() CUBO A Mathematical Journal Vol.17, No¯ 01, (11–27). March 2015 Periodic BVP for a class of nonlinear differential equation with a deviated argument and integrable impulses Alka Chadha and Dwijendra N Pandey Department of Mathematics, Indian Institute of Technology Roorkee, Roorkee-247667 alkachaddha03@gmail.com, dwij.iitk@gmail.com ABSTRACT This paper deals with periodic BVP for integer/fractional order differential equations with a deviated argument and integrable impulses in arbitrary Banach space X for which the impulses are not instantaneous. By utilizing fixed point theorems, we firstly establish the existence and uniqueness of the mild solution for the integer order dif- ferential system and secondly obtain the existence results for the mild solution to the fractional order differential system. Also at the end, we present some examples to show the effectiveness of the discussed abstract theory. RESUMEN Este art́ıculo estudia las ecuaciones diferenciales de orden entero/fraccional con condi- ciones de frontera periódicas con un argumento desviado e impulsos integrables en espacios de Banach arbitrarios X donde los pulsos no son instantáneos. Utilizando teo- remas de punto fijo, establecemos la existencia y unicidad de soluciones temperadas para los sistemas diferenciales de orden entero, y luego obtenemos resultados de exis- tencia para soluciones temperadas del sistema diferencial de orden fraccional. Además, presentamos un ejemplo para mostrar la efectividad de la teoŕıa abstracta discutida. Keywords and Phrases: Deviating arguments, Fixed point theorem, Impulsive differential equa- tion, Periodic BVP, Fractional calculus. 2010 AMS Mathematics Subject Classification: 34G20, 34K37, 34K45, 35R12, 45J05. 12 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) 1 Introduction: In a few decades, impulsive differential equations have received much attention of researchers mainly due to its demonstrated applications in widespread fields of science and engineering. Im- pulsive differential equations have played an important role in real world problems for describing a process which is characterized by the development of a sudden change in system’s state. Such processes are investigated in various fields such as biology, physics, control theory, population dy- namics, medicine and many others. Impulsive differential equations are an appropriate model to hereditary phenomena for which a delay argument arises in modelling equations. For more details on impulsive differential equation, we refer to the monographs [1],[2] and papers [3]-[12] and refer- ences given therein. A differential equation with boundary conditions arise in many areas of applied sciences, for example, chemical engineering, blood flow problems, thermoelasticity, population models, under- ground water flow and many others. For more details on differential equation with integral bound- ary conditions, we refer to [10, 17, 18, 19, 23, 27] and references given therein. On the other hand, fractional calculus have many applications in various areas of sciences and engineering for exam- ple, fluid dynamics, like fractal theory, diffusion in porous media and fractional biological neurons, traffic flow, polymer rheology. The fractional differential equation is an important tool to describe nonlinear oscillation of earthquake. For more study on fractional calclus, we refer to books [13]-[16]. In this work, we consider the periodic boundary value problems for integer order nonlinear differential equations in a Banach space X of the form with non-instantaneous integrable impulses u′(t) = f(t, u(t), u([h(u(t), t)])), t ∈ (sm, tm+1], m = 0, 1, 2, · · · , δ, (1.1) u(t) = ∫t tm Gm(s, u(s))ds, t ∈ (tm, sm], m = 1, 2, · · · , δ, δ ∈ N (1.2) u(0) = u(T). (1.3) Next, we consider the periodic boundary value problems for nonlinear fractional differential equa- tions in a Banach space X of the form with non-instantaneous integrable impulses cD q 0,tu(t) = I 2−q t f(t, u(t), u([h(u(t), t)])), t ∈ (sm, tm+1], 0 < q < 1, (1.4) u(t) = ∫t tm Gm(s, u(s))ds, t ∈ (tm, sm], m = 0, 1, 2, · · · , δ, δ ∈ N, (1.5) u(0) = u(T), (1.6) where 0 < T < ∞, cD q 0,t represents the Caputo fractional derivative of the order q with lower limit 0, 0 = t0 = s0 < t1 ≤ s1 ≤ t2 < · · · < tδ ≤ sδ ≤ tδ+1 = T are fixed numbers, Gm : (tm, sm] × X → X, m = 1, · · · , δ. The nonlinear X-valued functions f and h are appro- priate functions and satisfy some suitable conditions to be stated later. In this system of equations CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 13 (1.1)-(1.3) and (1.4)-(1.6), the impulses begin all of a sudden at the points ti and continue their proceeding on a finite interval [ti, si]. According to the authors in [4]-[5], there are many differ- ent inspirations for consideration of the problem (1.1)-(1.3) and (1.4)-(1.6). The hemodynamical equilibrium of a person is an example of such systems. One can prescribe some intravenous drugs (insulin) in the case of a decompensation (for example, low or high level of glucose). Since the introduction of the drugs in the bloodstream and the consequent absorption of the body are suc- cessive and continuous processes, we can describe this situation as an impulsive action which start suddenly and stays active on a finite time interval. The organization of the paper is as follows: In section 2, we give some basic definitions, assumptions, lemmas and theorems as preliminaries which will be used for proving our main results. In section 3, we prove the existence of a mild solution to the problem (1.1)-(1.3) and problem (1.4)-(1.6). Some examples are also presented at the end of the paper. 2 Preliminaries and Assumptions In this section, we discuss some basic definitions, preliminaries, theorem and lemmas which will be used for proving the required result. Let (X, ‖ · ‖) be a Banach space. Let C(J; X), where J = [0, T] denotes the space of all continuous X-valued functions on interval J which is a Banach space with the norm ‖ u‖C = supt∈J ‖ u(t)‖. The space of all Bochner integrable functions u : (0, T) → X represented by L1((0, T); X), is a Banach space with norm ‖ u‖1 = ∫T 0 ‖ u(t)‖dt. The Br(x, X) denotes the closed ball with center at x and radius r in X. To study the impulsive differential equation, we introduce the following space PC([0, T]; X) = {u : [0, T] → X; u ∈ C((tj, tj+1]; X), j = 0, 1, · · · , m, and ∃ u(t+j ) and u(t − j ), j = 1, · · · , m exist with u(t − j ) = u(tj)}. It is clear that PC([0, T]; X) is a Banach space with the norm ‖u‖PC = max t∈[0,T] ‖ u(t)‖. For a function u ∈ PC([0, T]; X) and j ∈ {0, 1, · · · , m}, we define the function ũj ∈ C([tj, tj+1]; X) such that ũj(t) = { u(t), for t ∈ (tj, tj+1], u(t+j ), for t = tj. (2.1) For B ⊂ PC([0, T]; X) and j ∈ {0, 1, · · · , m}, we have B̃j = {ũj : u ∈ B} and we have following Accoli-Arzelà type criteria. 14 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) Lemma 2.1. [4]. A set B ⊂ PC([0, T]; X) is relatively compact in PC([0, T]; X) if and only if each set B̃j(j = 1, 2, · · · , m) is relatively compact in C([tj, tj+1], X)(j = 0, 1, · · · , m). Now, we recall some basic definition. Definition 2.1. The Riemann-Liouville fractional integral of f with order q defined by I q 0,tf(t) = 1 Γ(q) ∫t 0 (t − s)q−1f(s)ds. (2.2) Definition 2.2. The fractional derivative of function f : [0, ∞) → R in the Riemann-Liouville sense with order q is defined by D q 0,tf(t) = dn dtn 1 Γ(n − q) ∫t 0 (t − s)n−q−1f(s)ds, t > 0, n − 1 < q < n. (2.3) Definition 2.3. The fractional derivative of function f : [0, ∞) → R in the Caputo sense of order q is defined by cD q 0,tf(t) = 1 Γ(n − q) ∫t 0 (t − s)n−q−1fn(s)ds, (2.4) for n − 1 < q < n, n ∈ N, t > 0, with the following property: cD q 0,t(I q 0,tf(t)) = f(t) − n−1∑ k=1 tk k! fk(0). (2.5) Before expressing and demonstrating the required main result, we present the following defi- nition of mild solution to the system (1.1)-(1.3) and (1.4)-(1.6). Lemma 2.2. For given continuous function f : [0, T] → X and Gm ∈ C([tm, sm], X), a function u ∈ PC([0, T]; X) is a mild solution for the impulsive periodic boundary value problem u′(t) = f(t), t ∈ (sm, tm+1], m = 0, 1, 2, · · · , δ, δ ∈ N (2.6) u(t) = ∫t tm Gm(s)ds, t ∈ (tm, sm], m = 1, 2, · · · , δ, (2.7) u(0) = u(T), (2.8) if and only if u(·) satisfies the following u(t) =    ∫sδ tδ Gδ(s)ds + ∫T sδ f(s)ds + ∫t 0 f(s)ds, t ∈ [0, t1], ∫sm tm Gm(s)ds + ∫t sm f(s)ds, t ∈ (sm, tm+1], ∫t tm Gm(s)ds, t ∈ (tm, sm], (2.9) for each m = 1, · · · , δ. CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 15 Lemma 2.3. For the continuous function f : [0, T] → X and Gm ∈ C([tm, sm], X), a function u ∈ PC([0, T]; X) is said to be a mild solution for the system cD q 0,tu(t) = I 2−q t f(t), t ∈ (sm, tm+1], 0 < q < 1, (2.10) u(t) = ∫t tm Gm(s)ds, t ∈ (tm, sm], m = 0, 1, 2, · · · , δ, δ ∈ N, (2.11) u(0) = u(T), (2.12) if and only if u(0) = u(T), u(t) = ∫t tm Gm(t), ∀ t ∈ (tm, sm], m = 1, · · · , δ and u(·) satisfies the following integral equations u(t) =    ∫sδ tδ Gδ(s)ds − ∫sδ 0 (sδ − s)f(s)ds + ∫T 0 (T − s)f(s)ds + ∫t 0 (t − s)f(s)ds, t ∈ [0, t1], ∫sm tm Gm(s)ds − ∫sm 0 (sm − s)f(s)ds + ∫t 0 (t − s)f(s)ds, t ∈ (sm, tm+1], (2.13) for each m = 1, · · · , δ. Further, we list the following assumption which will be used to establish the main result. Assumptions on f, h and Gm, (m = 1, · · · , δ) : (A1) The function f : [0, T] × X × X → X is continuous and there exist a positive constant Lf and 0 < γ1 ≤ 1 such that ‖f(t1, u1, v1) − f(t2, u2, v2)‖ ≤ Lf[|t1 − t2| γ1 + ‖u1 − u2‖X + ‖v1 − v2‖X], (2.14) for all (tj, uj, vj) ∈ [0, T] × X × X, j = 1, 2. (A2) h : X × [0, T] → [0, T] is continuous function and there exist positive constants Lh and 0 < γ2 ≤ 1 such that |h(u1, t1) − h(u2, t2)| ≤ Lh[‖u1 − u2‖X + |t1 − t2| γ2], (2.15) for each (uj, tj) ∈ X × [0, T], for j = 1, 2. (A3) Gm : [0, T] × X → X, m = 1, 2, · · · , δ, are continuous functions and there exist constants LGm > 0 such that ‖ Gm(t, x) − Gm(t, y)‖ ≤ LGm‖ x − y‖, (2.16) ‖Gm(t, u(t))‖ ≤ Km, (2.17) for all (t, x), (s, y) ∈ [0, T] × X, u ∈ X and Km > 0, m = 1, · · · , δ are constants. 16 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) 3 Existence Result In this section, we establish the existence of a mild solutions for the systems (1.1)-(1.3) and (1.4)- (1.6) by using fixed point theorems. Let Y0 = PC(J; X) = {y ∈ PC(J; X) : y ∈ C((tm, tm+1], X), m = 0, 1, · · · , δ and y(t−m) = y(tm), y(t + m) exist}. (3.1) and Y1 = {y ∈ Y0 : ‖y(t) − y(s)‖ ≤ L|t − s|, ∀ t ∈ [tm, tm+1], m = 0, 1, · · · , δ}. (3.2) Where L is an appropriate positive constant to be defined later. 3.1 Integer Order case Theorem 3.1. We assume that assumptions (A1) − (A3) are satisfied. If Θ = sup{ max m=1,··· ,δ [LGm(sm − tm) + Lf(1 + LhL)(tm+1 − sm)], LGδ(sδ − tδ) + Lf(1 + LhL)(T − sδ + t1)} < 1. (3.3) Then, the system (1.1)-(1.3) has a unique mild solution on the interval J. Proof. In order to transform the system (1.1)-(1.3) into a fixed point problem, we consider the map Q : S → S defined by Qu(t) =    ∫t tm Gm(s, u(s))ds, t ∈ (tm, sm], m = 1, · · · , δ, ∫sδ tδ Gδ(s, u(s))ds + ∫T sδ f(s, u(s), u([h(u(s), s)]))ds + ∫t 0 f(s, u(s), u([h(u(s), s)]))ds, t ∈ [0, t1], ∫sδ tδ Gm(s, u(s))ds + ∫t sm f(s, u(s), u([h(u(s), s)]))ds, t ∈ (sm, tm+1], (3.4) where S = {u ∈ Y0 ∩Y1 : ‖u‖PC ≤ R}. Clearly, S is a closed and bounded subset of Y1 and complete metric space. It is not difficult to show that Qu ∈ Y0. Now, it remains to show that Qu ∈ Y1. For u ∈ S and τ2, τ1 ∈ [0, t1] with τ1 < τ2, ‖Qu(τ2) − Qu(τ1)‖ ≤ ∫τ2 τ1 ‖f(s, u(s), u([h(u(s), s)]))‖ds, ≤ H|τ2 − τ1|. (3.5) where H = supt∈[0,T] ‖f(t, u(t), u([h(u(t), t)]))‖. Similarly, τ2, τ1 ∈ (tm, sm], m = 1, · · · , δ ‖Qu(τ2) − Qu(τ1)‖ ≤ ‖ ∫τ2 tm Gm(s, u(s))ds − ∫τ1 tm Gm(s, u(s))ds‖ ≤ Km|τ2 − τ1|, (3.6) CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 17 and for τ2, τ1 ∈ (sm, tm+1], m = 1, · · · , δ ‖Qu(τ2) − Qu(τ1)‖ ≤ H|τ2 − τ1|. (3.7) Therefore, we conclude that Qu ∈ Y1 with suitable constant L = min{ max m=1,··· ,δ Km, H}. Now, we show that Q(S) ⊆ S. For t ∈ [0, t1] and u ∈ S, we get ‖Qu(t)‖ ≤ ‖ ∫sδ tδ Gδ(s, u(s))ds‖ + ∫T sδ ‖f(s, u(s), u([h(u(s), s)]))‖ds + ∫t 0 ‖f(s, u(s), u([h(u(s), s)]))‖ds, ≤ Kδ(sδ − tδ) + H(T − sδ + t1) ≤ KδT + HT. (3.8) For t ∈ [sm, tm+1], m = 1, · · · , δ, ‖Qu(t)‖ ≤ Km(sm − tm) + H(tm+1 − sm) ≤ KmT + HT, (3.9) and for t ∈ (sm, tm], we have that ‖Qu(t)‖ ≤ KmT. We choose R = max[KδT +HT, sup m=1,··· ,δ {KmT + HT}]. Thus, we get that Q(S) ⊆ S. In the next step, we prove that Q is a contraction map. For w1, w2 ∈ S and t ∈ [0, t1], we get ‖Qw1(t) − Qw2(t)‖ ≤ [LGδ(sδ − tδ) + Lf(1 + LhL)(T − sδ + t1)] ×‖w1 − w2‖PC. (3.10) For t ∈ [sm, tm+1], m = 1, · · · , δ ‖Qw1(t) − Qw2(t)‖ ≤ [LGm(sm − tm) + Lf(1 + LhL)(tm+1 − sm)]‖w1 − w2‖PC, ≤ max m=1,··· ,δ [LGm(sm − tm) + Lf(1 + LhL)(tm+1 − sm)] ×‖w1 − w2‖PC, (3.11) and for t ∈ (tm, sm], we obtain that ‖Qw1(t) − Qw2(t)‖ ≤ max m=1,··· ,δ LGm(sm − tm) × ‖w1 − w2‖PC. (3.12) From the inequalities (3.10)-(3.12), we get ‖Qw1 − Qw2‖PC ≤ Θ‖w1 − w2‖PC. (3.13) Thus, by the inequality (3.3), we conclude that Q is a contraction on S and there exists a unique fixed point u ∈ S of the map Q. It is obvious that u is a mild solution for the system (1.1)-(1.3). Our second existence result is based on Krasnoselskii’s theorem. The statement of the theorem is given as: 18 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) Theorem 3.2. Let F ⊂ X be a closed convex and nonempty subset of X, where X is a Banach space. Let P1 and P2 be the operator such that (a) P1w1 + P2w2 ∈ F, whenever, w1, w2 ∈ F, (b) P1 is a contraction, (c) P2 is compact and continuous. Then, the map P = P1 + P2 has a fixed point x ∈ F i.e., x = P1x + P2x. Theorem 3.3. Assume that (A1) − (A3) are satisfied. Then, there exists a mild solution for the system (1.1)-(1.3) on J provided that Ξ = max{KGm(sm − tm); m = 1, · · · , δ} < 1. (3.14) Proof. We define the following operators Q1 : S → S which is decomposition of operator Q, by Q1u(t) =    ∫sδ tδ Gδ(s, u(s))ds, t ∈ [0, t1], ∫t sm Gm(s, u(s))ds, t ∈ (tm, sm], m = 1, · · · , δ, ∫sm tm Gm(sm, u(sm)), t ∈ (sm, tm+1] m = 1, · · · , δ. (3.15) and Q2 : S → S by Q2u(t) =    ∫T sδ f(s, u(s), u(h(u(s), s)))ds + ∫t 0 f(s, u(s), u(h(u(s), s)))ds, t ∈ [0, t1], 0, t ∈ (tm, sm], m = 1, · · · , δ, ∫t sm f(s, u(s), u(h(u(s), s)))ds, t ∈ (sm, tm+1] i = 1, · · · , δ. (3.16) We choose r such that max{ max m=1,··· ,δ (Km + H)T, (Kδ + H)T} < r. (3.17) Consider Br = {u ∈ Y0 ∩ Y1 : ‖u‖PC ≤ r}. (3.18) It is clear that the mappings Q1 and Q2 are well-defined. Now, we show the result in several steps. Step 1. For u, v ∈ Br and t ∈ [0, t1], we have ‖(Q1u + Q2v)(t)‖ ≤ ‖ ∫sδ tδ Gδ(s, u(s))ds‖ + ∫T sδ ‖f(s, v(s), v(h(v(s), s)))‖ds + ∫t 0 ‖f(s, v(s), v(h(v(s), s)))‖ds, ≤ Kδ(sδ − tδ) + H[T − sδ − t1] ≤ KδT + HT. (3.19) For t ∈ (sm, tm+1], m = 1, · · · , δ, ‖(Q1u + Q2v)(t)‖ ≤ ‖ ∫sm tm Gm(s, u(s))ds‖ + ∫t sm ‖f(s, u(s), u(h(u(s), s)))‖ds, ≤ Km(sm − tm) + H(tm+1 − sm) ≤ (Km + H)T, (3.20) CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 19 and for t ∈ [tm, sm], we have ‖(Q1u + Q2v)(t)‖ ≤ KmT. Thus, by the choice of r, we get that ‖(Q1u + Q2v)‖PC ≤ r, for all t ∈ [0, T]. (3.21) Hence, Q1u + Q2v ∈ Br. Step 2. We show that Q1 is contraction map. For w1, w2 ∈ Br and t ∈ [0, t1], ‖Q1w1(t) − Q2w2(t)‖ ≤ KGδ‖w1(sδ) − w2(sδ)‖ × |sδ − tδ| ≤ KGδ(sδ − tδ)‖w1 − w2‖PC. (3.22) For t ∈ (tm, sm], m = 1, · · · , δ, we get ‖Q1w1(t) − Q2w2(t)‖ ≤ KGm(sm − tm)‖w1 − w2‖PC, (3.23) and t ∈ (sm, tm+1], m = 1, · · · , δ ‖Q1w1(t) − Q2w2(t)‖ ≤ KGm(sm − tm)‖w1 − w2‖PC. (3.24) From the above inequalities, we conclude that ‖Q1w1 − Q2w2‖PC ≤ Ξ‖w1 − w2‖PC, (3.25) which gives that Q1 is a contraction. Step 3. Q2 is continuous map. Let {zp} ∞ p=1 be a sequence such that zp → z ∈ Br. For t ∈ [0, t1], ‖Q2zp(t) − Q2z(t)‖ ≤ ∫T sδ ‖f(s, zp(s), zp(h(zp(s), s))) − f(s, z(s), z(h(z(s), s)))‖ds + ∫t 0 ‖f(s, zp(s), zp(h(zp(s), s))) − f(s, z(s), z(h(z(s), s)))‖ds, by the continuity of f and h, we have that s ∈ [0, t] f(s, zp(s), zp(h(zp(s), s))) → f(s, z(s), z(h(z(s), s))), as p → ∞, (3.26) h(zp(s), s) → h(z(s), s), as p → ∞, (3.27) From the dominated convergence theorem, we get ‖Q2zp − Q2z‖PC → 0, as p → ∞, For t ∈ (tm, sm], m = 1, · · · , δ, ‖Q2zp(t) − Q2z(t)‖ = 0. Similarly, for t ∈ (sm, tm+1], m = 1, · · · , δ ‖Q2zp(t) − Q2z(t)‖ ≤ ∫t sm ‖f(s, zp(s), zp(h(zp(s), s))) − f(s, z(s), z(h(z(s), s)))‖ds, 20 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) by the continuity of f, h and the dominated convergence theorem, we deduce that ‖Q2zp − Q2z‖PC → 0, as p → ∞. Step 3. Q2 is compact. Since f is continuous map and ‖(Q2u)(t)‖ ≤ 2HT < r. This implies that Q2 is uniformly bounded on Br. Now, we show that Q2 maps bounded set into equicontinuous set of Br. For τ2 > τ1 ∈ [0, t1] and u ∈ Br, we have ‖Q2u(τ2) − Q2u(τ1)‖ ≤ LF(τ2 − τ1). (3.28) For τ2 > τ1 ∈ (tm, sm], we have ‖Q2u(τ2) − Q2u(τ1)‖ = 0. For τ2 > τ1 ∈ (sm, tm+1], m = 1, · · · , δ and u ∈ Br, we have ‖Q2u(τ2) − Q2u(τ1)‖ ≤ LF(τ2 − τ1). (3.29) Thus, we conclude that ‖Q2u(τ2) − Q2u(τ1)‖ → 0 as τ2 → τ1. Hence Q2 is equicontinuous. By the Steps (3) − (4) and Arzela-Ascoli theorem, we deduce that Q2 : Br → Br is continuous and compact i.e. completely continuous. Since Q1 is contraction and Q2 is completely continuous operator. Thus, Q = Q1 + Q2 has a fixed point by using Krasnoselskiis fixed point theorem which is just a mild solution for the system (1.1)-(1.3). The proof of the theorem is finished. 3.2 Fractional order case Now, we obtain the existence results for the problem (1.4)-(1.6) via fixed points theorems, the first existence result of the mild solution for problem (1.4)-(1.6) is obtained by using Banach fixed point theorem and second existence results is obtained by using Krasnoselskii’s fixed point theorem. Theorem 3.4. Assume that hypotheses (A1) − (A3) are fulfilled and Λ = sup{ max m=1,··· ,δ [LGm(sm − tm) + Lf(1 + LLh)(t 2 m+1 + s 2 m) 2 ], max m=1,··· .δ (sm − tm)LGm, LGδ(sδ − tδ) + Lf(1 + LLh)(T 2 + s2δ + t 2 1 2 } < 1. (3.30) Then, the problem (1.4)-(1.6) has at least one mild solution on [0, T]. Proof. We firstly define the operator Q : S → S by (Qu)(t) =    ∫sδ tδ Gδ(s, u(s))ds − ∫sδ 0 (sδ − s)f(s, u(s), u([h(u(s), s)]))ds + ∫T 0 (T − s)f(s, u(s), u([h(u(s), s)]))ds + ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds, t ∈ [0, t1], ∫t tm Gm(s, u(s))ds, t ∈ (tm, sm], m = 1, · · · , δ, ∫sm tm Gm(s, u(s))ds − ∫sm 0 (sm − s)f(s, u(s), u([h(u(s), s)]))ds + ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds, t ∈ (sm, tm+1], m = 1, · · · , δ. (3.31) CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 21 It is clear that Qu ∈ Y0. So it remains to show that Qu ∈ Y1. For u ∈ S and τ2, τ1 ∈ [0, t1] with τ1 ≤ τ2, we get ‖(Qu)(τ2) − (Qu)(τ1)‖ = ‖ ∫τ2 0 (τ2 − s)f(s, u(s), u([h(u(s), s)]))ds − ∫τ1 (τ1 − s)f(s, u(s), u([h(u(s), s)]))ds‖, ≤ ‖ ∫τ1 0 [(τ2 − s) − (τ1 − s)]f(s, u(s), u([h(u(s), s)]))ds‖ +‖ ∫τ2 τ1 (τ2 − s)f(s, u(s), u([h(u(s), s)]))ds‖, ≤ H(τ2 − τ1) 2 + H (τ2 − τ1) 2 2 , ≤ 2HT |τ2 − τ1|, (3.32) Similarly, for τ2, τ1 ∈ (sm, tm+1], m = 1, · · · , δ, ‖(Qu)(τ2) − (Qu)(τ1)‖ ≤ HT |τ2 − τ1| (3.33) and for τ2, τ1 ∈ (tm, sm], ‖(Qu)(τ2) − (Qu)(τ1)‖ ≤ Km|τ2 − τ1|. (3.34) Thus, from (3.32)-(3.34), we conclude that Qu ∈ Y1 with L = min{ max m=1,··· ,δ Km, 2HT, HT}. Hence Q is well defined on S. Next we show that Q(S) ⊆ S. For u ∈ S and t ∈ [0, t1], we get ‖Qu(t)‖ ≤ ‖ ∫sδ tδ Gδ(s, u(s))ds‖ + ∫sδ 0 (sδ − s)‖f(s, u(s), u([h(u(s), s)]))‖ds + ∫T 0 (T − s)‖f(s, u(s), u([h(u(s), s)]))‖ds + ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds, ≤ Kδ(sδ − tδ) + H(T2 + s2δ + t 2 1) 2 ≤ KδT + 3HT2 2 . (3.35) Similarly, for t ∈ (sm, tm+1], m = 1, · · · , δ, ‖Qu(t)‖ ≤ KmT + T 2H, (3.36) and for t ∈ (tm, sm], we get ‖Qu(t)‖ ≤ KmT. (3.37) We choose R = max{KδT + 3T2H 2 , supm=1,··· ,δ KmT +T 2H} such that ‖Qu(t)‖ ≤ R, for all t ∈ [0, T]. We now show that Q is a contraction map on S. For u∗, u∗∗ ∈ S and t ∈ [0, t1], we get 22 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) ‖(Qu∗)(t) − (Qu∗∗)(t)‖ ≤ ‖ ∫sδ tδ [Gδ(s, u ∗ (s)) − Gδ(s, u ∗∗ (s))]ds‖ + ∫sδ 0 (sδ − s)‖f(s, u ∗(s), u∗([h(u∗(s), s)])) − f(s, u∗∗(s), u∗∗([h(u∗∗(s), s)]))‖ds + ∫T 0 (T − s)‖f(s, u∗(s), u∗([h(u∗(s), s)])) − f(s, u∗∗(s), u∗∗([h(u∗∗(s), s)]))‖ds + ∫t 0 (t − s)‖f(s, u∗(s), u∗([h(u∗(s), s)])) − f(s, u∗∗(s), u∗∗([h(u∗∗(s), s)]))‖ds, ≤ [LGδ(sδ − tδ) + Lf(1 + LLh)(T 2 + s2δ + t 2 1 2 ]‖u∗ − u∗∗‖PC. (3.38) Similarly, for t ∈ (sm, tm+1], m = 1, · · · , δ ‖(Qu∗)(t) − (Qu∗∗)(t)‖ ≤ ‖ ∫sm tm [Gm(s, u ∗(s)) − Gm(s, u ∗∗(s))]ds‖ + ∫sm 0 (sm − s)‖f(s, u ∗ (s), u∗([h(u∗(s), s)])) − f(s, u∗∗(s), u∗∗([h(u∗∗(s), s)]))‖ds + ∫t 0 (t − s)‖f(s, u∗(s), u∗([h(u∗(s), s)])) − f(s, u∗∗(s), u∗∗([h(u∗∗(s), s)]))‖ds, ≤ max m=1,··· ,δ [LGm(sm − tm) + Lf(1 + LLh)(t 2 m+1 + s 2 m) 2 ]‖u∗ − u∗∗‖PC, (3.39) and for t ∈ (tm, sm], we get ‖(Qu∗)(t) − (Qu∗∗)(t)‖ ≤ max m=1,··· .δ LGm(sm − tm)‖u ∗ − u∗∗‖PC. (3.40) From the inequalities (3.38)-(3.40), we obtain ‖(Qu∗)(t) − (Qu∗∗)(t)‖ ≤ Λ‖u∗ − u∗∗‖PC. (3.41) Thus, by the inequality (3.30), we conclude that Q is a contraction on S i.e., there exists a unique fixed point of the map u ∈ S such that Qu(t) = u(t) for all t ∈ [0, T]. Hence problem (1.4)-(1.6) has a unique mild solution on [0, T]. Theorem 3.5. Assume that (A1)-(A3) are fulfilled and Ξ = max{LGm|sm − tm|; m = 1, · · · , δ} < 1. (3.42) Then, problem (1.4)-(1.6) has at least one mild solution on [0, T]. Proof. We consider the operators Q1 and Q2 on Bq,r = {u ∈ Y0 ∩ Y1 : ‖u‖PC ≤ r} defined by Q1u(t) =    ∫t tm Gm(s, u(s))ds, t ∈ (tm, sm], ∫sδ tδ Gδ(s, u(s))ds, t ∈ [0, t1] ∫sm tm Gm(s, u(s))ds, t ∈ (sm, tm+1], m = 1, · · · , δ, (3.43) CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 23 and Q2u(t) =    ∫T 0 (T − s)f(s, u(s), u([h(u(s), s)]))ds − ∫sδ 0 (sδ − s)f(s, u(s), u([h(u(s), s)]))ds + ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds, t ∈ [0, t1] 0, t ∈ (tm, sm], m = 1, · · · , δ, − ∫sm 0 (sm − s)f(s, u(s), u([h(u(s), s)]))ds + ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds, t ∈ (sm, tm+1], m = 1, · · · , δ. (3.44) where r is a positive constant such that max{ sup m=1,··· ,δ Km(sm − tm) + H(t2m+1 + s 2 m) 2 , Kδ(sδ − tδ) + H(T2 + s2δ + t 2 1) 2 } ≤ r. (3.45) For the purpose of convenience, we separate the proof into a few steps. Step 1. We show that Q1u + Q2u ∈ Bq,r for each u ∈ Bq,r. For t ∈ [0, t1], we have ‖Q1u(t) + Q2u(t)‖ ≤ ‖ ∫sδ tδ Gδ(s, u(s))ds‖ + ‖ ∫T 0 (T − s)f(s, u(s), u([h(u(s), s)]))ds‖ +‖ ∫sδ 0 (sδ − s)f(s, u(s), u([h(u(s), s)]))ds‖ + ‖ ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds‖ ≤ Kδ(sδ − tδ) + H(T2 + s2δ + t 2 1) 2 , (3.46) where H = supt∈[0,T] ‖f(t, u(t), u([h(u(t), t)]))‖. Similarly, for t ∈ (sm, tm+1], m = 1, · · · , δ ‖Q1u(t) + Q2u(t)‖ ≤ ‖ ∫sm tm Gm(tm, u(tm))‖ + ‖ ∫sm 0 (sm − s)f(s, u(s), u([h(u(s), s)]))ds‖ +‖ ∫t 0 (t − s)f(s, u(s), u([h(u(s), s)]))ds‖, ≤ Km(sm − tm) + H(t2m+1 + s 2 m) 2 , (3.47) and for t ∈ (tm, sm], m = 1, · · · , δ, ‖Q1u(t) + Q2u(t)‖ ≤ Km(sm − tm). (3.48) By inequality (3.45), we get ‖Q1u(t) + Q2u(t)‖ ≤ r for all t ∈ [0, T]. Hence, Q1u + Q2u ∈ Bq,r. Step 2. The map Q1 is contraction on Bq,r. From the step 2 of Theorem 3.3, we have that Q1 is a contraction on Bq,r. Step 3. The map Q2 is continuous on Bq,r. Let {up} ∞ p=1 be a sequence in Bq,r such that limp→∞ up = u ∈ Bq,r. For t ∈ (tm, sm], m = 24 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) 1, · · · , δ, it is obvious since Q2up(t) = 0. For t ∈ [0, t1], we get ‖(Q2up)(t) − (Q2u)(t)‖ ≤ ∫T 0 (T − s)‖f(s, up(s), up([h(up(s), s)])) − f(s, u(s), u([h(u(s), s)]))‖ds + ∫sδ 0 (sδ − s)‖f(s, up(s), up([h(up(s), s)])) − f(s, u(s), u([h(u(s), s)]))‖ds + ∫t 0 (t − s)‖f(s, up(s), up([h(up(s), s)])) − f(s, u(s), u([h(u(s), s)]))ds‖ds, by the continuity of f and Lebesgue dominated convergence theorem, we estimate ‖(Q2up)(t) − (Q2u)(t)‖ → 0, as p → ∞. (3.49) Similarly, t ∈ (sm, tm+1], m = 1, · · · , δ, ‖(Q2up)(t) − (Q2u)(t)‖ ≤ ∫sm 0 (sm − s)‖f(s, up(s), up([h(up(s), s)])) − f(s, u(s), u([h(u(s), s)]))‖ds + ∫t 0 (t − s)‖f(s, up(s), up([h(up(s), s)])) − f(s, u(s), u([h(u(s), s)]))‖ds by the continuity of f and Lebesgue dominated convergence theorem, we estimate ‖(Q2up)(t) − (Q2u)(t)‖ → 0, ∀ t ∈∈ (sm, tm+1] as p → ∞. (3.50) Hence, Q2 is continuous map on Bq,r. Step 4. Q2 is compact. Q2 is firstly uniformly bounded on Bq,r, since ‖Q2u‖PC ≤ r. We now prove that Q2 maps bounded set into equicontinuous set of Bq,r. For t ∈ (tm, sm], m = 1, · · · , δ, it is obvious. For τ2, τ1 ∈ [0, t1] with τ1 < τ2, we have ‖Q2(τ2) − Q2(τ1)‖ ≤ H(τ2 − τ1) 2 + H (τ2 − τ1) 2 2 . (3.51) For τ2, τ1 ∈ (sm, tm+1], m = 1, · · · , δ with τ2 > τ1, ‖Q2(τ2) − Q2(τ1)‖ ≤ H(τ2 − τ1) 2 + H (τ2 − τ1) 2 2 . (3.52) The right hand side of inequalities (3.51)-(3.52) tend to zero as τ2 → τ1. Thus, Q2(Bq,r) is equicontinuous. By Arzela-Ascoli theorem, we conclude that Q2 is completely continuous. Therefore, from the Krasnoselskiis fixed point theorem, we deduce that Q = Q1 + Q2 has a fixed point which is just a mild solution for the problem (1.4)-(1.6). CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 25 4 Examples For illustrating the application of the theory, we consider the following examples. Let us consider the following impulsive nonlinear Cauchy problems with boundary conditions as u′(t)(or cD 1/2 0,t u(t)) = 1 3 + t3 [ |u(t)| 6(1 + |u(t)|) + |u(1 3 u(t))| 1 + |u(1 3 u(t))| ] or (I 3/2 0,t 1 3 + t3 [ |u(t)| 6(1 + |u(t)|) + |u(1 3 u(t))| 1 + |u(1 3 u(t))| ]), t ∈ (0, 1] ∪ (2, 3] (4.1) u(t) = ∫t 1 |u(s)| (9s + 1)(1 + |u(s)|) ds, t ∈ (1, 2], (4.2) u(0) = u(3), (4.3) where 0 = s0 < t1 = 1 < s1 = 2 < t2 = 3 and J = [0, 3] and u ∈ C 1([0, 3], [0, 3]). Then, u ∈ CL([0, 3], [0, 3]). Here CL([0, 3], [0, 3]) = {u ∈ C([0, 3], [0, 3]) : |u(t) − u(s)|L ≤ L|t − s|, ∀ t, s ∈ [0, 3]} (4.4) and f(t, u(t), u(h(u(t), t))) = 1 3 + t3 [ |u(t)| 6(1 + |u(t)|) + |u(1 3 u(t))| 1 + |u(1 3 u(t))| ], (4.5) G1(t, u(t)) = |u(t)| (9t + 1)(1 + |u(t)|) . (4.6) It is easy to show that f and g satisfy the following condition ‖f(t, u1, v1) − f(t, u2, v2)‖ ≤ Lf[‖u1 − u2‖ + ‖v1 − v2‖L], u1, u2 ∈ [0, 3], (4.7) ‖G1(t, u1) − G1(t, u2)‖ ≤ 1 10 ‖u1 − u2‖, (4.8) ‖G1(t, u)‖ ≤ 1 9t + 1 = K1 ≤ 1 10 (4.9) Thus all the assumptions of Theorem 3.1/3.3 or 3.4/3.5 are fulfilled. Hence, there exists a mild solution for the problem (4.1). Acknowledgement. The authors would like to thank the referee for valuable comments and suggestions. The work of the first author is supported by the UGC (University Grants Commission, India) under Grant No (6405 − 11 − 061) and Indian Institute of Technology, Roorkee. nt. Received: December 2014. Accepted: January 2015. 26 Alka Chadha & Dwijendra N Pandey CUBO 17, 1 (2015) References [1] M. Benchohra, J. Henderson, S. K. Ntouyas, Impulsive Differential Equations and Inclusions, Contemporary Mathematics and Its Applications, Vol.2, Hindawi Publishing Corporation, New York, 2006. [2] V. Lakshmikantham, D. D. Bainov, P.S. Simeonov, Theory of Impulsive Differential Equations, World Scientific, Singapore-London, 1989. [3] P. Kumar, D. N. Pandey, D. Bahuguna, Impulsive boundary value problems for fractional differential equations with deviating arguments, J. Fract. Cal. Appl. 5 (2014), 146-155. [4] E. Hernández, D. O’Regan, On a new class of abstract impulsive differential equations, Proc. Amer. Math. Soc., 141 (2012), 1641-1649. [5] M. Pierri, D. O’ Regan, V. Rolnik, Existence of solutions for semilinear abstract differential equation with not instantaneous impulsive, App. Maths. Comput., 219 (2013), 6743-6749. [6] A. Chadha, D. N. Pandey, Existence of the mild solution for impulsive semilinear differential equation, Int. J. Partial. Diff. Equat., 2014 (2014), Art. ID 640931, pp-8. [7] J. R. Wang, X. Li, Periodic BVP for integer/fractional order nonlinear differential equations with non-instantaneous impulses, J. Appl. Math. Comput., 2014 (2014), pp-14. [8] M. Feckan, Y. Zhou, J. Wang, On the concept and existence of solution for impulsive fractional differential equations, Commun. Nonlinear Sci. Numer. Simulat., 17 (2012), 3050-3060. [9] K. Balachandran, F. C. Samuel, Existence of mild solutions for integrodifferential equations with impulsive conditions, Electr. J. Diff. Equat., 84 (2009), 1-9. [10] B. Ahmad, S. K. Ntouyas, A. Alsaedi, An existence result for fractional differential inclusions with nonlinear integral boundary conditions, J. Inequalities and Appl., 296 (2013), pp-9. [11] H. M. Ahmed, A. A. M. Hassan, A. S. Ghanem, Existence of mild solution for impulsive fractional differential equations with non-local conditions in Banach space, British Journal Mathematics Computer Science, 4 (2014) (6), 73-83. [12] Y. Liu, B. Ahmad, A study of impulsive multiterm fractional differential equations with single and multiple base points and applications, The Scient. World J., 2014 (2014), Art. ID- 194346, pp-28. [13] K. S. Miller, B. Ross, An Introduction to the Fractional Calculus and Fractional Differential Equation, John Wiley and Sons, Inc., New York, 1993. [14] S. G. Samko, A. A. Kilbas, O. I. Marichev, Fractional Integrals and Derivatives: Theory and Applications, Gordon and Breach Science Publisher, Yverdon, 1993. CUBO 17, 1 (2015) Periodic BVP for a class of nonlinear differential equation . . . 27 [15] A. A. Kilbas, H. M. Srivastava, J. J. Trujillo, Theory and Applications of Fractional Differ- ential Equations, Elsevier, Amsterdam, 2006. [16] I. Podlubny, Fractional Differential Equations, Mathe. Sci. Eng., vol. 198, Academic Press, San Diego 1999. [17] X. Liu, M. Jia, B. Wu, Existence and uniqueness of solution for fractional differential equations with integral boundary conditions. Elect. J. Qualit. Theory Diff. Equ. 2009 (2009) 69, pp-10 (2009). [18] A. Ahmad, A. Alsaedi, B. S. Alghamdi, Analytic approximation of solutions of the forced Duff- ing equation with integral boundary conditions, Nonlinear Analysis: Real World Applications, 9 (2008), 1727-1740. [19] B. Ahmad, J. J. Nieto, Existence results for nonlinear boundary value problems of fractional integrodifferential equations with integral boundary conditions, Boundary Val. Prob., 2009 (2009), Article ID-708576, pp-11. [20] M. Muslim, D. Bahuguna, Existence of solutions to neutral differential equations with deviated arguments, Elect. J. Qualit. Theory Diff. Equ., (2008) 27, 1-12. [21] R. Haloi, D. N. Pandey, D. Bahuguna, Existence of solutions to a non-autonomous abstract neutral differential equation with deviated argument, J. Nonlinear Evol. Equ. Appl., (2011) 5, 75-90. [22] L. E. El’sgol’ts, S. B. Norkin, Introduction to the Theory of Differential Equations with Devi- ating Arguments, Academic Press 1973. [23] Z.-W. Lv, J.-J. Xiao, Solutions to fractional differential equation with nonlocal initial condi- tions in Banach space, Adv. Diff. Equ., 2010 (2010), ID 340349, pp-10. [24] C. G. Gal, Nonlinear abstract differential equations with deviated argument, J. Math. Anal. Appl., 333 (2007), 971-983. [25] X. He, J. Xie, G. Chen, J. Shen, Integral BVPs for a class of first order impulsive functional differential equations, Int. J. Diff. Equ., 2010 (2010), Art. ID 908960, pp-11. [26] G. Song, Y. Zhao, X. Sun, Integral boundary value problems for first order impulsive integro- differential equations of mixed type, J. Comp. Appli. Math., 235 (2011), 2928-2935. [27] Y. K. Chang, A. Anguraj, P. Karthikeyan, Existence results for initial value problems with integral conditions for impulsive fractional differential equations, J. Fract. Cal. Appl., 2 (2011), 1-10. Introduction: Preliminaries and Assumptions Existence Result Integer Order case Fractional order case Examples