 Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑦 𝑠 𝑑 𝑠 . ( 3 . 2 2 ) be any element. Then, for each 𝜇 ( 𝑡 ) = m a x { | 𝑦 ( 𝑠 ) | 𝑟 𝑠 𝑡 } , 𝑡 𝐽 . ( 3 . 2 3 ) ,ThenWe consider the function defined byLet 𝑡 𝐽 , such that 𝑡 𝑡 , if 𝜇 ( 𝑡 ) 𝑀 𝑏 𝛼 𝑝 + Γ ( 𝛼 + 1 ) 𝑀 𝑞 Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝜇 ( 𝑠 ) 𝑑 𝑠 . ( 3 . 2 4 ) then by (3.22) we have, for 𝑡 [ 𝑟 , 0 ] (note 𝜇 ( 𝑡 ) = 𝜙 𝐶 )If 𝑦 m a x 𝜙 𝐶 , Λ 𝑦 . ( 3 . 2 6 ) then and the previous inequality holds.
By Lemma 2.7 we haveHenceThis shows that the set 𝑟 > 0 , 𝜙 [ 𝑟 , 0 ] × [ 0 , 𝜋 ] is bounded. As a consequence of Theorem 2.9, we deduce that the operator 𝑄 [ 0 , 𝑏 ] × has a fixed point which is a mild solution of the problem (3.1) and (3.2).

3.3. An Example

As an application of our results we consider the following partial functional differential equation of the formwhere 𝐸 = 𝐿 2 [ 0 , 𝜋 ] is continuous and 𝐴 𝐷 ( 𝐴 ) 𝐸 𝐸 is a given function.

Let Take 𝐷 ( 𝐴 ) = 𝑤 𝐸 , 𝑤 , 𝑤 a r e a b s o l u t e l y c o n t i n u o u s , 𝑤 𝐸 , 𝑤 ( 0 ) = 𝑤 ( 𝜋 ) = 0 . ( 3 . 2 9 ) and define 𝐴 𝑤 = 𝑛 = 1 𝑛 2 𝑤 , 𝑤 𝑛 𝑤 𝑛 , 𝑤 𝐷 ( 𝐴 ) , ( 3 . 3 0 ) by ( , ) with domainThenwhere 𝐴 . is the inner product in 𝐴 and 𝑇 ( 𝑡 ) , 𝑡 [ 0 , 𝑏 ] is the orthogonal set of eigenvectors in 𝐸 It is well known (see [29]) that 𝑇 ( 𝑡 ) 𝑤 = 𝑛 = 1 e x p ( 𝑛 2 𝑡 ) 𝑤 , 𝑤 𝑛 𝑤 𝑛 , 𝑤 𝐸 . ( 3 . 3 1 ) is the infinitesimal generator of an analytic semigroup 𝑇 ( 𝑡 ) in 𝑀 1 and is given bySince the analytic semigroup 𝜎 1 , 𝜎 2 [ 0 , 𝑏 ] + is compact, there exists a constant | 𝑄 ( 𝑡 , 𝑤 ( 𝑡 𝑟 , 𝑥 ) ) | 𝜎 1 ( 𝑡 ) | 𝑤 | + 𝜎 2 ( 𝑡 ) . ( 3 . 3 3 ) such thatAlso assume that there exist continuous functions [ 𝑟 , 𝑏 ] × [ 0 , 𝜋 ] . such thatWe can show that problem (3.1) and (3.2) is an abstract formulation of problem (3.27). Since all the conditions of Theorem 3.3 are satisfied, the problem (3.27) has a solution 𝐷 𝛼 on 𝑓 𝐽 × 𝐸

3.4. Existence Results for Infinite Delay

In the following we will extend the previous results to the case when the delay is infinite. More precisely we consider the following problemwhere 𝐴 𝐷 ( 𝐴 ) 𝐸 𝐸 is the standard Riemann-Liouville fractional derivative, { 𝑇 ( 𝑡 ) } 𝑡 0 is a continuous function, 𝜙 𝐸 the phase space [41], 𝜙 ( 0 ) = 0 is the infinitesimal generator of a strongly continuous semigroup ( 𝐸 , | | ) , 𝑡 𝐽 a continuous function with 𝑦 𝑡 and 𝑦 𝑡 ( 𝜃 ) = 𝑦 ( 𝑡 + 𝜃 ) , 𝜃 ( , 0 ] . ( 3 . 3 5 ) a real Banach space. For any 𝑏 = 𝑦 ( , 𝑏 ] 𝐸 𝑦 / 𝐽 𝐶 ( 𝐽 , 𝐸 ) , 𝑦 0 , ( 3 . 3 6 ) the function 𝑦 / 𝐽 is defined byConsider the following space:where 𝑏 is the restriction of 𝑏 to 𝑦 𝑏 = 𝑦 0 + s u p { | 𝑦 ( 𝑠 ) | 0 𝑠 𝑏 } , 𝑦 𝑏 . ( 3 . 3 7 ) Let 𝑦 𝑏 be the seminorm in 𝑦 0 = 𝜙 defined by

Definition 3.4. One says that a function ( 𝐻 4 ) is a mild solution of problem (3.34) if 𝑘 and

The first existence result is based on Banach's contraction principle.

Theorem 3.5. Assume the following.
( , 𝑏 ] . There exists a nonnegative constant 𝒩 𝑏 𝑏 such that Then there exists a unique mild solution of problem (3.34) on 𝜙

Proof. Transform the IVP (3.34) into a fixed point problem. Consider the operator 𝑥 ( 𝑡 ) = 𝜙 ( 𝑡 ) , 𝑡 ( , 0 ] , 0 , 𝑡 𝐽 . ( 3 . 4 1 ) defined byFor 𝑦 ( 𝑡 ) = 𝑧 ( 𝑡 ) + 𝑥 ( 𝑡 ) . ( 3 . 4 2 ) , we define the functionThen 𝑧 . SetIt is obvious that 1 𝑧 ( 𝑡 ) = Γ ( 𝛼 ) 𝑡 0 ( 𝑡 𝑠 ) 𝛼 1 𝑇 ( 𝑡 𝑠 ) 𝑓 𝑠 , 𝑧 𝑠 + 𝑥 𝑠 𝑑 𝑠 , 𝑡 𝐽 . ( 3 . 4 3 ) satisfies (3.38) if and only if 0 𝑏 = 𝑧 𝑏 𝑧 0 = 0 . ( 3 . 4 4 ) satisfies 𝑧 0 𝑏 andLetFor any 𝒫 0 𝑏 0 𝑏 , we haveThus 𝒩 is a Banach space. Let the operator 𝒫 defined byIt is obvious that 𝒫 𝑛 has a fixed point is equivalent to 𝑧 , 𝑧 0 𝑏 has a fixed point, and so we turn to proving that 𝒫 𝑛 ( 𝑧 ) 𝒫 𝑛 ( 𝑧 ) 𝑏 𝑘 𝑀 𝐾 𝑏 𝑛 𝑏 Γ ( 𝑛 𝛼 + 1 ) 𝑛 𝛼 𝑧 𝑧 𝑏 , ( 3 . 4 7 ) has a fixed point. As in Theorem 3.2, we show by induction that 𝒫 𝑛 satisfy for any 𝑛 , the following inequality:which yields the contraction of 𝑧 for sufficiently large values of 𝑦 ( 𝑡 ) = 𝑧 ( 𝑡 ) + 𝑥 ( 𝑡 ) , 𝑡 ( , 𝑏 ] . Therefore, by the Banach's contraction principle 𝒩 has a unique fixed point ( 𝐻 5 ) . Then { 𝑇 ( 𝑡 ) } 𝑡 𝐽 is a fixed point of the operator 𝑡 > 0 , which gives rise to a unique mild solution of the problem (3.34).

Next we give an existence result based upon the nonlinear alternative of Leray-Schauder type.

Theorem 3.6. Assume that the following hypotheses hold.
𝐻 6 The semigroup 𝑝 , 𝑞 𝐶 ( 𝐽 , + ) is compact for | 𝑓 ( 𝑡 , 𝑢 ) | 𝑝 ( 𝑡 ) + 𝑞 ( 𝑡 ) 𝑢 , f o r a . e . 𝑡 𝐽 , a n d e a c h 𝑢 . ( 3 . 4 8 ) .( (