Abstract
We study the existence and uniqueness of mild and classical solutions for a general class of abstract impulsive differential equations with state-dependent impulses. Some examples on partial differential equations are presented.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this paper, we study the existence and uniqueness of mild and classical solutions for a class of abstract impulsive differential equations of the form
where \(A:D(A)\subset X\mapsto X\) is the generator of an analytic semigroup of bounded linear operators \((T(t))_{t\ge 0}\) on a Banach space \((X,\parallel \cdot \parallel )\), \(0=t_{0}<t_{1}< t_{2}<\cdots<t_{N}< t_{N+1} =a\) are pre-fixed numbers and \(f(\cdot )\), \(g_{i}(\cdot )\), \(\sigma _{i}(\cdot )\), \( i=1,\ldots , N,\) are functions specified be later.
The study of state-dependent delay equations is motivated by applications and theory. Related ODEs on finite dimensional spaces we cite the early works by Driver [9, 10] and Aiello et al. [1], the survey by Hartung, Krisztin et al. [15], the papers by Hartung et.al. [16,17,18] and the references in these works. For the case PDEs and abstract differential equations with state-dependent delay, we mention [19, 20, 26, 36,37,38] and the recent interesting works by Krisztin and Rezounenko [25], Yunfei et al. [33], Kosovalic et al. [26, 27] and Hernandez et al. [24].
Concerning the theory of impulsive differential equations, their motivations and relevant developments, we cite the books by Bainov and Covachev [2], Lakshmikantham et al. [28], Samoilenko and Perestyuk [40] for the case of ordinary differential equations on finite dimensional space and Benchohra et al. [7] for abstract differential equations and partial differential equations. In addition, we cite the interesting papers [8, 11, 20,21,22,23, 29, 31, 34, 39, 43] and the references therein. Related differential equations with impulse at state-dependent moments and state-dependent delayed impulses, we refer the reader to [3,4,5,6, 13, 14, 30, 41].
Our work is motivated by the papers Hakl et al. [14] related partial differential equations with impulse at state-dependent moments and Li and Wu [30] on differential equations with state-dependent delayed impulses. Specifically, we study the existence and “uniqueness” of solutions for the problem (1.1)–(1.3) which is a highly not trivial problem since functions of the form \(u\mapsto u(\zeta (\cdot ,u(\cdot ))) \) are (in general) nonlinear and not Lipschitz on space of continuous or sectionally continuous functions. By noting that
when the involved functions are Lipschitz, we study the existence of solutions on spaces of sectionally Lipschitz functions, a hard problem in the semigroup framework and in the general field of partial differential equations. In addition, we note that the Lipschizianity of \(T(\cdot ) g_{i}(u(\sigma _{i}(u(t_{i}^{+}))))\) not depend on the Lipschizianity of \(g_{i}(\cdot )\) and \(u(\cdot )\), which introduce a extra difficulty in our studies.
This paper has four sections. The existence and uniqueness of a classical solution via the contraction mapping principle is proved in Theorems 2.1, 2.2 and Proposition 2.3. In Theorem 2.3 we prove the existence of a mild solution using the Schauder’s fixed point Theorem. The particular case in which \(\sigma _{i}(\cdot )\) and (or) \(\zeta (\cdot )\) have values in \([-p,0]\), is studied in Propositions 2.1 and 2.2. In the last section some examples on partial differential equations are presented.
We include now some notations and results used in this work. Let \((Z,\parallel \cdot \parallel _{Z})\) and \((W,\parallel \cdot \parallel _{W})\) be Banach spaces. We denote by \({\mathcal {L}}(Z,W)\) the space of bounded linear operators from Z into W endowed with operator norm denoted by \(\parallel \cdot \parallel _{{\mathcal {L}}(Z,W)} \) and we write \({\mathcal {L}}(Z)\) and \(\parallel \cdot \parallel _{{\mathcal {L}}(Z)} \) if \(Z = W\). Moreover, if \(X=Z=W\) we write simply \(\parallel \cdot \parallel \) for the norms \(\parallel \cdot \parallel _{ X} \) and \(\parallel \cdot \parallel _{{\mathcal {L}}(X)} \). In addition, \(B_{r}(z,Z)=\{y\in Z:\parallel y-z\parallel _{Z}\le r\}\).
Let \(J\subset {\mathbb {R}}\) be a bounded interval. The spaces C(J, Z) and \(C_\mathrm{Lip}(J,Z)\) and their norms denoted by \(\parallel \cdot \parallel _{C (J,Z)}\) and \(\parallel \cdot \parallel _{C_\mathrm{Lip}(J,Z)}\) are the usual. We only note that \(\parallel \cdot \parallel _{C_\mathrm{Lip}(J;Z)}\) is given by \(\parallel \cdot \parallel _{C_\mathrm{Lip}(J;Z)} =\parallel \cdot \parallel _{C(J;Z)} +[ \cdot ]_{C_\mathrm{Lip}(J;Z)} \) where \( [ \zeta ]_{C_\mathrm{Lip}(J;Z)} =\sup _{t,s\in J, t\ne s }\frac{\parallel \zeta (s)-\zeta (t)\parallel _{Z}}{\mid t-s\mid }\).
The notation \({{\mathcal {P}}{\mathcal {C}}}(Z)\) is used for the space formed by all the bounded functions \(u :[0, a] \mapsto Z\) such that \(u(\cdot )\) is continuous at \(t \ne t_{i},\) \( u(t_{i}^{-})= u(t_{i}) \) and \(u(t_{i}^{+}) \) exists for all \(i = 1, \ldots , N\), provided with the norm \(\parallel u\parallel _{{{\mathcal {P}}{\mathcal {C}}}(Z)}=\max _{i=0,1,\ldots ,N}\parallel u\parallel _{C((t_{i},t_{i+1}];Z)}\). In addition, \({{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(Z)\) represents the space of functions \(u\in {{\mathcal {P}}{\mathcal {C}}}(Z)\) such that \(u_{\mid _{(t_{i},_{i+1}]}}\in C_\mathrm{Lip}((t_{i},t_{i+1}];Z)\) for all \(i=0,1,\ldots t_{N+1}\), endowed with the norm \(\parallel u\parallel _{{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(Z)}=\max _{i=0,\ldots , N} \parallel u_{\mid _{(t_{i},t_{i+1}]}}\parallel _{C_\mathrm{Lip}((t_{i},_{i+1}];Z)}\).
We use the symbol \( {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(Z)\) for the set of all the functions \(u:[-p,a]\mapsto Z\) such that \( u_{\mid _{[-p,t_{1}]}}\in C([-p,t_{1}];Z)\) and \( u_{\mid _{[0,a]}}\in {{\mathcal {P}}{\mathcal {C}}}(Z)\). In addition, \({\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(Z)\) is the space formed by all the functions \(u:[-p,a]\mapsto Z\) such that \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(Z)\), \(u_{\mid _{[-p,0]}}\in {C}_\mathrm{Lip}([-p,0];Z)\) and \( u_{\mid _{[0,a]}}\in {\mathcal {P}}{\mathcal {C}}_\mathrm{Lip}(Z)\), endowed with the norm \(\parallel u\parallel _{{\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(Z)}=\max \{\parallel u_{\mid _{I_{i}}}\parallel _{C_\mathrm{Lip}( I_{i};Z)} : i=-1,0,\ldots , N\} .\)
For \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(Z)\) and \( i\in \{-1, 0, 1, \cdots , N\}\), we use the notation \({\tilde{u}}_{i} \) for the function \({\tilde{u}}_{i} \in C([t_{i},t_{i+1}];Z) \) given by \( {\widetilde{u}}_{i}(t) = u(t)\) for \(t\in (t_{i},t_{i+1}]\) and \( {\widetilde{u}}_{i}(t) =u(t_{i}^{+}) \) for \( t=t_{i}\). For \(B\subseteq {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(Z) \) and \( i\in \{-1,0, 1, \cdots , N\}\), \({\widetilde{B}}_{i}\) is the set \({\widetilde{B}}_{i}=\{ {\tilde{u}}_{i}: u\in B \}.\) We note the following Ascoli–Arzela type criteria.
Lemma 1.1
A set \(B\subseteq {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}(Z)}\) is relatively compact in \( {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}(Z)}\) if and only if each set \({\widetilde{B}}_{i}\) is relatively compact in \(C([t_{i},t_{i+1}],Z)\).
In this paper, \(X_{1}\) is the domain of A endowed with the norm \(\parallel x\parallel _{X_{1}}=\parallel x\parallel +\parallel Ax\parallel \) and \( C_{0}, C_{1}\) are positive constants such that \(\parallel AT(s)\parallel _{{\mathcal {L}}(X_{1},X)} \le C_{1}\), \(\parallel T(s)\parallel \le C_{0}\) and \(\parallel AT(t)\parallel \le \frac{ C_{1}}{t}\) for all \(s\in [0,a] \) and \(t\in (0,a]\).
Related the abstract Cauchy problem
we note that the function \(u\in C ([c,d];X) \) given by \( u(t) = T(t-c)x +\int _{c}^{t}T(t-s) \xi (s) {\text {d}}s\), is called mild solution of (1.4). In addition, a function \(v\in C ([c,d];X) \) is said to be a classical solution of (1.4) if \(v\in C^{1}((c,d];X) \cap C ((c,d];X_{1} ) \) and \(v(\cdot )\) satisfies (1.4) on (c, d].
2 Existence of solutions
In this section we present some results on the existence of solution for (1.1)–(1.3). To begin, we introduce the followings concepts of solution.
Definition 2.1
A function \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(X)\) is called a mild solution of the problem (1.1)–(1.3) if \(u_{0}=\varphi \), \( u(t_{i}^{+})= g_{i}(u(\sigma _{i}(u(t_{i}^{+}))))\) for all \( i=1,\ldots , N\) and
for all \(t\in (t_{i},t_{i+1}]\) and \(i=1,\ldots , N.\)
Definition 2.2
A function \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(X)\) is called a classical solution of (1.1)–(1.3) if \(u_{0}=\varphi \), \( u(t_{i}^{+})= g_{i}(u(\sigma _{i}(u(t_{i}^{+}))))\) for all \( i=1,\ldots , N\) and \(u(\cdot )\) satisfy (1.1).
In the remainder of this work, we assume that \((W,\parallel \cdot \parallel _{W})\) is Banach continuously embedded in \( (X,\parallel \cdot \parallel )\) such that \( A T(\cdot )\in L^{\infty }([0,a];{\mathcal {L}}(W,X))\). To prove our results, we introduce the following conditions.
- \({\mathbf{H}}_{\zeta }\) :
-
\(\zeta \in C_\mathrm{Lip}( [0,a]\times X ;[-p,a])\) and there is a function \(j:\{1,\ldots , N\}\mapsto \{-1,0,1,\ldots , N\}\) such that \(\zeta \in C_\mathrm{Lip}( I_{i}\times X ;I_{j(i)})\) and \(j(i)\le i\) for all \(i\in \{1,\ldots , N\}\).
- \({\mathbf{H}}_{\sigma _{{\mathbf{i}}} }\) :
-
There is a function \(q:\{1,\ldots , N\}\mapsto \{-1,0,1,\ldots , N\}\) such that \( q(i)\le i\) and \(\sigma _{i}\in C(X,I_{q(i)})\) for all \(i\in \{1,\ldots , N\}\). Next we write \([\sigma _{i}]_{C_\mathrm{Lip}}\) in place \([\sigma _{i}]_{C_\mathrm{Lip}( X;I_{q(i)})}\).
- \({\mathbf{H_{g,X}^{W}}}\) :
-
\(g_{i} \in C_\mathrm{Lip}(X; W)\) and \( {\mathcal {C}}_{X,W}(g_{i})= \parallel g_{i}\parallel _{ C(X; W)}<\infty \) for every \(i\in \{1,\ldots , N\}\). Next, \(L_{Z,W}(g_{i})\) denotes the Lipschitz constant of \(g_{i}(\cdot )\), \(L_{Z,W}(g)=\max _{i=1,\ldots , N}L_{Z,W}(g_{i})\) and \({\mathcal {C}}_{Z,W}(g)=\max _{i=1,\ldots , N} {\mathcal {C}}_{Z,W}(g_{i})\).
- \({\mathbf{H_{g}}}\) :
-
\(g_{i} \in C_\mathrm{Lip}(X;X)\) and \( {\mathcal {C}}_{X}(g_{i})= \parallel g_{i}\parallel _{ C(X; X)}<\infty \) for all \(i\in \{1,\ldots , N\}\). Next, \(L_{g_{i}}\) is the Lipschitz constant of \(g_{i}(\cdot )\), \(L_{g}=\max _{i=1,\ldots , N}L_{g_{i}}\) and \({\mathcal {C}}_{X}(g)=\max _{i=1,\ldots , N} {\mathcal {C}}_{X}(g_{i})\).
- \({\mathbf{H_{f}}} \) :
-
\(f\in C_\mathrm{Lip}([0,a]\times X; X)\) and \(C_{X}(f)= \parallel f\parallel _{ C([0,a]\times X; X)}<\infty \). Next, \(L_{f}\) denotes the Lipschitz constant of \(f(\cdot )\).
Notations 1
Next, for convenience, we write \([\zeta ]_{C_\mathrm{Lip}} \) in place \([\zeta ]_{C_\mathrm{Lip}( [0,a]\times X ; [-p,a])}\), \(b_{i}=t_{i+1}-t_{i}\), \(b=\max _{i= 1,\ldots , N }b_{i}\), \(i_{c}: W\mapsto X\) is the inclusion map and
The next useful result follows from the proof of [24, Lemma 1]. The proof is omitted.
Lemma 2.2
Assume that the conditions \({\mathbf{H}}_{\zeta }\), \({\mathbf{H}}_{\sigma _{{\mathbf{i}}}}\) are satisfied, \(u,v\in {\mathcal {B}}{\mathcal {P}}{\mathcal {C}}_\mathrm{Lip}(X)\) and \(u_{0}=v_{0} \). Then \( u(\zeta (\cdot ,u(\cdot ))) \in {\mathcal {P}}{\mathcal {C}}_\mathrm{Lip}(X) \) and
We can prove now our first result.
Theorem 2.1
Assume that the conditions \({\mathbf{\mathbf H}}_{\zeta }, {\mathbf{H}}_{\sigma _{{\mathbf{i}}} }\), \({\mathbf{H}}_{{\mathbf{g,X}}}^{{\mathbf{W}}}\) and \({\mathbf{H}}_{{\mathbf{f}}}\) are satisfied, \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X)\), \(\varphi \in C_\mathrm{Lip}([-p,0];X)\) and
Then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of the problem (1.1)–(1.3).
Proof
Let \(P:{\mathbb {R}}\mapsto {\mathbb {R}} \) be the polynomial given by
From (2.4) and noting that \( C_{0}b L_{f}(1+ [\zeta ]_{C_\mathrm{Lip}})+\Lambda _{X,W}L_{X,W}(g) <1\), we infer that \(P(\cdot )\) has a root \(R_{1}>0\) and there exists \(R>0\) such that \(P(R)<0\). From the definition of \(P(\cdot )\), we get
Let \({\mathcal {S}}(R)\) be the space \({\mathcal {S}}(R)= \{u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X) \, ;\, u_{0}=\varphi , \, [u_{\mid _{[0,a]}}]_{{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)}\le R\} ,\) endowed with the metric \(d(u,v)=\parallel u-v\parallel _{{\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X)} \) and \(\Gamma : {\mathcal {S}}(R) \mapsto {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X) \) be the map defined by \((\Gamma u)_{0} =\varphi \) and
for \(t\in (t_{i},t_{i+1}]\) and \(i=1,\ldots , N.\)
It’s easy to see that \({\mathcal {S}}(R)\) is closed in \({\mathcal {B}}{\mathcal {P}}{\mathcal {C}}(X) \) and that \(\Gamma (\cdot ) \) is well defined. Moreover, from Lemma 2.2, for \(i\in \{1,\ldots ,i\}\), \(t\in (t_{i},t_{i+1}) \) and \(h>0\) such that \(t+h\in (t_{i},t_{i+1}] \), we get
which implies that \( [ (\Gamma u)_{\mid _{I_{i}}} ]_{C_\mathrm{Lip}( I_{i};X)} \le \Phi _{X,W} + C_{0}b L_{f} [\zeta ]_{C_\mathrm{Lip}}( R+ R^{2})<R . \) In a similar way, we obtain that
From the above and noting that \( [\varphi ]_{C_\mathrm{Lip}([-p,0];X)}\le R\), we obtain that \( [ \Gamma u]_{ {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)}\le R\), which implies that \(\Gamma \) is a \({\mathcal {S}}(R)\)-valued function.
On the other hand, using (2.2), for \(u,v \in {\mathcal {S}}(R)\), \(i=1,\ldots , N\) and \(t\in (t_{i},t_{i+1}]\) we have that
In addition, for \(t\in [0,t_{1}]\) we note that \( \parallel \Gamma u (t)- \Gamma v(t)\parallel \le C_{0}bL_{f} (1+ R[\zeta ]_{C_\mathrm{Lip}}) d(u,v) .\) From the above estimates we infer that
Thus, \(\Gamma (\cdot )\) is a contraction and there exists a unique mild solution \(u\in {\mathcal {S}}(R)\) of (1.1)–(1.3).
We prove now that \(u(\cdot )\) is a classical solution. Let \({\widetilde{u}}_{i}\), \(i=1,\ldots N,\) be defined as in the introduction. It is easy to see that \({\widetilde{u}}_{i}(\cdot )\) is the mild solution of the problem
Since \( f(\cdot , u(\zeta (\cdot ,u(\cdot ))))\) is Lipschitz on \(I_{i}\) and the semigroup is analytic, from [35, Theorem 4.3.2] it follows that \({\widetilde{u}}_{i}\) is a classical solution of (2.8)–(2.9). The same argument prove that \({\widetilde{u}}_{0}\) is a classical solution of (2.8) on \([0,t_{1}]\) with initial condition \(u(0)=\varphi (0)\). From the above, we obtain that \(u(\cdot )\) is a classical solution of (1.1)–(1.3). \(\square \)
In the next result we establish the existence and uniqueness of a classical solution without to use condition \({\mathbf{H}}_{{\mathbf{g,X}}}^{{\mathbf{W}}}\). In place of this condition, we introduce the following one:
- \( {\mathbf{H}}_{{{g}}_{{{i}}},\sigma _{{{j}}}}\) :
-
\(\sigma _{i}\in C_\mathrm{Lip}(X,[-p,a])\) for all \(i\in \{1,\ldots , N\}\), \(\overline{\cup _{i=1}^{N}\sigma _{i}(X)}\subset \cup _{i=0}^{N}I_{i}\cup [-p,0]\), \(g_{i}\in C(X_{1};X_{1})\cap C_\mathrm{Lip}(X;X)\) and there are constants \(l_{g_{i}}, k_{g_{i}}\) such that \(\parallel Ag_{i}(x)\parallel \le l_{g_{i}}r+ k_{g_{i}} \) for all \(x\in B_{r}(0,X_{1})\), \(i\in \{1,\ldots , N\}\) and every \(r>0\).
Notations 2
If condition \( {\mathbf{H}}_{{{g}}_{{{i}}},\sigma _{{{j}}}}\) is verified, we use the notations \(l_{g}=\max _{i=1,\ldots }l_{g_{i}} \) and
Theorem 2.2
Assume that the conditions \({\mathbf{H}}_{\zeta } \), \( {\mathbf{H}}_{{{g}}_{{{i}}},\sigma _{{{j}}}}\), \({\mathbf{H}}_{\mathbf{g}}\) and \({\mathbf{H}}_{{\mathbf{f}}}\) are satisfied, X is a Hilbert space, A is self-adjoint, \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X)\), \(\varphi \in C_\mathrm{Lip}([-p,0];X)\cap C([-p,0];X_{1})\) and
Then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of the problem (1.1)–(1.3) such that \(A{\widetilde{u}}_{i} \in C([t_{i},t_{i+1}];X )\) for all \(i=1,\ldots , N \).
Proof
Let \(P:{\mathbb {R}}\mapsto {\mathbb {R}} \) be the polynomial given by
From (2.10) there exists \(R>0\) such that \(P(R)<0\) and
Let \({\mathcal {S}}(R)\) the space in the proof of Theorem 2.1 and \({\mathcal {S}}(\sigma _{i},R)\) be the space
endowed with the metric \(d(u,v)=\parallel u-v\parallel _{{{\mathcal {P}}{\mathcal {C}}} (X)} \). Let \(\Gamma :{\mathcal {S}}(\sigma _{i},R) \mapsto {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(X) \) be defined as in the proof of Theorem 2.1. Next we prove that \(\Gamma \) is a contraction on \({\mathcal {S}}(\sigma _{i},R)\).
Let \(u\in {\mathcal {S}}(\sigma _{i},R) \), \( i\in \{1,\ldots ,N\}\), \(t\in (t_{i},t_{i+1}) \) and \(h>0\) such that \(t+h\in (t_{i},t_{i+1}] \). Arguing as in the proof of Theorem 2.1 and noting that \(u(\sigma (u(t_{i}^{+})))\in X_{1}\), we see that
and hence, \( [ (\Gamma u)_{\mid _{I_{i}}} ]_{C_\mathrm{Lip}( I_{i};X)} \le \Upsilon + C_{0} l_{g_{i}}R+ C_{0}b L_{f} [\zeta ]_{C_\mathrm{Lip}}( R+ R^{2})\le R . \) In addition, it is easy to see that
From the above remarks we have that \( [ (\Gamma u)_{\mid _{[0,a]}} ]_{ {{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)}\le R\) which shows that \(\Gamma u\in {\mathcal {S}}(R) \). In addition, arguing as in the proof of Theorem 2.1 it follows that
From the above remarks, we have that \(\Gamma \) is a contraction on \( {\mathcal {S}}(R)\).
Next we show that \( \parallel A\Gamma u(t)\parallel \le R \) for all \(t\in \cup _{j=1}^{N}\sigma _{j}(X)\). Let \(t\in \cup _{j=1}^{N}\sigma _{j}(X)\) and assume that \(t\in (t_{i},t_{i+1}] \) for \(i\ge 1\). Using that \((T(t))_{t\ge 0}\) is analytic and that \(u(\sigma (u(t_{i}^{+})))\in X_{1}\) and \( \parallel A u(\sigma (u(t_{i}^{+}))) \parallel \le l_{g_{i}}R +k_{g_{i}} \), we note that
which implies that \( \parallel A\Gamma u(t)\parallel \le \Upsilon + C_{0} l_{g_{i}}R + C_{1}b L_{f} [\zeta ]_{C_\mathrm{Lip}}( R+ R^{2})\le R . \) If \(t\in I_{1}\) we see that
Thus, \( \parallel A\Gamma u(t)\parallel \le R\) for all \(t\in \cup _{i=1}^{N}\sigma _{i}(X)\) and \(\Gamma \) is a \({\mathcal {S}}(\sigma _{i},R)\)-valued function.
To finish the proof, we prove that \({\mathcal {S}}(\sigma _{i},R) \) is a closet subset of \( {\mathcal {S}}(R)\). Let \((u_{n})_{n\in {\mathbb {N}}}\) be a sequence in \({\mathcal {S}}(\sigma _{i},R) \) and \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X) \) such that \(u_{n}\rightarrow u\) as \(n\rightarrow \infty \). Let \(t\in \cup _{i=1}^{N}\sigma _{i}(X)\). Since \((Au_{n}(t))_{n\in {\mathbb {N}}}\) is bounded, there exists \(w\in X\) such that \(<Au_{n}(t), z>\rightarrow <w, z> \) as \(n\rightarrow \infty \) for all \(z\in X\). In particular, for \(v\in X_{1}\) we have that \(<Au_{n}(t),v>=<u_{n}(t),Av>\rightarrow <u(t),Av>\) as \(n\rightarrow \infty \), which implies that \(<w, v> = <u(t),Av> \) for all \(v\in X_{1}\). Using that A is self-adjoint, we obtain that \(u(t)\in X_{1} \), \(Au(t)=w\) and \(\parallel Au(t)\parallel =\parallel w\parallel \le \liminf _{n\rightarrow \infty } \parallel Au_{n}(t)\parallel \le R\), which completes the proof that \( {\mathcal {S}}(\sigma _{i},R)\) is closed.
From the above it follows that \(\Gamma \) is a contraction on \( {\mathcal {S}}(\sigma _{i},R)\) and there exists a unique mild solution \(u\in {\mathcal {S}}(\sigma _{i},R)\). The fact that \(u(\cdot )\) is a classical solution follows from the proof of Theorem 2.1. \(\square \)
The next result consider the case where \(\sigma _{i}(X)\subset [-p,0]\) for all \(i=1,\ldots ,N\). The proof use the ideas in the proof of Theorem 2.1 and we include a short proof for completeness.
Proposition 2.1
Let conditions \({\mathbf{H}}_{\mathbf{g}} \) and \({\mathbf{H}}_{\mathbf{f}}\) be holds. Assume \(\zeta \in C_\mathrm{Lip}([0,a]\times X;[-p,a])\), \(\sigma _{i}\in C_\mathrm{Lip}(X;[-p,0])\) for all \(i=1,\ldots ,N\), \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X)\), \(\varphi \in C_\mathrm{Lip}([-p,0];X)\), \(g_{i}(\varphi (\cdot ))\in C([-p,0];W)\) for all \(i=1,\ldots ,N\) and
where \( { \Phi _{X,W,\varphi }} = \Phi _{X,W}\max _{i=1,\ldots ,N}\parallel g_{i}(\varphi (\cdot ))\parallel _{C([-p,0];W)} + C_{0} ( {\mathcal {C}}_{X}(f) + bL_{f}) + [\varphi ]_{C_\mathrm{Lip}([-p,0];X)} \) \(+[T(\cdot )\varphi (0)]_{C_\mathrm{Lip}([-p,0];X)} \) and \(\Psi _{\varphi ,\sigma _{i},g_{i}}= C_{0}[\varphi ]_{C_\mathrm{Lip}([-p,0];X)}\max _{i=1,\ldots , N}[\sigma _{i}]_{C_\mathrm{Lip}}\). Then there exists a unique classical solution \(u \in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of the problem (1.1)–(1.3).
Proof
Let \(P:{\mathbb {R}}\mapsto {\mathbb {R}} \) be given by \( P(x) = \Phi _{X,W,\varphi } + (C_{0}b L_{f}(1+ [\zeta ]_{C_\mathrm{Lip}})+ L_{g}\Psi _{\varphi ,\sigma _{i},g_{i}}-1)x + C_{0}bL_{f} [\zeta ]_{C_\mathrm{Lip}} x^{2} . \) From (2.15) there exists \(R>0\) such that \(P(R)<0\). Let \({\mathcal {S}}(R)\) be defined as in the proof of Theorem 2.1 and \(\Gamma : {\mathcal {S}}(R) \mapsto {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X) \) be the map given by \(\Gamma u_{0} =\varphi \) and
Arguing as in the proof of Theorem 2.1, for \(i\in \{1,\ldots ,i\}\), \(t\in (t_{i},t_{i+1}) \) and \(h>0\) such that \(t+h\in (t_{i},t_{i+1}] \), it is easy to see that
which implies (from the definition of \(P(\cdot )\)) that \( [ (\Gamma u)_{\mid _{I_{i}}} ]_{C_\mathrm{Lip}( I_{i};X)} \le R . \) Similarly, we have that
From the above, \( [ (\Gamma u)_{\mid _{[0,a]}} ]_{ {{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)}\le R,\) which proves that \(\Gamma \) is a \({\mathcal {S}}(R)\)-valued function.
On the other hand, for \(u,v \in {\mathcal {S}}(R)\), \(i=1,\ldots , N\), \(t\in (t_{i},t_{i+1}]\) and \(s\in [0,t_{1}]\) we get
which allows us infer that \(\Gamma \) is a contraction and there exists a unique mild solution \(u\in {\mathcal {S}}(R)\) of the problem (1.1)–(1.3). The fact that \(u(\cdot )\) is a classical solution follows from the proof of Theorem 2.1. \(\square \)
In the next result, we assume that the functions \(\zeta (\cdot )\) and \(\sigma _{i}(\cdot )\) have values in \([-r,0]\).
Proposition 2.2
Suppose that the conditions \({\mathbf{H}}_{\mathbf{g}}\), \({\mathbf{H}}_{{\mathbf{f}}}\) are satisfied, \(\varphi \in C_\mathrm{Lip}([-p,0];X)\), \(\sigma _{i}\in C_\mathrm{Lip}(X;[-p,0])\) for all \(i=1,\ldots ,N\), \(\zeta \in C_\mathrm{Lip}([0,a]\times X;[-p,0])\) and
Then there exists a unique mild solution \(u\in {{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of (1.1)–(1.3).
Proof
Let \(\Gamma : {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X) \mapsto {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X) \) be defined as in the proof of Theorem 2.1, but using \(f(\tau ,\varphi (\zeta (\tau ,u(\tau ))))\) in place \( f(\tau ,u(\zeta (\tau ,u(\tau )))) \). In this case, for \(u,v\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X) \) we see that
which allows us to conclude that \(\Gamma \) is a contraction. \(\square \)
Next, we discuss briefly the case in which the functions \(f(\cdot )\) and \(g_{i}(\cdot )\) are locally bounded and (or) locally Lipschitz. For sake of clarity, we include the next conditions.
- \( {\mathcal {H}}_{g, X}^{W}\) :
-
For all \(i=1,\ldots ,N\), there is \({L}_{X,W}(g_{i},\cdot )\in C({\mathbb {R}}; {\mathbb {R}})\) such that \(\parallel g_{i}(x)-g_{i}(y)\parallel _{W}\le {L}_{X,W}(g_{i},r)\parallel x-y\parallel \) for all \(x,y\in B_{r}(0,X)\) and every \(r>0\). Next, \({L}_{X,W}(g,r)= \max _{i=1,\ldots , N} L_{X,W}(g_{i},r)\) and \({\mathcal {C}}_{X,W}(g_{i},r)=\parallel g_{i}\parallel _{C(B_{r}(0,X);W)} \).
- \( {\mathcal {H}}_{f}\) :
-
There is \({L}_{f} \in C({\mathbb {R}}; {\mathbb {R}})\) such that \(\parallel f(t,x)-f(s,y)\parallel \le {L}_{f}(r)(\mid t-s\mid + \parallel x-y\parallel ) \) for all \(x,y\in B_{r}(0,X)\), \(t,s\in [0,a]\) and \(r>0\). Next, for \(r>0\) we use the notation \({\mathcal {C}}_{X}(f,r)=\parallel f\parallel _{C([0,a]\times B_{r}(0,X);X)} \).
- \({\mathcal {H}}_{\mathbf{g}}\) :
-
There are functions \({L}_{g_{i}} \in C({\mathbb {R}}; {\mathbb {R}})\) such that \(\parallel g_{i}(x)-g_{i}(y)\parallel \le {L}_{g_{i}}(r)\parallel x-y\parallel \) for all \(x,y\in B_{r}(0,X)\) and \(r>0\). Next, \(L_{g}(r){=}\max _{i=1,\ldots , N}L_{g_{i}}(r)\), \({\mathcal {C}}_{X}(g)(r)=\max _{i=1,\ldots , N} {\mathcal {C}}_{X}(g_{i})(r)\) and \({\mathcal {C}}_{X}(g_{i},r)=\parallel g_{i}\parallel _{C(B_{r}(0,X);X)} \).
Notations 3
For \(r>0\), we define \( \Phi _{X,W}(r) = \Lambda _{X,W} {\mathcal {C}}_{X,W}(g,r) + C_{0}( C_{X}(f,r) + b L_{f}(r)) + [T(\cdot )\varphi (0)]_{C_\mathrm{Lip}([-p,0];X)} + [\varphi ]_{C_\mathrm{Lip}([-p,0];X)} .\)
The proof of Proposition 2.3 follows from the proof of Theorem 2.1.
Proposition 2.3
Let conditions \({\mathbf{H}}_{\zeta }, {\mathbf{H}}_{\sigma _{{\mathbf{i}}} }\), \( {\mathcal {H}}_{g,X}^{W}\) and \( {\mathcal {H}}_{f}\) be holds. Suppose that \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X)\), \(\varphi \in C_\mathrm{Lip}([-p,0];X)\) and there is \(r>0 \) such that (2.4) is satisfied with \( L_{f}(r),\) \( \Phi _{X,W} (r)\) and \( L_{X,W}(g,r)\) in place \( L_{f}, \Phi _{X,W}\) and \( L_{X,W}(g)\), and
Then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of the problem (1.1)–(1.3).
Proof
Let \(P:{\mathbb {R}}\mapsto {\mathbb {R}} \) be defined as in the proof of Theorem 2.1, but using \( L_{f}(r), \Phi _{X,W} (r)\) and \( L_{X,W}(g,r)\) in place \( L_{f}, \Phi _{X,W}\) and \( L_{X,W}(g)\). Arguing as in the proof of Theorem 2.1 we infer that there exists \(R>0\) such that
Let \( {\mathcal {S}}(R)\) be the space in the proof of Theorem 2.1 and \( {\mathcal {S}}(r,R)= \{u\in {\mathcal {S}}(R) : \, \parallel u \parallel _{ {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X)}\le r \} ,\) endowed with the metric \(d(u,v)=\parallel u-v\parallel _{{\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X)} \). Let \(\Gamma : {\mathcal {S}}(r,R) \mapsto {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X) \) be defined as in the proof of Theorem 2.1.
From the proof of Theorem 2.1 we infer that \(\Gamma \) is a contraction on \( {\mathcal {S}}( R)\). Moreover, for \(t\in I_{i}\) with \(i\ge 0\) it is easy to see that
which implies that \(\parallel \Gamma u \parallel _{ {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X)}\le r \) since \( r>\parallel \varphi \parallel _{C([-p,0];X)}\). Thus, \(\Gamma \) is a contraction on \( {\mathcal {S}}( r,R)\) and there exists a unique mild solution \(u\in {\mathcal {S}}(r,R)\) of (1.1)–(1.3). Finally, from [35, Theorem 4.3.2] we infer that \(u(\cdot )\) is a classical solution. \(\square \)
Corollary 2.1
Assume that the conditions \({\mathbf{H}}_{\zeta }, {\mathbf{H}}_{\sigma _{{\mathbf{i}}} }\), \( {\mathcal {H}}_{g,X}^{W}\) and \( {\mathcal {H}}_{f}\) are satisfied, the functions \(L_{f}(\cdot ),\) \({\mathcal {C}}_{X}(f,\cdot )\), \(L_{X,W}(g,\cdot )\) and \({\mathcal {C}}_{X,W}(g,\cdot )\) are non-decreasing, \(\varphi \in C_\mathrm{Lip}([-p,0];X)\), \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X)\), \( \limsup _{r\rightarrow \infty }\frac{1}{r} C_{0} (\parallel i_{c}\parallel _{{\mathcal {L}}(W,X)} {\mathcal {C}}_{X,W}(g,r) + b {\mathcal {C}}_{X}(f,r)) < 1\) and
Then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\cap B_{r}(0, {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X))\) of (1.1)–(1.3).
We establish now, without proof, a result similar to Theorem 2.2 for the case where \(f(\cdot )\) satisfy the condition \({\mathcal {H}}_{f}\).
Proposition 2.4
Suppose the conditions \({\mathbf{H}}_{\zeta } \), \({\mathbf{H}}_{g_{i},\sigma _{j}}\), \( {\mathcal {H}}_{g}\) and \( {\mathcal {H}}_{f}\) be holds, X is a Hilbert space, A is self-adjoint, \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X) \) and \(\varphi \in C_\mathrm{Lip}([-p,0];X)\cap C([-p,0];X_{1})\). If there is \(r>0\) such that the inequality (2.10) is valid with \(L_{f}(r)\), \(L_{g}(r)\), \({\mathcal {C}}_{X}(f,r) \) and \({\mathcal {C}}_{X}(g,r)\) in place \(L_{f}\), \(L_{g}\), \({\mathcal {C}}_{X}(f)\) and \({\mathcal {C}}_{X}(g)\), and \( C_{0}( \max \{ \parallel \varphi (0)\parallel ,{\mathcal {C}}_{X}(g,r)\} + b {\mathcal {C}}_{X}(f,r))\le r,\) then there exists a unique classical solution \(u \in B_{r}(0,{\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}} (X))\cap {{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X))\) of (1.1)–(1.3).
To complete this section, we study the existence of solution using the Schauder’s fixed point Theorem. The next lemma follows from the proof of [32, Proposition 4.2.1].
Lemma 2.3
Let \(\alpha \in (0,1)\), \(\xi \in L^{\infty }([b,c];X) \) and \(v:[b,c]\mapsto X\) be the function defined by \(v(t)=\int _{b}^{t}T(t-s)\xi (s){\text {d}}s\). Then \( { [v]_{ C^{\alpha }([b,c];X) } } \le \parallel \xi \parallel _{L^{\infty }([b,c];X)}({ (c-b)^{1-\alpha }}C_{0}+ \frac{C_{1}}{\alpha (1-\alpha )}) \).
Theorem 2.3
Assume that the conditions \({\mathbf{H}}_{\zeta } \) and \({\mathbf{H}}_{\sigma _{{\mathbf{i}}} }\) are satisfied, there is a Banach space \((Y,\parallel \cdot \parallel _{Y}) \hookrightarrow (X,\parallel \cdot \parallel )\) such that \(\parallel T(t)- I\parallel _{{\mathcal {L}}(Y,X)} \rightarrow 0\) as \(t\rightarrow 0\), \(g_{i} \in C(X;Y)\) for all i, \(f\in C([0,a]\times X;X)\), the functions \(g_{i}(\cdot ),f(\cdot )\) are bounded and \((T(t))_{t\ge 0}\) is compact. Then there exists a mild solution of the problem (1.1)–(1.3).
Proof
Let \({\mathcal {C}}_{X,Y}(g)=\max _{i=1,\ldots , N}\parallel g_{i}\parallel _{ C(X; Y)}\), \(C_{X}(f)= \parallel f\parallel _{ C([0,a]\times X; X)}\) and \(\alpha \in (0,1)\). Let \( {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X)=\{ u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(X):\, u_{0}=\varphi \}\) endowed with the metric \(d(u,v)=\parallel u- v\parallel _{{{\mathcal {B}}{\mathcal {P}}{\mathcal {C}}}(X)}\) and \(\Gamma : {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \mapsto {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(X) \) be defined as in the proof of Theorem 2.1.
It is easy to prove that \(\Gamma \) is continuous. Next, using Lemma 1.1, we show that \(\Gamma \) is completely continuous.
Let \(i\in \{1,\ldots , N\}\). From Lemma 2.3, for \(t\in (t_{i},t_{i+1})\), \(h>0\) with \(t+h\in (t_{i},t_{i+1}]\), we get
which shows that \(\{(\Gamma u)_{\mid _{I_{i}}} :u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \}\) is right equicontinuous at \(t\in (t_{i},t_{i+1})\). A similar argument prove that \(\{(\Gamma u)_{\mid _{I_{i}}} :u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \}\) is left equicontinuous at \(t=t_{i+1}\), which implies that \(\{(\Gamma u)_{\mid _{I_{i}}} :u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \}\) is equicontinuous on \(I_{i}\). In addition, for \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X)\) and \(0<h<\delta \) we note that
which proves that \( \widetilde{ \Gamma {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) }_{i}= \{(\widetilde{ \Gamma u})_{i} : u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \} \) is right equicontinuous at \(t_{i}\). From the above it follows that \(\{\widetilde{(\Gamma u)}_{{i}} :u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \}\) is equicontinuous on \(I_{i}\).
We prove now that \(\{(\widetilde{ \Gamma u})_{i} (t) :u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \}\) is relatively compact in X for all \(t\in [t_{i},t_{i+1}]\). Since the semigroup is compact, \((Y,\parallel \cdot \parallel _{Y}) \hookrightarrow (X,\parallel \cdot \parallel )\) and \(g_{i}(\cdot )\) is bounded with values in Y, we have that \(U= \{ g_{j}(u(\sigma _{j}(u(t_{j}^{+})))):u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X),\,j=1,\ldots ,N\} \) is relatively compact in X. For \(t\in (t_{i}, t_{i+1}]\) and \(0<\varepsilon <t-t_{i}\), we note that
and hence, \(\{(\widetilde{ \Gamma u})_{i} (t) :u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) \} \subset K_{\varepsilon }+ D_{\varepsilon },\) where \(K_{\varepsilon }\) is relatively compact and the diameter of \(D_{\varepsilon }\) converges to zero as \(\varepsilon \rightarrow 0\). This prove that the set \( \Gamma {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) (t) \) is relatively compact in X. Moreover, since \( \widetilde{ \Gamma {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) }_{i} (t_{i})= \{ g_{i}(u(\sigma _{i}(u(t_{i}^{+})))):u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X)\} \subset {\overline{U}}, \) we obtain that \( \widetilde{ \Gamma {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) }(t_{i})\) is relatively compact in X. From the above remarks we have that \( (\widetilde{ \Gamma {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) })_{i}\) is relatively compact in \(C([t_{i},t_{i+1}];X)\). Moreover, the same argument also prove that \( (\widetilde{\Gamma {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X)})_{1} =\{ (\Gamma u)_{\mid _{[0,t_{1}]}} : u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X))\}\) is relatively compact in \( C([0,t_{1}];X)\).
From the above and Lemma 1.1, it follows that \(\Gamma \) is completely continuous and noting that the functions \(f(\cdot ) \) and \( g_{i}(\cdot )\) are bounded, we infer that there exists \(r>0\) such that \( \Gamma ( {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X)) \subset B_{r}( 0, {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) )\). Thus, \(\Gamma \) is completely continuous from \( B_{r}(0, {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) )\) into \( B_{r}(0, {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) ) \) and there exists a mild solution \(u\in B_{r}(0, {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_{\varphi }(X) )\) of (1.1)–(1.3). \(\square \)
3 Examples
In this section, \(X= L^{2}(\Omega ;{\mathbb {R}})\) or \(X= C(\Omega ;{\mathbb {R}})\), \(\Omega \subset {\mathbb {R}}^{n}\) is a open set with smooth boundary and \(A:D(A)\subset X\mapsto X\) is the realization of an second order strongly elliptic operator. Next, we assume that \((T(t))_{t\ge }\) is the analytic semigroup generated by A, \(D(A)=\{u\in L^{2}(\Omega ):Au\in L^{2}(\Omega )\}\) if \(\Omega ={\mathbb {R}}^{n}\) and \(D(A)=W^{2,2}(\Omega )\cap W^{2,1}_{0}(\Omega )\) if \(\Omega \) is bounded. For sake of simplicity, we suppose that the conditions \({\mathbf{H}}_{\zeta }\) and \({\mathbf{H}}_{\sigma _{{\mathbf{i}}} }\) are satisfies, \(0\in \rho (A)\), \(\varphi \in C_\mathrm{Lip}([-p,0];X)\) and \(T(\cdot )\varphi (0)\in C_\mathrm{Lip}([0,a];X)\). In addition, \(X_{1}\) is the domain of A endowed with the norm \(\parallel x\parallel _{X_{1}}= \parallel Ax\parallel \) and \(C_{0},C_{1}\) are the constants in the introduction.
To begin, we study the impulsive problem
where \( \Omega = {\mathbb {R}}^{n} \), \(X= L^{2}(\Omega ;{\mathbb {R}})\), \(0=t_{0}< \cdots < t_{N+1} =a\) are pre-fixed, \(I_{i}= (t_{i},t_{i+1}]\), \(\beta _{1}\in C_\mathrm{Lip}([0,a]\times {\mathbb {R}};{\mathbb {R}})\), \(\beta _{1}(\cdot )\) is bounded, \(\beta _{2} \in C_\mathrm{Lip}([-p,a]; {\mathbb {R}})\) and \({\mathcal {L}}_{i}, A {\mathcal {L}}_{i} \in L^{2}(\Omega \times \Omega ,{\mathbb {R}})\). In addition, we assume that there is \(\gamma \in L^{p}(\Omega )\) such that
To represent this problem in the form (1.1)–(1.3) we define the functions \(g_{i}(\cdot )\) and \(f(\cdot )\) by \( g_{i}(t,x)(\xi ) = \int _{{\mathbb {R}}^{n}}{\mathcal {L}}(\xi ,y)x(y){\text {d}}y \) and \( f(t,x)(\xi ) {=} \beta _{1} (t,\xi ,x(\xi ))+ \beta _{2} (t) x(\xi )\). It is easy to see that \( \parallel A g_{i}(x)\parallel \le \parallel A{\mathcal {L}}_{i}\parallel _{L^{2}(\Omega \times \Omega ;{\mathbb {R}})}\parallel x\parallel \) and
Thus, we can apply Proposition 2.3 with \(L_{f}(r)= \parallel \gamma \parallel _{C_\mathrm{Lip}(\Omega )} + [\beta _{2}]_{C_\mathrm{Lip}}r + \parallel \beta _{2}\parallel _{C(\Omega )}\), \(C_{X}(f,r)= \parallel \beta _{1}\parallel _{C([0,a]\times \Omega \times \Omega ;{\mathbb {R}})}+ \parallel \beta _{2}\parallel _{C(\Omega )}r\), \(L_{X,X_{1}}(g_{i})=\parallel A{\mathcal {L}}_{i}\parallel _{L^{2}(\Omega \times \Omega ;{\mathbb {R}})}\), \(C_{X,X_{1}}(g,r)= \max _{i=1,\ldots ,n} \parallel A{\mathcal {L}}_{i}\parallel _{L^{2}(\Omega \times \Omega ;{\mathbb {R}})} r\) and \(L_{X,X_{1}}(g)=\sup _{i=1,\ldots ,n} \parallel A{\mathcal {L}}_{i}\parallel _{L^{2}(\Omega \times \Omega ;{\mathbb {R}})}\).
In the next result, we adopt the above notations and the notations in Remark 1. In addition, we say that \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}(X)\) is a classical solution of (3.1)–(3.3) if \(u(\cdot )\) is a classical solution of the associate problem (1.1)–(1.3) and we adopt a similar (for mild and classical solutions) in the following examples.
Proposition 3.5
If \(\max \{ \parallel \varphi \parallel _{C([-p,0];X)}, C_{0} b {\mathcal {C}}_{X}(f,r) + C_{1} {\mathcal {C}}_{X,W}(g,r)\} \le r\) and
for some \(r>0\), then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of (3.1)–(3.3).
We study now the problem
where \(\Omega \) is bounded, \(\beta _{1}\in C_\mathrm{Lip}([0,a]\times {\mathbb {R}};{\mathbb {R}})\) and \(\beta _{1}(\cdot )\) is bounded.
To apply Theorem 2.2, we assume \(X=L^{2}(\Omega )\), the condition \( {\mathbf{H}}_{g_{i},\sigma _{j}}\) is satisfied and we define \(g_{i}(\cdot )\) and \(f(\cdot )\) by \( f(t,x)(\xi ) = \int _{0}^{t} \beta _{1} (\tau ,x(\xi )) {\text {d}}\tau \) and \( g_{i}(t,x)(\xi )= \alpha _{i} x(\xi ).\) From the above,
for \(t,s\in [0,a]\), \(x,y\in X\) and \(z\in D(A)\), and the conditions in Theorem 2.2 are satisfied with \(L_{f} =(1+b) \parallel \beta _{1}\parallel _{C_\mathrm{Lip}([0,a]\times {\mathbb {R}};{\mathbb {R}} )} \), \(C_{X}(f)= a\parallel \beta _{1}\parallel _{C([0,a]\times {\mathbb {R}};{\mathbb {R}})}\), \(l_{g_{i}}= \mid \alpha _{i}\mid \), \(k_{g_{i}}=0\), \(L_{g}=\max _{i=1,\ldots ,N} \mid \alpha _{i}\mid \) and \( \Upsilon = 2 C_{0} C_{X}(f) + b(C_{0} +C_{1})L_{f} + [T(\cdot )\varphi (0)]_{C_\mathrm{Lip}([-p,0];X)} + [\varphi ]_{C_\mathrm{Lip}([-p,0];X)} .\) The next result follows from Theorem 2.2.
Proposition 3.6
Under the above conditions and notations, if the inequality (2.10) is verified, then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of (3.4)–(3.6).
We complete this section studying a problem motivated by equations arising in population dynamics. Consider the problem
To treat this problem, we assume \(X= C(\Omega ;{\mathbb {R}})\) and \(\alpha ,\alpha _{i}\in {\mathbb {R}}\) and we define \(g_{i}(\cdot )\) and \(f(\cdot )\) by \( g_{i}(t,x)(\xi )= \alpha _{i} x(\xi ) \) and \( f(t,x)(\xi ) = \alpha x(\xi )(1- x(\xi ))\). It is trivial to see that
for all \(t,s\in [0,a]\), \(x,y\in B_{r}(0;X)\) and \(z\in D(A)\). From Proposition 2.4, we get.
Proposition 3.7
Suppose that there is \(r>\parallel \varphi \parallel _{C([-p,0];X)}\) such that the inequality (2.10) is verified with \(L_{f}(r)\) in place \(L_{f}\) and \( C_{0} ( \parallel \varphi (0)\parallel + b \mid \alpha \mid (1+2r)) + C_{0}(\max _{i=1,\ldots ,N}l_{ g_{i}}r+ k_{g_{i}}) < r.\) Then there exists a unique classical solution \(u\in {\mathcal {B}}{{\mathcal {P}}{\mathcal {C}}}_\mathrm{Lip}(X)\) of (3.7)–(3.9).
References
Aiello, Walter, Freedman, H.I., Wu, J.: Analysis of a model representing stage-structured population growth with state-dependent time delay. SIAM J. Appl. Math. 52(3), 855–869 (1992)
Bainov, D., Covachev, V.: Impulsive Differential Equations with a Small Parameter. Series on Advances in Mathematics for Applied Sciences, 24. World Scientific Publishing Co., Inc., River Edge, NJ (1994)
Bajo, I., Liz, E.: Periodic boundary value problem for first order differential equations with impulses at variable times. J. Math. Anal. Appl. 204, 65–73 (1996)
Belley, J.M., Guen, R.: Periodic van der Pol equation with state dependent impulses. J. Math. Anal. Appl. 426(2), 995–1011 (2015)
Belley, J.M., Virgilio, M.: Periodic Duffing delay equations with state dependent impulses. J. Math. Anal. Appl. 306, 646–662 (2005)
Belley, J.M., Bondo, É.: Anti-periodic solutions of Liénard equations with state dependent impulses. J. Differ. Equ. 261(7), 4164–4187 (2016)
Benchohra, M., Henderson, J., Ntouyas, S.: Impulsive Differential Equations and Inclusions. Contemporary Mathematics and Its Applications, 2. Hindawi Publishing Corporation, New York (2006)
Chu, J., Nieto, J.: Impulsive periodic solutions of first-order singular differential equations. Bull. Lond. Math. Soc. 40(1), 143–150 (2008)
Driver, R.D.: A functional-differential system of neutral type arising in a two-body problem of classical electrodynamics. In: LaSalle, J., Lefschtz S. (eds.) International Symposium on Nonlinear Differential Equations and Nonlinear Mechanics, Academic Press, New York, pp. 474-484 (1963)
Driver, R.D.: A neutral system with state-dependent delay. J. Differ. Equ. 54, 73–86 (1984)
Fan, Z., Li, G.: Existence results for semilinear differential equations with nonlocal and impulsive conditions. J. Funct. Anal. 258(5), 1709–1727 (2010)
Frigon, M., O’Regan, D.: First order impulsive initial and periodic problems with variable moments. J. Math. Anal. Appl. 233, 730–739 (1999)
Gabor, Grzegorz: Differential inclusions with state-dependent impulses on the half-line: new Fréchet space of functions and structure of solution sets. J. Math. Anal. Appl. 446(2), 1427–1448 (2017)
Hakl, R., Pinto, M., Tkachenko, V., Trofimchuk, S.: Almost periodic evolution systems with impulse action at state-dependent moments. J. Math. Anal. Appl. 446(1), 1030–1045 (2017)
Hartung, F., Krisztin, T., Walther, H-O., Wu, J.: Functional differential equations with state-dependent delays: theory and applications. Handbook of differential equations: ordinary differential equations. Vol. III, 435-545, Handb. Differ. Equ
Hartung, F.: On differentiability of solutions with respect to parameters in neutral differential equations with state-dependent delays. Ann. Mat. Pura Appl. (4) 192(1), 17–47 (2013)
Hartung, F.: Differentiability of solutions with respect to the initial data in differential equations with state-dependent delays. J. Dyn. Differ. Equ. 23(4), 843–884 (2011)
Hartung, F., Turi, J.: On differentiability of solutions with respect to parameters in state-dependent delay equations. J. Differ. Equ. 135(2), 192–237 (1997)
Hernández, E., Prokopczyk, A., Ladeira, L.: A note on partial functional differential equations with state-dependent delay. Nonlinear Anal. Real World Appl. 7(4), 510–519 (2006)
Hernández, E., Pierri, M., Goncalves, G.: Existence results for an impulsive abstract partial differential equation with state-dependent delay. Comput. Math. Appl. 52(3–4), 411–420 (2006)
Hernández, E., Henríquez, H., Rabello, M.: Existence of solutions for a class of impulsive partial neutral functional differential equations. J. Math. Anal. Appl. 331(2), 1135–1158 (2007)
Hernández., Pierri, M., O’Regan, D.: On abstract differential equations with non instantaneous impulses. Topol. Methods Nonlinear Anal. 46(2), 1067-1088 (2015)
Hernández, ., O’Regan, D.: On a new class of abstract impulsive differential equations. Proc. Am. Math. Soc. 141(5), 1641-1649 (2013)
Hernandez, E., Pierri, M., Wu, J.: \( C^{1+\alpha }\)-strict solutions and wellposedness of abstract differential equations with state dependent delay. J. Differ. Equ. 261(12), 6856–6882 (2016)
Krisztin, T., Rezounenkob, A.: Parabolic partial differential equations with discrete state-dependent delay: classical solutions and solution manifold. J. Differ. Equa. 260(5), 4454–4472 (2016)
Kosovalic, N., Magpantay, F.M.G., Chen, Y., Wu, J.: Abstract algebraic-delay differential systems and age structured population dynamics. J. Differ. Equ. 255(3), 593–609 (2013)
Kosovalic, N., Chen, Y., Wu, J.: Algebraic-delay differential systems: \( C^{0}\)-extendable submanifolds and linearization. Am. Math. Soc. 369(5), 3387–3419 (2017). (To appear in Trans)
Lakshmikantham, V., Bainov, D.D., Simeonov, P.S.: Theory of impulsive differential equations. Series in Modern Applied Mathematics, 6. World Scientific Publishing Co., Inc., Teaneck, NJ (1989)
Kou, C., Zhang, S., Wu, S.: Stability analysis in terms of two measures for impulsive differential equations. J. Lond. Math. Soc. (2) 66(1), 142–152 (2002)
Li, X., Wu, Jianhong: Stability of nonlinear differential systems with state-dependent delayed impulses. Automatica J. IFAC 64, 63–69 (2016)
Liu, J.: Nonlinear impulsive evolution equations. Dyn. Contin. Discrete Impuls. Syst. 6(1), 77–85 (1999)
Lunardi, A.: Analytic semigroups and optimal regularity in parabolic problems, PNLDE Vol. 16, Birkhäauser Verlag, Basel (1995)
Lv, Y., Rong, Y., Yongzhen, P.: Smoothness of semiflows for parabolic partial differential equations with state-dependent delay. J. Differ. Equ. 260, 6201–6231 (2016)
Nieto, J., O’Regan, D.: Variational approach to impulsive differential equations. Nonlinear Anal. Real World Appl. 10(2), 680–690 (2009)
Pazy, A.: Semigroups of linear operators and applications to partial differential equations. Appl. Math. Sci. 44. Springer, New York (1983)
Rezounenko, A.V.: A condition on delay for differential equations with discrete state-dependent delay. J. Math. Anal. Appl. 385(1), 506–516 (2012)
Rezounenko, A.V.: Partial differential equations with discrete and distributed state-dependent delays. J. Math. Anal. Appl. 326(2), 1031–1045 (2007)
Rezounenko, A.V., Wu, J.: A non-local PDE model for population dynamics with state-selective delay: local theory and global attractors. J. Comput. Appl. Math. 190(1–2), 99–113 (2006)
Rogovchenko, Y.: Impulsive evolution systems: main results and new trends. Dyn. Contin. Discrete Impuls. Syst. 3(1), 57–88 (1997)
Samoilenko, A.M., Perestyuk, N.A.: Impulsive differential equations. With a preface by Yu. A. Mitropol’skii and a supplement by S. I. Trofimchuk. Translated from the Russian by Y. Chapovsky. World Scientific Series on Nonlinear Science. Series A: Monographs and Treatises, 14. World Scientific Publishing Co., Inc., River Edge, NJ (1995)
Tomecek, Jan: Periodic solution of differential equation with \(\phi \)-Laplacian and state-dependent impulses. J. Math. Anal. Appl. 450(2), 1029–1046 (2017)
Walther, Hans-Otto: The solution manifold and \(C^{1}\)-smoothness for differential equations with state-dependent delay. J. Differ. Equ. 195(1), 46–65 (2003)
Yu, J.S., Tang, X.H.: Global attractivity in a delay population model under impulsive perturbations. Bull. Lond. Math. Soc. 34(3), 319–328 (2002)
Acknowledgements
The authors wish to thank the referees and the editor responsible for this paper, for their valuable comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Hernández, E., Azevedo, K.A.G. & Gadotti, M.C. Existence and uniqueness of solution for abstract differential equations with state-dependent delayed impulses. J. Fixed Point Theory Appl. 21, 36 (2019). https://doi.org/10.1007/s11784-019-0675-1
Published:
DOI: https://doi.org/10.1007/s11784-019-0675-1