arXiv:1512.05772v1 [math.CA] 17 Dec 2015

Stability by Lyapunov functions of Caputo fractional differential equations with non-instantaneous impulses

Ravi Agarwala,c,∗ ,

a

S. Hristovab ,

D. O’Regand,c

Department of Mathematics, Texas A&M University-Kingsville, Kingsville, TX 78363, USA

b

Department of Applied Mathematics, Plovdiv University, Plovdiv, Bulgaria

d

School of Mathematics, Statistics and Applied Mathematics, National University of Ireland, Galway, Ireland

c

NAAM Research Group, King Abdulaziz University, Jeddah, Saudi Arabia ∗

a

e-mail address: [email protected]

b

e-mail address: [email protected]

d

e-mail address: [email protected]

corresponding author

Abstract The stability of the zero solution of a nonlinear Caputo fractional differential equation with noninstantaneous impulses is studied using Lyapunov like functions. The novelty of this paper is based on the new definition of the derivative of a Lyapunov like function along the given noninstantaneous impulsive fractional differential equations. On one side this definition is a natural generalization of Caputo fractional Dini derivative of a function and on the other side it allows us the assumption for Lyapunov functions to be weakened to continuity. By appropriate examples it is shown the natural relationship between the defined derivative of Lyapunov functions and Caputo derivative. Several sufficient conditions for uniform stability and asymptotic uniform stability of the zero solution, based on the new definition of the derivative of Lyapunov functions are established. Some examples are given to illustrate the results. Key words : stability, Caputo derivative, Lyapunov functions, non-instantaneous impulses, fractional differential equations. AMS Subject Classifications : 34A34, 34A08, 34D20

1.

INTRODUCTION

In the real world life there are many processes and phenomena that are characterized by rapid changes in their state, so called impulses. Mainly there are two types of impulses: - instantaneous impulses- the duration of these changes is relatively short compared to the overall duration of the whole process and the changes turn out to be irrelevant to the delopment of the studied process. The model of such kind of processes is given by impulsive differential equations (see, for example, [3], [20], [21], [22], [6], [23], [24], [25], [26], [28], the monographs [19], [30] and cited therein references); - noninstantaneous impulses - an impulsive action, which starts at an arbitrary fixed point and keeps active on a finite time interval. To achieve this aim E. Hernandez and D. O’Regan ([18]) introduced a new class of abstract differential equations for which the impulses are not instantaneous and investigated the existence of mild and classical solutions. Later several authors studied and obtained qualitative properties of the solutions such as existence ([40], [41], [48], [49]), stability properties ([44]), boundary value problems ([47]), periodic solutions ([14]). The stability of fractional order systems is quite recent. There are several approaches in the literature to study stability, one of which is the Lyapunov approach. Results on stability of fractional differential equations in the literature via Lyapunov functions could be divided into two main groups: - continuously differentiable Lyapunov functions with their Caputo fractional derivatives (see, for example, the papers [5], [27], [33], [36]). - continuous Lyapunov functions with their fractional Dini derivatives (see, for example, the papers [12], [31], [32]). Introducing impulses into fractional differential equations leads to complications on the stability properties of the solutions. In the literature there are several papers which examine existence and qualitative properties of fractional differential equations with instantaneous impulses (see, for example, [2], [7], [9], [10], [46]). There are also results concerning fractional differential equations with noninstantaneous impulses; see for example [1], [15], [16], [17], [29], [34], [38], [51], [52]. In the above cited papers the impulses start abruptly at some points and their action continues on a finite interval. As a motivation for the study of such kind of systems we consider the following simplified situation concerning the hemodynamical equilibrium of a person. In the case of a decompensation (for example, high or low levels of glucose) one can prescribe some intravenous drugs (insulin). Since the introduction of the drugs in the bloodstream and the consequent absorption for the body are gradual and continuous processes, we can interpret the situation as an impulsive action which starts abruptly and stays active on a finite time interval. In this paper the stability of the zero solution of the noninstantaneous impulsive nonlinear Caputo fractional differential equations is studied. We define in an appropriate way two types of fractional derivatives of Lyapunov functions: Fractional Dini derivative as well as Caputo fractional Dini derivative. Their applications is discussed. Several sufficient conditions for uniform stability and asymptotic uniform stability are obtained. Some examples illustrating the obtained results are given.

2.

NOTES ON FRACTIONAL CALCULUS

Fractional calculus generalizes the derivative and the integral of a function to a non-integer order [11, 13, 31, 42, 43] and there are several definitions of fractional derivatives and fractional integrals. In engineering, the fractional order q is often less than 1, so we restrict our attention to q ∈ (0, 1). 1: The Riemann–Liouville (RL) fractional derivative of order q ∈ (0, 1) of m(t) is given by (see, for example, 1.4.1.1 [11], or [42]) RL q t0 D m(t)

1 d = Γ (1 − q) dt

Zt

(t − s)−q m(s)ds,

t ≥ t0 .

t0

where Γ(.) denotes the Gamma function. 2: The Caputo fractional derivative of order q ∈ (0, 1) is defined by (see, for example, 1.4.1.3 [11]) Zt 1 c q (t − s)−q m0 (s)ds, t ≥ t0 . (1) t0 D m(t) = Γ (1 − q) t0

The Caputo and Riemann-Liouville formulations coincide when m(t0 ) = 0. The properties of the Caputo derivative are quite similar to those of ordinary derivatives. Also, the initial conditions of fractional differential equations with the Caputo derivative has a clear physical meaning and as a result the Caputo derivative is usually used in real applications. 3: The Grunwald−Letnikov fractional derivative is given by (see, for example, 1.4.1.2 [11]) [

t−t0

] h 1 X GL q (−1)r (qCr)m(t − rh), t0 D m(t) = lim q h→0 h

t ≥ t0 ,

r=0

and the Grunwald−Letnikov fractional Dini derivative by t−t0

[ h ] 1 X GL q (−1)r (qCr)m(t − rh), t0 D+ m(t) = lim sup q h→0+ h

t ≥ t0 ,

(2)

r=0

t−t0 0 where qCr are the Binomial coefficients and [ t−t h ] denotes the integer part of the fraction h . c q The relations between the three types of fractional derivatives are given by t0 D m(t) = RL D q [m(t) − m(t )] 0 t0

Proposition 1. (Theorem 2.25 [13]). Let m ∈ C 1 [t0 , b]. Then, for t ∈ (t0 , b], RL D q m(t) . t0

GL D q m(t) t0

=

−q

(t−t0 ) q Also, according to Lemma 3.4 ([13]), ct0 Dtq m(t) =RL t0 Dt m(t) − m(t0 ) Γ(1−q) holds. From the relation between the Caputo fractional derivative and the Grunwald − Letnikov fractional derivative using (2) we define the Caputo fractional Dini derivative as q c t0 D+ m(t)

=

GL q t0 D+ [m(t)

− m(t0 )],

i.e. t−t0

[ h ]  i X 1h q c r+1 D m(t) = lim sup m(t) − m(t ) − (−1) (qCr) m(t − rh) − m(t ) . 0 0 t0 + q h→0+ h r=1

(3)

Definition 1. ([12]) We say m ∈ C q ([t0 , T ], Rn ) if m(t) is differentiable (i.e. m0 (t) exists), the Caputo derivative ct0 Dq m(t) exists and satisfies (1) for t ∈ [t0 , T ]. Remark 1. Definition 1 coud be extended to any interval I ⊂ R+ . Remark 2. If m ∈ C q ([t0 , T ], Rn ) then

3.

c D q m(t) t0 +

=

c D q m(t). t0

NONINSTANTANEOUS IMPULSES IN FRACTIONAL DIFFERENTIAL EQUATIONS

∞ In this paper we will assume two increasing sequences of points {ti }∞ i=1 and {si }i=0 are given such that s0 = 0 < ti ≤ si < ti+1 , i = 1, 2, . . . , and limk→∞ tk = ∞. Let t0 ∈ ∪∞ k=0 [sk , tk+1 ) be a given arbitrary point. Without loss of generality we will assume that t0 ∈ [s0 , t1 ), i.e. 0 ≤ t0 < t1 . Define the sequence of points {τi }∞ i=0 by  t0 for k = 0, τk = sk for k ≥ 1.

Consider the initial value problem (IVP) for the system of noninstantaneous impulsive fractional differential equations (NIFrDE) with a Caputo derivative for 0 < q < 1, c q τk D x

= f (t, x) for t ∈ (τk , tk+1 ], k = 0, 1, 2, . . . ,

x(t) = φi (t, x(ti − 0))

for t ∈ (ti , si ], k = 1, 2, . . . ,

(4)

x(t0 ) = x0 n n n n where x, x0 ∈ Rn , f : ∪∞ k=0 [τk , tk+1 ] × R → R , φi : [ti , si ] × R → R , (i = 1, 2, 3, . . . ). We suppose that the function f (t, x) is smooth enough, such that for any initial value x ˜0 ∈ Rn the IVP for the system of Caputo fractional differential equations (FrDE) c q τk D x

= f (t, x)

for t ∈ [τk , tk+1 ]

with x(τk ) = x ˜0

(5)

has a solution x(t) = x(t; τk , x ˜0 ) ∈ C q ([τk , tk ], Rn ). Some sufficient conditions for global existence of solutions of (5) are given in [8], [31]. The IVP for FrDE (5) is equivalent to the integral equation Z t 1 (t − s)q−1 f (s, x(s))ds for t ∈ [τk , tk+1 ]. (6) x(t) = x ˜0 + Γ(q) τk Remark 3. The intervals (tk , sk ], k = 1, 2, . . . are called intervals of noninstantaneous impulses and the functions φk (t, x), k = 1, 2, . . . , are called noninstantaneous impulsive functions. Now t0 is the initial time. We will assume throughout this paper that the initial time t0 is not in an interval of noninstantaneous impulses, i.e. we will assume t0 ∈ ∪∞ k=0 [sk , tk+1 ). We will give a brief description of the solution of IVP for NIFrDE (4). The solution x(t; t0 , x0 ), t ≥ t0 of (4) is given by  Xk (t) for t ∈ (sk , tk+1 ], k = 0, 1, 2, . . . , x(t; t0 , x0 ) = φk (t, Xk−1 (tk − 0)) for t ∈ (tk , sk ], k = 1, 2, . . . , where

- X0 (t) is the solution of IVP for FrDE (5) for k = 0, t ∈ [t0 , t1 ], x ˜0 = x0 and X0 (t) satisfies (6) on [t0 , t1 ]; - X1 (t) is the solution of IVP for FrDE (5) for k = 1, t ∈ [s1 , t2 ], x ˜0 = φ1 (s1 , X0 (t1 − 0)), and X1 (t) satisfies (6) on [s1 , t2 ]; - X2 (t) is the solution of IVP for FrDE (5) for k = 2, t ∈ [s2 , t3 ], x ˜0 = φ2 (s2 , X1 (t2 − 0)), and X2 (t) satisfies (6) on [s2 , t3 ]; and so on. Also, the solution x(t) = x(t; t0 , x0 ), t ≥ t0 of (4) is given by  Rt 1 q−1 f (s, x(s))ds for t ∈ [t , t ], x0 + Γ(q)  0 1 t0 (t − s)     φ (t, x(t − 0)) for t ∈ (t  1 1 1 , s1 ],  Rt  1 q−1  φ1 (s1 , x(t1 − 0)) + Γ(q) s1 (t − s) f (s, x(s))ds for t ∈ [s1 , t2 ],      φ2 (t, x(t2 − 0)) for t ∈ (t2 , s2 ],  Rt 1 q−1 f (s, x(s))ds for t ∈ [s , t ], x(t) = φ2 (s2 , x(t2 − 0)) + Γ(q) (t − s) 2 3 s2    ..................................................................................     φk (t, x(tk − 0)) for t ∈ (tk , sk ],   Rt   1 q−1 f (s, x(s))ds for t ∈ [s , t  (t − s) φ (s , x(t − 0)) + k k k k k+1 ],  Γ(q) sk   .................................................................................. Remark 4. If tk = sk , k = 1, 2, . . . then the IVP for NIFrDE (4) reduces to an IVP for impulsive fractional differential equations studied in [2], [7], [9], [10]. In this case at any point of instantaneous impulse tk the amount of jump of the solution x(t) is given by ∆x(tk ) = x(tk + 0) − x(tk − 0) = Φk (x(tk − 0)) = φk (tk , x(tk − 0)) − x(tk − 0). Remark 5. In the case q = 1 the IVP for NIFrDE (4) reduces to an IVP for noninstantaneous impulsive differential equations studied in [34], [44], [47], [48]. Remark 6. In the case q = 1, tk = sk , k = 1, 2, . . . the IVP for NIFrDE (4) reduces to an IVP for impulsive differential equations (see for example the books [19], [30] and the cited references therein). Let J ⊂ R+ be a given interval. We introduce the following classes of functions
n [
n P C q (J) = {u ∈ C q (J ∩ ∪∞ [s , t ) , R ) C(J ∩ ∪∞ k k+1 k=0 k=1 (tk , sk ] , R ) : u(tk ) = u(tk − 0) = lim u(t) < ∞, u(tk + 0) = lim u(t) < ∞ t↑tk

t↓tk

for k : tk ∈ J, u(sk ) = u(sk − 0) = lim u(t) = u(sk + 0) = lim u(t) for k : sk ∈ J}, t↑sk t↓sk
n ∞ P C(J) = {u ∈ C(J ∩ ∪k=0 (tk , tk+1 ) , R ) : u(tk ) = u(tk − 0) = lim u(t) < ∞ and u(tk + 0) = lim u(t) < ∞ t↑tk

t↓tk

for k : tk ∈ J}. Remark 7. According to the above description any solution of (4) is from the class P C q ([t0 , b)), b ≤ ∞, i.e. any solution could have a discontinuity at points tk , k = 1, 2, . . . .

Example 1. Consider the IVP for the scalar NIFrDE c q τk D x

= Ax for t ∈ (τk , tk+1 ], k = 0, 1, 2, . . . ,

x(t) = Ψk (t, x(tk − 0)) for

t ∈ (tk , sk ], k = 1, 2, . . . ,

(7)

x(t0 ) = x0 , where x, x0 ∈ R, A is a constant. The solution of (7) is given by  x0 Eq (A(t − t0 )q )   Ψk (t,x(tk − 0))  x(t; 0, x0 ) =    Ψk sk , x(tk − 0) Eq A(t − sk )q

for t ∈ [t0 , t1 ] for t ∈ (tk , sk ], k = 1, 2, . . . for t ∈ [sk , tk+1 ], k = 1, 2, 3, . . .

where the Mittag − Leffler function (with one parameter) is defined by Eq (z) =

∞ P k=0

zk Γ(qk+1) .

If Ψk (t, x) = ak (t)x, ak : [tk , sk ] → R, k = 1, 2, 3, . . . , then the solution of NIFrDE (7) is given by   x0 Eq (A(t −t0 )q ) for t ∈ [0, t1 ],   Qk−1  q  q x0 Eq (AC )  i=1 ai (si )Eq (ACi ) ak (t)  x(t; 0, x0 ) = for t∈ (tk , sk ], k = 1, 2, . .. ,  Qk−1  q  q q x0 Eq (AC )ak (sk )  i=1 ai (si )Eq (ACi ) Eq (A(t − sk ) )    for t ∈ (sk , tk+1 ], k = 1, 2, . . . . where C = t1 − t0 and Ck = sk − tk ≥ 0, k = 1, 2, . . . . If A = 0 and Ψk (t, x) = ak (t)x, ak : [tk , sk ] → R, k = 1, 2, 3, . . . the solution of NIFrDE (7) is given by  for t ∈ [t0 , t1 ],   Q  x0  k−1 x0 ai (si ) ak (t) for t ∈ (tk , sk ], k = 1, 2, . . . x(t; 0, x0 ) = i=1   Q  k  x0 ai (si ) for t ∈ (sk , tk+1 ], k = 1, 2, . . . i=1



4.

DEFINITIONS CONCERNING STABILITY AND LYAPUNOV FUNCTIONS

The goal of the paper is to study the stability properties of the system NIFrDEs (4). In the definition below we denote by x(t; t0 , x0 ) ∈ P C q ([t0 , ∞), Rn ) any solution of (4). Definition 2. The zero solution of the IVP for NIFrDE (4) is said to be • stable if for every  > 0 and t0 ∈ ∪∞ k=0 [sk , tk+1 ) there exist δ = δ(, t0 ) > 0 such that for any x0 ∈ Rn the inequality ||x0 || < δ implies ||x(t; t0 , x0 )|| <  for t ≥ t0 ; • uniformly stable if for every  > 0 there exist δ = δ() > 0 such that for any initial n point t0 ∈ ∪∞ k=0 [sk , tk+1 ) and any inital value x0 ∈ R with ||x0 || < δ the inequality ||x(t; t0 , x0 )|| <  holds for t ≥ t0 ;

• uniformly attractive if for β > 0 : for every  > 0 there exist T = T () > 0 such that n for any initial pont t0 ∈ ∪∞ k=0 [sk , tk+1 ) and any initial value x0 ∈ R with ||x0 || < β the inequality ||x(t; t0 , x0 )|| <  holds for t ≥ t0 + T ; • uniformly asymptotically stable if the zero solution is uniformly stable and uniformly attractive. Example 2. Consider the scalar NIFrDE (7) where A ≤ 0 and ΨQ k (t, x) = ak (t)x and ak : [tk , sk ] → R, k = 1, 2, 3, . . . are such that supt∈[tk ,sk ] |ak (t)| ≤ Mk , ∞ i=1 Mi < ∞ where Mk > 0 are constants. According to Example 1 and the inequality 0 < Eq (A(T − τ )q ) ≤ 1 for T ≥ τ there exists a constant M > 0 such that |x(t; t0 , x0 )| ≤ M |x0 |

for t ≥ t0 .

(8)

Inequality (8) guarantees that the zero solution of (7) is uniformly stable. 

In this paper we will use the followings sets: K = {a ∈ C[R+ ,R+ ] : a is strictly increasing and a(0) = 0}, S(A) = {x ∈ Rn : ||x|| ≤ A},

A > 0.

We now introduce the class Λ of Lyapunov-like functions which will be used to investigate the stability of the zero solution of the system NIFrDE (4). Definition 3. Let J ∈ R+ be a given interval, and ∆ ⊂ Rn , 0 ∈ ∆ be a given set. We will say that the function V (t, x) : J × ∆ → R+ , V (t, 0) ≡ 0 belongs to the class Λ(J, ∆) if 1. The function V (t, x) is continuous on J/{tk ∈ J} × ∆ and it is locally Lipschitzian with respect to its second argument; 2. For each tk ∈ Int(J) and x ∈ ∆ there exist finite limits V (tk − 0, x) = lim V (t, x) < ∞, t↑tk

and V (tk + 0, x) = lim V (t, x) < ∞ t↓tk

and the following equalities are valid V (tk − 0, x) = V (tk , x). Remark 8. In the case when the Lyapunov function does not depend on the time t, i.e. V (t, x) = V (x) ∈ C[∆, R+ ], V (0) = 0, ∆ ⊂ Rn , 0 ∈ ∆, and the function V (x) is locally Lipschitzian then we will say V (x) ∈ ΛC (∆). Lyapunov-like functions used to discuss stability for differential equations require an appropriate definition of their derivatives along the studied differential equations. For fractional differential equations some authors (see, for example, [33], [36]) used the so called Caputo fractional derivative of Lyapunov function ct0 Dq V (t, x(t)) where x(t) is the unknown solution of the studied fractional differential equation. This approach requires the function to be smooth enough (at least continuously differentiable) and also some conditions involved are quite restrictive. Other authors used the so called Dini fractional derivative of Lyapunov function

([31], [32]). This is based on the Dini derivative of the Lyapunov function V (t, x) among the ordinary differential equation x0 = f (t, x) given by i 1h DV (t, x) = lim sup V (t, x) − V (t − h, x − hf (t, x) . (9) h→0 h The authors generalized (9) to the f (t, x), t ≥ t0 by c

Dini fractional derivative along the FrDE

q D+ V (t, x) = lim sup h→0

i 1h q V (t, x) − V (t − h, x − h f (t, x) . hq

c Dq x t0

=

(10)

This definition requires only the continuity of the Lyapunov function. In this paper we will use piecewise continuous Lyapunov functions from the above introduced class Λ([t0 , T ), ∆). We will introduce the derivative of Lyapunov function in two different ways and we will discuss their applications. We now define the generalized Caputo fractional Dini derivative of the Lyapunov-like function V (t, x) ∈ Λ([t0 , T ), ∆) along trajectories of solutions of IVP for the system NIFrDE (4). It is based on the Caputo fractional Dini derivative of a function m(t) given by (3). The Caputo fractional Dini derivative along trajectories of solutions of IVP for the system NIFrDE (4) is given by:  1 q c (4) D+ V (t, x; t0 , x0 ) = lim sup q V (t, x) − V (t0 , x0 ) h→0+ h [

−

t−t0

] h X

r+1

(−1)

  q qCr V (t − rh, x − h f (t, x)) − V (t0 , x0 )

(11)

r=1

for t ∈ (sk , tk+1 ) ∩ (t0 , T ), k = 0, 1, 2, . . . , where x, x0 ∈ ∆, and for any t ∈ (sk , tk+1 ) ∩ (t0 , T ) there exists ht > 0 such that t − h ∈ (sk , tk+1 ) ∩ (t0 , T ), x − hq f (t, x) ∈ ∆ for 0 < h ≤ ht . Remark 9. The generalized Caputo fractional Dini derivative of the Lyapunov function was introduced and used for studying stability properties of the zero solution of Caputo fractional differential equations in [4]. The formula (11) could be reduced to q c (4) D+ V

(t, x; t0 , x0 ) 

= lim sup h→0+

1 V (t, x) − hq

− V (t0 , x0 )

[

t−t0

] h X

r=1 )−q

(t − t0 Γ(1 − q)

 (−1)r+1 qCrV (t − rh, x − hq f (t, x))

(12)

for t ∈ (sk , tk+1 ) ∩ (t0 , T ), k = 0, 1, 2, . . . ,

Now, based on the Dini fractional derivative of a continuous Lyapunov function defined by (10), we will define the generalized Dini fractional derivative of the Lyapunov-like function V (t, x) ∈ Λ([t0 , T ), ∆) along trajectories of solutions of IVP for the system NIFrDE (4) by i 1h q c q D V (t, x) = lim sup V (t, x) − V (t − h, x − h f (t, x)) (4) + hq h→0+ (13) for t ∈ (sk , tk+1 ) ∩ (t0 , T ), k = 0, 1, 2, . . . ,

where x ∈ ∆, and for any t ∈ (sk , tk+1 ) ∩ (t0 , T ) there exists ht > 0 such that t − h ∈ (sk , tk+1 ) ∩ (t0 , T ), x − hq f (t, x) ∈ ∆ for 0 < h ≤ ht . Example 3. Let V ∈ Λ(R+ , R) be given by V (t, x) = m(t)g(x) where the function n m ∈ C 1 (∪∞ i=0 (sk , tk+1 ), R+ ), g : R → R+ is a locally Lipshitz function such that the limit Fr

Dq g(x) = lim sup h→0+

g(x) − g(x − hq f (t, x)) hq

exists for x ∈ Rn . First we apply the formula (13) to obtain the generalized Dini fractional derivative of the considered Lyapunov function. We obtain
i 1h q q c m(t)g(x) − m(t − h)g x − h f (t, x) D V (t, x) = lim sup (4) + hq h→0+ g(x) − g(x − hq f (t, x)) = m(t) lim sup hq h→0+ (14)  
m(t) − m(t − h)  + lim sup lim sup h1−q g x − hq f (t, x) h h→0+ h→0+ = m(t) F r Dq g(x) for t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . . Next we use (12) to obtain the generalized Caputo fractional Dini derivative of the (t−t0 )−q q function V . Let t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . . Apply equalities (2), RL t0 D 1 = Γ(1−q) , and [

limh→0+ sup h1q

t−t0 ] h

P

(−1)r qCr m(t − rh) =

r=0

q c (4) D+ V

RL D q t0



 m(t) to (12) and obtain

(t, x; t0 , x0 ) t−t0

[ h ] i X 1h q = lim sup q m(t)g(x) + g(x − h f (t, x)) (−1)r qCr m(t − rh) h h→0+ r=1

(t − t0 )−q − m(t0 )g(x0 ) Γ(1 − q) g(x) − g(x − hq f (t, x)) = m(t) lim sup hq h→0+ [

t−t0 ] h

(15)

i 1 X (−1)r qCr m(t − rh) + lim g(x − hq f (t, x)) lim sup q h→0 h h→0+ r=0

− m(t0 )g(x0 ) = m(t)

Fr

for

)−q

(t − t0 Γ(1 − q)

 (t − t0 )−q m(t) − m(t0 )g(x0 ) , Γ(1 − q) t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . .

Dq g(x) + g(x)

RL q t0 D



or q c (4) D+ V

(t, x; t0 , x0 )

= m(t)

Fr

for

   (t − t0 )−q m(t) + g(x) − g(x0 ) m(t0 ) , Γ(1 − q) t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . .

Dq g(x) + g(x)

C q t0 D



(16)

x 1.0

x 0.8 0.5

q=0.5

0.6

t

q=0.1

5

10

15

20

q=0.8

0.4

q=0.5

-0.5 0.2 -1.0 t 5

10

15

20

25

30

-1.5

Figure 1. Graphs of solutions of c q 0 D x + x(t) = 1 for various q.

Figure 2. Example 8. Graph of f (t).

Example 4. Let V ∈ Λ(R+ , R) be given by the equality V (t, x) = m(t)x2 where m ∈ C 1 (∪∞ i=0 (sk , tk+1 ), R+ ) and x ∈ R. In this case Fr

x2 − (x − hq f (t, x))2 hq h→0+ q = lim f (t, x)(2x − h f (t, x)) = 2xf (t, x)

Dq g(x) =

Fr

Dq (x2 ) = lim sup

(17)

h→0

First we apply the formula (13) to obtain the generalized Dini fractional derivative of the considered Lyapunov function. From (14) and (17) we obtain q c (4) D+ V

for t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . .

(t, x) = 2x m(t)f (t, x),

(18)

q Note c(4) D+ V (t, x) does not depend on the order q of the fractional differential equation. The behavior of solutions of fractional differential equations depends significantly on the order q. For example, let us consider the simple fractional differential equation c0 Dq x + x(t) = 1, x(0) = 0 whose solution is given by x(t) = tq Eq,1+q (−tq ). From Figure 1 it can be seen limt→∞ x(t) = a where a is different for different values of the order q of fractional differential equation. Next we use (12) to obtain the generalized Caputo fractional Dini derivative of the function V . Let t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . . From (15), (16) for g(x) = x2 and (17) we obtain q c (4) D+ V

(t, x; t0 , x0 ) =  (t − t0 )−q m2 (t) − (x0 )2 m2 (t0 ) , Γ(1 − q) t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . .

= 2x m2 (t)f (t, x) + x2 for

RL q t0 D



(19)

or q c (4) D+ V

(t, x; t0 , x0 )    (t − t0 )−q m2 (t) + x2 − (x0 )2 m2 (t0 ) , Γ(1 − q) t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . .

= 2x m2 (t)f (t, x) + x2 for

C q t0 D



(20)

q Note the generalized Caputo fractional Dini derivative c(4) D+ V (t, x; t0 , x0 ) depends significantly not only on the order q of the fractional differential equation but also on the initial data.

The derivative of the Lyapunov function in the well known case for first order impulsive differential equations (q = 1) is D+ V (t, x) = 2x m2 (t)f (t, x) + x2

dh 2 i m (t) , dt

t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . .

(21)

The generalized Caputo fractional Dini derivative given by formula (12) seems to be the natural generalization of the derivative of Lyapunov functions for ordinary differential equations (with or without impulses). 

COMPARISON RESULTS

5.

Again in this section we assume 0 < q < 1. Also, t0 ∈ ∪∞ k=0 [sk , tk+1 ) so without loss of generality we assume t0 ∈ [s0 , t1 ). We will obtain some comparison results for NIFrDE (4) using the definitions (12) and (13) for a derivative of Lyapunov-like function.

5.1.

Generalized Caputo fractional Dini derivative

In this section we will use the following result for fractional differential equations: Lemma 1. ([4]) Assume the following conditions are satisfied: 1. The function x∗ (t) = x(t; t0 , x0 ) ∈ C q ([t0 , T ], ∆), is a solution of the FrDE c q t0 D x

= f (t, x),

t ∈ [t0 , T ]

with x(t0 ) = x0

(22)

where ∆ ⊂ Rn , 0 ∈ ∆, x0 ∈ ∆ and t0 , T ∈ R+ , t0 < T are given constants. 2. The function V ∈ ΛC ([t0 , T ], ∆) and the inequality q c (22) D+ V

(t, x; t0 , x0 ) ≤ 0 for (t, x) ∈ [t0 , T ] × ∆

holds. Then the inequality V (t, x∗ (t)) ≤ V (t0 , x0 ) holds for t ∈ [t0 , T ]. Now we will prove some comparison results for noninstantaneous impulsive Caputo fractional differential equations. Lemma 2. (Comparison result for NIFrDE by generalized Caputo fractional Dini derivative). Let: 1. The function x∗ (t) = x(t; t0 , x0 ) ∈ P C q ([t0 , T ], ∆) is a solution of the NIFrDE (4) where ∆ ⊂ Rn , 0 ∈ ∆, x0 ∈ ∆ and t0 , T are given constants such that t0 ∈ [s0 , t1 ), T > t0 . 2. The function V ∈ Λ([t0 , T ], ∆) and (i) the inequality q c (4) D+ V

holds;

(t, x∗ (t); t0 , x0 ) ≤ 0

for t ∈ (t0 , T )

\

∪∞ k=0 (sk , tk+1 )

(23)

(ii) the inequalities V (t, x∗ (t)) ≤ V (tk − 0, x∗ (tk − 0)) for t ∈ [t0 , T ]

\

(tk , sk ] for k = 1, 2, 3, . . .

hold. Then the inequality V (t, x∗ (t)) ≤ V (t0 , x0 ) holds on [t0 , T ]. P r o o f: We use induction to prove Lemma 2. Let t ∈ [t0 , t1 ] ∩ [t0 , T ]. The function x∗ (t) ∈ C q ([t0 , t1 ] ∩ [t0 , T ], Rn ), satisfies FrDE (22) and from Lemma 1 (with T = t1 ) the inequality V (t, x∗ (t)) ≤ V (t0 , x0 ) holds on [t0 , t1 ] ∩ [t0 , T ]. Let T > t1 and t ∈ (t1 , s1 ]∩[t0 , T ]. From condition 2(ii) and the above we get V (t, x∗ (t)) ≤ V (t1 − 0, x∗ (t1 − 0)) = V (t1 , x∗ (t1 )) ≤ V (t0 , x0 ). Let T > s1 and t ∈ (s1 , t2 ] ∩ [t0 , T ]. Consider the function x1 (t) = x∗ (t) for t ∈ (s1 , t2 ] and x1 (s1 ) = x∗ (s1 ) = φ1 (s1 , x∗ (t1 − 0)). The function x1 (t) ∈ C q ([s1 , t2 ], Rn ) and satisfies IVP for FrDE (22) with t0 = s1 , x0 = x∗ (s1 ), and T = t2 . Using condition 2(i), Lemma 1 for the function x1 (t) and the above we obtain V (t, x∗ (t)) = V (t, x1 (t)) ≤ V (s1 , x1 (s1 )) = V (s1 , x∗ (s1 )) ≤ V (t0 , x0 ). Continue this process and an induction argument proves the claim of Lemma 2 is true for t ∈ [t0 , T ]. 

Lemma 3. ([4]). Let the following conditions be satisfied: 1. The function x∗ (t) = x(t; t0 , x0 ) ∈ C q ([t0 , T ], ∆) is a solution of the FrDE (22) where ∆ ⊂ Rn , 0 ∈ ∆, x0 ∈ ∆, t0 , T ∈ R+ , t0 < T are given constants. 2. The function V ∈ ΛC ([t0 , T ], ∆) is such that for any points t ∈ [t0 , T ], x ∈ ∆ the inequality q c (5) D+ V

(t, x; τ0 , x0 ) ≤ −c(||x||)

holds where c ∈ K. Then for t ∈ [t0 , T ] the inequality V (t, x∗ (t)) ≤ V (t0 , x0 ) −

1 Γ(q)

Z

t

(t − s)q−1 c(||x∗ (s)||)ds

(24)

t0

holds. Lemma 4. (Comparison result for NIFrDE, negative generalized Caputo fractional Dini derivtive). Assume the following conditions are satisfied: 1. The function x∗ (t) = x(t; t0 , x0 ) ∈ P C q ([t0 , T ], ∆) is a solution of the NIFrDE (4) where ∆ ⊂ Rn , 0 ∈ ∆, x0 ∈ ∆ and t0 , T are given constants such that t0 ∈ [s0 , t1 ), T > t0 . 2. The function V ∈ Λ([t0 , T ], ∆) and q (i) the inequality c(4) D+ V (t, x∗ (t); t0 , x0 ) ≤ −c(||x∗ (t)||) for t ∈ (t0 , T ) holds where c ∈ K;

T

∪∞ k=0 (sk , tk+1 )

(ii) for any k = 1, 2 . . . the inequalities V (t, x∗ (t)) ≤ V (tk − 0, x∗ (tk − 0)) for t ∈ [t0 , T ]

\ (tk , sk ]

hold. Then for t ∈ [t0 , T ] the inequality  Rt 1 q−1 c(||x∗ (s)||)ds, V (t0 , x0 ) − Γ(q)   t0 (t − s)    t ∈ [t0 , t1 ]   P  R ti+1  k−1 1 q−1  V (t0 , x0 ) − Γ(q) c(||x∗ (s)||)ds  i=0 τi (ti+1 − s)   Rt q−1 c(||x∗ (s)||)ds , V (t, x∗ (t)) ≤ + (t − s) sk     t ∈ (sk , tk+1 ] ∩ (t0 , T ), k ≥ 1   Pk−1 R ti+1  1  q−1  V (t0 , x0 ) − Γ(q) i=0 τi (ti+1 − s) c(||x∗ (s)||)ds,    t ∈ (tk , sk ] ∩ [t0 , T ], k ≥ 1 holds where

 τk =

t0 for k = 0, sk for k ≥ 1.

P r o o f: Let t ∈ [t0 , t1 ] ∩ [t0 , T ]. The function x∗ (t) ∈ C q ([t0 , t1 ] ∩ [t0 , T ], ∆) and satisfies the IVP (22) for T = t1 . From Lemma 3 the inequality V (t, x∗ (t)) ≤ V (t0 , x0 ) − R t for FrDE 1 q−1 c(||x∗ (s)||)ds holds, i.e. the claim of Lemma 4 is true on [t0 , t1 ] ∩ [t0 , T ]. Γ(q) t0 (t − s) Let T > t1 and t ∈ (t1 , s1 ] ∩ [t0 , T ]. From condition 2(ii) and the above we get V (t, x∗ (t)) ≤ V (t1 − 0, x∗ (t1 − 0)) = V (t1 , x∗ (t1 )) Z t1 1 ≤ V (t0 , x0 ) − (t1 − s)q−1 c(||x∗ (s)||)ds. Γ(q) t0 Let T > s1 and t ∈ (s1 , t2 ] ∩ [t0 , T ]. Consider the function x1 (t) = x∗ (t) for t ∈ (s1 , t2 ] and x1 (s1 ) = x∗ (s1 ) = φ1 (s1 , x∗ (t1 − 0)). The function x1 (t) ∈ C q ([s1 , t2 ], Rn ) and satisfies IVP for FrDE (22) with t0 = s1 , x0 = x∗ (s1 ) and T = t2 . Using condition 2(i), Lemma 3 for the function x1 (t), and the above we obtain Z t 1 ∗ (t − s)q−1 c(||x∗ (s)||)ds V (t, x (t)) = V (t, x1 (t)) ≤ V (s1 + 0, x1 (s1 )) − Γ(q) s1 Z t 1 ∗ = V (s1 , x (s1 )) − (t − s)q−1 c(||x∗ (s)||)ds Γ(q) s1 Z t1 Z t 1 1 q−1 ∗ ≤ V (t0 , x0 ) − (t1 − s) c(||x (s)||)ds − (t − s)q−1 c(||x∗ (s)||)ds. Γ(q) t0 Γ(q) s1 Therefore, the claim of Lemma 4 is true on (s1 , t2 ] ∩ [t0 , T ]. Let T > t2 and t ∈ (t2 , s2 ] ∩ [t0 , T ]. From condition 2(ii) and the above we obtain V (t, x∗ (t)) ≤ V (t2 − 0, x∗ (t2 − 0)) = V (t2 , x∗ (t2 )) Z t1 Z t2 1 1 q−1 ∗ (t1 − s) c(||x (s)||)ds − (t2 − s)q−1 c(||x∗ (s)||)ds. ≤ V (t0 , x0 ) − Γ(q) t0 Γ(q) s1 Continue this process and an induction argument proves the claim is true for t ∈ [t0 , T ]. 

Remark 10. The results of Lemma 2 and Lemma 4 are true on the half line (recall [4] that Lemma 1 and Lemma 3 extend to the half line). Remark 11. The results of Lemma 2 and Lemma 4 will be similar with slight changes of condition 2(ii) if the initial time t0 is in a interval of noninstantaneous impulses, i.e. t0 ∈ ∪∞ k=1 (tk , sk ].

5.2.

Generalized Dini fractional derivative

Now we present the analogue results of Section 5.1 using (10). Lemma 5. ([32]). Let V ∈ C(R+ × Rn , R+ ) and V (t, x) be locally Lipschitzian in x. Assume that c q D+ V (t, x) ≤ g(t, V (t, x)), (t, x) ∈ R+ × Rn , q where g ∈ C(R2+ , R) and c D+ V (t, x) is defined by (10). Suppose that the maximal solution r(t; t0 , u0 ) of IVP c q u(t0 ) = u0 ≥ 0 t0 D u = g(t, u),

exists on [t0 , ∞). Then V (t0 , u0 ) ≤ u0 implies V (t, x(t)) ≤ r(t) for t ≥ t0 where x(t) = x(t; t0 , x0 ) is any solution of IVP (22) existing on [t0 , ∞). q Corollary 1. Let V ∈ C(R+ × Rn , R+ ) and V (t, x) be locally Lipschitz in x and c D+ V (t, x) ≤ n 0 for (t, x) ∈ R+ × R . Then V (t, x(t)) ≤ V (t0 , x0 ) for t ≥ t0 where x(t) = x(t; t0 , x0 ) is any solution of IVP (22) existing on [t0 , ∞).

Now we will give a comparison result for NIFrDE (4) by application of generalized Dini fractional derivative of Lyapunov function. Lemma 6. (Comparison result for NIFrDE by generalized Dini fractional derivative). Assume the conditions of Lemma 2 are satisfied where the inequality (23) is replaced by \ q c ∗ D V (t, x (t)) ≤ 0 for t ∈ [t , ∞) ∪∞ (25) 0 k=0 (sk , tk+1 ). (4) + Then the inequality V (t, x∗ (t)) ≤ V (t0 , x0 ) holds on [t0 , ∞). P r o o f: The proof of Lemma 6 is similar to that in Lemma 2 where instead of Lemma 1 we apply Corollary 1. If V (x) = result:

xT x

=

Pn

2 k=1 xk

∈

ΛC (Rn ),



x = (x1 , x2 , . . . , xn

)T

we obtain the following

Corollary 2. (Comparison result by a quadratic Lyapunov function). Let the function x∗ (t) = x(t; t0 , x0 ) ∈ P C q ([t0 , ∞), Rn ) be a solution of the NIFrDE (4) such that
T T (i) x∗ (t) f (t, x∗ (t)) ≤ 0 for t ∈ [t0 , ∞) (sk , tk+1 ), k = 0, 1, 2, . . . . (ii) for any k = 1, 2 . . . the inequalities ||x∗ (t)|| ≤ ||x∗ (τk − 0)|| for t ∈ [t0 , ∞) hold.

\ (tk , sk ] with τk = max{t0 , tk }

Then the inequality ||x∗ (t)|| ≤ ||x0 || holds on [t0 , ∞). P r o o f: The proof follows from Lemma 6 applied with V (x) = xT x = q c (4) D+ V

=

n X

(x) = lim sup h→0+

n

n

i=1

i=1

Pn

2 k=1 xk

and note

2 i 1 hX 2 X q (i) x − x − h f (t, x) i i hq (26)

2xi f (i) (t, x) = 2 xT f (t, x)

i=1

for t ∈ [t0 , ∞)

\

(sk , tk+1 ), k = 0, 1, 2, . . . 

Example 5. Consider the scalar NIFrDE (7) where A < 0, and Ψk (t, x) = ak (t)x, and ak : [tk , sk ] → R, k = 1, 2, 3, . . . are such that supt∈[tk ,sk ] |ak (t)| ≤ 1. Its solution x∗ (t) = x(t; 0, x0 ) is given in Example 1. Using 0 < Eq (A(T − τ )q ) ≤ 1 for T ≥ τ it follows that the conditions of Corollary 2 are satisfied and therefore |x∗ (t)| ≤ |x0 | for t ≥ 0.  ∞ Remark 12. In this paper we assumed an infinite number of points {ti }∞ i=1 , {si }i=1 with 0 < ti ≤ si < ti+1 and limk→∞ tk = ∞. However it is worth noting that the results in Section 5 (and elsewhere) hold true if we only consider a finite of points {ti }pi=1 , {si }pi=1 with 0 < ti ≤ si < ti+1 , i = 1, ...., p with tp+1 = T .

MAIN RESULTS

6.

We will obtain sufficient conditions for stability of the zero solution of nonlinear impulsive Caputo fractional differential equations. Again we assume 0 < q < 1. We say conditions (H) are satisfied if : n ∞ (H1) The function f ∈ C(∪∞ k=0 [sk , tk+1 ], R ), f (t, 0) ≡ 0 for t ∈ ∪k=0 [sk , tk+1 ] is such n ∞ ˜ ˜ that for any initial point (t0 , x ˜0 ) ∈ ∪k=0 [sk , tk+1 ) × R , sk ≤ t0 < tp+1 , p is a nonzero integer, the IVP for the system of FrDE (5) with τk = t˜0 has a solution x(t; t˜0 , x ˜0 ) ∈ C q ([t˜0 , tp+1 ], Rn ). n n (H2) The functions φk ∈ C([tk , sk ] × R , R ), φk (t, 0) ≡ 0 for t ∈ [tk , sk ], k = 1, 2, . . . . Remark 13. Conditions (H) guarantee the existence of a solution x(t; t0 , x0 ) ∈ P C q ([t0 , ∞), Rn ) n of NIFrDE (4) for any initial data (t0 , x0 ) ∈ ∪∞ k=0 [sk , tk+1 ) × R . If (H) is satisfied then (4) has a zero solution.

6.1.

Stability by generalized Caputo fractional Dini derivative

Theorem 1. (Stability). Let the following conditions be satisfied: 1. Condition (H) is satisfied. 2. There exists a function V ∈ Λ(R+ , Rn ) such that V (t, 0) = 0 and (i) the inequality q c (4) D+ V

holds;

(t, x; t0 , x0 ) ≤ 0

n for t ∈ ∪∞ k=0 (sk , tk+1 ), x, x0 ∈ R

(27)

(ii) for any point x ∈ Rn and any t ∈ (tk , sk ], k = 1, 2, 3, . . . the inequality V (t, φk (t, x)) ≤ V (tk − 0, x) holds; (iii) b(||x||) ≤ V (t, x) for t ∈ R+ , x ∈ Rn , where b ∈ K. Then the zero solution of the NIFrDE (4) is stable. P r o o f: Let  > 0 and t0 ∈ ∪∞ k=0 [sk , tk+1 ) be arbitrary given numbers. There exists a p ∈ {0, 1, ...} with t0 ∈ [sp , tp+1 ). Without loss of generality assume p = 0. Since V (t0 , 0) = 0 there exists δ1 = δ1 (t0 , ε) > 0 such that V (t0 , x) < b(ε) for ||x|| < δ1 . Let x0 ∈ Rn with ||x0 || < δ1 . Then V (t0 , x0 ) < b(ε). Consider any solution x∗ (t) = x(t; t0 , x0 ) ∈ P C q ([t0 , ∞), Rn ) of NIFrDE (4). From inequality (27) it follows that \ q c ∗ ∪∞ k=0 (sk , tk+1 ), (4) D+ V (t, x (t); t0 , x0 ) ≤ 0 for t ∈ (t0 , ∞) i.e. condition 2(i) of T Lemma 2 (with T = ∞, see Remark 10) is satisfied. Let t ∈ (tk , sk ] [t0 , ∞), k = 1, 2, 3, . . . . From condition 2(ii) of Theorem 1 we get V (t, x∗ (t)) = V (t, φk (t, x∗ (tk ))) ≤ V (tk − 0, x∗ (tk )) = V (tk − 0, x∗ (tk − 0)). Therefore, condition 2(ii) of Lemma 2 is fulfilled. From Lemma 2 applied to the solution x∗ (t) with T = ∞ (see Remark 10) and condition 2(iii) we obtain b(||x∗ (t)||) ≤ V (t, x∗ (t)) ≤ V (t0 , x0 ) < b(), so the result follows. 

Theorem 2. (Uniform stability). Let the following conditions be satisfied: 1. Condition (H) is satisfied. 2. There exists a function V ∈ Λ(R+ , Rn ) such that (i) the inequality is satisfied q c (4) D+ V

(t, x; t0 , x0 ) ≤ 0

for t ∈ ∪∞ k=0 (sk , tk+1 ), x, x0 ∈ S(λ)

(28)

where λ > 0 is a given number; (ii) for any point t ∈ (tk , sk ], k = 1, 2, 3, . . . and any x ∈ S(λ) the inequality V (t, φk (t, x)) ≤ V (tk − 0, x) holds; (iii) b(||x||) ≤ V (t, x) ≤ a(||x||) for t ∈ R+ , x ∈ Rn , where a, b ∈ K. Then the zero solution of NIFrDE (4) is uniformly stable. P r o o f: Let  ∈ (0, λ] and t0 ∈ ∪∞ k=0 [sk , tk+1 ) be arbitrary given numbers. There exists p ∈ {0, 1, ...} with t0 ∈ [sp , tp+1 ). Without loss of generality assume p = 0. Let δ1 < min{, b()}. From a ∈ K there exists δ2 = δ2 () > 0 so if s < δ2 then a(s) < δ1 . Let δ = min(, δ2 ). Choose the initial value x0 ∈ Rn such that ||x0 || < δ and let x∗ (t) = x(t; t0 , x0 ) ∈ P C q ([t0 , ∞), Rn ) be a solution of the IVP for NIFrDE (4). We now prove that ||x∗ (t)|| < , t ≥ t0 . (29)

Assume inequality (29) is not true and let t∗ = inf {t > t0 : ||x∗ (t)|| ≥ }. Then ||x∗ (t)|| <  for t ∈ [t0 , t∗ )

and ||x∗ (t∗ )|| = .

(30)

If t∗ 6= tk , k = 1, 2, . . . or if t∗ = tp for some natural number p and ||x∗ (tp − 0)|| =  then (30) is true. If for a natural number p we have t∗ = tp and ||x∗ (tp − 0)|| < , then according to Lemma 2 for T = tp and ∆ = S(λ) we obtain V (t, x∗ (t)) ≤ V (t0 , x0 ) for t ∈ [t0 , tp ]. Then for all t ∈ (tp , sp ] from condition 2(iii) we get b(||x∗ (t)||) ≤ V (t, x∗ (t)) = V (t, φp (t, x∗ (tp − 0))) ≤ V (tp − 0, x∗ (tp − 0)) ≤ V (t0 , x0 ) ≤ a(δ) < δ1 < b(ε). Thus ||x∗ (t)|| < b−1 (δ1 ) <  for t ∈ (tp , sp ], and this contradicts the choice of t∗ . Therefore, (30) holds. Then, x∗ (t) ∈ S(λ) on [t0 , t∗ ] and conditions 2(i) and 2(ii) of Lemma 2 are satisfied on [t0 , t∗ ]. From Lemma 2 applied to the solution x∗ (t) with T = t∗ and ∆ = S(λ) we get V (t, x∗ (t)) ≤ V (t0 , x0 ) on [t0 , t∗ ]. Then applying condition 2 (iii) of Theorem 2 we obtain b(ε) = b(||x∗ (t∗ )||) ≤ V (t∗ , x∗ (t∗ )) ≤ V (t0 , x0 ) ≤ a(δ) < δ1 < b(ε). The contradiction proves (29) and therefore, the zero solution of NIFrDE (4) is uniformly stable. 

Now we present some sufficient conditions for the uniform asymptotic stability of the zero solution of the NIFrDE. Theorem 3. (Uniform asymptotic stability). Let the following conditions be satisfied: 1. Condition (H) is satisfied. 2. There exists a positive constant M < ∞ such that

P∞

i=1 (si

− ti ) ≤ M.

3. There exists a function V ∈ Λ(R+ , Rn ) such that (i) the inequality is satisfied q c (4) D+ V

(t, x; t0 , x0 ) ≤ −c(||x||)

for t ∈ ∪∞ k=0 (sk , tk+1 ), x, x0 ∈ S(λ)

(31)

where λ > 0 is a given number, c ∈ K; (ii) for any point t ∈ (tk , sk ], k = 1, 2, 3, . . . and any x ∈ S(λ) the inequality V (t, φk (t, x)) ≤ V (tk − 0, x) holds; (iii) b(||x||) ≤ V (t, x) ≤ a(||x||) for t ∈ R+ , x ∈ Rn , where a, b ∈ K. Then the zero solution of NIFrDE (4) is uniformly asymptotically stable. P r o o f: From Theorem 2 the zero solution of the NIFrDE (4) is uniformly stable. Therefore, for the number λ there exists α = α(λ) ∈ (0, λ) such that for any t˜0 ∈ ∪∞ k=0 [sk , tk+1 ) and x ˜0 ∈ Rn the inequality ||˜ x0 || < α implies ||x(t; t˜0 , x ˜0 )|| < λ for t ≥ t˜0

(32)

where x(t; t˜0 , x ˜0 ) is any solution of the NIFrDE (4) (with initial data (t˜0 , x ˜0 )). Now we will prove that the zero solution of the fractional differential equations (4) is uniformly attractive. Consider the constant β ∈ (0, α] such that a(β) ≤ b(α). Let  ∈ (0, λ] and t0 ∈ ∪∞ k=0 [sk , tk+1 ) be arbitrary given numbers. There exists a p ∈ {0, 1, ...} with t0 ∈ [sp , tp+1 ). Without loss of generality assume p = 0.

Let the point x0 ∈ Rn , ||x0 || < β and x∗ (t) = x(t; t0 , x0 ) be any solution of (4). Then b(||x0 ||) ≤ a(||x0 ||) < a(β) < b(α), i.e. ||x0 || < α and according to (32) the inequality ||x∗ (t)|| < λ for t ≥ t0

(33)

holds, i.e. the solution x∗ (t) ∈ S(λ) on [t0 , ∞). q Choose a constant γ = γ() ∈ (0, ] such that a(γ) < b(). Let T > q a(α) qΓ(q) c(γ) + M and m be a natural number such that sm < t0 + T ≤ tm+1 . Note T depends only on ε but not on t0 . We now prove that ||x∗ (t)|| <  for t ≥ t0 + T. (34) Assume ||x∗ (t)|| ≥ γ for every t ∈ [t0 , t0 + T ].

(35)

Then from Lemma 4 (applied to the interval [t0 , t0 + T ] and ∆ = S(λ)), conditions 2 and 3 (ii) of Theorem 3, inequality aq + bq ≥ (a + b)q for a, b > 0 and the choice of T we get V (t0 + T, x∗ (t0 + T )) m−1 Z 1  X ti+1 ≤ V (t0 , x0 ) − (ti+1 − s)q−1 c(||x∗ (s)||)ds Γ(q) τ i i=0 Z t0 +T  q−1 ∗ + (t0 + T − s) c(||x (s)||)ds sm m−1 Z ti+1

c(γ)  X ≤ a(||x0 ||) − Γ(q) i=0

< a(α) −

(ti+1 − s) τi

c(γ)  (t1 − t0 )q + qΓ(q)

q−1

Z

t0 +T

ds +

(t0 + T − s)q−1 ds



sm m−1 X

(ti+1 − si )q + (T + t0 − sm )q



i=1

m−1 q X c(γ)  ≤ a(α) − (t1 − t0 ) + (ti+1 − si ) + (T + t0 − sm ) qΓ(q) i=1

m q q c(γ)  X c(γ)  = a(α) − − (si − ti ) + T ≤ a(α) − − M + T < 0. qΓ(q) qΓ(q) i=1

The above contradiction proves there exists t∗ ∈ [t0 , t0 + T ] such that ||x∗ (t∗ )|| < γ. Let the natural number p be such that tp−1 ≤ t∗ < tp . Case 1. Let t ∈ [t∗ , tp ]. If sp−1 < t∗ < tp then for t ∈ [t∗ , tp ] the function x∗ (t) ∈ C q ([t∗ , tp ], Rn ) and from Lemma Rt 1 q−1 c(||x∗ (s)||)ds ≤ V (t∗ , x∗ (t∗ )). 3 we get V (t, x∗ (t)) ≤ V (t∗ , x∗ (t∗ )) − Γ(q) t∗ (t − s) ∗ ∗ If tp−1 < t ≤ sp−1 then for t ∈ [t , tp ] the function x∗ (t) ∈ P C q ([t∗ , tp ], Rn ) and from Lemma 4 we get V (t, x∗ (t)) ≤ V (t∗ , x∗ (t∗ )). Case 2. For any t > t∗ , t ∈ (sk , tk+1 ], k = p, p + 1, . . . , from Lemma 4 for ∆ = S(λ) we obtain V (t, x∗ (t)) Z k−1 Z ti+1 X 1  tp q−1 ∗ ≤ V (t , x (t )) − (t − s) c(||x (s)||)ds + (ti+1 − s)q−1 c(||x∗ (s)||)ds Γ(q) t∗ s i i=p Z t  + (t − s)q−1 c(||x∗ (s)||)ds ≤ V (t∗ , x∗ (t∗ )). ∗

∗

∗

sk

Case 3. For any t > t∗ , t ∈ (tk , sk ], k = p + 1, p + 2, . . . , from Lemma 4 for ∆ = Rn we obtain V (t, x∗ (t)) 1  ≤ V (t , x (t )) − Γ(q) ∗

∗

∗

Z

tp

q−1

(t − s)

∗

c(||x (s)||)ds +

t∗

k−1 Z X i=p

ti+1

q−1

(ti+1 − s)

 c(||x (s)||)ds ∗

si

≤ V (t∗ , x∗ (t∗ )). Therefore, for t ≥ t∗ the following inequality is satisfied: V (t, x∗ (t)) ≤ V (t∗ , x∗ (t∗ )).

(36)

Then for any t ≥ t∗ applying (36), condition 3(iii) and inequality (33) we get the inequalities b(||x∗ (t)||) ≤ V (t, x∗ (t)) ≤ V (t∗ , x∗ (t∗ )) ≤ a(||x∗ (t∗ )||) ≤ a(γ) < b(). Therefore, the inequality (34) holds for all t ≥ t∗ (hence for t ≥ t0 + T ). 

Remark 14. If the initial time t0 is in an interval of noninstantaneous impulses, i.e. t0 ∈ ∪∞ k=1 (tk , sk ] then the results of Theorem 1, Theorem 2 and Theorem 3 will be similar with slight changes in Definition 2 and condition 2(ii) (Theorems 1,2) or condition 3(ii)(Theorem 3).

6.2.

Stability by generalized Dini fractional derivative

The proofs below are similar to those in Theorem 1 and Theorem 2 where Lemma 6 is used instead of Lemma 2. Theorem 4. (Stability). Let the conditions of Theorem 1 be satisfied where the inequality (27) is replaced by q c (4) D+ V

(t, x) ≤ 0

for t ∈ (sk , tk+1 ], k = 0, 1, 2, . . . , x ∈ Rn .

(37)

Then the zero solution of the NIFrDE (4) is stable. Theorem 5. (Uniform stability). Let the conditions of Theorem 2 be satisfied where the inequality (28) is replaced by q c (4) D+ V

(t, x) ≤ 0

for t ∈ (sk , tk+1 ], k = 0, 1, 2, . . . , x ∈ S(λ).

(38)

Then the zero solution of the IFrDE (4) is uniformly stable. We will give the comparison result with the quadratic Lyapunov function V (x) = xT x = n 2 C k=1 xk ∈ Λ (R ), x = (x1 , x2 , . . . , xn ).

Pn

Corollary 3. (Stability by a quadratic function). Let condition (H) be satisfied and (i) xT f (t, x) ≤ 0 for t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . , x ∈ S(λ);
T (ii) for any k = 1, 2, 3, . . . and x ∈ S(λ) for t ∈ [sk , tk ] the inequality φk (t, x) φk (t, x) ≤ xT x for t ∈ (tk , sk ] holds;

(iii) b(||x||) ≤ xT x ≤ a(||x||) for t ∈ R+ , x ∈ Rn , where a, b ∈ K. Then the zero solution of (4) is uniformly stable. Remark 15. If tk = sk and y = φk (tk , y), y ∈ Rn for all k = 1, 2 . . . system (4) reduces to a system of fractional differential equations. For the reduced (4) in [50] the author defines a generalized fractional-order derivative ( Dini-like derivative) in the Caputo sense based on the fractional-order Dini derivative in the Caputo sense ([31], [32]) and some sufficient conditions for the stability with initial time difference are obtained.

7. 7.1.

APPLICATIONS Quadratic Lyapunov function

We will apply the quadratic Lyaunov function and its fractional derivative. Example 6. Consider the scalar NIFrDE (7) where A ≤ 0 and φk (t, x) = ak (t)x, ak : [tk , sk ] → R, k = 1, 2, 3, . . . are such that supt∈[tk ,sk ] |ak (t)| ≤ 1. Consider the quadratic Lyapunov function V (x) = x2 . Then xf (t, x) = Ax2 ≤ 0 for t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . , x ∈  2  2 S(λ), i.e. condition (i) of Corollary 3 is satisfied. Also, φk (t, x) = ak (t) x2 ≤ x2 for t ∈ [tk , sk ], i.e. condition (ii) is satisfied. From Corollary 3 the zero solution of the scalar NIFrDE (7) is uniformly stable (this was proved directly in Example 2). 

Example 7. Let the points tk , sk , k = 0, 1, 2, . . . , be such that 0 ≤ tk < sk < tk+1 , limk→∞ tk = ∞. Consider the scalar noninstantaneous impulsive Caputo fractional differential equation c q sk D x

= −a(t)x(1 + x2 ) for t ∈ (sk , tk+1 ], k = 0, 1, 2, . . . ,

x(t) = ck (t)x(tk − 0) for t ∈ (tk , sk ],

k = 1, 2, 3, . . . ,

(39)

x(0) = x0 where x ∈ R, a(t) ∈ C(∪∞ k=0 (sk , tk+1 ], R+ ), ck (t) ∈ C([tk , sk ], [−1, 1]), k = 1, 2, . . . . Consider the function V (t, x) = x2 . Then xf (t, x) = −a(t)x2 (1 + x2 ) ≤ 0, i.e. condition  2 (i) of Corollary 3 is satisfied. Also, ck (t)x = (ck (t)x)2 ≤ x2 for t ∈ (sk , tk ], k = 1, 2, 3, . . . , i.e. condition (ii) is satisfied. From Corollary 3, the trivial solution of NIFrDE (39) is uniformly stable. 

7.2.

General Lyapunov function

Now we will apply a general Lyapunov function and its generalized Caputo fractional Dini derivative. Example 8. Let points sk = (4k + 1) π2 , tk = (4k − 1) π2 , k = 1, 2, . . . , s0 = 0. Consider the following initial value problem for the scalar noninstantaneous impulsive Caputo fractional differential equation c 0.5 sk D x(t)

= xf (t),

x(t) = ck (t)x(tk − 0), x(0) = x0 ,

t ∈ (sk , tk+1 ], k = 0, 1, 2, . . . , t ∈ [tk , sk ], k = 1, 2, . . . ,

(40)

−2 √ + tπ

√

tE2,1.5 (−t2 )

where x ∈ R, ck ∈ C([tk , sk ], [−1, 1]), f (t) = 0.5 , k = 0, 1, 2, . . . . 2−sin(t)
2 2 Let V (t, x) = x . Then x xf (t) = x f (t). Since the sign of the function f (t) changes (see Figure 2) Corollary 3 and the quadratic Lyapunov function are not applicable to the fractional equation (40). Let V (t, x) = (2 − sin(t))x2 . Apply generalized Dini fractional derivative given by (13), Example 4 and formula (18) and we get 0.5 2 c (40) D+ V (t, x) = 2x (2 − sin(t))f (t). 0.5 V (t, x) are changeable. Therefore, The sign of the function f (t) and the derivative c(40) D+ the application of fractional Dini derivative (13) does not give us a conclusion about stability properties of NIFrDE (40). Now apply generalized Caputo fractional Dini derivative to the considered Lyapunov √ function. According to Example 4 for t ∈ (sk , tk+1 ), k = 0, 1, 2, . . . and Γ(0.5) = π we obtain

2(x0 )2 − sin(t)) − √ tπ (41)  2  2 √ 2(x0 ) 2 2 2 = 2x (2 − sin(t))f (t) + x √ − tE2,1.5 (−t ) − √ ≤ 0. tπ tπ    Also, for t ∈ [tk , sk ], k = 1, 2, . . . we get V (t, ck (t)x) = 2 − sin(t) (ck (t)x)2 ≤ 2 −      sin(t) x2 ≤ 2 − sin(tk ) x2 = 2 − sin((4k − 1) π2 ) x2 = V (tk − 0, x), i.e. condition 2(ii) of Theorem 1 is satisfied. According to Theorem 1 the zero solution of (40) is stable. c 0.5 (40) D+ V

(t, x; 0, x0 ) = 2x2 (2 − sin(t))f (t) + x2

RL 0.5 0 D (2



8.

CONCLUSSIONS

Piecewise continuous scalar Lyapunov functions are applied to study stability, uniform stability and asymptotic uniform stability of the zero solution of nonlinear Caputo fractional differential equations with not instantaneous impulses. Two types of derivatives of Lyapunov function among NIFrDE are introduced and their applications are discussed. Several sufficient conditions for stability, uniform stability and uniform asymptotic stability of the zero solution of nonlinear NIFrDE are obtained. The results are illustrated with several examples. Note the above description and all consideration in the paper could be generalized when additionally an instantaneous jump at the points sk are given and the right side part of the fractional equation is changing (so called variable structure), i.e. problem (4) can be generalized to c q sk D x

= fk (t, x) for t ∈ (sk , tk+1 ], k = 0, 1, 2, . . .

x(t) = φi (t, x(ti − 0))

for t ∈ (ti , si ], i = 1, 2, . . . ,

x(si + 0) = Gk (x(si − 0))

for i = 1, 2, . . . ,

x(t0 ) = x0 . Acknowledgments. Research was partially supported by the Fund NPD, Plovdiv University, No. MU15-FMIIT-008.

References [1] S. Abbas, M. Benchohra, J. Trujillo, Upper and lower solution method for partial fractional differential inclusions with not instantaneous impulses, Progr. Fract. Differ. Appl., 11, 1, (2015), 11–22. [2] R. Agarwal, M. Benchohra, B. A. Slimani, Existence results for differential equations with fractional order and impulses, Mem. Differ. Equ. Math. Phys., 44, (2008), 1–21. [3] R. Agarwal, S. Hristova, Strict stability in terms of two measures for impulsive differential equations with supremum, Appl. Anal. , 91, 7, (2012), 1379–1392. [4] R. Agarwal, D. O’Regan, S. Hristova, Stability of Caputo fractional differential equations by Lyapunov functions, Appl. Math. (accepted). [5] N. Aguila-Camacho, M. A. Duarte-Mermoud, J. A. Gallegos, Lyapunov functions for fractional order systems, Comm. Nonlinear Sci. Numer. Simul., 19, (2014), 2951–2957. [6] D. Bainov, S. Hristova, The method of quasilinearization for the periodic boundary value problem for systems of impulsive differential equations, Appl. Math. Comput., 117, 1, (2001), 73–85. [7] K. Balachandran, S. Kiruthika, Existence of solutions of abstract fractional impulsive semilinear evolution equations, Electron. J. Qual. Theory Differ. Equ., 4 , (2010), 1–12. [8] D. Baleanu, O.G. Mustafa, On the global existence of solutions to a class of fractional differential equations, Comput. Math. Appl., 59, (2010), 1835–1841. [9] M. Benchohra, D. Seba, Impulsive fractional differential equations in Banach spaces, Electron. J. Qual. Theory Differ. Equ., (Special Edition I), 8, (2009), 1–14. [10] M. Benchohra, B. A. Slimani, Existence and uniqueness of solutions to impulsive fractional differential equations,Electronic J. Differential Equations, 2009, 10, (2009), 1–11. [11] Sh. Das, Functional Fractional Calculus, Springer-Verlag Berlin Heidelberg, 2011. [12] J. V. Devi , F.A. Mc Rae, Z. Drici, Variational Lyapunov method for fractional differential equations, Comput. Math. Appl., 64, (2012), 2982–2989. [13] K. Diethelm, The Analysis of Fractional Differential Equations, Springer-Verlag Berlin Heidelberg, 2010. [14] M. Feckan, J.R. Wang, Y. Zhou, Periodic solutions for nonlinear evolution equations with non-istantaneous impulses, Nonauton. Dyn. Syst., 1, (2014), 93–101. [15] G. R. Gautam, J. Dabas, Mild solution for fractional functional integro-differential equation with not instantaneous impulse, Malaya J. Mat., 2, 3, (2014), 428-437. [16] G. R. Guantam, J. Dabas, Mild solutions for a class of neutral fractional functional differential equations with not instantaneous impulses, Apl. Math. Comput., 259, (2015), 480–489. [17] G. R. Guantam, J. Dabas, Existence results on fractional functional differential equations with not instantaneous impulses, Int. J. Adv. Appl. Math. Mech., 1, 3, (2014), 11–21.

[18] E. Hernandez, D. O’Regan, On a new class of abstract impulsive differential equations, Proc. Amer. Math. Soc., 141, (2013), 1641–1649. [19] S. Hristova, Qualitative investigations and approximate methods for impulsive equations, Nova Sci. Publ. Inc., New York, 2009. [20] S. Hristova, Integral stability in terms of two measures for impulsive functional differential equations, Math. Comput, Modell,, 51, 12, (2010), 100-108. [21] S. Hristova, Stability on a cone in terms of two measures for impulsive differential equations with supremum, Appl. Math. Lett., 23, 5, (2010), 508–511. [22] S. Hristova, Razumikhin method and cone valued Lyapunov functions for impulsive differential equations with supremum, Inter. J. Dynam. Syst. Diff. Eq., 2, 3-4, (2009), 223–236. [23] S. Hristova, Lipschitz stability for impulsive differential equations with supremum, International Electronic Journal of Pure and Applied Mathematics, 1, 4, (2010), 345–358. [24] S. Hristova, D. Bainov, Periodic solutions of quasilinear nonautonomous systems with impulses, Bull. Austral. Math. Soc., 31, (1985), 185–198. [25] S. Hristova, D. Bainov, Application of Liapunov’s functions for studying the boundedness of solutions of systems with impulses, COMPEL, 5, 1, (1986), 23–40. [26] S. Hristova, K. Stefanova, Practical stability of impulsive differential equations with supremum by integral inequalities, Eur. J. Pure Appl. Math., 5, 1, (2012), 30–44. [27] J.B. Hu, G.P. Lu, S.B. Zhang, L.-D. Zhao, Lyapunov stability theorem about fractional system without and with delay, Commun. Nonlinear Sci. Numer. Simul., 20, (2015), 905– 913. [28] P. Eloe, S. Hristova, Method of the quasilinearization for nonlinear impulsive differential equations with linear boundary conditions, EJQTDE, 10, (2002), 1-14. [29] P. Kumar, D. N. Pandey, D. Bahuguna, On a new class of abstract impulsive functional differential equations of fractional order, J. Nonlinear Sci. Appl., 7, (2014), 102–114. [30] V. Lakshmikantham, D.D. Bainov, P.S. Simeonov, Theory of Impulsive Differential Equations, World Scientific, Singapore, 1989. [31] V. Lakshmikantham, S. Leela, J.V. Devi, Theory of Fractional Dynamical Systems, Cambridge Scientific Publishers, 2009. [32] V. Lakshmikantham, S. Leela, M. Sambandham, Lyapunov theory for fractional differential equations, Commun. Appl. Anal., 12, 4, (2008), 365–376. [33] Y. Li, Y. Chen, I. Podlubny, Stability of fractional-order nonlinear dynamic systems: Lyapunov direct method and generalized Mittag-Leffler stability, Comput. Math. Appl., 59, (2010), 1810–1821. [34] P. Li, Ch. Xu, Boundary value problems of fractional order differential equation with integral boundary conditions and not instantaneous impulses, J. Funct. Spaces, 2015, 2015, Article ID 954925.

[35] C. Li, D. Qian, Y. Chen, On Riemann-Liouville and Caputo Derivatives, Discr. Dynam. Nat, Soc., 2011, Article ID 562494. [36] C.P. Li, F.R. Zhang, A survey on the stability of fractional differential equations, Eur. Phys. J., Special Topics, 193, (2011), 27–47. [37] Y. M. Liao, J. R. Wang, A note on stability of impulsive differential equations, Boundary Value Probl., 2014, 2014:67. [38] Z.Lin, J.R. Wang, W. Wei, Multipoint BVPs for generalized impulsive fractional differential equations, Appl. Math. Comput., 258, (2015), 608-616. [39] Z. Lin, W. Wei, J. R. Wang, Existence and stability results for impulsive integro-differential equatons, Facta Univer. (Nis), ser. Math. Inform., 29, No 2 (2014), 119130. [40] D.N. Pandey, S. Das, N. Sukavanam, Existence of solutions for a second order neutral differential equation with state dependent delay and not instantaneous impulses, Intern. J. Nonlinear Sci., 18, 2, (2014), 145–155. [41] M. Pierri, D. O’Regan, V. Rolnik, Existence of solutions for semi-linear abstract differential equations with not instantaneous impulses, Appl. Math. Comput., 219, (2013), 6743–6749. [42] I. Podlubny, Fractional Differential Equations, Academic Press, San Diego, 1999. [43] G. Samko, A.A. Kilbas, O.I. Marichev, Fractional Integrals and Derivatives: Theory and Applications, Gordon and Breach, 1993. [44] A. Sood, S. K. Srivastava, On Stability of Differential Systems with Noninstantaneous Impulses, Math. Probl. Eng., 2015, 2015, Article ID 691687. [45] J.C. Trigeassou, N. Maamri, J. Sabatier, A. Oustaloup, A Lyapunov approach to the stability of fractional differential equations, Signal Processing, 91, (2011), 437–445. [46] G. Wang, B. Ahmad, L. Zhang, J. Nieto, Comments on the concept of existence of solution for impulsive fractional differential equations, Commun. Nonlinear Sci. Numer. Simulat., 19, (2014), 401–403. [47] J. R. Wang, X. Li, Periodic BVP for integer/fractional order nonlinear differential equations with non-instantaneous impulses, J. Appl. Math. Comput., 46, 1-2, (2014), 321–334. [48] J. Wang, Z. Lin, A class of impulsive nonautonomous differential equations and Ulam − Hyers−Rassias stability, Math. Meth. Appl. Sci, 38, 5, (2015), 868–880. [49] Z. Wang, Y. Zang Existence and stability of solutions to nonlinear impulsive differential equations in β-normed space, Elect. J. Diff. Eq. 2014, 83, (2014), 1–10. [50] C. Yakar, Fractional differential equations in terms of comparison results and Lyapunov stability with initial time difference, Abstr. Appl. Anal., 2010, 2010, Article ID 762857, Hindawi Publishing Corporation. [51] Z. Yan, F. Lu, Existence results for a new class of fractional impulsive partial neutral stochastic integro-differential equations with ifinite delays, J. Appl. Anal. Comput., 5, 4, (2015), 329–346.

[52] X. Yu, Existence and β-Ulam-Hyers stability for a class of fractional differential equations with non instantaneous impulses, Adv. Diff. Eq., 2015, (2015):104.