Applied Mathematics and Computation xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Applied Mathematics and Computation journal homepage: www.elsevier.com/locate/amc
Analytic study on linear neutral fractional differential equations q Xian-Feng Zhou ⇑, Fuli Yang, Wei Jiang School of Mathematical Sciences, Anhui University, Hefei 230039, China
a r t i c l e
i n f o
a b s t r a c t This paper is devoted to investigating linear neutral fractional differential equations with constant coefficients. The existence, uniqueness and iterative formula of the solution are obtained. Meanwhile, dependence of the solution on initial value and the general solution represented by a fundamental solution matrix are discussed. Several examples are given to illustrate the applications of our results. Some conclusions in the literature are extended greatly. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Linear fractional differential equation Neutral Existence Uniqueness Iterative formula
1. Introduction Fractional differential equations have gained increasing attention because of their varied applications in various fields of applied sciences and engineering. For example, memory and hereditary of various material and process can be properly described as fractional differential systems. At present, a great deal of effort were spent in linear and nonlinear fractional differential equations. Most of the works focused on the existence, uniqueness of solutions of initial value problem for fractional differential equations, stability, controllability and observability for fractional differential systems, etc, see the monograph [1,2,4,5,3], the papers [16,8,9,11,6,7,13,14,12,15,17–23,10,24] and the references therein. Usually, the existence and uniqueness of the solution for nonlinear fractional differential equations are obtained by the fixed point theory [6,7], the classical method which is involved of comparison principle [8,9], variational iteration and numerical approach [10,11], Cauchy problem for evolution equations were discussed in [13,14,12], etc. Linear fractional differential equations (or systems) with Caputo’s derivative have been discussed in [11,15,17,16,18–21]. In [18], the authors studied the existence, uniqueness and dependence of the solution of the fractional differential equation
Da0 yðtÞ ¼ ayðt sÞ þ byðtÞ þ f ðtÞ; yðtÞ ¼ /ðtÞ;
t > 0;
ð1:1Þ
t 2 ½s; 0;
ð1:2Þ a
where a; b are constants, f is continuous on ½0; T; T > 0, the initial function / is continuous on ½s; 0, and D0 is the Caputo’s derivative of order a. In [19], the author considered the general solution of the commensurate fractional order equation
q This paper is supported by Natural Science Foundation of China (11371027 and 11471015), Program of Natural Science Research in Anhui Universities (KJ2011A020 and KJ2012A032), Research Fund for Doctoral Program of Higher Education of China (20123401120001 and 20133401120013). ⇑ Corresponding author. E-mail address:
[email protected] (X.-F. Zhou).
http://dx.doi.org/10.1016/j.amc.2014.12.056 0096-3003/Ó 2014 Elsevier Inc. All rights reserved.
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
2
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
a
d a xðtÞ ¼ AxðtÞ; dt
0 < t 6 a;
ð1:3Þ
and the existence, uniqueness of the incommensurate fractional order linear Eq. (1.3), where x 2 Rn ; a > 0, A is n n matrix. da is the Caputo fractional derivative of order a and 0 < a 6 1. dt a Motivated by the papers [18,19], this paper is concerned with the linear neutral fractional differential equation
Da0 xðtÞ ¼ AxðtÞ þ Bxðt sÞ þ C Da0 xðt sÞ þ f ðtÞ;
t P 0;
ð1:4Þ
s 6 t 6 0;
xðtÞ ¼ /ðtÞ;
ð1:5Þ
where 0 < a < 1; xðtÞ 2 Rn is a n dimensional vector, A; B; C are n n constant matrices, C – 0, the initial function /ðtÞ 2 Cð½s; 0; Rn Þ is continuous, s > 0 is delay, Da0 denotes the Caputo’s derivative of order a with the lower limit 0, f ðtÞ 2 Rn and is continuous on ½0; þ1Þ or f ðtÞ 2 Cð½0; 1; Rn Þ. Obviously, if n ¼ 1 and C ¼ 0, then Eqs. (1.4) and (1.5) can be reduced to Eqs. (1.1) and (1.2); if s ¼ 0; B ¼ 0 and C ¼ 0, then Eqs. (1.4) and (1.5) can be reduced to Eq. (1.3). Throughout this paper, Dat0 f ðtÞ or c Dat0 f ðtÞ denotes the Caputo’s derivative of order a with the lower limit t 0 for function f ;r Dat0 f ðtÞdenotes the Riemann–Liouville’s derivative of order a with the lower limit t 0 for function f ; Iat0 f ðtÞ denotes the integral of order a with lower limit t 0 for function f ; f ðsÞ ¼ L½f ðtÞ; s denotes the Laplace transform of the function f ðtÞ, and ‘‘j j’’ denotes the norm of ‘‘’’ in Rn or in Rnn . 2. Preliminaries In this section, we mainly recall some definitions and lemmas which will be used later. Definition 2.1 [1]. The fractional integral of order a with the lower limit 0 for function f is defined as
Ia0 f ðtÞ ¼
1 CðaÞ
Z
t
ðt sÞa1 f ðsÞds;
t > 0;
a > 0;
ð2:1Þ
0
provided the right-hand side is pointwise defined on ½0; þ1Þ, where C is the gamma function. Definition 2.2 [1]. The Caputo’s derivative of order a with the lower limit 0 for a function f : ½0; 1Þ ! R can be written as 1 Da0 f ðtÞ ¼ Cðn aÞ
¼
Rt 0
ðt sÞna1 f
a ðnÞ ðtÞ; In 0 f
ðnÞ
ðsÞds
ð2:2Þ
0 < n 1 < a 6 n:
t > 0;
Particularly, when 0 < a < 1, it holds
Da0 f ðtÞ ¼
1 Cð1 aÞ
Z 0
t
a ðt sÞa f ðsÞds ¼ I1 0 f ðtÞ; 0
0
t > 0:
ð2:3Þ
The Laplace transform of the Caputo’s fractional derivative Da0 f ðtÞ is
L½Da0 f ðtÞ; s ¼
Z
þ1
0
est ðDa0 f ðtÞÞdt ¼ sa f ðsÞ
n1 X ðkÞ sak1 f ð0Þ;
n 1 < a 6 n;
ð2:4Þ
k¼0
where f ðsÞ is the Laplace transform of f ðtÞ. Especially, for 0 < a < 1, it holds
Z
þ1
0
est ðDa0 f ðtÞÞdt ¼ sa f ðsÞ sa1 f ð0Þ:
ð2:5Þ
Definition 2.3 [1]. The two-parameter Mittag–Leffler function is defined as
Ea;b ðzÞ ¼
1 X
zk ; Cðak þ bÞ k¼0
a > 0; b > 0:
ð2:6Þ
Especially,
Ea;1 ðzÞ ¼ Ea ðzÞ ¼
1 X
zk ; Cðak þ 1Þ k¼0
a > 0; b > 0:
ð2:7Þ
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
3
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
The Laplace transform of Mittag–Leffler function is
h i Z ðkÞ L t akþb1 Ea;b ðata Þ; s ¼
1
0
k!sab
ðkÞ
est takþb1 Ea;b ðat a Þdt ¼
ðsa aÞkþ1
;
1
ReðsÞ > jaja :
ð2:8Þ
where ReðsÞ denotes the real part of s. In addition, the Laplace transform of ta1 is
L½t a1 ; s ¼ CðaÞsa ;
ReðsÞ > 0:
ð2:9Þ
Definition 2.4. A function xðtÞ is said to be a solution of initial value problem (IVP in short) Eqs. (1.4) and (1.5) on ½s; T 0 , if there exists a constant T 0 > 0 such that (i) xðtÞ 2 Cð½s; T 0 ; Rn Þ for t 2 ½s; T 0 ; (ii) xðtÞ satisfies
Da0 ðxðtÞ Cxðt sÞÞ ¼ AxðtÞ þ Bxðt sÞ þ f ðtÞ;
t 2 ½0; T 0
ð2:10Þ
and xðtÞ ¼ /ðtÞ for t 2 ½s; 0 almost everywhere.
Lemma 2.1 [4]. Let 0 < a < 1. Then it holds
Ia0 ðc Da0 xðtÞÞ ¼ xðtÞ xð0Þ:
ð2:11Þ
Lemma 2.2 [4]. There exists a link between the Riemann–Liouville and the Caputo’s fractional derivative of order a . Namely, c
n1 X
ðkÞ
f ðt 0 Þ ðt t 0 Þka ; C ðk a þ 1Þ k¼0
Dat0 f ðtÞ¼r Dat0 f ðtÞ
t > t0 ;
n 1 < ReðaÞ < n:
ð2:12Þ
t a ;
ð2:13Þ
Especially, for 0 < a < 1 and t0 ¼ 0, it holds c
Da0 f ðtÞ ¼
Z
1 Cð1 aÞ
t 0
ðt sÞa f ðsÞds¼r Da0 f ðtÞ 0
f ð0Þ
Cð1 aÞ
t > 0:
Lemma 2.3 [25]. Suppose that b > 0; aðtÞ is a nonnegative nondecreasing function locally integrable on 0 6 t < T ðT 6 þ1Þ and gðtÞ is nonnegative nondecreasing continuous function defined on 0 6 t < T; gðtÞ 6 M (constant), and suppose uðtÞ is nonnegative and locally integrable on 0 6 t < T with
uðtÞ 6 aðtÞ þ gðtÞ
Z
t
ðt uÞb1 uðsÞds;
ð2:14Þ
0
on this interval. Then
uðtÞ 6 aðtÞEb gðtÞCðbÞtb :
ð2:15Þ
3. Existence, uniqueness and iterative formula of the solution In this section, we will study the existence, uniqueness of the initial value problem (1.4) and (1.5). First, we give a lemma. Lemma 3.1. Eqs. (1.4) and (1.5) is equivalent to the following integral equation
xðtÞ ¼ /ð0Þ C/ðsÞ þ Cxðt sÞ þ xðtÞ ¼ /ðtÞ;
s 6 t 6 0:
1 CðaÞ
Z
t
ðt uÞa1 ½AxðuÞ þ Bxðu sÞ þ f ðuÞdu;
t P 0;
ð3:1Þ
0
ð3:2Þ
Proof. On one hand, if xðtÞ is a solution of Eqs. (1.4) and (1.5), then applying Ia0 on both side of Eq. (1.4) and using Lemma 2.1 yields
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
4
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
Ia0 Da0 ðxðtÞ Cxðt sÞ ¼ Ia0 ðAxðtÞ þ Bxðt sÞ þ f ðtÞÞ:
ð3:3Þ
That is
xðtÞ ¼ /ð0Þ C/ðsÞ þ Cxðt sÞ þ
1 CðaÞ
Z
t
ðt uÞa1 ½AxðuÞ þ Bxðu sÞ þ f ðuÞdu;
t P 0:
ð3:4Þ
0
Thus xðtÞ is a solution of Eq. (3.1) and (3.2). On the other hand, if xðtÞ is a solution of Eqs. (3.1) and (3.2), then applying c Da0 on both side of Eq. (3.1) yields
Da0 xðtÞ Da0 ðCxðt sÞÞ n o Rt ¼ Da0 Cð1aÞ 0 ðt uÞa1 ½AxðuÞ þ Bxðu sÞ þ f ðuÞdu ¼ Da0 Ia0 ½AxðtÞ þ Bxðt sÞ þ f ðtÞ :
ð3:5Þ
By the formula (2.13), we have
Da0 xðtÞ Da0 ðCxðt sÞÞ ¼
r
t a Da0 ðIa0 ½AxðtÞ þ Bxðt sÞ þ f ðtÞÞ Ia0 ðAxðtÞ þ Bxðt sÞ þ f ðtÞÞ t¼0 Cð1 aÞ
ð3:6Þ
t ¼ AxðtÞ þ Bxðt sÞ þ f ðtÞ ½Ia0 ðAxðtÞ þ Bxðt sÞ þ f ðtÞÞt¼0 Cð1 aÞ : a
Since xðtÞ is continuous t 2 ½s; 1Þ and f ðtÞ is continuous on ½0; 1Þ, thus ½Ia0 ðAxðtÞ þ Bxðt sÞ þ f ðtÞÞt¼0 ¼ 0. Hence, by (3.6), xðtÞ is also a solution of Eqs. (1.4) and (1.5). This completes the proof. h The explicit solution of Eqs. (1.4) and (1.5) is given by the following theorem. Theorem 3.1. If xðtÞ is a solution of Eqs. (1.4) and (1.5), then xðtÞ can be written as
xðtÞ ¼ Ea ðAta Þ½/ð0Þ C/ðsÞ þ Cxðt sÞ þ þ
Z
Z
t
Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu
0 t
ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu:
ð3:7Þ
0
Proof. By Lemma 3.1, Eqs. (1.4) and (1.5) is equivalent to the following integral equation
xðtÞ ¼ /ð0Þ C/ðsÞ þ Cxðt sÞ þ xðtÞ ¼ /ðtÞ;
1 CðaÞ
Z
t
ðt uÞa1 ½AxðuÞ þ Bxðu sÞ þ f ðuÞdu;
t > 0:
ð3:8Þ
0
t 2 ½s; 0:
ð3:9Þ
Since
Z
t
ðt uÞa1 ½AxðuÞ þ Bxðu sÞ þ f ðuÞdu ¼ t a1 ½AxðtÞ þ Bxðt sÞ þ f ðtÞ;
ð3:10Þ
0
we get
xðtÞ ¼ /ð0Þ C/ðsÞ þ Cxðt sÞ þ
1
CðaÞ
t a1 ½AxðtÞ þ Bxðt sÞ þ f ðtÞ;
ð3:11Þ
where ⁄ denotes convolution. Applying the Laplace transform on both side of Eq. (3.11) and using the formula (2.9) yields
xðsÞ ¼ L½/ð0Þ C/ðsÞ þ Cxðt sÞ; s þ
1
CðaÞ
CðaÞsa ðAxðsÞ þ L½Bxðt sÞ þ f ðtÞ; sÞ;
ð3:12Þ
where xðsÞ is Laplace transform of xðtÞ. Reducing Eq. (3.12) yields
xðsÞ ¼ sa ðsa I AÞ1 L½/ð0Þ C/ðsÞ þ Cxðt sÞ; s þ ðsa I AÞ1 L½Bxðt sÞ þ f ðtÞ; s ¼ L½/ð0Þ C/ðsÞ þ Cxðt sÞ; s þ Aðsa I AÞ1 L½/ð0Þ C/ðsÞ þ Cxðt sÞ; s þ ðsa I AÞ1 L½Bxðt sÞ þ f ðtÞ; sÞ: ð3:13Þ By the formula (2.8), we get
xðsÞ ¼ L½/ð0Þ C/ðsÞ þ Cxðt sÞ; s þ L½Ata1 Ea;a ðAt a Þ; s L½/ð0Þ C/ðsÞ þ Cxðt sÞ; s þ L½ta1 Ea;a ðAt a Þ; s L½Bxðt sÞ þ f ðtÞ; s:
ð3:14Þ
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
5
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
Applying the Laplace inverse transform and Convolution Theorem yields
xðtÞ ¼ /ð0Þ C/ðsÞ þ Cxðt sÞ þ ðAt a1 Ea;a ðAta ÞÞ ð/ð0Þ C/ðsÞ þ Cxðt sÞÞ þ ðta1 Ea;a ðAt a ÞÞ ðBxðt sÞ þ f ðtÞÞ
¼ /ð0Þ C/ðsÞ þ Cxðt sÞ þ þ
Z
t
Aðt uÞa1 Ea;a ðAðt uÞa Þð/ð0Þ C/ðsÞ þ Cxðu sÞÞdu
0
Z
t
ðt uÞa1 Ea;a ðAðt uÞa ÞðBxðu sÞ þ f ðuÞÞdu:
0
Z t Z t ¼ /ð0Þ I þ Aðt uÞa1 Ea;a ðAðt uÞa Þdu þ Cxðt sÞ C/ðsÞ I þ Aðt uÞa1 Ea;a ðAðt uÞa Þdu þ
0
Z
t
Aðt uÞ
a1
a
Ea;a ðAðt uÞ ÞCxðu sÞdu þ
Z
0
0 t
ðt uÞ
a1
Ea;a ðAðt uÞa Þ½ðBxðu sÞ þ f ðuÞdu:
ð3:15Þ
0
Noting that
Z
t
Aðt uÞa1 Ea;a ðAðt uÞa Þdu ¼
0
Z
t
Aðt uÞa1
0
¼
k 1 X ðAðt uÞa Þ k¼0
Cðka þ aÞ
du ¼
1 X
Akþ1 Cðka þ aÞ k¼0
Z
t
ðt uÞkaþa1 du
0
1 X
1 X Akþ1 t ðkþ1Þa Ak t k a ; ¼ Cððk þ 1Þa þ 1Þ k¼1 Cðka þ 1Þ k¼0
ð3:16Þ
we get
Iþ
Z
t
Aðt uÞa1 Ea;a ðAðt uÞa Þdu ¼ I þ
0
1 X
1 X Ak t k a Ak t k a ¼ ¼ Ea;1 ðAta Þ ¼ Ea ðAt a Þ: Cðka þ 1Þ k¼0 Cðka þ 1Þ k¼1
Inserting (3.17) into (3.15) yields the formula (3.7). This completes the proof.
ð3:17Þ
h
Denote the solution of Eqs. (1.4) and (1.5) by: /ðtÞ ¼ /0 ðtÞ for t 2 ½s; 0; xðtÞ ¼ /i ðtÞ for t 2 ½ði 1Þs; is; the formula (3.7), we have, for t 2 ½0; s, that
Z t /1 ðtÞ ¼ Ea ðAta Þ½/ð0Þ C/ðsÞ þ C/ðt sÞ; t 2 ½0; s þ Aðt uÞa1 Ea;a ðAðt uÞa ÞC/ðu sÞdu 0 Z t þ ðt uÞa1 Ea;a ðAðt uÞa Þ½B/ðu sÞ þ f ðuÞdu:
i ¼ 1; 2; . . . ; k. By
ð3:18Þ
0
Furthermore, we have the following corollary. Corollary 3.1. Suppose that the solution of Eqs. (1.4) and (1.5) exists on ½s; ks; k is a positive integer and k P 2. Then for t 2 ½ðk 1Þs; ks, the iterative formula for the solution /i ðtÞ of Eqs. (1.4) and (1.5) defined on ½ði 1Þs; is; i ¼ 2; . . . k, can be written as
/k ðtÞ ¼ Ea ðAta Þ½/ð0Þ C/ðsÞ þ C/k1 ðt sÞ k2 X R ðjþ1Þs þ Aðt uÞa1 Ea;a ðAðt uÞa ÞC/j ðu sÞdu js j¼0 k2 X R ðjþ1Þs þ ðt uÞa1 Ea;a ðAðt uÞa Þ B/j ðu sÞ þ f ðuÞ du js
ð3:19Þ
j¼0
þ
Rt
þ
Rt
ðk1Þs
Aðt uÞa1 Ea;a ðAðt uÞa ÞC/k1 ðu sÞdu
ðk1Þs
ðt uÞa1 Ea;a ðAðt uÞa Þ½B/k1 ðu sÞ þ f ðuÞdu;
where /0 ðtÞ ¼ /ðtÞ for t 2 ½s; 0. Proof. For t 2 ½s; 2s, we have by the formula (3.7) that
xðtÞ ¼ /2 ðtÞ ¼ Ea ðAta Þ½/ð0Þ C/ðsÞ þ C/1 ðt sÞ Rs þ 0 Aðt uÞa1 Ea;a ðAðt uÞa ÞC/ðu sÞdu Rs þ 0 ðt uÞa1 Ea;a ðAðt uÞa Þ½B/ðu sÞ þ f ðuÞdu Rt þ s Aðt uÞa1 Ea;a ðAðt uÞa ÞC/1 ðu sÞdu Rt þ s ðt uÞa1 Ea;a ðAðt uÞa Þ½B/1 ðu sÞ þ f ðuÞdu:
ð3:20Þ
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
6
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
For t 2 ½ðk 1Þs; ks, k P 1, we have,
xðtÞ ¼ /k ðtÞ
¼ Ea ðAta Þ½/ð0Þ þ C/ðsÞ þ C/k1 ðt sÞ Rt þ 0 Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu Rt þ 0 ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu
¼ Ea ðAta Þ½/ð0Þ þ C/ðsÞ þ C/k1 ðt sÞ Rs þ 0 Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu Rs þ 0 ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu R 2s þ s Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu R 2s þ s ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu
ð3:21Þ
þ R ðk1Þs þ ðk2Þs Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu R ðk1Þs þ ðk2Þs ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu Rt þ ðk1Þs Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu Rt þ ðk1Þs ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu: It is easy to obtain that
Rs ¼
R0s 0
Rs ¼
R0s 0
Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu Aðt uÞa1 Ea;a ðAðt uÞa ÞC/0 ðu sÞdu; ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu ðt uÞa1 Ea;a ðAðt uÞa Þ½B/0 ðu sÞ þ f ðuÞdu;
R 2s ¼
s
Aðt uÞa1 Ea;a ðAðt uÞa Þ½Cxðu sÞdu
s
Aðt uÞa1 Ea;a ðAðt uÞa ÞC/1 ðu sÞdu;
R 2s R 2s s
¼
R 2s
ðt uÞa1 Ea;a ðAðt uÞa Þ½B/1 ðu sÞ þ f ðuÞdu; ......... s
ðk2Þ
Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu
ðk2Þ
Aðt uÞa1 Ea;a ðAðt uÞa ÞC/k2 ðu sÞdu;
R ðk1Þs R ðk1Þs
¼
ðk2Þ
ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu
ðk2Þ
ðt uÞa1 Ea;a ðAðt uÞa Þ½B/k2 ðu sÞ þ f ðuÞdu;
R ðk1Þs Rt
¼
ðk1Þs
Aðt uÞa1 Ea;a ðAðt uÞa ÞCxðu sÞdu
ðk1Þs
Aðt uÞa1 Ea;a ðAðt uÞa ÞC/k1 ðu sÞdu:
Rt Rt
¼
ðk1Þs
ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu
ðk1Þs
ðt uÞa1 Ea;a ðAðt uÞa Þ½B/k1 ðu sÞ þ f ðuÞdu:
Rt
ð3:23Þ
ð3:24Þ
ðt uÞa1 Ea;a ðAðt uÞa Þ½Bxðu sÞ þ f ðuÞdu
R ðk1Þs ¼
ð3:22Þ
ð3:25Þ
ð3:26Þ
ð3:27Þ
ð3:28Þ
ð3:29Þ
Inserting (3.22)–(3.29) into (3.21) yields the iterative formula (3.19). This completes the proof. h Remark 3.1. (i) For the case C ¼ 0 in Eqs. (1.4) and (1.5), Theorem 3.1 and Corollary 3.1 are also valid. (ii) By the formulae (3.7) and (3.19), we know if the initial function /0 ðtÞ defined on ½s; 0 and f ðtÞ defined on ½0; 1 are continuous, then /k ðtÞ is continuous on ½ðk 1Þs; ks for any positive integer k. By Corollary 3.1, we can easily get the following existence theorem. Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
7
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
Theorem 3.2. Suppose that f ðtÞ is continuous on ½0; þ1Þ and /ðtÞ is continuous on ½s; 0. Then there exists a unique solution of Eqs. (1.4) and (1.5) defined on ½0; þ1Þ. Remark 3.2. Theorem 3.2 is an extension of Theorem 1 in [18], which weakens the conditions and extends the conclusion of Theorem 1 in [18]. Example 3.1. Consider the analytic solution of the neutral linear fractional differential equation 1
1
D20 xðtÞ ¼ xðt 1Þ D20 xðt 1Þ þ t; 1
xðtÞ ¼ t3 ;
t P 0;
ð3:30Þ
t 2 ½1; 0:
ð3:31Þ 1 3
Comparing with Eqs. (1.4) and (1.5), it hold that a ¼ 12 ; s ¼ 1; A ¼ 0; B ¼ 1; C ¼ 1; f ðtÞ ¼ t and /ðtÞ ¼ t . 1 Note that /0 ðtÞ ¼ /ðtÞ ¼ t3 . By the formulae (3.7) and (3.19), we have
Rt 1 1 ¼ ½/ð0Þ /ð1Þ /0 ðt 1Þ þ 0 ðt uÞ2 E1;1 ð0ðt uÞ2 Þð/0 ðu 1Þ þ uÞdu 22 Rt Rt 1 1 1 1 ¼ 1 ðt 1Þ3 þ C1ð1Þ 0 ðt uÞ2 ðu 1Þ3 du þ C1ð1Þ 0 ðt uÞ2 udu
/1 ðtÞ
2
1 3
1 2
ð3:32Þ
2
1 3
3 2
¼ 1 ðt 1Þ þ I0 ðt 1Þ þ C1ð5Þ t ;
t 2 ½0; 1
2
and
/2 ðtÞ ¼ ½/ð0Þ /ð1Þ /1 ðt 1Þ þ
1 Cð12Þ
Z
1
1
ðt uÞ2 ð/1 ðu 1Þ þ uÞdu þ
0
1 Cð12Þ
Z
t
1
ðt uÞ2 ð/1 ðu 1Þ þ uÞdu; t 2 ½1; 2:
1
ð3:33Þ Remark 3.3. It is necessary to mention that the conditions of Theorem 3.2 is sufficient, not necessary. Example 3.2. Consider the neutral linear fractional differential equation 1
1
1
D20 xðtÞ ¼ xðt 1Þ þ D20 xðt 1Þ þ t6 ; xðtÞ ¼ t; here f ðtÞ ¼ t
16
t P 0;
ð3:34Þ
t 2 ½1; 0;
ð3:35Þ
is not continuous in ½0; 1. Using the iterative formula (3.19), we get the solution of Eq. (3.34), (3.35)
xðtÞ ¼ /1 ðtÞ ¼ t þ
1 3 1 1 Cð5Þ 1 t2 3 t 2 þ 64 t 3 ; 5 Cð 2 Þ Cð2Þ Cð3Þ
t 2 ½0; 1;
ð3:36Þ
which is continuous on ½0; 1. 4. Dependence of the solution on the initial value In this section, we consider the dependence on initial value for the solution of Eqs. (1.4) and (1.5). The main result is given by the following theorem. Theorem 4.1. Suppose that xðtÞ and yðtÞ are the solutions of Eq. (1.4) subject to initial value xðtÞ ¼ /ðtÞ and yðtÞ ¼ wðtÞ on ½s; 0, respectively, which exist on the interval ½0; ks for some k P 1 . Then it holds
jxðtÞ yðtÞj 6 Ak j/ wj;
t 2 ½ðk 1Þs; ks;
h i Pk1 a a where A0 ¼ 1; Ak ¼ 1 þ jcj þ jcjAk1 þ CðajBjþ1Þ j¼0 ððk jÞsÞ Aj Ea ðjAjðksÞ Þ, and j/ wj ¼ sup j/ðhÞ wðhÞjRn .
ð4:1Þ
s6h60
Proof. By the assumptions, we have
Da0 yðtÞ ¼ AyðtÞ þ Byðt sÞ þ CDa0 yðt sÞ þ f ðtÞ; yðtÞ ¼ wðtÞ;
t P 0;
t 2 ½s; 0:
ð4:2Þ ð4:3Þ
Combining Eqs. (4.2) and (4.3) with Eqs. (1.4) and (1.5) yields
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
8
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
Da0 ðxðtÞ yðtÞÞ ¼ AðxðtÞ yðtÞÞ þ Bðxðt sÞ yðt sÞÞ þ C Da0 ðxðtÞ yðtÞÞ; xðtÞ yðtÞ ¼ /ðtÞ wðtÞ;
t P 0;
t 2 ½s; 0:
ð4:4Þ ð4:5Þ
Evidently, Eqs. (4.4) and (4.5) is equivalent to the following integral equation
xðtÞ yðtÞ ¼ ð/ð0Þ wð0ÞÞ Cð/ðsÞ wðsÞÞ þ Cðxðt sÞ yðt sÞÞ þ
1 CðaÞ
Z
t
ðt uÞa1 ½AðxðuÞ yðuÞÞ
0
þ Bðxðu sÞ yðu sÞÞdu:
ð4:6Þ
Now we prove the validity of inequality (4.1) by induction method on k. ðiÞ First, we prove that inequality (4.1) holds for k = 1. If t 2 ½0; s, then we have by Eq. (4.6) that
jxðtÞ yðtÞj 6 j/ wj þ jCjj/ wj þ jCjj/ wj Rt þ Cð1aÞ 0 ðt uÞa1 jAjjxðuÞ yðuÞjdu Rt þ Cð1aÞ 0 ðt uÞa1 jBjj/ wjdu: By the fact that
Rt 0
ð4:7Þ
ðt uÞa1 du ¼ a1 ta 6 a1 sa for t 2 ½0; s, we have
jxðtÞ yðtÞj 6 1 þ jcj þ jcjA0 þ
Z t jBj jAj sa j/ wj þ ðt uÞa1 jxðuÞ yðuÞjdu: Cða þ 1Þ CðaÞ 0
ð4:8Þ
jBj sa Ea ðjAjsa Þj/ wj; Cða þ 1Þ
ð4:9Þ
By Lemma 2.3, we have
jxðtÞ yðtÞj 6 1 þ jcj þ jcjA0 þ
t 2 ½0; s:
Thus, the inequality (4.1) holds for k = 1. ðiiÞ Assume that the inequality (4.1) holds for 2; ; k 1. That is
jxðtÞ yðtÞj 6 Aj j/ wj;
j ¼ 2; 3; . . . ; k 1:
ð4:10Þ
Now we prove the inequality (4.1) holds for k. Now we prove that the inequality (4.1) holds for k. If t 2 ½ðk 1Þs; ks, we have by Eq. (4.6) that
xðtÞ yðtÞ
¼ ð/ð0Þ wð0ÞÞ Cð/ðsÞ wðsÞÞ þ Cðxðt sÞ yðt sÞÞ Rt þ Cð1aÞ 0 ðt uÞa1 AðxðuÞ yðuÞÞdu Rt þ Cð1aÞ 0 ðt uÞa1 Bðxðu sÞ yðu sÞÞdu ¼ ð/ð0Þ wð0ÞÞ Cð/ðsÞ wðsÞÞ þ Cðxðt sÞ yðt sÞÞ Rt þ Cð1aÞ 0 ðt uÞa1 AðxðuÞ yðuÞÞdu Rs þ Cð1aÞ 0 ðt uÞa1 Bðxðu sÞ yðu sÞÞdu R 2s þ Cð1aÞ s ðt uÞa1 Bðxðu sÞ yðu sÞÞdu þ þ Cð1aÞ þ Cð1aÞ
R ðk1Þs ðk2Þs
ðt uÞa1 Bðxðu sÞ yðu sÞÞdu
Rt
ðk1Þs
ðt uÞa1 Bðxðu sÞ yðu sÞÞdu:
ð4:11Þ
Thus, it follows that
jxðtÞ yðtÞj 6 ð1 þ jcjÞj/ wj þ jcjjxðt sÞ yðt sÞj Rt þ Cð1aÞ 0 ðt uÞa1 jAjjxðuÞ yðuÞjdu Rs þ Cð1aÞ 0 ðt uÞa1 jBjjxðu sÞ yðu sÞjdu R 2s þ Cð1aÞ s ðt uÞa1 jBjjxðu sÞ yðu sÞjdu
ð4:12Þ
þ R ðk1Þs þ Cð1aÞ ðk2Þs ðt uÞa1 jBjjxðu sÞ yðu sÞjdu Rt þ Cð1aÞ ðk1Þs ðt uÞa1 jBjjxðu sÞ yðu sÞjdu: Since t 2 ½ðk 1Þs; ks, it holds that t s 2 ½ðk 2Þs; ðk 1Þs. By the inequalities (4.9) and (4.10), we have
jxðt sÞ yðt sÞj 6 Ak1 j/ wj for
t 2 ½ðk 1Þs; ks;
ð4:13Þ
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
jxðu sÞ yðu sÞj 6 j/ wj for
u 2 ½0; s;
jxðu sÞ yðu sÞj 6 A1 j/ wj for
9
ð4:14Þ
u 2 ½s; 2s;
ð4:15Þ
jxðu sÞ yðu sÞj 6 Ak2 j/ wj for
u 2 ½ðk 2Þs; ðk 1Þs;
ð4:16Þ
jxðu sÞ yðu sÞj 6 Ak1 j/ wj for
u 2 ½ðk 1Þs; ks:
ð4:17Þ
By computation, it holds, for t 2 ½ðk 1Þs; ks, that
Z s
ðt uÞa1 du ¼
0
R 2s s
1
a
1
½ta ðt sÞa 6
a
ðksÞ ;
a
ð4:18Þ
ðt uÞa1 du ¼ a1 ½ðt sÞa ðt 2sÞa a
6 a1 ðt sÞa 6 a1 ððk 1ÞsÞ ;
ð4:19Þ
R ðk1Þs ðk2Þs
a a ðt uÞa1 du ¼ a1 ðt ðk 2ÞsÞ ðt ðk 1ÞsÞ
ð4:20Þ
a
6 a1 ðt ðk 2ÞsÞ 6 1a ð2sÞa ; Z
t
ðt uÞa1 du ¼
ðk1Þs
1
a
a
ðt ðk 1ÞsÞ 6
1
a
sa :
ð4:21Þ
Inserting the inequalities (4.13)–(4.21) into (4.12) yields
6
jxðtÞ yðtÞj h a a 1 þ jcj þ jcjAk1 þ CðajBjþ1Þ ðksÞ þ ððk 1ÞsÞ A1 þ þ ð2sÞa Ak2 þ sa Ak1 j/ wj R t ðt uÞa1 jxðuÞ yðuÞjdu þ CjAj ða Þ 0
"
# Z t k1 X jBj jAj a ¼ 1 þ jcj þ jcjAk1 þ ððk jÞsÞ Aj j/ wj þ ðt uÞa1 jxðuÞ yðuÞjdu: Cða þ 1Þ j¼0 CðaÞ 0
ð4:22Þ
By Lemma 2.3, we have
"
# k1 X jBj a a jxðtÞ yðtÞj 6 1 þ jcj þ jcjAk1 þ ððk jÞsÞ Aj Ea ðjAjðksÞ Þj/ wj: Cða þ 1Þ j¼0 Therefore, the inequality (4.1) is valid for k. This completes the proof.
ð4:23Þ
h
5. The general solution by a fundamental solution matrix In this section, the representation of the general solution of Eqs. (1.4) and (1.5) is given by a fundamental solution matrix of Eq. (1.4). For the case a ¼ 1, the general solution of Eqs. (1.4) and (1.5) is coincide with the general solution of first-order linear neutral differential equations [26,27]. Let’s begin with some definitions and Lemmas. Lemma 5.1. Let 0 < n 1 < a < n; s > 0 be a constant, xðtÞ be continuous on ½s; 1Þ and xðkÞ ðtÞ exist for k ¼ 1; 2; ; n 1. Then the Laplace transform of c Da0 xðt sÞ is given by
L c Da0 xðt sÞ; s ¼ sa ess xðsÞ þ sa ess
Z
0
est xðtÞdt
s
n1 X sak1 xðkÞ ðt sÞ t¼0 ;
ð5:1Þ
k¼0
xðsÞ is the Laplace transform of xðtÞ. Namely, L½xðtÞ; s ¼ xðsÞ. where Especially, when 0 < a < 1, it holds
L½c Da0 xðt sÞ; s ¼ sa ess xðsÞ þ sa ess
Z
0
s
est xðtÞdt sa1 xðt sÞjt¼0 :
ð5:2Þ
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
10
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
Proof. Denote gðtÞ ¼ xðt sÞ, then c
a ðnÞ Da0 xðt sÞ ¼ In 0 g ðtÞ ¼
1 Cðn aÞ
Z
t
ðt uÞna1 g ðnÞ ðuÞdu ¼
0
1
Cðn aÞ
tna1 g ðnÞ ðtÞ:
ð5:3Þ
By the Formula (2.9) and applying the Laplace transform on both side of (5.3), we get
L½c Da0 xðt sÞ; s ¼ sðnaÞ L g ðnÞ ðtÞ; s :
ð5:4Þ
Obviously, it holds n1 X L g ðnÞ ðtÞ; s ¼ sn gðsÞ snk1 g ðkÞ ð0Þ:
ð5:5Þ
k¼0
Then we have
gðsÞ ¼ L½gðtÞ; s ¼ L½xðt sÞ; s ¼
Z
þ1
est xðt sÞdt ¼
Z
þ1
esðuþsÞ xðuÞdu ¼ ess
s
0
Z
0
est xðtÞdt þ ess xðsÞ:
ð5:6Þ
s
Since g ðkÞ ð0Þ ¼ xðkÞ ðt sÞjt¼0 , so we get
" # n1 X L c Da0 xðt sÞ; s ¼ sðnaÞ L g ðnÞ ðtÞ; s ¼ sðnaÞ sn gðsÞ snk1 g ðkÞ ð0Þ ( Z ðnaÞ ¼s sn ess
k¼0
0 st
e
xðtÞdt þ e
s
¼ sa ess xðsÞ þ sa ess
) X n1 nk1 ðkÞ s x ðt sÞjt¼0 xðsÞ
ss
k¼0
Z
0
est xðtÞdt
s
n1 X
sak1 xðkÞ ðt sÞjt¼0 :
ð5:7Þ
k¼0
If 0 < a < 1, then
L c Da0 xðt sÞ; s ¼ sa ess xðsÞ þ sa ess This completes the proof.
Z
0
s
est xðtÞdt sa1 xðt sÞjt¼0 :
ð5:8Þ
h
Applying the Laplace transform on both side of Eq. (1.4) yields
a s I A Bess Csa ess xðsÞ ¼ sa1 ðxð0Þ CxðsÞÞ þ Bess Z 1 est f ðtÞdt: þ
Z
0
est xðtÞdt þ Csa ess
s
Z
0
est xðtÞdt
s
ð5:9Þ
0
Definition 5.1. The matrix ðsa I A Bess Csa ess Þ in (5.9) is said to be the characteristic matrix of Eqs. (1.4) and (1.5), and the equation
detðsa I A Bess Csa ess Þ ¼ 0;
ð5:10Þ
is said to be the characteristic equation of Eqs. (1.4) and (1.5). Definition 5.2. If n order square matrix XðtÞ satisfies
Da0 XðtÞ ¼ AXðtÞ þ BXðt sÞ þ C Da0 Xðt sÞ;
t>0
ð5:11Þ
XðtÞ ¼ I;
t ¼ 0;
ð5:12Þ
XðtÞ ¼ 0;
t < 0;
ð5:13Þ
then we say XðtÞ is a fundamental solution matrix of Eq. (1.4), where I denotes the unit matrix with order n. Lemma 5.2. Let XðtÞ be a fundamental solution matrix of Eq. (1.4). Then it holds that Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
ðsa I A Bess Csa ess ÞXðsÞ ¼ sa1 I;
11
ð5:14Þ
where XðsÞ is the Laplace transform of XðtÞ, and I denotes the unit matrix with order n. Proof. By Lemma 5.1 and applying the Laplace transform on both side of Eq. (5.11) yields the equality (5.14) immediately. h Now we will introduce a representation of the general solution of Eqs. (1.4) and (1.5) through a fundamental solution matrix of Eq. (1.4). Theorem 5.1. If XðtÞ is a fundamental solution matrix of Eq. (1.4), then the general solution of Eqs. (1.4) and (1.5) can be written as
xðtÞ ¼ XðtÞ½/ð0Þ C/ðsÞ ih i R ts h a1 a I ^ xðhÞ þ f ðh þ sÞ dh þ s c D1 B/ðhÞ 0 Xðt s hÞ þ CðaÞ ðt s hÞ R0 _ h sÞC/ðhÞdh þ C /ðt ^ sÞxðt sÞ; t 2 ½0; þ1Þ; þ s Xðt where xðhÞ ¼
0; 1;
h P 0; ^ , /ðhÞ ¼ h < 0;
/ðhÞ; /ð0Þ;
ð5:15Þ
s 6 h 6 0; . h > 0:
Proof. Applying the Laplace transform on Eq. (1.4) yields
sa I A Bess Csa ess xðsÞ ¼ sa1 ðxð0Þ CxðsÞÞ R0 R0 þBess s est xðtÞdt þ Csa ess s est xðtÞdt þ L½f ðtÞ; s:
ð5:16Þ
_ s þ Xð0Þ. By Lemma 5.2 and Denoting YðtÞ ¼ Cð1aÞ ta1 XðtÞ ¼ Ia0 XðtÞ, we Note that L½ta1 ; s ¼ CðaÞsa and sL½XðsÞ; s ¼ L½XðtÞ; have
R R xðsÞ ¼ s1a XðsÞ sa1 ð/ð0Þ C/ðsÞÞ þ Bess 0s est /ðtÞdt þ Csa ess 0s est /ðtÞdt þ L½f ðtÞ; s R0 R0 ¼ XðsÞð/ð0Þ C/ðsÞÞ þ s1a XðsÞBess s est /ðtÞdt þ sXðsÞCess s est /ðtÞdt þ s1a XðsÞL½f ðtÞ; s h i R R0 0 ¼ XðsÞð/ð0Þ C/ðsÞÞ þ sL Cð1aÞ t a1 XðtÞ;s Bess s est /ðtÞdt þ sXðsÞCess s est /ðtÞdt þ s1a XðsÞL½f ðtÞ; s
n h i o
h i R0 R0 _ _ _ ¼ XðsÞð/ð0Þ C/ðsÞÞ þ L½YðtÞ; s þ Yð0Þ Bess s est /ðtÞdt þ L XðtÞ; s þ Xð0Þ Cess s est /ðtÞdt þ L YðtÞ; s þ Yð0Þ L½f ðtÞ;s:
ð5:17Þ
By the definition of YðtÞ, we have Yð0Þ ¼ 0. It follows
Z 0 Z 0 h i n h i o _ _ s Bess s þ Xð0Þ Cess xðtÞ ¼ XðsÞð/ð0Þ C/ðsÞÞ þ L YðtÞ; est /ðtÞdt þ L XðtÞ; est /ðtÞdt s s h i _ s L½f ðtÞ; s: þ L YðtÞ;
ð5:18Þ
^ By the definitions of xðhÞ and /ðhÞ, we get
Bess
Z
0
est /ðtÞdt ¼ Bess
s
¼B
Z
Z
1
^ xðtÞdt ¼ Bess est /ðtÞ
s þ1
Z
þ1
^ sÞxðh sÞdh esðhsÞ /ðh
0
h i ^ sÞxðh sÞdh ¼ L B/ðh ^ sÞxðh sÞ; s esh /ðh
ð5:19Þ
0
and
Cess
Z
0
h i ^ sÞxðh sÞ; s : est /ðtÞdt ¼ L C /ðh
ð5:20Þ
s
Inserting (5.19) and (5.20) into (5.18) yields
h i h i _ ^ sÞxðh sÞ þ f ðtÞ; s xðsÞ ¼ XðsÞð/ð0Þ C/ðsÞÞ þ L YðtÞ; s L B/ðh
h i _ ^ sÞxðh sÞ; s : þ L½XðtÞ; s þ Xð0Þ L C /ðh
ð5:21Þ
By the inverse Laplace transform and Convolution Theorem, we get
h i _ B/ðt _ ½C /ðt ^ sÞxðt sÞ þ f ðtÞ þ XðtÞ ^ sÞxðt sÞ þ Xð0Þ C /ðt ^ sÞxðt sÞ xðtÞ ¼ XðtÞð/ð0Þ C/ðsÞÞ þ YðtÞ
h i h i Rt _ R _ ^ u sÞxðt u sÞ þ f ðt uÞ du þ t Xðt ^ sÞxðu sÞ du þ C /ðt ^ sÞxðt sÞ : ¼ XðtÞð/ð0Þ C/ðsÞÞ þ 0 YðuÞ B/ðt uÞ C /ðu 0 ð5:22Þ
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
12
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
In view of YðtÞ ¼ Cð1aÞ ta1 XðtÞ ¼ Ia0 XðtÞ and Yð0Þ ¼ 0, we have
d Xð0Þ a1 1 a c 1a _ YðtÞ ¼ ðIa0 XðtÞÞ ¼ r D1 t ¼ 0 XðtÞ ¼ D0 XðtÞ þ dt CðaÞ CðaÞ
Z
t
_ þ ðt uÞa1 XðuÞdu
0
Xð0Þ
CðaÞ
ta1 :
ð5:23Þ
^ and xðhÞ yields Set u s ¼ h. Using the definitions of /ðhÞ
h i _ uÞ C /ðu ^ sÞxðu sÞ du Xðt h i R ts _ h sÞ C /ðhÞ ^ xðhÞ dh ¼ s Xðt R0 _ h sÞC/ðhÞdh: ¼ s Xðt Rt 0
ð5:24Þ
Inserting (5.23) and (5.24) into (5.22) yields
i Rt h a I a1 xðtÞ ¼ XðtÞ½/ð0Þ C/ðsÞ þ 0 c D1 0 XðuÞ þ CðaÞ u h i R _ ^ u sÞxðt u sÞ þ f ðt uÞ du þ 0 Xðt ^ sÞxðt sÞ: h sÞC/ðhÞdh þ C /ðt B/ðt s
ð5:25Þ
Setting t s u ¼ h, we have
R t hc
ih i a I a1 B/ðt ^ u sÞxðt u sÞ þ f ðt uÞ du D1 0 XðuÞ þ CðaÞ u ih i R ts h a1 a I ^ xðhÞ þ f ðh þ sÞ dh: ¼ s c D1 B/ðhÞ 0 Xðt s hÞ þ CðaÞ ðt s hÞ 0
ð5:26Þ
Inserting (5.26) into (5.25) yields
xðtÞ ¼ XðtÞ½/ð0Þ C/ðsÞ ih i R ts h a1 a I ^ xðhÞ þ f ðh þ sÞ dh þ s c D1 B/ðhÞ 0 Xðt s hÞ þ CðaÞ ðt s hÞ R0 _ h sÞC/ðhÞdh þ C /ðt ^ sÞxðt sÞ: þ s Xðt This completes the proof.
ð5:27Þ
h
Remark 5.1. For the case a ¼ 1, the general solution of Eqs. (1.4) and (1.5) can be written as
xðtÞ ¼ XðtÞ½/ð0Þ C/ðsÞ R ts þ s Xðt s hÞðB/ðhÞ þ f ðh þ sÞÞdh R0 _ h sÞC/ðhÞdh þ C/ðt sÞ; Xðt þ s
ð5:28Þ t 2 ½0; s
and
xðtÞ ¼ XðtÞ½/ð0Þ C/ðsÞ R0 þ s Xðt s hÞ½B/ðhÞ þ f ðh þ sÞdh R ts þ 0 Xðt s hÞf ðh þ sÞdh R0 _ h sÞC/ðhÞdh; t 2 ½s; þ1Þ: Xðt þ
ð5:29Þ
s
For t 2 ½s; þ1Þ and the case a ¼ 1, the solution (5.29) of Eqs. (1.4) and (1.5) is coincide with the general solution of first-order linear nonhomogeneous neutral differential equations [26,27]. However, for the case a ¼ 1 and c – 0, Theorem 11 in [28] is not coincide with the general solution. Acknowledgement The authors would like to thank the reviewers for their valuable comments on this paper, which helps to improve the quality of this paper. References [1] [2] [3] [4] [5] [6] [7] [8]
I. Podlubny, Fractional Differential Equations, Academic Press, New York, 1993. K.S. Miller, B. Ross, An Introduction to the Fractional Calculus and Fractional Differential Equations, Wiley, New York, 1993. Y. Zhou, Basic Theory of Fractional Differential Equations, World Scientific, Singapore, 2014. A.A. Kilbas, Hari M. Srivastava, Juan J. Trujillo, Theory and Applications of Fractional Differential Equations, Elsevier Science B.V., Amsterdan, 2006. V. Lakshmikantham, S. Leela, J. Vasundhara Devi, Theory of Fractional Dynamic Systems, Cambridge Scientific Pub., Cambridge, UK, 2009. Y. Zhou, F. Jiao, J. Li, Existence and uniqueness for p-type fractional neutral differential equations, Nonlinear Anal. 71 (2009) 2724–2733. Y. Zhou, F. Jiao, J. Li, Existence and uniqueness for fractional neutral differential equations with infinite delay, Nonlinear Anal. 71 (2009) 3249–3256. V. Lakshmikantham, A.S. Vatsala, Basic theory of fractional differential equations, Nonlinear Anal. 69 (2008) 2677–2682.
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056
X.-F. Zhou et al. / Applied Mathematics and Computation xxx (2015) xxx–xxx
13
[9] V. Lakshmikantham, Theory of fractional functional differential equations, Nonlinear Anal. 69 (2008) 3337–3343. [10] Z. Odibat, S. Momani, Application of variational iteration method to nonlinear differential equations of fractional order, Int. J. Nonlinear Sci. Numer. Simul. 7 (1) (2006) 27–34. [11] B. Bonilla, M. Rivero, J.J. Trujillo, On systems of linear fractional differential equations with constant coefficients, Appl. Math. Comput. 187 (1) (2007) 68–78. [12] Y. Zhou, L. Zhang, X.H. Shen, Existence of mild solutions for fractional evolution equations, J. Int. Equs. Appl. 25 (2013) 557–585. [13] Y. Zhou, F. Jiao, J. Pecaric, On the Cauchy problem for fractional functional differential equations in Banach spaces, Topological Methods Nonlinear Anal. 42 (2013) 119–136. [14] Y. Zhou, X.H. Shen, L. Zhang, Cauchy problem for fractional evolution equations with Caputo derivative, Eur. Phys. J. Special Topics 222 (2013) 1747– 1764. [15] M.M. Khader, Talaat S. EI Danaf, A.S. Hendy, A computational matrix method for solving systems of high order fractional differential equations, Appl. Math. Model. 37 (2013) 4035–4050. [16] W. Deng, C. Li, J. Lü, Stability analysis of linear fractional differential system with multiple time delays, Nonlinear Dyn. 48 (2007) 409–416. [17] H. Trinh, M. Aldeen, On the stability of linear systems with delayed perturbation, IEEE Trans. Automat. Contr. 139 (1994) 1948–1951. [18] M.L. Morgado, N.J. Ford, P.M. Lima, Analysis and numerical methods for fractional differential equations with delay, J. Comput. Appl. Math. 252 (2013) 159–168. [19] Z.M. Odibat, Analytic study on linear systems of fractional differential equations, Comput. Math. Appl. 59 (2010) 1171–1183. [20] M.S. Tavazoei, M. Haeri, A note on the stability of fractional order systems, Math. Comput. Simul. 79 (5) (2009) 1566–1576. [21] S. Momani, Z. Odibat, Numerical approach to differential equations of fractional order, J. Comput. Appl. Math. 207 (1) (2007) 96–110. [22] K. Diethelm, N. Ford, A. Freed, A predictor-corrector approach for the numerical solution of fractional differential equations, Nonlinear Dyn. 29 (2002) 3–22. [23] Z. Odibat, S. Momani, H. Xu, A reliable algorithm of homotopy analysis method for solving nonlinear fractional differential equations, Appl. Math. Model. 34 (2010) 593–600. [24] J. Wang, M. fec˘kan, Y. Zhou, Fractional order iterative functional differential equations with parameter, Appl. Math. Model. 37 (2013) 6055–6067. [25] H. Ye, J. Gao, Y. Ding, A generalized Gronwall inequality and its application to a fractional differential equation, J. Math. Anal. Appl. 328 (2007) 1075– 1081. [26] J.K. Hale, S.M. Verduyn Lunel, Introduction to Functional Differential Equatons, Springer-Verlag, New York, NY, USA, 1993. [27] X.F. Zhou, W. Jiang, L.G. Hu, Stability analysis of n-dimensional neutral differential equations with multiple delays, J. Math. Res. Appl. 33 (2013) 631– 652, http://dx.doi.org/10.3770/j.issn:2095-2651.2013.06.001.
. [28] H. Zhang, J. Cao, W. Jiang, General solution of linear fractional neutral differential difference equations, Discrete Dyn. Nature Soc. (2013) 7. Article ID 489521.
Please cite this article in press as: X.-F. Zhou et al., Analytic study on linear neutral fractional differential equations, Appl. Math. Comput. (2015), http://dx.doi.org/10.1016/j.amc.2014.12.056