647
Views
0
CrossRef citations to date
0
Altmetric
Applied & Interdisciplinary Mathematics

Existence results of mild solutions for some stochastic integrodifferential equations with state-dependent delay and noninstantaneous impulses in Hilbert spaces

, , & ORCID Icon | (Reviewing editor:)
Article: 2141173 | Received 09 Aug 2022, Accepted 25 Oct 2022, Published online: 01 Dec 2022

ABSTRACT

The primary objective of this work is to investigate the existence of mild solutions for some neutral stochastic integrodifferential systems in Hilbert spaces. These systems have a state-dependent delay and noninstantaneous impulses. In order to achieve this objective, the neutral stochastic integrodifferential system that has been proposed with state-dependent delay and noninstantaneous impulses is converted into an equivalent fixed point problem by means of an appropriate integral operator.Together with the theory of stochastic analysis, we develop several existence results that are founded on the theory of resolvent operators in the sense of Grimmer, Hausdorff’s measures of noncompactness, and fixed point theorems.In the last part of this paper, specific examples are used to show how the general conclusions reached in this paper are made.

Public Interest Statement

In physics, automatic control, neural networks, infectious diseases, population growth, and cell production, stochastic differential equations with state-dependent delays arise. Some of these models are quite distinct from one another, and the majority of them do not appear to have an easy construction method. The delay may be defined implicitly by a functional, integral, or differential equation and should frequently be considered as part of the state variables. In most cases, the delay is not given explicitly as a function of what appears to be the natural state variable. Most of the time, the delay is not given as a direct function of what seems to be the natural state variable.

1 Introduction

The theory of state-dependent delay differential equations is the subject of a lot of research because it has real-world applications and because the qualitative theory is different and more complex than the usual ones for differential equations with delay. As a direct result of this fact, there has been a lot of interest in the study of the various properties that are associated with this class of equations, see for instance(Anguraj et al., Citation2007; Chadha & Pandey, Citation2015; Das, Pandey, Sukavanam et al., Citation2016a, Citation2016b; Dineshkumar et al., Citation2021; Dos Santos, Cuevas et al., Citation2011; Dos Santos, Mallika Arjunan et al., Citation2011; Hernández et al., Citation2018, Citation2006; Hernández, Wu, Chadha et al., Citation2020; Hernández, Wu, Fernandes et al., Citation2020; Sakthivel & Anandhi, Citation2010; Sakthivel & Ren, Citation2013; Suganya, Baleanu et al., Citation2015; Suganya, Mallika Arjunan et al., Citation2015; Valliammal et al., Citation2019; Vijayakumar et al., Citation2013) and the references therein. On the other hand when mathematics is used to model multiple things that happen in the real world or are the result of human actions, dynamical systems with some randomness are the result. Stochastic differential equations use randomness to give a mathematical description of an event. Over the course of the past few years, stochastic differential equations in both finite and infinite dimensions have garnered a significant amount of attention in a wide variety of research areas. This is due to the fact that they are capable of describing a diverse selection of occurrences in the fields of population dynamics, physics, electrical engineering, ecology, medicine, biology, and other scientific and engineering disciplines. See, (Deng et al., Citation2018; Y. Guo et al., Citation2019; Ma et al., Citation2020; Yuchen Guo et al., Citation2020) and references therein for a comprehensive introduction to stochastic differential equations and the applications of these equations.

Impulsive differential equation theory can accurately simulate many practical scenarios such as biological phenomena involving thresholds, bursting rhythm models in medicine and biology, optimal control models in economics, and frequency modulated systems. Impulses of evolution processes formulated by impulsive differential equations are generally abrupt and instantaneous, i.e. the perturbations (impulses) start abruptly and their duration is negligible compared to the process’s duration. In many cases, instant impulses do not adequately reflect evolution’s dynamics. A good example of this would be medication that takes into account an individual’s hemodynamic balance. The process of a drug being absorbed into the body is one that happens gradually and in a continual fashion. As a result, instantaneous impulses are unable to adequately characterize such a process.Hernandez and O’Regan (Hernandez & O’Regan, Citation2013) give a new example of impulsive behavior that starts at a random time and ends after a certain amount of time.

On the other hand, taking drugs has an effect on memory. Because of this, the newly proposed category of impulsive conditions in (Hernandez & O’Regan, Citation2013) does not fully or very well explain this phenomenon. When used in this situation, the theory of fractional calculus is a powerful way to explain why this happened. There has been a recent impulse in the scientific community to develop rigorous mathematical models and to access historic information on input. In this approach, the fractional derivative is one such concept that is more effective in complicated models with heredity properties. Several applications can be seen in the modeling of human liver (Baleanu, Jajarmi et al., Citation2020), hearing loss due to Mumps virus (Tuan et al., Citation2020), COVID-19 transmission (Baleanu, Mohammadi et al., Citation2020), etc.

Stochastic differential equations with state-dependent delay have become a popular topic. Several authors have looked at both qualitative and quantitative results for different types of stochastic systems that have state-dependent delays. In (Chaudhary & Pandey, Citation2019), Chaudhary and Pandey provided a framework for study the mild solution to fractional neutral stochastic integrodifferential systems with state dependent delay and noninstantaneous impulses in Hilbert spaces. Thiagu in (Thiagu, Citation2021) focused the approximate controllability of second order (q(1,2]) fractional impulsive stochastic differential system with nonlocal, state-dependent delay and Poisson umps in Hilbert spaces. Diop et al. in (Diop, Diop et al., Citation2022) investigated optimal control problems for a class of stochastic functional integral-differential equations in Hilbert spaces. A class of multi-term time-fractional random integrodifferential equations with state-dependent delay was investigated by Diop et al. in (Diop, Frederico et al., Citation2022). Yan and Lu in (Yan & Lu, Citation2018), studied a new class of fractional impulsive stochastic partial integrodifferential control systems with state-dependent delay and their optimal controls in a Hilbert space. Huang et al in (Huang et al., Citation2018), considered the existence and controllability of mild solutions for a class of second-order neutral impulsive stochastic evolution integrodifferential equations with state-dependent delay in a real separable Hilbert space among. However, no work has been reported in the literature regarding the existence of mild solutions of stochastic integrodifferential equations with state-dependent delay and noninstantaneous impulses in Hilbert spaces. Motivated by these facts, the purpose of this paper is to study the existence results for a class of impulsive neutral stochastic integrodifferential equations with state dependent delay, of the form

(1) d[ϑ(t)D(t,ϑρ(t,ϑt))] ={A[ϑ(t)D(t,ϑρ(t,ϑt))]+0tG(ts)[ϑ(s)D(s,ϑρ(s,ϑs))]ds}dt+F(t,ϑρ(t,ϑt))dt+0tG(t,s,ϑρ(s,ϑs))dW(s),ti=0n(si,ti+1],n, ϑ(t)=Si(t,ϑρ(t,ϑt)),ti=1n(ti,si], ϑ(t)=ϕ(t),t(,0],(1)

where, the state ϑ() takes values in a separable real Hilbert space H,(,),.

A:D (A)HH is a closed linear operator which generates a semigroup (T(t))t0, G(t) is a closed linear operator with domain D(G(t))D(A). Further, the history function ϑt:(,0]H representes the time history of the function ϑ from to the present time t and defined by ϑt(θ)=ϑ(t+θ) for θ(,0], and belongs to the abstract phase space B; w(t) is a brownian motion on a real separable Hilbert space (K,(,)K,K). Let nN, 0=t0=s0<t1<s1<t2\break<<tn<sn<tn+1<, are prefixed numbers.The functions D:[0,T]×BH, ρ:[0,T]×B\break(,T], F:[0,T]×BH, Si:[ti,si]×BH and G:Δ×BL20(K,H) are appropriate functions to be specified later, where Δ={(t,s):0s<tT}.

Three existence results for the system (1) are derived in this paper using resolvent operator theory in the sense of Grimmer and fixed point techniques, which are discussed in detail in the paper. We obtain sufficient conditions to prove the existence and uniqueness of the mild solution of system (1) in the first result by employing the Banach fixed point theorem under Lipschitz conditions on nonlinear terms. The second and third existence results, on the other hand, have demonstrated that mild solutions can be found by applying the Darbo and Darbo-Sadovskii fixed point theorems on nonlinear terms.

The following is how the paper is organized: In Section 2, we give notations, definitions and auxiliary outcomes on stochastic integral with respect to Wiener process (Prato & Zabczyk, Citation1992), measure of non compactness (Deimling, Citation1985) and resolvent operators theory for integro-differential equations (Grimmer, Citation1982). In Section 3, we use fixed point theorems to obtain the existence results for Equationequation (1). Section 4 concludes with examples that demonstrate the results that were obtained.

2 Preliminaries

Throughout this paper H,,H,H and K,,K,K are two real separable Hilbert spaces. Let (Ω,F,{Ft}t0,P) be a complete probability space with a normal filtration {Ft}t0 satisfying the usual conditions(i.e. it is increasing and right-continuous, while F0 contains all P-null sets). Let w(t):t0 denote a K-valued Wiener process defined on the probability space (Ω,F,{Ft}t0,P), with covariance operator Q, that is,

E(w(t),xKw(s),yK)=(ts)Qx,yK,

for all x,yK, where Q is a positive, self adjoint, trace class operator on K. In particular, we denote by w(t) a K valued Q Wiener process with respect to {Ft}t0. To define stochastic integrals with respect to the Q Wiener process w(t), we introduce the subspace K0=Q12K of K endowed with the inner product

u,vK0=Q12u,Q12vK

as a Hilbert space. We assume that there exists a complete orthonormal system {ei} in K, a bounded sequence of positive real numbers {λi} such that Qei=λiei, i=1,2,, and a sequence {wi(t)}i>1 of independent standard Brownian motions such that w(t)=i=1+λiwi(t)ei for t0 and Ft=Ftw, where Ftw is the σ algebra generated by w(s):0st. Let L20=L2(K0,H) be the space of all Hilbert-Schmidt operators from K0 to H. It turns out to be a separable Hilbert space equipped with the norm vL202=tr(vQ12)(vQ12) for any vL20. Obviously, for any bounded operator vL20, this norm reduces to vL202=trvQv=i=1+λivei2.

Let L2(Ω,F,P;H)L2(Ω;H) denote the Banach space of strongly-measurable, square integrable random variables equipped with norm

vL2(Ω,H)=EvH212.

The phase space B described by Hale and Kato in (Hale & Kato, Citation1978) is taken into consideration in order to deal with the infinite delay. Phase space B must possess the characteristics specified in the lemma below.

Lemma 2.1 (Hino et al., Citation1991) The abstract phase space is a seminormed linear space (B,.) of F0-measurable functions which maps (,0] into H and satisfies the following axioms:

1. If v:(,T]H is a continuous function on [0,T] such that v|[0,T]C([0,T],H) and v0B, then for every t[0,T], the following condition hold:

(a) vt is in B,

(b)v(t)∥≤KvtB,

(c)vtBH(t)v0B+N(t)supv(s)∥:0st..Where K is a positive constant, H:[0,+)[0,+) is locally bounded, N:[0,+)[0,+) is continuous, H and N are independent of v().

2. The phase space B is complete.

3. The function tϕt is well defined from the set R()=ρ(s,ψ):(s,ψ)[0,T]×B into B and there exists a continuous and bounded function

Jϕ:R()(0,+) such that ϕtBJϕ(t)\breakϕB, for every tR().

Lemma 2.2 (Hernandez et al., Citation2006) Let v:(,T]H be a function such that v0=ϕ and v|[0,T]PC([0,T],H). Then

vtB(HT+Jϕ)ϕB+NTsupv(ς)∥:ς[0,max{0,t}],tR()[0,T]

where Jϕ=suptR()Jϕ(t), HT=supt[0,T]H(t), NT=supt[0,T]N(t).

Lemma 2.3 (Prato & Zabczyk, Citation1992) For any q1 and for arbitrary L20(K,H)- valued predictable process z(),

sups[0,t]E0sz(r)dW(r)H2qCq0tEz(r)L202drq,t[0,+),

where Cq=(q(2q1))q.

Before going into greater detail about the main findings of this study, we’d like to provide a brief overview of the Hausdorff measure of noncompactness.

Definition 2.1 (Banas & Goebel, Citation1980) Let Q be bounded subset of normed space Y. The Hausdorff measure of noncompactness(shortly MNC) is defined by

β(Q)=inf{ϵ 0:Qhasafinitecoverbyballsofradiuslessthanϵ}.

Lemma 2.4 (Banas & Goebel, Citation1980) The Hausdorff measure verifies the following proprieties: let Q,Q1,Q2 be bounded subset of a normed space Y.

(a) Q1Q2 then β(Q1)β(Q2) (monotonicity);

(b) β(Q)=β(Q)

(c) β(Q1+Q2)β(Q1)+β(Q2),where Q1+Q2={x+y:xQ1,yQ2};

(d) β(λQ)|λ|β(Q) for any λR;

(e) β({a}Q)=β(Q) for every aY;

(f) β(Q)=0 if and only if Q is relatively compact;

(g) For any map G:D(G)YY which is Lipschitz continuous with Lipschitz constant α, we have

β(G(Q))αβ(Q),foranysubsetQD(G).

Definition 2.2 A continuous and bounded mapK:UXX is called β-contraction if there exists a constant 0<α<1 such that βK(E)αβ(E); for any noncompact bounded subset EU; where X is a Banach space.

The subsequently fixed point theorems will be employed in the proofs of our outcomes.

Lemma 2.5 (Darbo) If QX is closed and convex and 0Q, the continuous map K:QQ is a β-contraction and the set {ϑQ:ϑ=αK(ϑ)} is bounded for 0<α<1, then the map K has at least one fixed point in Q.

Lemma 2.6 (Darbo-Sadovskii) If QX be closed, bounded and convex. If the continuous map K:QQ is a β-contraction, then K has a fixed point in D.

In the sequel, we denote βC and βPC to denote the Hausdorff measure of noncompactness of C([0,T],H) and PC([0,T],H) respectively.

Next, to be able to access the existence of mild solutions for (1) we need to introduce partial integrodifferential equations and resolvent operators that will be used to develop the main results. Let X and M be Banach spaces. We denote by L(X,M) the Banach space of bounded linear operators from X to M endowed with the operator norm, and we abbreviate this notation to L(X) when X=M.

In what follows, X1 is a Banach space, A and G(t) are closed linear operators on X1, X2 is the Banach space D(A) endowed with the graph norm zX2=∥Az+zforzX2 and C(R+,X2) denotes the space of continuous functions from R+ into X2. For further purposes, let us consider the following system

(2) ϑ (t)=Aϑ(t)+0tG(ts)ϑ(s)dsfort[0,T],ϑ(0)=ϑ0X1.(2)

Definition 2.3 (Grimmer, Citation1982) A bounded linear operator valued function R(t)L(X1), t0 is called the resolvent operator for system (2) if it satisfies the following conditions:

(i) R(0)=Id and R(t)L(X1)Deδt for some constants D and δ.

(ii) For each xX1, R(t)x is continuous for t0.

(iii) R(t)L(X2) for t0. For xX2,R()xC1(R+,X1)C(R+,X2) and

R(t)x=AR(t)x+0tG(ts)R(s)xds
=R(t)Ax+0tR(ts)G(s)xds,fort0.

In what follows, we suppose the following assumptions.

(R1) A is the infinitesimal generator of a C0-semigroup (S(t))t0 on X1.

(R2) For all t0, G(t) is a closed linear operator from D(A) to X1 and G(t)L(X2,X1). For any zX2, the map tG(t)z is bounded, differentiable and the derivative tG(t)z is bounded and uniformly continuous on R+.

The next theorem gives sufficient conditions ensuring the existence of the resolvent operators for EquationEq. (2).

Theorem 2.7 (Grimmer, Citation1982) Assume that (R1) and (R2) hold. Then, there exists a unique resolvent operator of Eq. (2).

For more details on the theory of resolvent operators, one can see, (Desch et al., Citation1984; Grimmer, Citation1982).

Lemma 2.8 (Desch et al., Citation1984) Let assumptions (R1) and (R2) hold. The corresponding resolvent operator R() is compact for t>0 if and only if the C0-semigroup S() is compact for t>0.

Lemma 2.9 (Liang et al., Citation2008) Let the assumptions (R1) and (R2) be satisfied. Then there exists a constant L0=L0(T) such that

R(t+)R()R(t)L(H)L0,0tT.

Motivated by Grimmer (Grimmer, Citation1982), we present the concept of mild solutions for the non-instantaneous impulses, neutral stochastic integrodifferential equation(1).

Definition 2.4 A stochastic process ϑ:(,T]H is said to be a mild solution to the problem (1) if

1. ϑ(t) is measurable and Ft-adapted for each t0.

2. ϑ(t) has Cadlag paths on t[0,T], ϑ0=ϕB satisfying ϑ0L20(Ω,H), v|[0,T]PC([0,T],H) such that ϑ satisfies the following integral equation

(3) ϑ(t)=ϕ(t),t(,0];R(t)ϕ(0)D(0,ϕ(0))+D(t,ϑρ(t,ϑt))+0tR(ts)F(s,ϑρ(s,ϑs))ds+0tR(ts)0sG(s,ς,ϑρ(ς,ϑς))dW(ς)ds,t[0,t1];Si(t,ϑρ(t,ϑt)),ti=1n(ti,si];R(tsi)Si(si,ϑρ(si,ϑsi))D(si,ϑρ(si,ϑsi))+D(t,ϑρ(t,ϑt))+sitR(ts)F(s,ϑρ(s,ϑs))ds+sitR(ts)0sG(s,ς,ϑρ(ς,ϑς))dW(ς)ds,ti=1n(si,ti+1].(3)

3 Main results

Consider the space BT of all functions ϑ:(,T]H such that ϑ0B and ϑ|[0,T]PC([0,T],H) with seminorm .T defined by

ϑT=∥ϑ0B+sups[0,T]E(ϑ(s)2)12,ϑBT.

Now consider the operator Ψ:BTBT defined by

(4) (Ψϑ)(t)=ϕ(t),t(,0];R(t)ϕ(0)D(0,ϕ(0))+D(t,ϑρ(t,ϑt))+0tR(ts)F(s,ϑρ(s,ϑs))ds+0tR(ts)0sG(s,ς,ϑρ(ς,ϑς))dW(ς)ds,t[0,t1];Si(t,ϑρ(t,ϑt)),ti=1n(ti,si];R(tsi)Si(si,ϑρ(si,ϑsi))D(si,ϑρ(si,ϑsi))+D(t,ϑρ(t,ϑt))+sitR(ts)F(s,ϑρ(s,ϑs))ds+sitR(ts)0sG(s,ς,ϑρ(ς,ϑς))dW(ς)ds,ti=1n(si,ti+1].(4)

Let ϕˉ():(,T]H be the function defined by

ϕˉ(t)=ϕ(t),t(,0];R(t)ϕ(0),t[0,T].

Therefore ϕˉ0=ϕ.

If ϑ satisfies Equation (3), we can decompose ϑ(t)=ϕˉ(t)+y(t), t(,T] if and only if y0=0 and

(5) y(t)=R(t)D(0,ϕ(0))+Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)+0tR(ts)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds+0tR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds,t[0,t1];Sit,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt),ti=1n(ti,si];R(tsi)[Si(si,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi))D(si,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi))]+D(t,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt))+sitR(ts)F(s,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys))ds+sitR(ts)0sG(s,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς))dW(ς)ds,ti=1n(si,ti+1].(5)

Define BT0=yBT:y0=0B. For any yBT0,

yT=∥y0B+sups[0,T](Ey(s)2)1/2=sups[0,T](Ey(s)2)1/2,yBT.

It can be shown that BT0,BT0 is a Banach space. For each r>0, set

Br=yBT0:Ey2r

then for each r, Br is a bounded closed convex set in BT0.

Next, by using Lemma 2.2 and relation (a+b)22a2+2b2, we have

(6) ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)B22ϕˉρ(t,ϕˉt+yt)B2+yρ(t,ϕˉt+yt)B24(HT+Jϕ)2Eϕˉ0B2+NT2sup0sTEϕˉ(s)2+(HT+Jϕ)2Ey0B2+NT2sup0sTEy(s)24NT2sup0sTEy(s)2+NT2M˜2Eϕ(0)2+(HT+Jϕ)2ϕB24NT2r+NT2M˜2K2+(HT+Jϕ)2ϕB2=r˜,t[0,T].(6)

Now, we define the operator Π:BT0BT0 by

(7) Π(y)(t)=R(t)D(0,ϕ(0))+Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)+0tR(ts)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds+0tR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds,t[0,t1];Sit,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt),ti=1n(ti,si];R(tsi)[Si(si,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi))D(si,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi))]+D(t,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt))+sitR(ts)F(s,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys))ds+sitR(ts)0sG(s,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς))dW(ς)ds,ti=1n(si,ti+1].(7)

Throughout the paper, M˜=supt[0,T]R(t)2.

3.1 Existence of mild solution under lipschitz conditions

In the following, we give the first existence result for EquationEq.(1) under Lipschitz conditions on nonlinear terms. The proof is based on Banach fixed point theorem under the following assumptions:

(I) The functions D:[0,T]×BH, F:[0,T]×B\breakH, G:Δ×BL20(K,H) are continuous and there exists constant M1,M2,M3>0 such that

ED(t,v1)D(t,v2)H2M1v1v2B2,EF(t,v1)F(t,v2)H2M2v1v2B2,0tEG(t,s,v1)G(t,s,v2)L202dsM3v1v2B2

for all t[0,T], and v1,v2B.

(II) The function Si:(ti,si]×BH, i=1,2,,n are continuous and there exists Li>0, i=1,2,,n such that

ESi(t,v1)Si(t,v2)H2Liv1v2B2

for all t[0,T] and v1,v2B.

Theorem 3.1 Let ϑ0L20(Ω,H) and suppose that (R1), (R2), (I) and (II) are satisfied. Then there exists a unique mild solution of (1) provided that

(8) NT2sup1in(1+(M˜)2)Li+2M1+(M˜)2(M2+M3)T<14.(8)

Proof. To prove this result, it is enough to prove that the map Π has a unique fixed point. For y,zBT0 and t[0,t1], we have

(9) E(Πy)(t)(Πz)(t)H23EDt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)Dt,ϕˉρ(t,ϕˉt+zt)+zρ(t,ϕˉt+zt)H2+3E0tR(ts)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)Fs,ϕˉρ(s,ϕˉs+zs)+zρ(s,ϕˉs+zs)dsH2+3E0tR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)Gs,ς,ϕˉρ(ς,ϕˉς+zς)+zρ(ς,ϕˉς+zς)dW(ς)dsH23M1yρ(t,ϕˉt+yt)zρ(t,ϕˉt+zt)B2+(3M˜2M2)0tyρ(s,ϕˉs+ys)zρ(s,ϕˉs+zs)B2ds+(3M˜2M3)0tyρ(s,ϕˉs+ys)zρ(s,ϕˉs+zs)B2ds3M1+(M˜)2(M2+M3)t1×NT2Ey(t)z(t)H24NT2M1+(M˜)2(M2+M3)t1Ey(t)z(t)H2.(9)

For ti=1n(ti,si], we have

E(Πy)(t)(Πz)(t)H2ESit,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)Sit,ϕˉρ(t,ϕˉt+zt)+zρ(t,ϕˉt+zt)H2Liyρ(t,ϕˉt+yt)zρ(t,ϕˉt+zt)B2LiNT2Ey(t)z(t)H2.

Similarly for ti=1n(si,ti+1], we get

E(Πy)(t)(Πz)(t)H24NT2(M˜)2Li+2M1+(M˜)2(M2+M3)(ti+1si)Ey(t)z(t)H2.

Thus for all t[0,T], we obtain that

E(Πy)(t)(Πz)(t)H24NT2sup1in(1+(M˜)2)Li+2M1+(M˜)2(M2+M3)TEy(t)z(t)H2.

In accordance with condition (3.1), we can conclude that the Π is a contraction map, and that it has a a unique fixed point yBT0 . Since v(t)=ϕˉ(t)+y(t) for t(,T], therefore v is the unique mild solution of Equation (1).

3.2 Existence of mild solution under non-Lipschitz conditions

We establish the existence results for EquationEq.(1) under the case that the nonlinear terms satisfies non-Lipschitz continuous conditions. We will employ Darbo and Darbo-Sadovskii’s fixed point theorems. We introduce the following assumptions:

(B0) The resolvent operator (R(t))t0 compact for t>0

(B1) The function D:[0,T]×BH is continuous and there exists a positive constante M1 such that

ED(t,v)H2M11+vB2,

for all t[0,T] and vB.

(B2) The function F:[0,T]×BH satisfies the following condition:

(i) F verifies the Caratheodory’s condition i.e. F(t,):BH is continuous for all t[0,T] and the function F(,v):[0,T]H is strongly measurable for all vB.

(ii) There exists a continuous nondecreasing function nF:[0,](0,) and a positive integrable function χFL1([0,T],R+) such that

EF(t,v)H2χF(t)nFvB2.

(B3) The function G:Δ×BL20(K,H) satisfies the following condition

(i) G verifies the Caratheodory’s condition i.e. G(t,s,):BL20(K,H) is continuous for all (t,s)Δ and the function G(,,v):ΔL20(K,H) is strongly measurable for all vB.

(ii) There exists a continuous nondecreasing function nG:[0,](0,) and a positive integrable function χGL1([0,T],R+) such that

0tEG(t,s,v)L20(K,H)2dsχG(t)nGvB2.

(B4) The function Si:(ti,si]×BH, i=1,2,,n are continuous and there exists Li>0, i=1,2,,n such that

ESi(t,v)H2Li1+vB2,

for all t(ti,si] and vB.

Theorem 3.2 Let ϑ0L20(Ω,H) and suppose that (R1), (R2), (I)-(II) and (B0)-(B4) are satisfied. If 1dsnF(s)+nG(s)= and

(10) 4NT2max1in(1+8M˜2)Li+41+2M˜2M1<1.(10)

Then EquationEquation (1) admits a mild solution on [0,T].

Proof. The following are steps we take to verify the outcome.

Step 1. the set yPC:y=αΠyfor0<α<1 is bounded

Let yˉ be a solution of y=αΠy for 0<α<1. We have

(11) sup0stϕˉρ(s,ϕˉs+yˉs)+yˉρ(s,ϕˉs+yˉs)B24[NT2Eyˉt2+NT2M˜2K2+(HT+Jϕ)2ϕB2](11)

where

yˉt2=sup0styˉ(s)H2.

Take

(12) q(t)=4NT2Eyˉt2+NT2M˜2K2+(HT+Jϕ)2ϕB2.(12)

For t[0,t1], using Lemma 2.3, we have that

Ey¯(t)H2 4ER(t)D(0,ϕ(0))2+ED(t,ϕ¯ρ(t,ϕ¯t+y¯t)+y¯ρ(t,ϕ¯t+y¯t))2+E0t1ptR(ts)F(s,ϕ¯ρ(s,ϕ¯s+y¯s)+y¯ρ(s,ϕ¯s+y¯s))1ptds2+E0t1ptR(ts)[0sG(s,ς,ϕ¯ρ(ς,ϕ¯ς+y¯ς)+y¯ρ(ς,ϕ¯ς+y¯ς))dW(ς)]1ptds24{M2ED(0,ϕ(0))2+ED(t,ϕ¯ρ(t,ϕ¯t+y¯t)+y¯ρ(t,ϕ¯t+y¯t))2+M21ptt11pt0tEs,ϕ¯ρ(s,ϕ¯s+y¯s)+ y¯ρ(s,ϕ¯s+y¯s)21ptds+M21ptt11pt0tE0sG(s,ς,ϕ¯ρ(ς,ϕ¯ς+y¯ς)+y¯ρ(ς,ϕ¯ς+y¯ς))dW(ς)2021ptds}4{M2M1(1+ϕ2)+M1(1+ϕ¯ρ(t,ϕ¯t+y¯t)+y¯ρ(t,ϕ¯t+y¯t)2)+M21ptT0t1ptχF(s)1ptnF(q(s))1ptds +M21ptT0t 0s1ptEG(s,ς,ϕ¯ρ(ς,ϕ¯ς+y¯ς)+y¯ρ(ς,ϕ¯ς+y¯ς))2021ptdς1ptds} 4{M2M1(1+ϕ2)+M1(1+ϕ¯ρ(t,ϕ¯t+y¯t)+y¯ρ(t,ϕ¯t+y¯t)2) +M21ptTχF(t)1ptnF(ϕ¯ρ(t,ϕ¯t+y¯t)+y¯ρ(t,ϕ¯t+y¯t)2)+M21ptT0t1ptχG(s)1ptnG(q(s))1ptds}.

For ti=1n(ti,si], we have

Eyˉ(t)2Li1+ϕˉρ(t,ϕˉt+yˉt)+yˉρ(t,ϕˉt+yˉt)B2Li(1+q(t)).

Similarly for ti=1n(si,ti+1], we have

Eyˉ(t)24ER(tsi)Sisi,ϕˉρ(si,ϕˉsi+yˉsi)+yˉρ(si,ϕˉsi+yˉsi)Dsi,ϕˉρ(si,ϕˉsi+yˉsi)+yˉρ(si,ϕˉsi+yˉsi)2+EDt,ϕˉρ(t,ϕˉt+yˉt)+yˉρ(t,ϕˉt+yˉt)2+EsitR(ts)Fs,ϕˉρ(s,ϕˉs+yˉs)+yˉρ(s,ϕˉs+yˉs)ds2+EsitR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yˉς)+yˉρ(ς,ϕˉς+yˉς)dW(ς)ds2
8M˜2[Li(1+q(si))+M1(1+q(si))]+4M1(1+q(t))+4M˜2(ti+1si)sitχF(s)nF(q(s))ds+4M˜2(ti+1si)0tχG(s)nG(q(s))ds.

Therefore for all t[0,T], we obtain that

Eyˉ(t)2C+Liq(t)+8M˜2Liq(t)+M1q(t)+4M1q(t)+4M˜2T0tχF(s)nF(q(s))ds+4M˜2T0tχG(s)nG(q(s))ds,

where

C=4M˜2M11+ϕB2+(1+8M˜2)Li+42M˜2+1M1.

Putting these values in EquationEquation (12), we get

q(t)4NT2C+NT2M˜2K2+(HT+Jϕ)2ϕB2+4NT2(1+8M˜2)Li+4(1+2M˜2)M1q(t)+16M˜2NT2T0tχF(s)nF(q(s))ds+16M˜2NT2T0tχG(s)nG(q(s))ds.

Since

α1=4NT2max1iN(1+8M˜2)Li+4(1+2M˜2)M1<1,

we have

q(t)α21α1+α21α10tχF(s)nF(q(s))ds+α31α10tχG(s)nG(q(s))ds

where α2=4NT2C+NT2M˜2K2+(HT+Jϕ)2ϕB2, and α3=16M˜2NT2T.

Let α(t)=α21α1+redα21α10tχF(s)nF(q(s))ds+α31α10tχG(s)nG(q(s))ds

with α(0)=α21α1. Therefore we get

α (t)α31α1χF(t)nF(q(t))+χG(t)nG(q(t))α31α1maxχF(t),χG(t)×nF(q(t))+nG(q(t))

and from which we derive that

α(0)α(t)dunF(u)+nG(u)0Tα31α1maxχF(t),χG(t)dt<.

This provides that the function α(t) are bounded on [0,T]. Therefore, the functions q(t) are bounded and y() are bounded on [0,T].

Step 2. Π:BT0BT0 is continuous

Let {y(k)}rNBT0 be a sequence such that {y(k)}yBT0 as k. Then there exists a number 0>0 such that y(k)(t)∥≤0 for all k and a.s. t[0,T]. Also using EquationEquation (6), we have

ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)B2r˜.

Now using Lemma 2.1, we obtain that

yρ(t,ϕˉt+yt(k))(k)yρ(t,ϕˉt+yt)B22(N(t))2sups[0,t]y(k)(s)y(s)2+(H(t))2y0(k)y0B2=2(N(t))2sups[0,t]y(k)(s)y(s)22NT2y(k)y20,ask.

By continuity of functions F(t,) for a.e t[0,T] and G(t,s,) for a.e (t,s)Δ, we get that

limkFt,ϕˉρ(t,ϕˉt+yt(k))+yρ(t,ϕˉt+yt(k))(k)=Ft,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt),
limkGt,s,ϕˉρ(t,ϕˉt+yt(k))+yρ(t,s,ϕˉt+yt(k))(k)=Gt,s,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt).

Thanks to Lebesgue dominated convergence theorem, we have that for t[0,t1]

E(Πy(k))(t)(Πy)(t)23EDt,ϕˉρ(t,ϕˉt+yt(k))+yρ(t,ϕˉt+yt(k))(k)Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)2+E0tR(ts)Fs,ϕˉρ(s,ϕˉs+ys(k))+yρ(s,ϕˉs+ys(k))(k)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds2+E0tR(ts)(0s(Gs,ς,ϕˉρ(ς,ϕˉς+yς(k))+yρ(ς,ϕˉς+yς(k))(k)Gs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς))dW(ς))ds2
3EDt,ϕˉρ(t,ϕˉt+yt(k))+yρ(t,ϕˉt+yt(k))(k)Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)2+M˜20tEFs,ϕˉρ(s,ϕˉs+ys(k))+yρ(s,ϕˉs+ys(k))(k)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)2ds+M˜20t0sEGs,ς,ϕˉρ(ς,ϕˉς+yς(k))+yρ(ς,ϕˉς+yς(k))(k)Gs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)L202dςds0ask.

For ti=1n(ti,si], we have that

E(Πy(k))(t)(Πy)(t)2=ESit,ϕˉρ(t,ϕˉt+yt(k))+yρ(k)(t,ϕˉt+yt(k))Sit,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)20ask.

For ti=1n(si,ti+1], we also have that

E(Πy(k))(t)(Πy)(t)24ER(tsi)Sisi,ϕˉρ(si,ϕˉsi+ysi(k))+yρ(si,ϕˉsi+ysi(k))(k)Sisi,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi)Dsi,ϕˉρ(si,ϕˉsi+ysi(k))+yρ(si,ϕˉsi+ysi(k))(k)+Dsi,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi)2+EDt,ϕˉρ(t,ϕˉt+yt(k))+yρ(t,ϕˉt+yt(k))(k)Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)2+EsitR(ts)Fs,ϕˉρ(s,ϕˉs+ys(k))+yρ(s,ϕˉs+ys(k))(k)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds2+EsitR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yς(k))+yρ(ς,ϕˉς+yς(k))(k)Gs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds2
4M˜2ESisi,ϕˉρ(si,ϕˉsi+ysi(k))+yρ(si,ϕˉsi+ysi(k))(k)Sisi,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi)2+EDsi,ϕˉρ(si,ϕˉsi+ysi(k))+yρ(si,ϕˉsi+ysi(k))(k)Dsi,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi)2+EDt,ϕˉρ(t,ϕˉt+yt(k))+yρ(t,ϕˉt+yt(k))(k)Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)2+M˜2sitEFs,ϕˉρ(s,ϕˉs+ys(k))+yρ(s,ϕˉs+ys(k))(k)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)2ds+M˜2sit0sEGs,ς,ϕˉρ(ς,ϕˉς+yς(k))+yρ(ς,ϕˉς+yς(k))(k)Gs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)L202dςds0ask.

It turns out that we can conclude that the nonlinear mapping Π:BT0BT0 is continuous.

Step 3. II is a β-contraction.

To prove this, decompose Π as follows Π=Π1+Π2 for t[0,T], where

(Π1y)(t)=R(t)D(0,ϕ(0))+Dt,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt)t[0,t1];Sit,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt),ti=1n(ti,si],R(tsi)[Si(si,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi));D(si,ϕˉρ(si,ϕˉsi+ysi)+yρ(si,ϕˉsi+ysi))]+D(t,ϕˉρ(t,ϕˉt+yt)+yρ(t,ϕˉt+yt))ti=1n(si,ti+1];

and(Π2y)(t)=0tR(ts)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds+0tR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds,t[0,t1];0,ti=1n(ti,si];sitR(ts)F(s,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys))ds+sitR(ts)0sG(s,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς))dW(ς)ds,ti=1n(si,ti+1]. Claim 1: Π1 is Lipschitz continuous.

For t[0,t1], using assumption (A1), Lemmas 2.2 and u,vBT0, we have

E(Π1u)(t)(Π1v)(t)2EDt,ϕˉρ(t,ϕˉt+ut)+uρ(t,ϕˉt+ut)Dt,ϕˉρ(t,ϕˉt+vt)+vρ(t,ϕˉt+vt)2M1uρ(t,ϕˉt+ut)vρ(t,ϕˉt+vt)22NT2M1supt[0,t1]Eu(t)v(t)2.

For ti=1n(ti,si], we have

E(Π1u)(t)(Π1v)(t)2ESit,ϕˉρ(t,ϕˉt+ut)+uρ(t,ϕˉt+ut)Sit,ϕˉρ(t,ϕˉt+vt)+vρ(t,ϕˉt+vt)2Liuρ(t,ϕˉt+ut)vρ(t,ϕˉt+vt)2NT2Lisupt(ti,si]Eu(t)v(t)2.

Similarly for ti=1n(si,ti+1], we have

E(Π1u)(t)(Π1v)(t)23ER(tsi)Sisi,ϕˉρ(si,ϕˉsi+usi)+uρ(si,ϕˉsi+usi)Sisi,ϕˉρ(si,ϕˉsi+vsi)+vρ(si,ϕˉsi+vsi)2
+D(si,ϕˉρ(si,ϕˉsi+usi)+uρ(si,ϕˉsi+usi))D(si,ϕˉρ(si,ϕˉsi+vsi)+vρ(si,ϕˉsi+vsi))2+ED(t,ϕˉρ(t,ϕˉt+ut)+uρ(t,ϕˉt+ut))D(t,ϕˉρ(t,ϕˉt+vt)+vρ(t,ϕˉt+vt))23M˜2[Liuρ(si,ϕˉsi+usi)vρ(si,ϕˉsi+vsi)2+M1uρ(si,ϕˉsi+usi)vρ(si,ϕˉsi+vsi)2+M1uρ(t,ϕˉt+ut)vρ(t,ϕˉt+vt)23M˜2Liuρ(si,ϕˉsi+usi)vρ(si,ϕˉsi+vsi)2+M1uρ(si,ϕˉsi+usi)vρ(si,ϕˉsi+vsi)2+M1uρ(t,ϕˉt+ut)vρ(t,ϕˉt+vt)23NT2M˜2Li+1+M˜2M1}supt(si,ti+1]Eu(t)v(t)2.

Thus for all t[0,T], we have

E(Π1u)(t)(Π1v)(t)23NT2(1+M˜2)Li+1+M˜2M1supt[0,T]Eu(t)v(t)2.

Taking the supremum over t, we obtain

Π1uΠ1v2κuv2

where κ=3NT2(1+M˜2)Li+1+M˜2M1. It turns out from (10) that κ<1 and thus we get Π1 is Lipschitz continuous.

Claim 2: {Π2y,yBT0} is an equicontinuous family of functions on [0,T].

Let 0<h1<h2t1. For each yBT0, we get that

(13) E(Π2y)(h2)(Π2y)(h1)24Eh1h2R(h2s)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds2+E0h1R(h2s)R(h1s)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds2+Eh1h2R(h2s)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds2+E0h1R(h2s)R(h1s)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds24M˜2nF(r˜)h1h2χF(s)ds+nF(r˜)0h1R(h2s)R(h1s)L(H)2χF(s)ds+M˜2nG(r˜)h1h2χG(s)ds+nG(r˜)0h1R(h2s)R(h1s)L(H)2χG(s)ds(13)

Similarly, for any h1,h2(si,ti+1], h1<h2, i=1,,n and each yBT0, we obtain that

(14) E(Π2y)(h2)(Π2y)(h1)24M˜2nF(r˜)h1h2χF(s)ds+nF(r˜)sih1R(h2s)R(h1s)L(H)2χF(s)ds+M˜2nG(r˜)h1h2χG(s)ds+nG(r˜)sih1R(h2s)R(h1s)L(H)2χG(s)ds.(14)

By the facts that sχF(s) and sχG(s) are Lebesgue integrable and the continuity of (R(t))t0 for t>0 in the operator norm topology, we derive that the right hand side of (13) and (14) goes to zero as h2h10 independently of y. Thus, the set {Π2y,yBT0} is equicontinuous.

Claim 3: Π2 maps BT0 onto a precompact set in BT0.

Let ti=1n(si,ti+1] be fixed and \isin be a real number satisfying 0t.

Define the operators Π2,Π~2 on BT0 by:

(Π2\isiny)(t)=R(\isin)0t\isinR(ts\isin)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds+R(\isin)0t\isinR(ts\isin)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds

and

(Π2\isiny)(t)=0t\isinR(ts)Fs,ϕˉρ(s,ϕˉs+ys)+yρ(s,ϕˉs+ys)ds+0t\isinR(ts)0sGs,ς,ϕˉρ(ς,ϕˉς+yς)+yρ(ς,ϕˉς+yς)dW(ς)ds.

In view of the compactness of R(\isin), it is clear that the set (Π2\isiny)(t),yBT0 is precompact for every \isin(0,t). Moreover, for each yBT0, by using Lemma 2.9, we have that

E(Π21pty)(t)(Π2y)(t)22E0tR(ts)F(s,ϕ¯ρ(s,ϕ¯s+ys)+yρ(s,ϕ¯s+ys))1ptdsR()0tR(ts)F(s,ϕ¯ρ(s,ϕ¯s+ys)+yρ(s,ϕ¯s+ys))1ptds2+ER()0tR(ts)[0s1ptG(s,ς,ϕ¯ρ(ς,ϕ¯ς+yς)+yρ(ς,ϕ¯ς+yς))dW(ς)]ds0tR(ts)[0s1ptG(s,ς,ϕ¯ρ(ς,ϕ¯ς+yς)+yρ(ς,ϕ¯ς+yς))dW(ς)]ds2} 21ptnF(r˜)0tR()R(ts)R(ts)(H)21ptχF(s)1ptds+21ptnG(r˜)0tR()R(ts)R(ts)(H)21ptχG(s)1ptds 21ptnF(r˜)(L0)0t1ptχF(s)1ptds+21ptnG(r˜)(L0)0t1ptχG(s)1ptds01ptas1pt0.

Hence, by total boundedness, we derive that the set (Π2\isiny)(t),yBT0 is precompact for every \isin, \isin(0,t). Applying this idea again, we have that

E(Π21pty)(t)(Π21pty)(t)22E0tR(ts)F(s,ϕ¯ρ(s,ϕ¯s+ys)+yρ(s,ϕ¯s+ys))1ptds0tR(ts)F(s,ϕ¯ρ(s,ϕ¯s+ys)+yρ(s,ϕ¯s+ys))1ptds2+E0tR(ts)[0s1ptG(s,ς,ϕ¯ρ(ς,ϕ¯ς+yς)+yρ(ς,ϕ¯ς+yς))dW(ς)]ds0tR(ts)[0s1ptG(s,ς,ϕ¯ρ(ς,ϕ¯ς+yς)+yρ(ς,ϕ¯ς+yς))dW(ς)]ds2} 21ptnF(r˜)1pt(M2)tt1ptχF(s)1ptds+21ptnG(r˜)(M2)tt1ptχG(s)1ptds01ptas0. 

It turns out that there are precompact sets arbitrary close to the set {(Π2y)(t):yBT0}. Hence {(Π2y)(t):yBT0} is precompact. The case t(0,t1] is similar, so we omit it.

Let O be an arbitrary bounded subset of BT0. Then using precompactness of Π2, we obtain that

(15) βPC(Π(O))=βPC(Π1(O)+Π2(O))βPC(Π1(O))+βPC(Π2(O))βPC(Π1(O))κβPC(O).(15)

Hence Π is β contraction.

Therefore, by the Darbo fixed point theorem, the operator Π has at least one fixed point yBT0. Let v(t)=ϕˉ(t)+y(t), t(,T]. Then, v is a fixed point of the operator Π, i.e. v is a mild solution of the problem (1).

Now, we employ the Darbo-Sadovskii fixed point theorem (Lemma 2.6) to obtain another existence result.

Theorem 3.3 Let ϑ0L20(Ω,H) and assume that the assumptions (R1), (R2), (I)(II) and (B0)(B4) are satisfied. If

(16) 4NT2max1iN(1+8M˜2)Li+42M˜2+1M1+4M˜2TlimsupςnF(ς)ς0TχF(s)ds+4M˜2TlimsupςnG(ς)ς0TχG(s)ds<1.(16)

Then EquationEquation (1) has at least one mild solution on [0,T].

Proof. In view of the proof of Theorem 3.2, it turns out that the nonlinear mapping Π:BT0BT0 given by EquationEqu. (7) is continuous.

In what follows, we demonstrate that there exists r>0 such that Π(Br)Br. We argue by contradiction. Suppose that it is not true, then for any r>0, there exist y˜Br and t˜[0,T] such that r<E(Πy˜)(t˜). Therefore

for t˜[0,t1] and y˜Br, we have that

r<E(Πy˜)(t˜)∥≤4M˜2M11+ϕB2+M11+ϕˉρ(t˜,ϕˉt˜+y˜t˜)+yρ(t˜,ϕˉt˜+y˜t˜)B2+M˜2t10t1χF(s)nFϕˉρ(s,ϕˉs+y˜s)+y˜ρ(s,ϕˉs+y˜s)B2ds+M˜2t10t1χG(s)nGϕˉρ(s,ϕˉs+y˜s)+y˜ρ(s,ϕˉs+y˜s)B2ds.

For t˜i=1n(ti,si] and y˜Br, we have that

r<E(Πy˜)(t˜)2=Li1+ϕˉρ(t˜,ϕˉt˜+yt˜)+y˜ρ(t˜,ϕˉt˜+y˜t˜)B2.

Likewise, for t˜i=1n(si,ti+1], we get that

r<E(Πy˜)(t˜)28M˜2Li1+ϕˉρ(si,ϕˉsi+y˜si)+yρ(si,ϕˉsi+y˜si)B2+M11+ϕˉρ(si,ϕˉsi+y˜si)+yρ(si,ϕˉsi+y˜si)B2+4M11+ϕˉρ(t˜,ϕˉt˜+yt˜)+y˜ρ(t˜,ϕˉt˜+y˜t˜)B2+M˜24(ti+1si)siti+1χF(s)nFϕˉρ(s,ϕˉs+ys˜)+y˜ρ(s,ϕˉs+y˜s)B2ds+M˜24(ti+1si)siti+1χG(s)nGϕˉρ(s,ϕˉs+ys˜)+y˜ρ(s,ϕˉs+y˜s)B2ds.

From the above estimations, for all t˜[0,T], we have that

r<E(Πy˜)(t˜)2C+Li4NT2r+4NT2M˜2K2+(HT+Jϕ)2ϕB2+8M˜2Li4NT2r+4NT2M˜2K2+(HT+Jϕ)2ϕB2+8M˜2M14NT2r+4NT2M˜2K2+(HT+Jϕ)2ϕB2+4M14NT2r+4NT2M˜2K2+(HT+Jϕ)2ϕB2+4M˜2T0TχF(s)nF4NT2r+NT2M˜2K2+(HT+Jϕ)2ϕB2ds+4M˜2T0TχG(s)nG4NT2r+NT2M˜2K2+(HT+Jϕ)2ϕB2ds.

By dividing both sides by r and letting r, we obtain that

1<4NT2max1iN(1+8M˜2)Li+42M˜2+1M1+4M˜2TlimsupςnF(ς)ς0TχF(s)ds+4M˜2TlimsupςnG(ς)ς0TχG(s)ds<1

This is a contradiction, therefore there exists a positive constant r>0 such that Π(Br)Br. Arguing likewise the proof of the Theorem 3.2, one can conclude that the system (1) has a mild solution.

4 Illustratives Examples

Example 1. We Consider the following neutral stochastic problem

(17) tz(t,ζ)te2(st)zsρ(z(t)),ζ25ds=2ζ2z(t,ζ)te2(st)zsρ(z(t)),ζ25ds+0ta(ts)2ζ2z(s,ζ)se2(rs)zrρ(z(s)),ζ25drds+te2(st)zsρ(z(t)),ζ20ds+0tsin(ts)se2(ks)zzρ(z(s)),x20dw(k)dtds,(t,ζ)i=0n(si,ti+1]×[0,π]z(t,ζ)=te2(st)zsρ(z(t)),x30ds,(t,ζ)i=1n(ti,si]×[0,π]z(t,0)=z(t,π)=0,t0z(t,ζ)=ϕ(t,ζ),t(,0],ζ[0,π],(17)

where, ϕB, 0=t0=s0<t1<s1<t2<<tn<sn<tn+1=1 are fixed real numbers, w(t) is a standard cylindrical wiener process defined on a complete probability (Ω,F,P), a:R+R+.

Let H=L2(0,π) with the norm . Consider the operator Au=2uζ2 with the domain

D(A)=uH|uanduζareabsolutelycontinuous,2uζ2H,u(0)=u(ζ)=0.

The spectrum of A consists of the eigenvalues n2 for nN, with associated eigenvectors en:=2πsin(nx˜),(n=1,2,3,). Furthermore, the set {en:nN} is an orthogonal basis in H. Then

Ax˜=n=1n2x˜,enen,x˜H.

It is well known that A is the infinitesimal generator of a strongly continuous semigroup {T(t)}t0 on H, which is compact and is given by

T(t)x˜=n=1en2tx˜,enen,x˜H.

Thus (R1) is true.

Let G(t):D(A)HH, t0, be the operator defined by

G(t)(u)=a(t)Au,t0,uD(A).

Let p[1,), r[0,) and h:(,r)R be a nonnegative borel measurable function with satisfies such that

(i) h is locally integrable function;

(ii) There exists a nonnegative locally bounded function ρ on (,0] such that h(+τ)ρ()h(τ) for all 0 and τ(,r)/N, where N(,r) is a set with Lebesgue measure 0.

More details on the choice of h can be found in (Hino et al., Citation1991). Now, let B=Cr×L2(h,H) be the space of all functions ν:(,0]X such that ν|[r,0]C[r,0],H, ν() is Lebesgue measurable on (,r) with seminorm

νB=supθ[r,0]ν(θ)+rh(θ)ν(θ)pdθ1p.

Thanks to [38, Theorem 1.3.8], it turns out that B is a phase space and for r=0 and p=2, we derive thatB=C0×L2(h,H) with

K=1,H(t)=ρ(t)andN(t)=1+t0h(θ)dθ12fort0.

Set ϕ(θ)(ζ)=ϕ(θ,ζ)B and define

ϑ(t)(ζ)=z(t,ζ),
ρ(t,ϕ)(ζ)=ρ(ϕ(0)),
D(t,ϕ)(ζ)=0e2sϕ25ds,
F(t,ϕ)(ζ)=0e2sϕ20ds,
0tG(t,s,ϕ)(ζ)ds=0tsin(ts)0e2τϕ20dτds,
Si(t,ϕ)(ζ)=0e2sϕ30ds.

Then EquationEq. (17), takes the following abstract form

(18) d[ϑ(t)D(t,ϑρ(t,ϑt))]=A[ϑ(t)D(t,ϑρ(t,ϑt))]+0tG(ts)[ϑ(s)D(s,ϑρ(s,ϑs))]dsdt+F(t,ϑρ(t,ϑt))dt+0tG(t,s,ϑρ(s,ϑs))dW(s),ti=0n(si,ti+1],nN,ϑ(t)=Si(t,ϑρ(t,ϑt)),ti=1n(ti,si],ϑ(t)=ϕ(t)B,t(,0].(18)

Moreover, if a is bounded and C1–function such that a is bounded and uniformly continuous, then (R2) is satisfied and hence, by Theorem 2.7, EquationEq. (2) has a resolvent operator R(t)t0 on H.

Simple computations yield that the functions D,F,G and Si satisfies assumptions (I)(II). Hence, from Theorem 3.1, the EquationEquation (17) has a unique mild solution on [0,1].

Example 2. Consider the following impulsive noninstantaneous stochastic integro-differential equation:

(19) ty(t,ζ)tP1t,st,x,y(s,ρ(y(t),ζ))ds=2ζ2y(t,ζ)tP1t,st,x,y(s,ρ(y(t),ζ))ds+0ta(ts)2ζ2y(s,ζ)sP1s,rs,x,y(r,ρ(y(s),ζ))drds+tP2t,st,x,y(s,ρ(y(t),ζ))ds+0tsP3s,zs,x,ϑ(z,ρ(y(s),ζ))dW(z)dtds,(t,ζ)i=0n(si,ti+1]×[0,π];y(t,ζ)=tP4t,st,x,y(s,ρ(y(t),ζ))ds,(t,ζ)i=1n(ti,si]×[0,π];y(t,0)=y(t,π)=0,t0y(t,ζ)=ϕ(t,ζ),t(,0],(19)

where ϕB, 0=t0=s0<t1<s1<t2<<tn<sn<tn+1=1 are fixed real numbers. The function ρ is a continuous function, a:R+R+. Moreover, we consider the functions Pi:R4R, i=1,2,3,4 are continuous functions and there exists continuous functions εi, κi:RR, i=1,2,3,4 such that

(20) |Pi(t,s,u,v)|εi(t)κi(s)|v|,(t,s,u,v)R4,(20)

with εi=0(κi(s))2g(s)ds12<. Define the operator A and phase space B=Cr×L2(h,H) as in Example 1.

Set ϕ(τ)(ζ)=ϕ(τ,ζ)B, define

ϑ(t)(ζ)=y(t,ζ),
ρ(t,ϕ)(ζ)=ρ(ϕ(0)),
D(t,ν)(ζ)=0P1t,s,x,ν(s)(ζ)ds,
F(t,ν)(ζ)=0P2t,s,x,ν(s)(ζ)ds,
0tG(t,s,ν)(ζ)ds=0t0P3s,z,x,ν(s)(ζ)dW(s)ds
Si(t,ν)(ζ)=0P4t,s,x,ν(s)(ζ)ds.

Then EquationEquation (19) can be written in the form of system (1). By using (20), we obtain

EF(t,ν)2=E0π|0P2t,s,x,ν(s)(ζ)ds|2dx122E0π0ε2(t)κ2(s)|ν(s)(ζ)|ds2dx122Eε20(κ2(s))2h(s)ds120h(s)ν(s)2ds122χF(t)νB2.

for all (t,ν)[0,1]×B, with χF(t)=ε2Ξ22. Likewise, one can obtain that

ED(t,ν)2M1νB2,EG(t,s,ν)2χG(t)νB2,ESi(t,ν)2LiνB2.

for all (t,ν)[0,1]×B. Therefore all the assumptions (A1)(A2) and (B1)(B2) are satisfied and

nF(s)=nG(s)=s,1dsnF(s)+nG(s)=.

In addition, if EquationEquations (10) or (16) holds, then using Theorem 3.2 or Theorem 3.3, we can say that the problem (19) has a mild solution on [0,1].

5 Conclusion

We have obtained various types of existence results for a class of neutral stochastic integro-differential systems with state-dependent delays and noninstantaneous impulses in Hilbert spaces as a result of applying functional analysis and the stochastic analysis method. In future, we plan to improved system (1) through the use of stochastic processes driven by fractional Brownian motions or G-Brownian motions. These processes can be applied to more complex situations in order to improve the overall performance of the system. We also intend to integrate a numerical treatement with different criteria of our outcomes.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors received no direct funding for this research.

Notes on contributors

Mamadou Niang

Mamadou Niang and Mohamed Didiya arePh.D. student at the Gaston Berger University, Saint-Louis, sénégal. Their main research area is stochastic (hybrid) non- linear differential equations with delay.

Amadou Diop

Amadou Diop received the Ph.D. degree in applied mathematics from the Gaston Berger University, Saint-Louis, sénégal in August 2020. His research interest include stochastic differential equations.

Mamadou Abdoul Diop

Mamadou Abdoul Diop is presently working as a professor, Department of mathematics, University of Gaston Berger . He has vast experience in research and has done several publications in different peer-reviewed and reputed journal. His research interests include nonlinear analysis, control theory and stochastic differential equations.

Amadou Diop received the Ph.D. degree in applied mathematics from the Gaston Berger University, Saint-Louis, sénégal in August 2020. His research interest include stochastic differential equations.

References

  • Anguraj, A., Mallika Arjunan, M., & Hernandez, E. (2007). Existence results for an impulsive neutral functional differential equation with state-dependent delay. Applicable Analysis, 86(7), 861–22. https://doi.org/10.1080/00036810701354995
  • Baleanu, D., Jajarmi, A., Mohammadi, H., & Rezapour, S. (2020). A new study on the mathematical modelling of human liver with Caputo–Fabrizio fractional derivative. Chaos, Solitons, and Fractals, 134, 109705. https://doi.org/10.1016/j.chaos.2020.109705
  • Baleanu, D., Mohammadi, H., & Rezapour, S. (2020). Analysis of the model of HIV-1 infection of CD4+ T-cell with a new approach of fractional derivative. Advances in Difference Equations, 71. https://doi.org/10.1186/s13662-020-02544-w
  • Banas, J., & Goebel, K. (1980). Measure of noncompactness in banach spaces. In Lecture notes in pure and applied mathematics. New York: Marcel Dekker 60.
  • Chadha, A., & Pandey, D. N. (2015). Existence of mild solutions for a fractional equation with state-dependent delay via resolvent operators. Nonlinear Studies, 22, 71–85.
  • Chaudhary, R., & Pandey, D. N. (2019). Existence results for a class of impulsive neutral fractional stochastic integro-differential systems with state dependent delay. Stoch. Analysis and Applications, 37(5), 865–892. https://doi.org/10.1080/07362994.2019.1621181
  • Das, S., Pandey, D. N., & Sukavanam, N. (2016a). Approximate controllability of a second order neutral differential equation with state dependent delay. Differential Equations and Dynamical Systems, 24(2), 201–214. https://doi.org/10.1007/s12591-014-0218-6
  • Das, S., Pandey, D. N., & Sukavanam, N. (2016b). Approximate controllability of a second-order neutral stochastic differential equation with state dependent delay. Nonlinear Analysis: Modelling and Control, 21(6), 751–769. https://doi.org/10.15388/NA.2016.6.2
  • Deimling, K. (1985). Nonlinear functional analysis. Springer-Verlag.
  • Deng, S., Shu, X.-B., & Mao, J. (2018). Existence and exponential stability for impulsive neutral stochastic functional differential equations driven by fBm with noncompact semigroup via Mönch fixed point. Journal of Mathematical Analysis and Applications, 467(1), 398–420. https://doi.org/10.1016/j.jmaa.2018.07.002
  • Desch, W., Grimmer, R., & Schappacher, W. (1984). Some considerations for linear integrodifferential equations, J. Mathematical Analysis and Applications, 104(1), 219–234. https://doi.org/10.1016/0022-247X(84)90044-1
  • Dineshkumar, C., Udhayakumar, R., Vijayakumar, V., & Nisar, K. S. (2021). Results on approximate controllability of neutral integrodifferential stochastic system with state-dependent delay. Numerical Methods for Partial Differential Equations, 237, 1–15. https://doi.org/10.1002/num.22698
  • Diop, A., Diop, M. A., Ezzinbi, K., & Guindo, P. D. A. (2022). Optimal controls problems for some impulsive stochastic integro-differential equations with state-dependent delay. In Stochastics, 1–35. https://doi.org/10.1080/17442508.2022.2029446
  • Diop, A., Frederico, G. S., & Sousa, J. (2022). On controllability for a class of multi-term time-fractional random differential equations with state-dependent delay. Annals of Functional Analysis, 13(2), 1–23. https://doi.org/10.1007/s43034-022-00165-w
  • Dos Santos, J. P. C., Cuevas, C., & de Andrade, B. (2011). Existence results for a fractional equation with state-dependent delay. Advances in Difference Equations, 2011, 642013. https://doi.org/10.1155/2011/642013
  • Dos Santos, J. P. C., Mallika Arjunan, M., & Cuevas, C. (2011). Existence results for fractional neutral integro-differential equations with state-dependent delay. Computers & Mathematics with Applications, 62(3), 1275–1283. https://doi.org/10.1016/j.camwa.2011.03.048
  • Grimmer, R. C. (1982). Resolvent operators for integral equations in a Banach space, T. Transactions of the American Mathematical Society, 273(1), 333–349. https://doi.org/10.1090/S0002-9947-1982-0664046-4
  • Guo, Y., Chen, M., Shu, X.-B., & Fei, X. (2020). The existence and hyers-Ulam stability of solution for almost periodical fractional stochastic differential equation with fBm. Stochastic Analysis and Applications. https://doi.org/10.1080/07362994.2020.1824677
  • Guo, Y., Shu, X. B., Li, Y., & Xu, F. (2019). The existence and Hyers–Ulam stability of solution for an impulsive Riemann–Liouville fractional neutral functional stochastic differential equation with infinite delay of order 1 < β < 2. Boundary Value Problems, (1), 1–18. https://doi.org/10.1186/s13661-019-1172-6
  • Hale, J. K., & Kato, J. (1978). Phase space for retarded equations with infinite delay. Funkcialaj Ekvacioj, 21, 11–41.
  • Hernández, E., Azevedo, K. A. G., & O’Regan, D. (2018). On second order differential equations with state-dependent delay. Applicable Analysis, 97(15), 2610–2617. https://doi.org/10.1080/00036811.2017.1382685
  • Hernandez, E., & O’Regan, D. (2013). On a new class of abstract impulsive differential equations. Proceedings of the American Mathematical Society, 141(5), 1641–1649. https://doi.org/10.1090/S0002-9939-2012-11613-2
  • Hernandez, E., Prokopczyk, A., & Ladeira, L. (2006). A note on partial functional differential equations with state-dependent delay. Nonlinear Analysis: Real World Applications, 7(4), 510–519. https://doi.org/10.1016/j.nonrwa.2005.03.014
  • Hernández, E., Prokopczyk, A., & Ladeira, L. (2006). A note on partial functional differential equations with state-dependent delay. Nonlinear Analysis: Real World Applications, 7(4), 510–519. https://doi.org/10.1016/j.nonrwa.2005.03.014
  • Hernández, E., Wu, J., & Chadha, A. (2020). Existence, uniqueness and approximate controllability of abstract differential equations with state-dependent delay. Journal of Differential Equations, 269(10), 8701–8735. https://doi.org/10.1016/j.jde.2020.06.030
  • Hernández, E., Wu, J., & Fernandes, D. (2020). Existence and uniqueness of solutions for abstract neutral differential equations with state-dependent delay. Applied Mathematics & Optimization, 81(1), 89–111. https://doi.org/10.1007/s00245-018-9477-x
  • Hino, Y., Murakami, S., & Naito, T. (1991). Functional differential equations with infinite delay, lecture notes in mathematics. Springer-Verlag.
  • Huang, H., Wu, Z., Hu, L., Wei, Z., & Wang, L. (2018). Existence and controllability of second-order neutral impulsive stochastic evolution integro-differential equations with state-dependent delay. Journal of Fixed Point Theory and Applications, 20(1), 1–27. https://doi.org/10.1007/s11784-018-0484-y
  • Liang, J., Liu, J. H., & Xiao, T. J. (2008). Nonlocal problems for integrodifferential equations. dynamics of continuous. Discrete & Impulsive Systems. Series A, 15(6), 815–824.
  • Ma, X., Shu, X. B., & Mao, J. Z. (2020). Existence of almost periodic solutions for fractional impulsive neutral stochastic differential equations with infinite delay. Stochastics and Dynamics, 20(1), 31. https://doi.org/10.1142/S0219493720500033
  • Prato, G. D., & Zabczyk, J. (1992). Stochastic equations in infinite dimensions, encyclopedia of mathematics and its applications (Vol. 44). Cambridge University Press.
  • Sakthivel, R., & Anandhi, E. R. (2010). Approximate controllability of impulsive differential equations with state-dependent delay. International Journal of Control, 83(2), 387–393. https://doi.org/10.1080/00207170903171348
  • Sakthivel, R., & Ren, Y. (2013). Approximate controllability of fractional differential equations with state-dependent delay. Results Maths, 63(3–4), 949–963. https://doi.org/10.1007/s00025-012-0245-y
  • Suganya, S., Baleanu, D., Kalamani, P., & Mallika Arjunan, M. (2015). On fractional neutral integro-differential systems with state-dependent delay and non-instantaneous impulses. Advances in Difference Equations, 1–39. https://doi.org/10.1186/s13662-015-0709-y
  • Suganya, S., Mallika Arjunan, M., & Trujillo, J. J. (2015). Existence results for an impulsive fractional integro-differential equation with state-dependent delay. Applied Mathematics and Computation, 266, 54–69. https://doi.org/10.1016/j.amc.2015.05.031
  • Thiagu, K. (2021). On approximate controllability of second order fractional impulsive stochastic differential system with nonlocal, state-dependent delay and Poisson jumps. American Journal of Applied Mathematics, 9(2), 52–63. https://doi.org/10.11648/j.ajam.20210902.13
  • Tuan, N., Mohammadi, H., & Rezapour, S. (2020). A mathematical model for COVID-19 transmission by using the Caputo fractional derivative. Chaos, Solitons, and Fractals, 140, 110107. https://doi.org/10.1016/j.chaos.2020.110107
  • Valliammal, V., Ravichandran, C., Hammouch, Z., & Baskonus, H. M. (2019). A new investigation on fractional-ordered neutral differential systems with state-dependent delay. International Journal of Nonlinear Sciences and Numerical Simulation, 20(7–8), 803–809. https://doi.org/10.1515/ijnsns-2018-0362
  • Vijayakumar, V., Ravichandran, C., & Murugesu, R. (2013). Approximate controllability for a class of fractional neutral integro-differential inclusions with state-dependent delay. Nonlinear Studies, 20, 511–530.
  • Yan, Z., & Lu, F. (2018). Solvability and optimal controls of a fractional impulsive stochastic partial integro-differential equation with state-dependent delay. Acta Applicandae Mathematicae, 155(1), 57–84. https://doi.org/10.1007/s10440-017-0145-y