Abstract
This paper studies the controllability of the initial value problems of linear and semilinear second-order impulsive systems. Necessary and sufficient conditions of controllability for linear problems are obtained, and a new rank criterion is presented. We also show semilinear problems are controllable via Krasnoselskii’s fixed point theorem. Finally, two examples are provided to verify the theoretically results.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Many evolution processes in science and technology, such as mechanics, population dynamics, pharmacokinetics, industrial robotics, biotechnology, economics and so on, may change their state rapidly, or the duration of the change is negligible. We describe these processes with impulsive effects by impulsive differential equations and the theory of impulsive differential equations is an important branch of differential equation theory; see [1] and the references therein.
Control theory is an important branch in applied mathematics and engineering and modern control theory was developed by Kalman. Roughly speaking, the object of control theory is to find a control function that can steer the state function to the desired result at the end (terminal). Numerous papers are devoted to the controllability of differential equations in Banach space [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22], such as exact controllability, approximate controllability and null controllability, and the main techniques are based on fixed point theorems [3, 4, 14, 18, 23], variational methods [5, 24], semigroup theory [2, 8], and so on.
Second-order systems capture the dynamic behavior of many natural phenomena and have applications in many fields such as mathematical physics, electrical power systems, quantum mechanics, biology, long transmission lines and finance [25, 26]. Numerous papers focus on the controllability of second-order impulsive systems (see [2, 3, 6, 7, 11, 27] for cosine family theory and [2, 8, 10, 28] where the corresponding operators of the cosine family are compact). However, as noted by Travis and Webb [29], some of these results work only to finite-dimensional spaces. We refer the reader also to [3, 30] for other results on the controllability of second-order impulsive systems.
For the controllability of initial value problems for second-order differential equations
many authors consider the controllability of the solution x(t) i.e., one finds a control u which makes the state function x(t) arrive at the value that we wish at the terminal. As mentioned in [7], it is unreasonable to regard the damped term \(x'(t)\) in the controllability. Recently, the authors in [7, 10, 11, 27] consider the controllability of x(t) and \(x'(t)\).
In [7], Li et al. consider the approximate controllability of system (1.1). Let \(J=[0,b]\), the state \(x(\cdot )\) takes values in a Banach space X, \(u(\cdot )\in L^{2}(J, U)\) is the control function where U is a Banach space, the definition of controllability defined as follows: Systems (1.1) are said to be approximately controllable on J if \({\overline{D}}=X\times X\), where \(D=\{{(x(b,x_{0},y_{0},u),y(b,x_{0},y_{0},u))}:u\in L^{2}(J,U)\}\), \(y(\cdot ,x_{0},y_{0},u)=x'(\cdot ,x_{0},y_{0},u)\) and \(x(\cdot ,x_{0},y_{0},u)\) is a mild solution of (1.1).
Their aim is to pick a control function u which controls both x(t) and \(x'(t)\). In [10, 11, 27], the following two assumptions are used,
(A1) The linear operator \(G_{1}:L^{2}(J,U)\rightarrow X\), defined by
has an invertible operator \(G_{1}^{-1}\) which takes the values in \(L^{2}(J,U)/\ker G_{1}\) and there exists positive constant \(M_{1}\) such that \(\Vert G_{1}^{-1}\Vert \le M_{1}\).
(A2) The linear operator \(G_{2}:L^{2}(J,U)\rightarrow X\), defined by
has an invertible operator \(G_{2}^{-1}\) which takes the values in \(L^{2}(J,U)/\ker G_{2}\) and there exists positive constant \(M_{2}\) such that \(\Vert G_{2}^{-1}\Vert \le M_{2}\).
As pointed by Balachandran and Kim [31] the control function defined in [11, 27] can not steer the value of the state function to what we want at the terminal unless the condition
(H) \(G_{1}G_{2}^{-1}=G_{2}G_{1}^{-1}=0\) is satisfied.
For the second-order systems in finite dimensional space, (A1) or (A2) will lead to a contradiction with the definition of controllability. Since if we assume system (1.1) is controllable. Then for any \((x_{1},y_{1})\in X\times X\), there exists a control \(u_{1}\) such that \(x(b)=x_{1}\), and \(x'(b)=y_{1}\) under the control \(u_{1}\). For another point \((x_{1},y_{2})\), since \(y_{1}\ne y_{2}\), there exists a control \(u_{2}\) such that \(x(b)=x_{1}\), and \(x'(b)=y_{2}\) under the control \(u_{2}\) as well. Then if \(u_{1}=u_{2}\), we have \(y_{1}=y_{2}\), a contradiction; if \(u_{1}\ne u_{2}\), since A is the infinitesimal generator of a strongly continuous cosine family C(t) on X, hence, the Cauchy problem (1.1) is well posed. Then from the expression of the solution for (1.1),
and we get
and
Combining these two equalities with conditions (A1), we find
a contradiction to the assumption \(u_{1}\ne u_{2}\). Hence, if assumptions (A1) or (A2) hold, we cannot obtain the controllability result of system (1.1) under the definition of controllability defined in [7]. In view of this, we introduce a weaker definition of controllability in Sect. 2.
To the best of our knowledge, there are only a few articles on the controllability of second-order linear systems, and we note that, for finite-dimensional linear systems, all the concepts of controllability are equivalent (exact controllability, approximate controllability and null controllability). In this paper, we consider the controllability of the following initial value problems for second-order impulsive differential equations
and semilinear second-order impulsive differential equations
where A, \(B_{1}\) and \(B_{2}\) are constant \(n\times n\) matrices satisfying \(AB_{1}=B_{1}A\), \(AB_{2}=B_{2}A\), \(B_{1}B_{2}=B_{2}B_{1}\), \(0=t_{0}<t_{1}<\cdots<t_{k}<t_{k+1}=b\) are impulsive points, \(u\in L^{2}(J,{\mathbb {R}}^{n})\) is a control function, and \(f\in C(J\times {\mathbb {R}}^{n};{\mathbb {R}}^{n})\).
The contributions of this paper are as follows:
-
(1)
We introduce a weaker definition of controllability with respect to the state function x(t) and the damped term \(x'(t)\).
-
(2)
We present a new algebraic method to obtain a rank criterion, and a rank criterion of controllability for second-order impulsive linear systems is given.
-
(3)
Based on the controllability of the linear systems, we give a sufficient condition to guarantee the controllability of the semilinear second-order impulsive systems.
The paper is structured in the following way. In Sect. 2, we give a weaker definition of controllability and some associated notations and essential lemmas. In Sect. 3, instead of converting a second-order system into a first order system, we obtain a new rank criterion of controllability of system (1.2) by direct analysis of the second-order system itself. In Sect. 4, we give a sufficient condition of the controllability of the system (1.3). Finally, in Sect. 5, some examples are provided to illustrate the suitability of our results.
2 Preliminaries
In this section, we modify the definition of controllability and list some notations and properties needed to establish our main results.
Let\(PC(J,{\mathbb {R}}^{n})\) denote the Banach space of piecewise continuous functions on the interval J, that is \(PC(J,{\mathbb {R}}^{n})=\{v:J\rightarrow {\mathbb {R}}^{n}|u\in C((t_{k-1},t_{k}],{\mathbb {R}}^{n})\) for \(k\in \{1, \ldots ,m+1\}\) and there exists \(v(t_{k}^{-})\) and \(v(t_{k}^{+}), k\in \{1, \ldots ,m\}\) with \(v(t_{k})=v(t_{k}^{-}) \}\) equipped with the Chebyshev PC-norm \(||v||_{PC}:=\sup \{||v(t)||:t\in J\}\). Let \(PC^{1}(J,{\mathbb {R}}^{n}):=\{x\in PC(J,{\mathbb {R}}^{n}):x'\in PC(J,{\mathbb {R}})\}\) equipped with the norm \(\Vert x\Vert _{PC^{1}}=\max \{\Vert x\Vert _{PC},\Vert x'\Vert _{PC}\}\). Obviously, \(PC(I,{\mathbb {R}}^{n})\) endowed with the norm \(\Vert \cdot \Vert _{PC^{1}}\) is also a Banach space. We use the notation
Let \(m=i(t,0)\) denote the number of impulsive points on (0, t), and assume \(AB=BA\).
Definition 2.1
The system (1.2) is said to be exact controllability in \({\mathbb {R}}^{n}\), if for each pair \((x_{0},y_{0})\in {\mathbb {R}}^{n}\times {\mathbb {R}}^{n}\), there exists a pair of control functions \((u_{1}(\cdot ),u_{2}(\cdot ))\in L^{2}([0,b],{\mathbb {R}}^{n})\times L^{2}([0,b],{\mathbb {R}}^{n})\) such that for any \((x_{1},y_{1})\in {\mathbb {R}}^{n}\times {\mathbb {R}}^{n}\),
here \(x(\cdot )\) is the solution of (1.2) under the control \(u_{1}\), \(y(\cdot )\) is the solution of (1.2) under the control \(u_{2}\), and \(y'(t)=dy(t)/dt\).
Remark 2.2
In [4, 5, 9, 14, 18], the definition of controllability imply that one find a control function which steer the state function \(x(\cdot )\) to the target value, and in [7, 10, 11, 27], which imply that one find a control function which steer both the state function \(x(\cdot )\) and damped term \(x'(\cdot )\) to the value we wanted. However, Definition 2.1 indicates that one pick a pair of control functions \((u_{0},u_{1})\) such that \(u_{0}\) control the state function x(t) and \(u_{1}\) control the damped term \(y'(t)\). Notice that at this moment except for a constant difference, the antiderivative of damped term \(y'(t)\) may be different with the state function x(t).
The following Lemmas is crucial to our proof of main results.
Lemma 2.3
(see [32]) Let \(|\cdot |\) be a norm on \({\mathbb {R}}^{n}\) and B be an \(n\times n\) matrix. Then for any \(\varepsilon >0\) there exist \(T_{B,\varepsilon }\ge 1\) such that \(||B^{k}||\le T_{B,\varepsilon }(\rho (B)+\varepsilon )^{k}\), where \(\rho (B)\) is the spectral radius of B.
Lemma 2.4
(Krasnoselskii’s fixed point theorem) Let B be a bounded closed and convex subset of a Banach space X and let \(F_{1},F_{2}\) be maps of B into X such that \(F_{1}x+F_{2}y\in B\) for every \(x,y\in B\). If \(F_{1}\) is a contraction and \(F_{2}\) is compact and continuous, then the equation \(F_{1}x+F_{2}x=x\) has a solution on B.
Lemma 2.5
(PC-type Ascoli–Arzela theorem, see [33]) Let \(Q\subset PC(\Omega ,X)\) where X is a Banach space. Then Q is a relatively compact subset of \(PC(\Omega ,X)\) if, (a) Q is uniformly bounded subset of \(PC(\Omega ,X)\); (b) Q is equicontinuous in \((t_{i},t_{i+1})\), \(i=0,1,\ldots ,k\); and (c) \(Q(t)=\{v(t)|v\in Q,t\in \Omega \backslash \{t_{i}\},i=0,1,\ldots ,k\}\), \(Q(t_{i}^{+})=\{v(t_{i}^{+})|v\in Q\}\) and \(Q(t_{i}^{-})=\{v(t_{i}^{-})|v\in Q\}\) are relatively compact subsets of X.
Lemma 2.6
(see [34]) For \(t\in (t_{m},t_{m+1}]\), \(m=0,1,\ldots ,k\), the solution of (1.2) is given by
where \(W(A,t,x_{0},y_{0})\) is the solution of the homogeneous initial value problem of (1.2), and
Consider the notation
then the solution of (1.2) can be expressed by
Lemma 2.7
For any \(t_{m}<\tau _{1}\le \tau _{2}\le b\), and \(t_{m}<t\le t_{m+1}\le b\), we have
and
where \(\theta _{1}\) and \(\theta _{2}\) are positive constants, and \(Q_{m}'(t,s)\) denotes the function that takes derivative with respect to t.
Proof
Since
combining this with the Lemma 2.3, we have
and
According to the definition of \(Q_{m}\), inequality (2.3), and the mean value theorem, we find
where \(\varsigma _{0},\varsigma _{1,j},\ldots ,\varsigma _{m-i-1,j},\varsigma _{m-i}\) are selected by the mean value theorem located in \([-b,b]\), \(T_{\varepsilon }=\max \{T_{A,\varepsilon },T_{A_{1},\varepsilon },T_{A_{2},\varepsilon }\}\). Similarly, by virtue of the definition of \(Q_{m}\), inequality (2.2), and the mean value theorem, we have
where \(\xi _{0},\xi _{1,j},\ldots ,\xi _{m-i-1,j},\xi _{m-i}\) are selected by the mean value theorem located in \([-b,b]\). \(\square \)
3 The controllability of linear systems
In this section, we present some controllability criteria for systems (1.2) by using an algebraic method.
Theorem 3.1
The following statements are equivalent: \(1^{\circ }\)The system (1.2) is exact controllability; \(2^{\circ }\) The matrix \(\Gamma _{0}^{b}=\int _{0}^{b}Q_{k}(b,s)BB^{*}Q_{k}^{*}(b,s)ds\) and \(\Lambda _{0}^{b}=\int _{0}^{b}Q'_{k}(b,s)BB^{*}Q_{k}'^{*}(b,s)ds\) are nonsingular; \(3^{\circ }\) There at least exists a pair of integers \(0\le i\le k\), \(0\le j\le k\) such that both \(\int _{t_{i}}^{t_{i+1}}W_{i}(b,s)BB^{*}W_{i}^{*}(b,s)ds\) and \(\int _{t_{j}}^{t_{j+1}}W_{j}'(b,s)BB^{*}W_{j}'^{*}(b,s)ds\) are nonsingular.
Proof
First, we show the equivalence of \(1^{\circ }\) and \(2^{\circ }\). Assume the systems are exact controllability. We show that the matrix \(\Gamma _{0}^{b}\) and \(\Lambda _{0}^{b}\) both are nonsingular. If the result is not true, then at least one of matrices \(\Gamma _{0}^{b}\) and \(\Lambda _{0}^{b}\) is singular. Suppose \(\Gamma _{0}^{b}\) is singular. Then there exists a nonzero vector \({\overline{x}}_{0}\in {\mathbb {R}}^{n}\) such that
Hence we have
that is
On the other hand, since the systems are exact controllability, then because of the definition of exactly controllability, there exists a pair of control functions \((u_{1},u_{2})\) such that for \({\overline{x}}_{0}+W(A,b,x_{0},y_{0})\in {\mathbb {R}}^{n}\), the solution \(x(\cdot )\) of systems (1.2) under the control \(u_{1}(\cdot )\) arrives at \({\overline{x}}_{0}+W(A,b,x_{0},y_{0})\in {\mathbb {R}}^{n}\) at the terminal b, i.e.
Now (3.1) with (3.2) allows us to affirm that
which implies \({\overline{x}}_{0}={\textbf{0}}\) and this contradicts the hypothesis. Hence \(\Gamma _{0}^{b}\) is nonsingular. In a similar way, we obtain that \(\Lambda _{0}^{b}\) is nonsingular,
If both the matrices \(\Gamma _{0}^{b}\) and \(\Lambda _{0}^{b}\) are nonsingular, we prove that systems (1.2) are exactly controllability, that is for any fixed \((x_{1},y_{1})\in {\mathbb {R}}^{n}\times {\mathbb {R}}^{n}\), we show that there exists a pair of control functions \((u_{1}(\cdot ),u_{2}(\cdot ))\in L^{2}([0,b],{\mathbb {R}}^{n})\times L^{2}([0,b],{\mathbb {R}}^{n})\) such that the solution x(t) of systems (1.2) satisfies \(x(b)=x_{1}\) under the control \(u_{1}(\cdot )\) and \(y'(b)=y_{1}\) under the control \(u_{2}(\cdot )\). We choose the control functions by
and
Then we have
Obviously, \(x(b)=x_{1}\). Similarly, under the control function \(u_{2}(t)\), \(y'(t)\) satisfies
and we have \(y'(b)=y_{1}\). Hence the systems (1.2) are exact controllability.
Next, we show the equivalence of \(2^{\circ }\) and \(3^{\circ }\). Assume the matrix \(\Gamma _{0}^{b}\) is singular. Then there exists a nonzero vector \({\overline{x}}_{0}\in {\mathbb {R}}^{n}\) such that
that is
which is equivalent to
that is \(\int _{t_{i}}^{t_{i+1}}W_{i}(b,s)BB^{*}W_{i}^{*}(b,s)ds\) is singular for all \(0\le i\le k\). Hence \(\Gamma _{0}^{b}\) is nonsingular iff there at least exists a constant \(0\le i\le k\) such that \(\int _{t_{i}}^{t_{i+1}}W_{i}(b,s)BB^{*}W_{i}^{*}(b,s)ds\) is nonsingular.
By the same argument, we also can show that \(\Lambda _{0}^{b}\) is nonsingular iff there at least exists a constant \(0\le j\le k\) such that the matrix \(\int _{t_{j}}^{t_{j+1}}W_{j}'(b,s)BB^{*}W_{j}'^{*}(b,s)ds\) is nonsingular. \(\square \)
Remark 3.2
Theorem 3.1 shows that initial value problems of second-order linear impulsive systems (1.2) are controllable iff there exist constants \(\lambda >0\) and \(\gamma >0\) such that for all \(x\in {\mathbb {R}}^{n}\),
and
Then
Since the conditions which guarantee the controllability in Theorem 3.1 are formal and are hard to verify. In what follows, we give a new rank criterion of controllability of systems (1.2). For convenience in writing, in what follows, we use the notation
Theorem 3.3
Systems (1.2) are exact controllability iff there exists a pair of integers \(l_{1},l_{2}\in \{0,1,2,\ldots ,k\}\) such that
and
Proof
Theorem 3.1 shows that systems (1.2) are exact controllability iff there is a pair of integers \(0\le i\le k\), \(0\le j\le k\) such that both \(\int _{t_{i}}^{t_{i+1}}W_{i}(b,s)BB^{*}W_{i}^{*}(b,s)ds\) and \(\int _{t_{j}}^{t_{j+1}}W_{j}'(b,s)BB^{*}W_{j}'^{*}(b,s)ds\) are nonsingular. We subdivide the proof into several cases.
Case 1 If \(i=j=k\), that is both \(\int _{t_{k}}^{b}W_{k}(b,s)BB^{*}W_{k}^{*}(b,s)ds\) and \(\int _{t_{k}}^{b}W_{k}'(b,s)BB^{*}W_{k}'^{*}(b,s)ds\) are nonsingular. Now \(\int _{t_{k}}^{b}W_{k}(b,s)BB^{*}W_{k}^{*}(b,s)ds\) is singular iff there exists a nonzero vector \(x_{0}\in {\mathbb {R}}^{n}\) such that \(x_{0}^{T}W_{k}(b,s)B={\textbf{0}}\) for all \(t_{k}\le s<b\). Since A is a nonsingular matrix, we have Rank \(W_{k}(b,s)=n\), hence Rank \(B<n\). Likewise, we can show that \(\int _{t_{k}}^{b}W_{k}'(b,s)BB^{*}W_{k}'^{*}(b,s)ds\) is singular iff Rank \(B<n\). Hence both \(\int _{t_{k}}^{b}W_{k}(b,s)BB^{*}W_{k}^{*}(b,s)ds\) and \(\int _{t_{k}}^{b}W_{k}'(b,s)BB^{*}W_{k}'^{*}(b,s)ds\) are nonsingular iff Rank \(B=n\).
Case 2 If \(i=j=k-1\), we show that both \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) and
\(\int _{t_{k-1}}^{t_{k}}W_{k-1}'(b,s)BB^{*}W_{k-1}'^{*}(b,s)ds\) are nonsingular iff
Assume \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) is nonsingular. Then we show that
If this is not true, that is Rank \(\begin{pmatrix} A_{1}B&A_{2}B \end{pmatrix} <n\), then there exists a nonzero vector \(x_{0}\in {\mathbb {R}}^{n}\) such that
i.e.,
Hence, for all \(t_{k-1}< s\le t_{k}\),
which implies \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) is singular. This contradicts the hypothesis. Therefore, Rank \(\begin{pmatrix} A_{1}B&A_{2}B \end{pmatrix} =n\).
Assume \( \text {Rank} \begin{pmatrix} A_{1}B&A_{2}B \end{pmatrix} =n\). We will show that \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) is nonsingular. Assume \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) is singular. First, we prove that there exists a number sequence \((\lambda _{1},\lambda _{2})\in (t_{k-1},t_{k}]^{2}\), where \(\lambda _{1}\ne \lambda _{2}\), such that the matrix
is nonsingular. Suppose for every \((\lambda _{1},\lambda _{2})\in (t_{k-1},t_{k}]^{2}\), we have
Take \(\lambda _{2}=t_{k}\) in (3.7), since \(|\Psi (b-\lambda _{2})|\ne 0\), we find
however, by the Jordan decomposition, we find zero is not an eigenvalue of \(\Psi (b-2t_{k}+\lambda _{1})-\Psi (b-\lambda _{1})\), hence, (3.8) is not valid, that is there exists a number sequence \((\lambda _{1},\lambda _{2})\in (t_{k-1},t_{k}]^{2}\) such that (3.6) is nonsingular.
Suppose \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) is singular. Then there exists a nonzero vector \(x_{0}\in {\mathbb {R}}^{n}\) such that
that is
which implies
Take \((\lambda _{1},\lambda _{2})\in (t_{k-1},t_{k}]^{2}\) such that (3.6) is nonsingular, and then by (3.9), we find
therefore,
that is
which contradict the hypothesis.
Thus \(\int _{t_{k-1}}^{t_{k}}W_{k-1}(b,s)BB^{*}W_{k-1}^{*}(b,s)ds\) is nonsingular iff
Using the same argument we can establish that \(\int _{t_{k-1}}^{t_{k}}W_{k-1}'(b,s)BB^{*}W_{k-1}'^{*}(b,s)ds\) is nonsingular iff
Case 3 If \(i=j=k-2\), we show that both \(\int _{t_{k-2}}^{t_{k-1}}W_{k-2}(b,s)BB^{*}W_{k-2}^{*}(b,s)ds\) and \(\int _{t_{k-2}}^{t_{k-1}}W_{k-2}'(b,s)BB^{*}W_{k-2}'^{*}(b,s)ds\) are nonsingular iff
To do this, we first show an auxiliary result. For \(s_{1},s_{2},s_{3}\in (t_{k-2},t_{k-1}]\), let
and
We claim that
or
where \(x_{0}=(\lambda _{1}I,\lambda _{2}I,\lambda _{3}I)^{T}\in {\mathbb {R}}^{3n\times n}\), implies \(x_{0}={\textbf{0}}_{3n\times n}\). We only show that (3.10) implies that \(x_{0}={\textbf{0}}_{3n\times n}\), and the other case can be treated similarly. For any \(s\in (t_{k-2},t_{k-1})\) and \(\varepsilon _{1}, \varepsilon _{2}>0\) small enough, let
By the first row of equality (3.10), we have
take the second and fourth derivatives with respect to s in (3.12) respectively and we have
Combine (3.12), (3.13) and (3.14) to obtain
and since \((\Psi (b-s),\Psi (b-(1+\varepsilon _{1})s),\Psi (b-(1+\varepsilon _{2})s))^{T}\) is a nonzero vector therefore
which implies that at least one of \(\lambda _{1}\), \(\lambda _{2}\), and \(\lambda _{3}\) is zero.
Let \(\lambda _{1}=0\), then by the second row of (3.10), we find
Take the second derivative with respect to s in (3.15) to get
Combine (3.15) with (3.16) and we have
since
hence,
which implies that at least one of \(\lambda _{2}\) and \(\lambda _{3}\) is zero.
Let \(\lambda _{2}=0\), then by the third row of (3.10), we have
obviously, (3.17) implies \(\lambda _{3}=0\). Thus, \(x_{0}={\textbf{0}}_{3n\times n}\).
Suppose \(\int _{t_{k-2}}^{t_{k-1}}W_{k-2}(b,s)BB^{*}W^{*}_{k-2}(b,s)ds\) is nonsingular and assume Rank \(\begin{pmatrix} A_{1}^{2}B&A_{1}A_{2}B&A_{2}^{2}B \end{pmatrix} <n\). Then there exists a nonzero vector \(x_{0}\in {\mathbb {R}}^{n}\) such that
that is, for all \(s\in (t_{k-2},t_{k-1}]\),
which contradicts the fact that \(\int _{t_{k-2}}^{t_{k-1}}W_{k-2}(b,s)BB^{*}W^{*}_{k-2}(b,s)ds\) is nonsingular.
Suppose Rank \(\begin{pmatrix} A_{1}^{2}B&A_{1}A_{2}B&A_{2}^{2}B \end{pmatrix} =n\) and assume
is singular. There exists a nonzero vector \(x_{0}\in {\mathbb {R}}^{n}\) such that
which implies
For \(s_{1},s_{2},s_{3}\) selected in the auxiliary result, by equation (3.18), we have
that is
then by the auxiliary result, we find
which implies
which contradict the hypothesis.
Thus \(\int _{t_{k-2}}^{t_{k-1}}W_{k-2}(b,s)BB^{*}W_{k-2}^{*}(b,s)ds\) is nonsingular iff
In similar way, we obtain \(\int _{t_{k-2}}^{t_{k-1}}W_{k-2}'(b,s)BB^{*}W_{k-2}'^{*}(b,s)ds\) is nonsingular iff
Thus, we have case 3.
Similarly, taking the proof method of auxiliary result in case 3, for any \(0\le m\le k-2\), we can show an auxiliary result for \(i=j=m\), i.e.
or
implies \(x_{0}={\textbf{0}}_{(k-m+1)n\times n}\), where \(\Sigma _{k-m+1}\) is constructed the same way as \(\Sigma _{3}\). Making use of this auxiliary result and proceeding as the technique in case 3, we can show that for \(i=j=0\), both \(\int _{t_{0}}^{t_{1}}W_{0}(b,s)BB^{*}W_{0}^{*}(b,s)ds\) and \(\int _{t_{0}}^{t_{1}}W_{0}'(b,s)BB^{*}W_{0}'^{*}(b,s)ds\) are nonsingular iff
By Theorem 3.1, obviously, if there exists an integer \(l\in \{0,1,2,\ldots ,k\}\) such that
then system (1.2) is controllable.
On the other hand, if system (1.2) is controllable. Then, by Theorem 3.1, there at least exists a pair of integers \(0\le {\mathfrak {i}}\le k\), \(0\le {\mathfrak {j}}\le k\) such that both \(\int _{t_{{\mathfrak {i}}}}^{t_{{\mathfrak {i}}+1}}W_{{\mathfrak {i}}}(b,s)BB^{*}W_{{\mathfrak {i}}}^{*}(b,s)ds\) and \(\int _{t_{{\mathfrak {j}}}}^{t_{{\mathfrak {j}}+1}}W_{{\mathfrak {j}}}'(b,s)BB^{*}W_{{\mathfrak {j}}}'^{*}(b,s)ds\) are nonsingular, that is
and
\(\square \)
4 The controllability of semilinear systems
In this section, we consider the controllability of the initial value problems of second-order semilinear systems (1.3).
For convenience in writing, let us introduce the notation \(\Vert B\Vert =K\), and the following assumptions.
\((H_{1})\) \(f:J\times {\mathbb {R}}^{n}\rightarrow {\mathbb {R}}^{n}\) is a continuous function and there exist a positive constant L such that
for every \(x,y\in {\mathbb {R}}^{n}\), and \(N=\max _{t\in [0,b]} \Vert f(t,0)\Vert \)
\((H_{2})\) The linear systems (1.2) are exactly controllable.
\((H_{3})\) Let
Theorem 4.1
Let \(x_{0},y_{0}\in {\mathbb {R}}^{n}\) and assume the condition \((H_{1})\)-\((H_{3})\) are satisfied. Then the initial value problems of semilinear second-order impulsive systems (1.3) are exactly controllable provided that
and
where \({\mu =\max \{\frac{1}{\rho (A)^{1/2}},\frac{1}{\rho (A)^{3/2}}\}}\).
Proof
From Lemma 2.6, for \(t\in (t_{k},b]\), (1.3) are equivalent to the integral equation
In light of \((H_{3})\), we choose \(\varepsilon >0\) small enough such that
Combine with Lemma 2.3 and it follows that for \(t\in (t_{k},b]\),
and
where
\(T_{\varepsilon }=\max \{T_{A,\varepsilon }, T_{A_{1},\varepsilon },T_{A_{2},\varepsilon }\}\) and \(\varepsilon >0\) small enough.
We show the controllability of the solutions of (1.3). Define the feedback control function
and the operator \({\mathcal {F}}:PC(J,{\mathbb {R}}^{n})\rightarrow PC(J,{\mathbb {R}}^{n})\) as follows,
Let
and we use the notation
Now \({\mathcal {F}}=F_{1}+F_{2}\).
We subdivide the proof into several steps.
Step 1 We show that for every \(x,y\in B_{r}\), \(F_{1}x+F_{2}y\in B_{r}\). In fact, for every \(x,y\in B_{r}\), from (4.3), (4.5), (4.7) and \((H_{1})\), we have
where
By inequality (4.1), we can pick
Then, we have
that is
Step 2 We claim that \(F_{1}:B_{r}\rightarrow PC(J,{\mathbb {R}}^{n})\) is a contraction mapping. For every \(x,y\in B_{r}\), by (3.5), (4.1), (4.3), and \((H_{1})\), we have
so \(F_{1}\) is a contraction mapping.
Step 3 We show that \(F_{2}\) is compact and continuous. For any \(x,y\in B_{r}\), by the inequality (4.3), we have
therefore, \(F_{2}:B_{r}\rightarrow PC(J,{\mathbb {R}})\) is continuous. To check the compactness of \(F_{2}\), we prove that \(F_{2}\) is uniformly bounded and equicontinuous. In fact, for any \(x\in B_{r}\), by the inequality (4.3), we have
that is \(F_{2}B_{r}=\{F_{2}x\big |x\in B_{r}\}\) is uniformly bounded. Next, we show that \(F_{2}\) is equicontinuous. For any \(t_{k}<\tau _{1}\le \tau _{2}\le b\), by Lemma 2.7, and (4.3), we have
therefore, \(F_{2}B_{r}\) is the equicontinuous family of functions in \(PC(J,{\mathbb {R}}^{n})\). From Lemma 2.5, \(F_{2}B_{r}\) is relatively compact in \(PC(J,{\mathbb {R}}^{n})\).
From Krasnoselskii’s fixed point theorem, we obtain that F has a fixed point x in \(B_{r}\), which is the solution of (1.3) and satisfies \(x(b)=x_{1}\).
In what follows, we show the controllability of the derivative of solutions for systems (1.3). Define the feedback control function
and the operator \({\mathcal {H}}:PC^{1}(J,{\mathbb {R}}^{n})\rightarrow PC^{1}(J,{\mathbb {R}}^{n})\) as follows,
Let
and \({\mathcal {H}}=H_{1}+H_{2}\). Let
We show that \({\mathcal {H}}:D_{\ell }\rightarrow PC^{1}(J,{\mathbb {R}}^{n})\) has a fixed point. Proceeding as before, we subdivide the proof into several steps.
Step 1 We show that \(H_{1}x+H_{2}y\in D_{\ell }\), for any \(x,y\in D_{\ell }\).
In fact, for any \(x,y\in D_{\ell }\), proceeding as in the proof for the operator \({\mathcal {F}}\), and by inequalities (3.5), (4.3)–(4.6) and condition \((H_{1})\), we have
where
By inequality (4.2), we can pick
then, we have
It follows that for any \(x,y\in D_{r}\), \(H_{1}x+H_{2}y\in D_{r}\).
Similarly, making use of (3.5), (4.4), (4.6), (4.7) and \((H_{1})\), we get
where
By inequality (4.2), we can pick
Then, we have
Take r be the maximum of the right hand of (4.8) and (4.9), then we obtain
that is
Step 2 We state that \(H_{1}\) is a contraction.
For any \(x,y\in D_{r}\), by inequalities (3.5), (4.2), (4.4), and \((H_{1})\), we have
and
hence according to (4.2), there exists \(\varepsilon >0\) small enough such that
and
Therefore, \(H_{1}\) is a contraction mapping.
Step 3 We show that \(H_{2}\) is compact and continuous. Since, for every \(x,y\in D_{\ell }\), by (4.3) and (4.4), we have
and
therefore, \(H_{2}:D_{\ell }\rightarrow PC^{1}(J,{\mathbb {R}})\) is continuous. To check the compactness of \(H_{2}\), we consider the mapping
For every \(x\in D_{\ell }\), by inequality (4.4) and \((H_{1})\), we obtain
which implies \((H_{2}D_{\ell })'=\{(H_{2}x)'\big |x\in D_{\ell }\}\) is uniformly bounded in \(PC(J,{\mathbb {R}})\). We prove that for any \(x\in D_{\ell }\), \((H_{2}x)'\) is equicontinuous. In fact, for any \(t_{k}<\tau _{1}<\tau _{2}\le b\), in term of inequality (4.4), \((H_{1})\) and Lemma 2.7, we have
therefore, \((H_{2}D_{\ell })'\) is the equicontinuous family of functions in \(PC(J,{\mathbb {R}}^{n})\). From Lemma 2.5, \((H_{2}D_{\ell })'\) is relatively compact in \(PC(J,{\mathbb {R}}^{n})\). Hence, for any sequence \(\{x_{n}\}\subset D_{\ell }\), there exists a subsequence of \(\{x_{n}\}\), again denoted by \(\{x_{n}\}\), such that
Obviously,
for any \(x\in PC^{1}(J,{\mathbb {R}}^{n})\). Let \({\overline{\phi }}\) be the antiderivative of \(\phi \), combining this inequality with (4.11), we have
as n is large enough, which implies that for any \(\{H_{2}x_{n}\}\subset H_{2}D_{\ell }\), there exists a subsequence \(\{H_{2}x_{n_{k}}\}\) which is convergence in \(PC^{1}(J,{\mathbb {R}})\) . Thus, \(H_{2}:D_{\ell }\rightarrow PC^{1}(J,{\mathbb {R}}^{n})\) is a compact and continuous operator.
Hence, by the Krasnoselskii’s fixed point theorem, we obtain that \({\mathcal {H}}\) has a fixed point x in \(D_{\ell }\) which is the solution of (1.3) and satisfies \(x'(b)=y_{1}\).
In conclusion, second-order impulsive systems (1.3) are exactly controllable.
\(\square \)
5 Examples
In this section, we give some examples to illustrate the effectiveness of our results.
Example 5.1
For the simplicity of calculation, we consider the controllability of systems (1.2) with
and \(0=t_{0}<1=t_{1}< 2=t_{2}=b\). Obviously, B is nonsingular, then Theorem 3.1 holds for \(l=0\), that is system (1.2) is controllable. For the sake of convenience in calculating, we consider \(x_{1}=(30~40)^{T}\), then we show that we can choose a control function \(u_{1}(t)\) such that, under \(u_{1}(t)\), \(x(2)=(30~40)^\textrm{T}\). By Theorem 3.1 in [34], we obtain that, for \(t\in (1,2]\), the solution \(W(A,t,x_{0},y_{0})\) of the homogeneous initial value problems of (1.2) is expressed as follows,
By the calculation, we find
here
and
Hence, by (3.3), we can define the control function \(u_{1}(t)\) by a piecewise function,piecewise function,
here \(W(A,t,x_{0},y_{0})\) is expressed by (5.1), \(\Gamma _{0}^{2}\) is expressed by (5.2), \(x_{1}\) is the state we want to arrive. Therefore, under the control \(u_{1}\), we have \(x(2)=x_{1}\), see Fig. 1. Similarly, take \(y_{1}=(0~0)^\textrm{T}\), then we can take the control \(u_{2}\) as follows,
and under this control, we can steer the derivative of the solution of systems (1.2) to \((0~0)^{T}\) at terminal, see Fig. 2.
Example 5.2
Consider the systems (1.3) with
\(0=t_{0}<\frac{1}{2}=t_{1}< 1=t_{2}=b\), and \(f(t,x(t))=\frac{1}{37}\sin x(t)\). Since B is nonsingular, by Theorem 3.3, we find condition \((H_{2})\) is satisfied. Obviously, \((H_{1})\) is satisfied with \(L=\frac{1}{37}\). Then we show that (4.1) and (4.2) hold. Define a new matrix norm \(\Vert \cdot \Vert '\) by
where
and \(\Vert A\Vert =\max \limits _{1\le j\le n}\sum \nolimits _{i=1}^{n}|a_{ij}|\). According the Theorem 2.2.8 of [35], we find for \(A_{1}=I+\frac{B_{1}+B_{2}}{2}\),
Let \(\varepsilon =\frac{49}{100}\), then we have, for all \(x\in {\mathbb {R}}^{n}\),
hence,
combining this with (5.3), we have
that is \(T_{A_{1},\frac{49}{100}}=\frac{100}{98}\). Obviously, \(T_{A,\varepsilon }=1\), therefore, \(T_{\frac{49}{100}}=\frac{100}{98}\). With a simple calculation, we find
here
hence, for all \(x\in {\mathbb {R}}^{n}\),
that is we can pick \(\gamma =\frac{1}{10}\). Hence we obtain
Similarly, we can get
here
hence, for all \(x\in {\mathbb {R}}^{n}\),
i.e., we can pick \(\lambda =\frac{1}{10}\). Put these constants into (4.2), we obtain
Therefore, all the assumptions of Theorem 4.1 are satisfied, that is the systems (1.3) are exactly controllable.
6 Conclusion
In this paper, we introduce a new definition of controllability, and obtain some sufficient and necessary conditions of second-order linear impulsive systems, the rank criterion of impulsive systems is obtained as well. Then, we present some sufficient conditions of nonlinear impulsive systems provided the linear systems are controllable. Finally, some examples are presented to illustrate our results.
References
Lakshmikantham, V., Bainov, D.D., Simeonov, P.S.: Theory of Impulsive Differential Equations, vol. 6. World Scientific, Singapore (1989)
Sakthivel, R., Ren, Y., Mahmudov, N.I.: Approximate controllability of second-order stochastic differential equations with impulsive effects. Mod. Phys. Lett. B 24(14), 1559–1572 (2010)
Sakthivel, R., Mahmudov, N.I., Kim, J.H.: On controllability of second-order nonlinear impulsive differential systems. Nonlinear Anal. Theory Methods Appl. 71(1–2), 45–52 (2009)
Wang, J., Fan, Z., Zhou, Y.: Nonlocal controllability of semilinear dynamic systems with fractional derivative in Banach spaces. J. Optim. Theory Appl. 154(1), 292–302 (2012)
Fan, Z., Dong, Q., Li, G.: Approximate controllability for semilinear composite fractional relaxation equations. Fract. Calculus Appl. Anal. 19(1), 267–284 (2016)
Kumar, S., Tomar, N.K.: Mild solution and controllability of second-order non-local retarded semilinear systems. IMA J. Math. Control Inf. 37(1), 39–49 (2020)
Li, M., Ma, J.: Approximate controllability of second-order impulsive functional differential system with infinite delay in Banach spaces. J. Appl. Anal. Comput. 6(2), 492–514 (2016)
Mahmudov, N.I., McKibben, M.A.: Approximate controllability of second-order neutral stochastic evolution equations. Dyn. Contin. Discrete Impuls. Syst. Ser. B 13(5), 619 (2006)
Henríquez, H.R., Hernández, E.: Approximate controllability of second-order distributed implicit functional systems. Nonlinear Anal. Theory Methods Appl. 70(2), 1023–1039 (2009)
Kang, J.R., Kwun, Y.C., Park, J.Y.: Controllability of the second-order differential inclusion in Banach spaces. J. Math. Anal. Appl. 285(2), 537–550 (2003)
Chang, Y.K., Li, W.T.: Controllability of second-order differential and integro-differential inclusions in Banach spaces. J. Optim. Theory Appl. 129(1), 77–87 (2006)
Sakthivel, R., Ganesh, R., Ren, Y., Anthoni, S.M.: Approximate controllability of nonlinear fractional dynamical systems. Commun. Nonlinear Sci. Numer. Simul. 18, 3498–3508 (2013)
He, B., Zhou, H., Kou, C.: The controllability of fractional damped dynamical systems with control delay. Commun. Nonlinear Sci. Numer. Simul. 32, 190–198 (2016)
Chen, P., Zhang, X., Li, Y.: Approximate controllability of non-autonomous evolution system with nonlocal conditions. J. Dyn. Control Syst. 26(1), 1–16 (2020)
Yang, W., Wang, Y., Guan, Z., Wen, C.: Controllability of impulsive singularly perturbed systems and its application to a class of multiplex networks. Nonlinear Anal. Hybrid Syst. 31, 123–134 (2019)
Zhao, D., Liu, Y., Li, X.: Controllability for a class of semilinear fractional evolution systems via resolvent operators. Commun. Pure Appl. Anal. 18(1), 455 (2019)
Yan, J., Hu, B., Guan, Z., Cheng, X., Li, T.: Controllability analysis of complex-valued impulsive systems with time-varying delays. Commun. Nonlinear Sci. Numer. Simul. 83, 105070 (2020)
Chen, P., Zhang, X., Li, Y.: Existence and approximate controllability of fractional evolution equations with nonlocal conditions via resolvent operators. Fract. Calc. Appl. Anal. 23(1), 268–291 (2020)
Zhu, C., Li, X., Cao, J.: Finite-time \(H^{\infty }\) dynamic output feedback control for nonlinear impulsive switched systems. Nonlinear Anal. Hybrid Syst. 39, 011975 (2021)
Arora, S., Manil Mohan, T., Dabas, J.: Approximate controllability of the non-autonomous impulsive evolution equation with state-dependent delay in Banach spaces. Nonlinear Anal. Hybrid Syst. 39, 100989 (2021)
Bashirov, A.E., Ghahramanlou, N.: On partial approximate controllability of semilinear systems. Cogent Eng. 1(1), 965947 (2014)
Bashirov, A.E., Jneid, M.: On partial complete controllability of semilinear systems. Abstr. Appl. Anal. 2013, 521052 (2013)
Singh, S., Arora, S., Mohan, M., Dabas, J.: Approximate controllability of second order impulsive systems with state-dependent delay in Banach spaces. Evol. Equ. Control Theory 11(1), 67 (2022)
Mahmudov, N.I.: Finite-approximate controllability of fractional evolution equations: variational approach. Fract. Calc. Appl. Anal. 21(4), 919–936 (2018)
Liu, K.: Stability of Infinite Dimensional Stochastic Differential Equations with Applications. Chapman and Hall/CRC, London (2005)
Da Prato, G., Zabczyk, J.: Second Order Partial Differential Equations in Hilbert Spaces, vol. 293. Cambridge University Press, Cambridge (2002)
Chalishajar, D.N.: Controllability of second order impulsive neutral functional differential inclusions with infinite delay. J. Optim. Theory Appl. 154(2), 672–684 (2012)
Chang, Y.K., Li, W.T., Nieto, J.J.: Controllability of evolution differential inclusions in Banach spaces. Nonlinear Anal. Theory Methods Appl. 67(2), 623–632 (2007)
Travis, C.C., Webb, G.F. : Compactness,regularity,and uniform continuity properties of strongly continuous cosine families. Houston J. Math. 3(4) (1977)
Chalishajar, D.N., Chalishajar, H.D., Acharya, F.S.: Controllability of second order neutral impulsive differential inclusions with nonlocal conditions. Dyn. Contin. Discrete Impuls. Syst. Ser. A Math. Anal. 19, 107–134 (2012)
Balachandran, K., Kim., J.H.: Remarks on the paper controllability of second order differential inclusion in Banach spaces [J. Math. Anal. Appl. 285 (2003) 537-550]. J. Math. Anal. Appl. 324(1), 746–749 (2006)
Wang, J., Fečkan, M., Tian, Y.: Stability analysis for a general class of non-instantaneous impulsive differential equations. Mediterr. J. Math. 14(2), 1–21 (2017)
Wei, W., Xiang, X., Peng, Y.: Nonlinear impulsive integro-differential equations of mixed type and optimal controls. Optimization 55(1–2), 141–156 (2006)
Wen, Q., Wang, J., O’Regan, D.: Stability analysis of second order impulsive differential equations. Qual. Theory Dyn. Syst. 21(2), 54 (2022)
Ortega, J.M., Rheinboldt, W.C.: Iterative Solution of Nonlinear Equations in Several Variables. SIAM, Philadelphia (2000)
Acknowledgements
We would like to thank the referee for his/her important comments.
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.
This work is partially supported by the National Natural Science Foundation of China (12161015), Training Object of High Level and Innovative Talents of Guizhou Province ((2016)4006), Guizhou Data Driven Modeling Learning and Optimization Innovation Team ([2020]5016), the Slovak Research and Development Agency under the contract No. APVV-18-0308, and the Slovak Grant Agency VEGA No. 1/0358/20 and No. 2/0127/20.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Wen, Q., Fečkan, M. & Wang, J. The Controllability for Second-Order Semilinear Impulsive Systems. Qual. Theory Dyn. Syst. 22, 10 (2023). https://doi.org/10.1007/s12346-022-00717-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12346-022-00717-4