- Research
- Open access
- Published:
Razumikhin-type theorems for impulsive differential equations with piecewise constant argument of generalized type
Advances in Difference Equations volume 2018, Article number: 267 (2018)
Abstract
In this paper, we focus on developing Razumikhin technique for stability analysis of impulsive differential equations with piecewise constant argument. Based on the Lyapunov–Razumikhin method and impulsive control theory, we obtain some Razumikhin-type theorems on uniform stability, uniform asymptotic stability, and global exponential stability, which are rarely reported in the literature. The significance and novelty of the results lie in that the stability criteria admit the existence of piecewise constant argument and impulses, which may be either slight at infinity or persistently large. Examples are given to illustrate the effectiveness and advantage of the theoretical results.
1 Introduction
Qualitative theories of impulsive differential equations (IDEs) have been investigated by many researchers in the past three decades due to their potential applications in many fields such as biology, engineering, economics, physics, and so on. Among these theories, the stability problem is of great importance. By now a large number of results on stability problem for various IDE have been obtained by some classical methods and techniques; see [1–9] and the references therein.
In the 1980s, differential equations with piecewise constant argument (DEPCA) that contain deviation of arguments were initially proposed for investigation by Cooke, Wiener, Busenberg, and Shah [10–12]. Later, many interesting results have been obtained and applied efficiently to approximation of solutions and various models in biology, electronics, and mechanics [13–17]. Such equations represent a hybrid of continuous and discrete dynamical systems and combine the properties of both differential and difference equations. Akhmet [18–20] generalized the concept of DEPCA by considering arbitrary piecewise constant functions as arguments; the proposed approach overcomes the limitations in the previously used method of study, namely reduction to discrete equations. Afterward, the results of the theory have been further developed [21, 22] and applied for qualitative analysis and control problem of real models, for example, in neural network models with or without impulsive perturbations [23–33], which have great significance in solving engineering and electronic problems.
Razumikhin technique was originally proposed by Razumikhin [34, 35] for delay differential equations (DDE) and was generalized by other researchers to functional differential equations (FDE) and impulsive functional differential equations (IFDE) [36–44]. The idea of Razumikhin technique is to build a relationship between history and current states using Lyapunov functions, so it is usually called the Lyapunov–Razumikhin method. This method avoids the construction of complicated Lyapunov functionals and provides a technically efficient way to study stability problems for delayed systems or impulsive delayed systems. Considering that DEPCA is a delayed-type system, Akhmet et al. [41] investigated the stability of DEPCA and established some Razumikhin-type theorems on uniform stability and asymptotical stability and applied the results to a logistic equation, whereas impulsive perturbations were not taken into consideration. To the best of our knowledge, there have been few results on stability analysis obtained by the Lyapunov–Razumikhin method for impulsive DEPCA.
Motivated by this discussion, in this paper, we develop the Lyapunov–Razumikhin method for stability of impulsive differential equations with piecewise constant argument and establish some Razumikhin-type theorems on uniform stability, uniform asymptotic stability, and global exponential stability, which are rarely reported in the literature. To overcome the difficulties created by piecewise constant argument and impulses, which may be persistently large [44], as we will see, more complicated and interesting analysis is demanded, in which a (persistent) impulsive control plays an important role to achieve stability. This paper is organized as follows. In Sect. 2, we introduce some basic notations, lemmas, and definitions. In Sect. 3, we present the main theoretical results. In Sect. 4, we give some practical examples to illustrate the effectiveness and novelty of our results. Finally, we conclude the paper in Sect. 5.
2 Preliminaries
Let \(\mathbb{R}\) be the set of real numbers, \(\mathbb{R}_{+}\) the set of positive real numbers, \(\mathbb{Z}_{+}\) the set of non-negative integers, and \(\mathbb{R}^{n}\) the n-dimensional real space equipped with the Euclidean norm \(|\cdot |\). Fix a real-valued sequence \(\{\theta_{k}\}\) such that \(0=\theta_{0}<\theta_{1}<\cdots <\theta _{k}<\cdots \) with \(\theta_{k}\rightarrow \infty \) as \(k\rightarrow \infty \).
We use the following sets of functions:
-
\(\Omega_{1}=\{\varphi (s)\in C(\mathbb{R}_{+},\mathbb{R}_{+}), \mbox{strictly increasing}, \varphi (0)=0, 0<\varphi (s)<s, s>0\}\)
-
\(\Omega_{2}=\{\varphi (s)\in C(\mathbb{R}_{+},\mathbb{R}_{+}), \mbox{strictly increasing}, \varphi (0)=0\}\)
-
\(\Omega_{3}=\{\varphi (s)\in C(\mathbb{R}_{+},\mathbb{R}_{+}), \mbox{strictly increasing}, \varphi (0)=0, \varphi (s)>s, s>0\}\)
Consider the following system with impulses and piecewise constant argument:
where \(x\in \mathbb{R}^{n}\) is the state vector, \(x(\theta^{-}_{k})= \lim_{t\rightarrow \theta^{-}_{k}}x(t)\), and \(\beta (t)=\theta_{k}\) for \(t\in [\theta_{k}, \theta_{k+1})\), \(k\in \mathbb{Z}_{+}\), is the so-called piecewise constant argument.
We need the following assumptions [41]:
- (\(A_{1}\)):
-
\(f(t,x,y):\mathbb{R}_{+}\times \mathbb{R}^{n}\times \mathbb{R}^{n}\rightarrow \mathbb{R}^{n}\) is piecewise continuous with respect to t and is right-continuous at the possible discontinuous points \(\theta_{k}\), \(k\in \mathbb{Z}_{+}-\{0\}\); \(f(t,0,0)=0\) for all \(t\geq 0\), and f satisfies the Lipschitz condition
$$ \bigl\vert f(t,x_{1},y_{1})-f(t,x_{2},y_{2}) \bigr\vert \leq l\bigl(\vert x_{1}-x_{2}\vert +\vert y_{1}-y_{2}\vert \bigr) $$for all \(t\in \mathbb{R}_{+}\) and \(x_{1}, x_{2}, y_{1}, y_{2} \in \mathbb{R}^{n}\), where \(l>0\) is a constant;
- (\(A_{2}\)):
-
\(I_{k}(t,x):\mathbb{R}_{+}\times \mathbb{R}^{n}\rightarrow \mathbb{R}^{n}\) is continuous with respect to t and \(x, I_{k}(\theta _{k},0)=0\), and \(x_{1}\neq x_{2}\) implies \(x_{1}+I_{k}(\theta_{k}, x _{1})\neq x_{2}+I_{k}(\theta_{k}, x_{2})\) for all \(k\in \mathbb{Z} _{+}-\{0\}\).
- (\(A_{3}\)):
-
there exists a positive constant θ such that \(\theta_{k+1}-\theta_{k}\leq \theta \) for all \(k\in \mathbb{Z}_{+}\);
- (\(A_{4}\)):
-
\(l\theta [1+(1+l\theta)e^{l\theta }]<1\);
- (\(A_{5}\)):
-
\(3l\theta e^{l\theta }<1\).
Notation 1
([41])
\(K(l)=\frac{1}{1-l\theta [1+(1+l\theta)e^{l \theta }]}\).
Lemma 1
([41])
Under assumptions (\(A_{1}\))–(\(A_{5}\)),
for all \(t\geq 0\).
Definition 1
A function \(x(t)\) is called a solution of (1) on \([t_{0}, \infty)\) if
-
(i)
\(x(t)\) is continuous on each \([\theta_{k}, \theta_{k+1})\subseteq [t_{0}, \infty)\) and is right-continuous at \(t=\theta_{k}\), \(k\in \mathbb{Z}_{+}\);
-
(ii)
the derivative \(x'(t)\) exists for \(t\in [t_{0}, \infty)\) with the possible exception of the points \(\theta_{k}\), \(k\in \mathbb{Z}_{+}\), where the right-hand derivatives exist;
-
(iii)
system (1) is satisfied by \(x(t)\) on \([t_{0}, \infty)\).
We give the following statement assertion on the existence and uniqueness of solutions of the initial value problem (1).
Theorem 1
Assume that conditions (\(A_{1}\))–(\(A_{5}\)) are fulfilled. Then, for every \((t_{0}, x_{0})\in \mathbb{R}_{+}\times \mathbb{R}^{n}\), there exists a unique solution \(x(t)=x(t, t_{0}, x _{0})\) of (1) on \([t_{0}, \infty)\) such that \(x(t_{0})=x_{0}\).
Proof
Existence. Without loss of generality, we assume that \(\theta_{k}\leq t_{0}<\theta_{k+1}\) for some \(k\in \mathbb{Z}_{+}\). Define the norm \(\|x(t)\|=\max_{[\theta_{k}, \theta_{k+1}]}|x(t)|\), take \(x_{0}(t)=x_{0}\), \(t\in [\theta_{k}, \theta_{k+1}]\), and the sequence
We easily obtain that
Thus, (\(A_{5}\)) implies that the sequence \(\{x_{m}(t)\}\) uniformly converges to a unique function \(x(t)=x(t,t_{0},x_{0})\), and it is exactly the unique solution of the integral equation
on \([\theta_{k}, \theta_{k+1})\), which is equivalent to (1) on \([\theta_{k}, \theta_{k+1})\), and \(x(\theta_{k+1}^{-})\) exists. Moreover, (\(A_{2}\)) implies that \(x(\theta_{k+1})=x(\theta_{k+1}^{-})+I _{k+1}(\theta_{k+1}, x(\theta_{k+1}^{-}))\) also exists. Taking \(x(\theta_{k+1})\) as the new initial value, by the same arguments as before we can get the solution \(x(t)\) of (1) on \([\theta_{k+1}, \theta_{k+2})\). Since \(\theta_{k}\rightarrow \infty \) as \(k\rightarrow \infty \), induction completes the proof.
Uniqueness. Denote by \(x_{j}(t)=x(t, t_{0}, x_{0}^{j})\), \(x_{j}(t_{0})=x _{0}^{j}\), \(j=1,2\), solutions of (1), where \(\theta_{k}\leq t_{0}< \theta_{k+1}\). We will show that, for each \(t\in [\theta_{k}, \theta _{k+1})\), \(x_{0}^{1}\neq x_{0}^{2}\) implies \(x_{1}(t)\neq x_{2}(t)\). We have
Hence
The Gronwall–Bellman lemma yields that
Particularly,
Thus
Assume on the contrary that there exists \(t\in [\theta_{k}, \theta _{k+1})\) such that \(x_{1}(t)=x_{2}(t)\). Then
Inequalities (2)–(3) and (\(A_{5}\)) imply that
a contradiction. Especially, if \(x_{1}(\theta_{k+1}^{-})\neq x_{2}( \theta_{k+1}^{-})\), then (\(A_{2}\)) implies that \(x_{1}(\theta_{k+1}) \neq x_{2}(\theta_{k+1})\), and by the same arguments as before we can conclude that also \(x_{1}(t)\neq x_{2}(t)\) on \([\theta_{k+1}, \theta _{k+2})\), and induction completes the proof of uniqueness. Thus, the proof of Theorem 1 is complete. □
Remark 1
From the proof of Theorem 1 we can see that every solution of system (1) exists uniquely and is piecewise continuous on \([t_{0}, \infty)\). Moreover, every solution \(x(t)\) is right-continuous at the possible discontinuous points \(\theta_{k}\), and \(x(\theta_{k} ^{-})\) exists. In addition, system (1) obviously has the zero solution.
Definition 2
A function \(V:\mathbb{R}_{+}\times \mathbb{R} ^{n}\rightarrow \mathbb{R}_{+}\) is said to belong to the class \(v_{0}\) if
-
(i)
V is continuous on \([\theta_{k}, \theta_{k+1})\times \mathbb{R} ^{n}\), \(k\in \mathbb{Z}_{+}\) and \(V(t, 0)\equiv 0\) for all \(t\in \mathbb{R}_{+}\);
-
(ii)
V is continuously differentiable on \([\theta_{k}, \theta_{k+1}) \times \mathbb{R}^{n}\), \(k\in \mathbb{Z}_{+}\) and for each \(x\in \mathbb{R}^{n}\), its right-hand derivative exists at \(t=\theta_{k}, k \in \mathbb{Z}_{+}\).
Definition 3
Given a piecewise continuously differentiable Lyapunov function \(V(t,x)\in v_{0}\), the upper right-hand derivative of V with respect to system (1) is defined by
for all \(t\in \mathbb{R}_{+}\) and \(x,y\in \mathbb{R}^{n}\). In particular, for \(t\neq \theta_{k}\) and \(x,y\in \mathbb{R}^{n}\), we have
Besides, several basic definitions such as uniform stability, uniform asymptotical stability, and global weak exponential stability are the same as those in [1, 7], and so we omit them.
3 Main results
In this section, under the same assumptions, we obtain the stability of the zero solution of system (1) based on the Lyapunov–Razumikhin method. Firstly, we present some uniform stability results.
Theorem 2
Assume that there exist functions \(V\in v_{0}\), \(u,v \in \Omega_{2}\), such that
-
(i)
\(u(|x|)\leq V(t,x)\leq v(|x|)\), \((t,x)\in [t_{0},\infty)\times \mathbb{R}^{n}\);
-
(ii)
for all \(t\in [\theta_{k},\theta_{k+1})\), \(k\in \mathbb{Z}_{+}\), and \(x, y\in \mathbb{R}^{n}\), \(V(\beta (t),y)\leq V(t,x)\) implies that
$$ D^{+}V(t,x,y)\leq 0; $$ -
(iii)
for all \(k\in \mathbb{Z}_{+}-\{0\}\) and \(x\in \mathbb{R}^{n}\), we have \(V(\theta_{k},x+I_{k}(\theta_{k},x))\leq (1+b_{k})V(\theta ^{-}_{k},x)\), where \(b_{k}\geq 0\) with \(\sum^{\infty }_{k=1} b_{k}< \infty\).
Then the zero solution of (1) is uniformly stable.
Theorem 3
Assume that there exist functions \(V\in v_{0}\), \(u,v \in \Omega_{2}\), \(\psi \in \Omega_{1}\), and \(W\in \Omega_{2}\) such that
-
(i)
\(u(|x|)\leq V(t,x)\leq v(|x|)\), \((t,x)\in [t_{0},\infty)\times \mathbb{R}^{n}\);
-
(ii)
for all \(t\in [\theta_{k},\theta_{k+1})\), \(k\in \mathbb{Z}_{+}\), and \(x, y\in \mathbb{R}^{n}\), \(V(\beta (t),y)\leq \psi^{-1}(V(t,x))\) implies that
$$ D^{+}V(t,x,y)\leq g(t)W\bigl(V(t,x)\bigr), $$where \(g:[t_{0},\infty)\rightarrow \mathbb{R}_{+}\) is locally integrable;
-
(iii)
for all \(k\in \mathbb{Z}_{+}-\{0\}\) and \(x\in \mathbb{R}^{n}\), \(V( \theta_{k},x+I_{k}(\theta_{k},x))\leq \psi (V(\theta^{-}_{k},x))\);
-
(iv)
for all \(k\in \mathbb{Z}_{+}\), \(\inf_{\mu \in \mathbb{R}_{+}} \int^{\mu }_{\psi (\mu)}\frac{ds}{W(s)}>\int^{\theta_{k+1}}_{\theta _{k}}g(s)\,ds\).
Then the zero solution of (1) is uniformly stable.
Remark 2
We may observe that the two theorems generalize the existing corresponding results, and their proofs can be formulated by combining the corresponding theorems in [38] and [41], and so we omit them.
Theorem 4
Assume that there exist functions \(V\in v_{0}\), \(u,v \in \Omega_{2}\), \(\psi \in \Omega_{3}\), and \(W\in \Omega_{2}\) such that
-
(i)
\(u(|x|)\leq V(t,x)\leq v(|x|)\), \((t,x)\in [t_{0},\infty)\times \mathbb{R}^{n}\);
-
(ii)
for all \(t\in [\theta_{k},\theta_{k+1})\), \(k\in \mathbb{Z}_{+}\), and \(x, y\in \mathbb{R}^{n}\), \(V(\beta (t),y)< \psi (V(t,x))\) implies that
$$ D^{+}V(t,x,y)\leq -g(t)W\bigl(V(t,x)\bigr), $$where \(g:[t_{0},\infty)\rightarrow \mathbb{R}_{+}\) is locally integrable;
-
(iii)
for all \(k\in \mathbb{Z}_{+}-\{0\}\) and \(x\in \mathbb{R}^{n}\), \(V( \theta_{k},x+I_{k}(\theta_{k},x))\leq \psi (V(\theta^{-}_{k},x))\);
-
(iv)
for all \(k\in \mathbb{Z}_{+}\), \(\sup_{\nu \in \mathbb{R}_{+}} \int^{\psi (\nu)}_{\nu }\frac{ds}{W(s)}<\int^{\theta_{k+1}}_{\theta _{k}}g(s)\,ds\).
Then the zero solution of (1) is uniformly stable.
Proof
For given \(\varepsilon >0\), we may choose \(\delta >0\) such that \(\psi (v(\delta))< u(\varepsilon)\). For any \(t_{0}\geq 0\) and \(|x_{0}|<\delta \), we shall show that \(|x(t)|<\varepsilon\), \(t\geq t _{0}\). To show that this δ is the needed one, we consider two cases where \(t_{0}=\theta_{i}\) for some \(i\in \mathbb{Z}_{+}\) and another one where \(t_{0}\neq \theta_{j}\) for all \(j\in \mathbb{Z}_{+}\).
First, let \(t_{0}=\theta_{m-1}\) for some \(m\in \mathbb{Z}_{+}-\{0\}\). For convenience, we take \(V(t)=V(t,x(t))\), \(D^{+}V(t)=D^{+}V(t,x(t),x( \beta (t)))\). We first claim that
Clearly, \(V(t_{0})\leq v(\delta)\). If (4) does not hold, then there exist points \(t_{1}\) and \(t_{2}, t_{0}\leq t_{1}< t_{2}<\theta_{m}\), such that \(V(t_{1})=v(\delta)\) and \(V(t)>v(\delta)\) for \(t\in (t_{1},t _{2}]\). Applying the mean-value theorem, we get
for some \(\hat{t}\in (t_{1},t_{2})\). Since \(\psi (V(\hat{t}))>V( \hat{t})>v(\delta)\geq V(t_{0})=V(\beta (\hat{t}))\), it follows from condition (ii) that \(D^{+}V(\hat{t})\leq -g(\hat{t})W(V(\hat{t}))<0\), which contradicts (5), and so (4) holds. From (4) and condition (iii) we obtain
Using the same argument as before, we can prove that
We now further claim that
Suppose not, that is, \(V(\theta^{-}_{m+1})>v(\delta)\). There are two cases: (a) \(V(t)>v(\delta)\) for all \(t\in [\theta_{m},\theta_{m+1})\), and (b) there exists \(t\in [\theta_{m},\theta_{m+1})\) such that \(V(t)\leq v(\delta)\). For case (a), we have \(\psi (V(t))>\psi (v( \delta))\geq V(\beta (t))\) for all \(t\in [\theta_{m},\theta_{m+1})\). By condition (ii) we get
Integrating this inequality yields
which is a contradiction with condition (iv). For case (b), we set
Obviously, \(V(\bar{t})=v(\delta)\) for \(\bar{t}<\theta_{m+1}\). Then there exists \(\tilde{t}\in (\bar{t},\theta_{m+1})\) such that \(D^{+}V(\tilde{t})>0\), whereas since \(\psi (V(\tilde{t}))>\psi (v( \delta))\geq V(\beta (\tilde{t}))\), condition (ii) implies that \(D^{+}V(\tilde{t})<0\), a contradiction. By now, we get the following statement:
By the same argument as in the proofs of (6) and (7), in general, we can deduce that
Hence from condition (i) and (8) we have
which implies that \(|x(t)|<\varepsilon\), \(t\geq t_{0}\).
Now, let \(t_{0}\geq 0\) with \(t_{0}\neq \theta_{i}\) for any \(i\in \mathbb{Z}_{+}\). By the idea in [41] and Lemma 1 we take \(\delta_{1}=\frac{ \delta }{K(l)}\), where δ satisfies \(\psi (v(\delta))< u( \varepsilon)\). Then \(|x_{0}|<\delta_{1}\) implies \(|x(t)|<\varepsilon\), \(t\geq t_{0}\). We see that the evaluation of \(\delta_{1}\) is independent of \(t_{0}\). The proof of Theorem 4 is complete. □
Remark 3
It should be noted that Theorem 4 allows for significant increases in V at impulse times, which may be persistently large (\(\psi (s)>s\) for \(s>0\)) and appropriately controlled by the length of impulsive intervals. So, the length of impulsive intervals cannot be too small; in other words, the disturbed impulses cannot happen too frequently, with the same idea as that in [44]; yet the impulse conditions presented in Theorem 4 are more general than that in [44] and can be verified more easily and conveniently.
Now, we present uniform asymptotical stability and exponential stability results.
Theorem 5
Assume that there exist functions \(V\in v_{0}\), \(u,v \in \Omega_{2}\), \(\psi \in \Omega_{3}\), and \(W\in \Omega_{2}\) such that
-
(i)
\(u(|x|)\leq V(t,x)\leq v(|x|)\), \((t,x)\in [t_{0},\infty)\times \mathbb{R}^{n}\);
-
(ii)
for all \(k\in \mathbb{Z}_{+}-\{0\}\) and \(x\in \mathbb{R}^{n}\), \(V( \theta_{k},x+I_{k}(\theta_{k},x))\leq (1+b_{k})V(\theta^{-}_{k},x)\), where \(b_{k}\geq 0\) with \(\bar{M}=\sum^{\infty }_{k=1} b_{k}<\infty\).
-
(iii)
for all \(t\in [\theta_{k},\theta_{k+1})\), \(k\in \mathbb{Z}_{+}\) and \(x, y\in \mathbb{R}^{n}\), \(V(\beta (t),y)<\psi (V(t,x))\) implies that
$$ D^{+}V(t,x,y)\leq -W\bigl(\vert x\vert \bigr), $$where \(\psi (s)>Ms\) for \(s>0\) with \(M=\prod^{\infty }_{k=1}(1+b_{k})\).
Then the zero solution of (1) is uniformly asymptotically stable.
Proof
Clearly, the conditions of Theorem 5 imply the uniform stability by Theorem 2.
First, let \(t_{0}=\theta_{m-1}\) for some \(m\in \mathbb{Z}_{+}-\{0\}\). We take \(V(t)=V(t,x(t))\), \(D^{+}V(t)=D^{+}V(t,x(t), x(\beta (t)))\) for convenience. For given \(\rho >0\), we choose \(\delta >0\) such that \(Mv(\delta)=u(\rho)\), and \(|x(t_{0})|<\delta \) implies that, for \(t\geq t_{0}\),
In what follows, we show that, for arbitrary ε, \(0<\varepsilon <\rho \), there exists \(T=T(\varepsilon)>0\) such that \(|x(t)|\leq \varepsilon \) for \(t\geq t_{0}+T\) if \(|x(t_{0})|<\delta \).
Obviously, there exists a number \(a>0\) such that \(\psi (s)-Ms>a\) for \(M^{-1}u(\varepsilon)\leq s\leq Mv(\delta)\). Let \(N=N(\varepsilon)\) be the smallest positive integer such that \(M^{-1}(u(\varepsilon)+Na) \geq Mv(\delta)\). Choose \(t_{k}=k(\frac{Mv(\delta)(1+\bar{M})}{ \gamma }+\theta)+\theta_{m-1}\), \(k=1,2,\ldots,N\), where
We will prove that
We have \(V(t)\leq Mv(\delta)\leq M^{-1}(u(\varepsilon)+Na)\leq u( \varepsilon)+Na\) for \(t\geq t_{0}=\theta_{m-1}\). Hence (9) holds for \(k=0\). Now, suppose that (9) holds for some \(0< k< N\). Let us show that
Let \(J_{k}=[\beta (t_{k})+\theta, t_{k+1}]\). We first claim that there exists \(t^{*}\in J_{k}\) such that
Otherwise, \(V(t)>M^{-1}[u(\varepsilon)+(N-k-1)a]\) for all \(t\in J _{k}\). On the other side, \(V(t)\leq u(\varepsilon)+(N-k)a\) for \(t\geq t_{k}\) implies that
Hence, for \(t\in J_{k}\),
It follows from hypothesis (iii) and the definition of γ that
We get
a contradiction, and so (11) holds.
Let \(q=\min \{i\in \mathbb{Z}_{+}:\theta_{i}>t^{*}\}\). We claim that
If (12) does not hold, then there exists \(\bar{t}\in (t^{*}, \theta _{q})\) such that
Thus, there exists \(\tilde{t}\in (t^{*}, \bar{t})\) such that \(\tilde{t}\neq \theta_{i}\), \(D^{+}V(\tilde{t})>0\), and \(V(\tilde{t})>M ^{-1}[u(\varepsilon)+(N-k-1)a]\). However,
implies that \(D^{+}V(\tilde{t})\leq -\gamma <0\), a contradiction, so (12) holds.
From (12) and (ii) we have
Therefore, for all \(t\in [t^{*}, \theta_{q}]\),
Similarly, we can show that, for all \(t\in [\theta_{q}, \theta_{q+1}]\),
and by a simple induction we conclude that, for \(t\in [\theta_{q+i}, \theta_{q+i+1}]\), \(i=0,1,2,\ldots \) ,
Thus \(V(t)\leq u(\varepsilon)+(N-k-1)a\) for \(t\geq t^{*}\), and so (10) holds.
For \(k=N\), we get
Hence \(|x(t)|\leq \varepsilon \) for \(t\geq t_{0}+T\), where \(T=N[\frac{Mv( \delta)(1+\bar{M})}{\gamma }+\theta ]\), proving the uniform asymptotic stability for \(t_{0}=\theta_{m-1}\), \(m\in \mathbb{Z}_{+}-\{0\}\).
Now, let \(t_{0}\neq \theta_{i}\) for any \(i\in \mathbb{Z}_{+}\). Similar to the arguments in Theorem 4, taking \(\delta_{1}=\frac{\delta }{K( \rho)}\), we obtain that \(|x(t_{0})|<\delta_{1}\) implies \(|x(t)|< \varepsilon\), \(t\geq t_{0}+T\). The proof of Theorem 5 is complete. □
Theorem 6
Assume that there exist functions \(V\in v_{0}\), \(u,v,W \in \Omega_{2}\) and constants \(\eta >0\), \(\mu >1\) such that
-
(i)
\(u(|x|)\leq V(t,x)\leq v(|x|)\), \((t,x)\in [t_{0},\infty)\times \mathbb{R}^{n}\);
-
(ii)
for all \(t\in [\theta_{k},\theta_{k+1})\), \(k\in \mathbb{Z}_{+}\) and \(x, y\in \mathbb{R}^{n}\), \(e^{\eta \beta (t)}V(\beta (t),y)< \mu e^{ \eta t}V(t,x)\) implies that
$$ D^{+}V(t,x,y)\leq -g(t)W\bigl(V(t,x)\bigr), $$where \(g:[t_{0},\infty)\rightarrow \mathbb{R}_{+}\) is locally integrable;
-
(iii)
for all \(k\in \mathbb{Z}_{+}-\{0\}\) and \(x\in \mathbb{R}^{n}\), \(V( \theta_{k},x+I_{k}(\theta_{k},x))\leq \mu V(\theta^{-}_{k},x)\);
-
(iv)
\(\inf_{t\in \mathbb{R}_{+}}g(t)\inf_{t\in \mathbb{R}_{+}} \frac{W(t)}{t}>\frac{1}{\tau }\ln \mu\), \(\tau =\inf_{k\in \mathbb{Z} _{+}}\{\theta_{k+1}-\theta_{k}\}\).
Then the zero solution of (1) is globally weakly exponentially stable.
Proof
Set
Given a constant ε such that
for any \(t_{0}\geq 0\), we shall prove that \(u(|x(t)|)\leq \phi (|x _{0}|)e^{-\varepsilon (t-t_{0})}\), \(t\geq t_{0}\), where \(\phi ( \mathbb{R}_{+},\mathbb{R}_{+})\in \Omega_{2}\). We also consider two cases where \(t_{0}=\theta_{i}\) for some \(i\in \mathbb{Z}_{+}\) and where \(t_{0}\neq \theta_{j}\) for all \(j\in \mathbb{Z}_{+}\).
First, let \(t_{0}=\theta_{m-1}\) for some \(m\in \mathbb{Z}_{+}-\{0\}\). We take \(V(t)=V(t,x(t))\), \(D^{+}V(t)=D^{+}V(t,x(t), x(\beta (t)))\) for convenience and define \(\Phi (t)=V(t)e^{\varepsilon (t-t_{0})}\), \(t \geq t_{0}\). We will further show that
Clearly, \(\Phi (t_{0})=V(t_{0})\leq v(|x_{0}|)\). If (15) does not hold, then there exist \(t_{1}\) and \(t_{2}, t_{0}\leq t_{1}< t_{2}<\theta _{m}\), such that \(\Phi (t_{1})=v(|x_{0}|)\) and \(\Phi (t)> v(|x_{0}|)\) for \(t\in (t_{1}, t_{2}]\). Applying the mean-value theorem, we get
for some \(\hat{t}\in (t_{1},t_{2})\), and \(\Phi (\hat{t})>v(|x_{0}|) \geq \Phi (t_{0})=\Phi (\beta (\hat{t}))\). From the definition of Φ, together with (14), we have
It follows from condition (ii), (13), and (14) that
which contradicts (16), and so (15) holds. From (15) and condition (iii) we obtain
Using the same argument as before, we can prove that
We now further claim that
Suppose not, that is, \(\Phi (\theta^{-}_{m+1})>v(|x_{0}|)\). There are two cases: (a) \(\Phi (t)>v(|x_{0}|)\) for all \(t\in [\theta_{m}, \theta_{m+1})\), and (b) there exists some \(t\in [\theta_{m},\theta _{m+1})\) such that \(\Phi (t)\leq v(|x_{0}|)\). In case (a), we have \(\mu \Phi (t)>\mu v(|x_{0}|)\geq \Phi (\beta (t))\) for all \(t\in [ \theta_{m},\theta_{m+1})\), which implies that \(\mu e^{\eta t}V(t)>e ^{\eta \beta (t)}V(\beta (t))\), \(t\in [\theta_{m},\theta_{m+1})\). By condition (ii) we get
Integrating this inequality yields that
which is a contradiction with (14). For case (b), we set
Obviously, \(\bar{t}<\theta_{m+1}\) and \(\Phi (\bar{t})=v(|x_{0}|)\). Then there exists \(\tilde{t}\in (\bar{t},\theta_{m+1})\) such that \(D^{+}\Phi (\tilde{t})>0\), whereas for all \(t\in [\bar{t},\theta_{m+1})\), we have \(\mu \Phi (t)\geq \mu v(|x_{0}|) \geq \Phi (\beta (t))\), which implies \(\mu V(t)e^{\eta t}\geq V( \beta (t))e^{\eta \beta (t)}\), and by condition (ii) we get \(D^{+}\Phi (t)<0\), \(t\in [\bar{t}, \theta_{m+1})\), which is also a contradiction. By now, we get the following statement:
By the same argument as in the proofs of (17) and (18), in general, we can deduce that
Hence we have \(\Phi (t)\leq \mu v(|x_{0}|)\) for all \(t\geq t_{0}\), which implies that
Now, let \(t_{0}\neq \theta_{i}\) for any \(i\in \mathbb{Z}_{+}\). Similarly to the arguments in Theorem 4, we can obtain
The proof of Theorem 6 is complete. □
4 Examples
In this section, we give some examples to illustrate the theoretical results obtained in the previous section.
Example 1
Consider the system
where \(a>0\), \(p>1\), \(b\in \mathbb{R}\) are some constants, satisfying \(a-|b|p>0\), \(\theta_{k+1}-\theta_{k}>\frac{\ln p}{a-|b|p}\).
Let \(V(t,x)=|x|\), \(\psi (s)=ps\), \(W(s)=s\). Then
and for all \(t\neq \theta_{k}\), \(V(\beta (t),x(\beta (t)))< pV(t,x)\), which means that \(|x(\beta (t))|< p|x(t)|\) implies
where \(g(t)=a-|b|p>0\). For all \(k\in \mathbb{Z}_{+}\),
Therefore by Theorem 4 the zero solution of system (20) is uniformly stable.
In particular, let \(a=e^{0.3}+1\), \(p=e^{0.3}\), \(b=1\), and \(\theta_{k}=0.6k\). Then the simulation of system (20) is shown in Fig. 1, which illustrates the stability of zero solution.
Example 2
Consider the system
where \(a(t)\) and \(b(t)\) are continuous functions on \([0,+\infty)\) satisfying \(a(t)\geq a>0\) and \(|b(t)|\leq b\), \(I_{k}(t,x)\in C( \mathbb{R}_{+}\times \mathbb{R}^{n},\mathbb{R}^{n})\) satisfy \(|x+I_{k}(t,x)|\leq (1+b_{k})|x|\) for \(t\in \mathbb{R}_{+}\), \(x\in \mathbb{R}^{n}\), with \(b_{k}\geq 0\) and \(\sum^{\infty }_{k=1}b_{k}< \infty \). Also suppose that \(a-bM>0\) with \(M=\prod^{\infty }_{k=1}(1+b _{k})\).
Let \(V(t,x)=|x|, \psi (s)=qMs\), where \(q>1\) is such that \(a-bqM>0\). Then
and for all \(t\neq \theta_{k}\), \(V(\beta (t),x(\beta (t)))<\psi (V(t,x))\), which means \(|x(\beta (t))|< qM|x(t)|\), implies that
Therefore by Theorem 5 the zero solution of system (21) is uniformly asymptotically stable.
Example 3
Consider the system
where \(a(t)\) and \(b(t)\) are continuous functions on \([0,+\infty)\), \(p>1\) is a given constant, and there exist constants \(\gamma >0\) and \(\eta >0\) such that \(e^{\beta (t)}(a(t)-\gamma)\geq p|b(t)|e^{\eta t}\) for all \(t\geq 0\). Also, suppose that \(\gamma (\theta_{k+1}-\theta _{k})>\ln p\) for all \(k\in \mathbb{Z}_{+}\).
Let \(V(t,x)=|x|\), \(W(t)=\gamma t\), and \(g(t)\equiv 1\). Then
and for all \(t\neq \theta_{k}\), \(e^{\eta \beta (t)}V(\beta (t),x(\beta (t)))< pe^{\eta t}V(t,x)\), which means \(e^{\eta \beta (t)}|x(\beta (t))|< pe ^{\eta t}|x(t)|\), implies that
Also,
Therefore by Theorem 6 the zero solution of system (22) is globally weakly exponentially stable.
In particular, let \(p=e^{0.5}\), \(\gamma =1\), \(\eta =0.2\), \(b(t)=\sin t\), \(\theta_{k}=0.6k\), and \(a(t)=1+ |\sin t|e^{0.2t+0.5-0.6k}\), \(t\in [\theta _{k}, \theta_{k+1})\), \(k\in \mathbb{Z}_{+}\). Then the simulation of system (22) is shown in Fig. 2, which illustrates the global exponential stability of zero solution. In addition, if we let \(\theta_{k}=0.4k\) and other parameters be fixed, then it will go against the required conditions. In this case, Fig. 3 tells us that the system becomes unstable, which shows the sharpness of our results.
Remark 4
On one hand, the stability of the systems in the examples may not be obtained by the results in existing references due to the existence of both impulse and piecewise constant argument. So the results in this paper are more general than those in the references. On the other hand, it should be noted that in the three systems, the sequence \(\{\theta_{k}\}\) needs to satisfy conditions (A3), (A4), and (A5). I particular, in Examples 1 and 3, considering the presence of persisting impulses, we give the lower bound of θ (e.g., \(\theta_{k+1}-\theta_{k}>\frac{\ln p}{a-|b|p}\) in Example 1), and, in fact, the range of the parameter θ is constructed for the stability of zero solution in Examples 1 and 3.
5 Conclusion
In this paper, we have derived several stability theorems for nonlinear systems with impulses and piecewise constant argument by employing the- Lyapunov-Razumikhin method and impulsive control theory. Examples are also given to show the effectiveness and novelty of the results. The theoretical results obtained can be applied to study the stability problem of many nonlinear impulsive models with piecewise constant argument such as neural networks, population models, and other biological models. However, our results are based on the fact that the piecewise constant argument in systems is of retarded type, so it is interesting to develop the Lyapunov–Razumikhin technique to impulsive systems with piecewise constant argument of advanced type or hybrid type, which requires further research in the future.
Abbreviations
- IDE:
-
impulsive differential equation
- DEPCA:
-
differential equation with piecewise constant argument
- DDE:
-
delay differential equation
- FDE:
-
functional differential equation
- IFDE:
-
impulsive functional differential equation
References
Lakshmikantham, V., Bainov, D., Simeonov, P.: Theory of Impulsive Differential Equations. World Scientific, Singapore (1989)
Samoilenko, A., Perestyuk, N.: Impulsive Differential Equations. World Scientific Series on Nonlinear Science Series A Monogr. Treatises, vol. 14. World Scientific, Singapore (1995)
Akhmet, M.: Principles of Discontinuous Dynamical Systems. Springer, New York (2010)
Xi, Q.: Global exponential stability for a class of generalized delayed neural networks with impulses. Math. Methods Appl. Sci. 34, 1414–1420 (2011)
Bonotto, E., Jimenez, M.: On impulsive semidynamical systems: minimal, recurrent and almost periodic motions. Topol. Methods Nonlinear Anal. 44, 121–141 (2014)
Li, X., Bohner, M., Wang, C.: Impulsive differential equations: periodic solutions and applications. Automatica 52, 173–178 (2015)
Li, X., Wu, J.: Stability of nonlinear differential systems with state-dependent delayed impulses. Automatica 64, 63–69 (2016)
Li, X., Cao, J.: An impulsive delay inequality involving unbounded time-varying delay and applications. IEEE Trans. Autom. Control 62, 3618–3625 (2017)
Li, X., Song, S.: Stabilization of delay systems: delay-dependent impulsive control. IEEE Trans. Autom. Control 62(1), 406–411 (2017)
Busenberg, S., Cooke, K.: Models of vertically transmitted disease with sequential continuous dynamics. In: Lakshmikantham, V. (ed.) Nonlinear Phenomena in Mathematical Sciences, pp. 179–189. Academic Press, New York (1982)
Shah, S., Wiener, J.: Advanced differential equations with piecewise constant argument deviations. Int. J. Math. Sci. 6, 671–703 (1983)
Cooke, K., Wiener, J.: Retarded differential equations with piecewise constant delays. J. Math. Anal. Appl. 99, 265–297 (1984)
Györi, I., Ladas, G.: Oscillation Theory of Delay Differential Equations with Applications. Oxford University Press, New York (1991)
Wiener, J.: Generalized Solutions of Functional Differential Equations. World Scientific, Singapore (1993)
Dai, L., Singh, M.: On oscillatory motion of spring-mass systems subjected to piecewise constant forces. J. Sound Vib. 173, 217–232 (1994)
Muroya, Y.: Persistence, contractivity and global stability in logistic equations with piecewise constant delays. J. Math. Anal. Appl. 270, 602–635 (2002)
Györi, I., Hartung, F.: On numerical approximation using differential equations with piecewise-constant arguments. Period. Math. Hung. 56, 55–69 (2008)
Akhmet, M.: Integral manifolds of the differential equations with piecewise constant argument of generalized type. Nonlinear Anal. 66, 367–383 (2007)
Akhmet, M.: On the reduction principle for differential equations with piecewise constant argument of generalized type. J. Math. Anal. Appl. 336, 646–663 (2007)
Akhmet, M.: Stability of differential equations with piecewise constant arguments of generalized type. Nonlinear Anal. 68, 794–803 (2008)
Xia, Y., Huang, Z., Han, M.: Existence of almost periodic solutions for forced perturbed systems with piecewise constant argument. J. Math. Anal. Appl. 333, 798–816 (2007)
Huang, Z., Wang, X., Xia, Y.: A topological approach to the existence of solutions for nonlinear differential equations with piecewise constant argument. Chaos Solitons Fractals 39, 1121–1131 (2009)
Akhmet, M., Yılmaz, E.: Impulsive Hopfield-type neural network system with piecewise constant argument. Nonlinear Anal., Real World Appl. 11, 2584–2593 (2010)
Akhmet, M., Aruğaslan, D., Yılmaz, E.: Stability analysis of recurrent neural networks with piecewise constant argument of generalized type. Neural Netw. 23, 805–811 (2010)
Akhmet, M., Aruğaslan, D., Yılmaz, E.: Stability in cellular neural networks with a piecewise constant argument. J. Comput. Appl. Math. 233, 2365–2373 (2010)
Bao, G., Wen, S., Zeng, Z.: Robust stability analysis of interval fuzzy Cohen–Grossberg neural networks with piecewise constant argument of generalized type. Neural Netw. 33, 32–41 (2012)
Akhmet, M., Yılmaz, E.: Global exponential stability of neural networks with nonsmooth and impact activations. Neural Netw. 34, 18–27 (2012)
Abbas, S., Xia, Y.: Almost automorphic solutions of impulsive cellular neural networks with piecewise constant argument. Neural Process. Lett. 42(3), 1–12 (2014)
Xi, Q.: Global exponential stability of Cohen–Grossberg neural networks with piecewise constant argument of generalized type and impulses. Neural Comput. 28, 229–255 (2016)
Wan, L., Wu, A.: Stabilization control of generalized type neural networks with piecewise constant argument. J. Nonlinear Sci. Appl. 9, 3580–3599 (2016)
Chiu, K.: Exponential stability and periodic solutions of impulsive neural network models with piecewise constant argument. Acta Appl. Math. 2, 1–28 (2017). https://doi.org/10.1007/s10440-017-0108-3
Wu, A., Liu, L., Huang, T., Zeng, Z.: Mittag-Leffler stability of fractional-order neural networks in the presence of generalized piecewise constant arguments. Neural Netw. 85, 118–127 (2017)
Yu, T., Cao, D.: Stability analysis of impulsive neural networks with piecewise constant arguments. Neural Process. Lett. 47, 153–165 (2018). https://doi.org/10.1007/s11063-017-9638-y
Razumikhin, B.: On the stability of systems with a delay. Prikl. Mat. Meh. 20, 500–512 (1956) (translated into English in Journal of Applied Mathematics and Mechanics)
Razumikhin, B.: Applications of Lyapunov’s method to problems in the stability of systems with a delay. Autom. Remote Control 21, 515–520 (1960)
Seifert, G.: Lyapunov–Razumikhin conditions for stability and boundedness of functional differential equations of Volterra type. J. Differ. Equ. 14, 424–430 (1973)
Hale, J., Lunel, S.: Introduction to Functional Differential Equations. Springer, Berlin (1993)
Shen, J.: Razumiknin techniques in impulsive functional differential equations. Nonlinear Anal. TMA 36, 119–130 (1999)
Jankovic, M.: Control Lyapunov–Razumikhin functions and robust stabilization of time delay systems. IEEE Trans. Autom. Control 46, 1048–1060 (2001)
Stamova, I., Stamov, G.: Lyapunov–Razumikhin method for impulsive functional differential equations and applications to the population dynamics. J. Comput. Appl. Math. 130, 163–171 (2001)
Akhmet, M., Aruğaslan, D.: Lyapunov–Razumikhin method for differential equations with piecewise constant argument. Discrete Contin. Dyn. Syst. 25(2), 457–466 (2009)
Wang, Q., Zhu, Q.: Razumiknin-type stability criteria for differential equations with delayed impulses. Electron. J. Qual. Theory Differ. Equ. 14, 1 (2013)
Li, X., Deng, F.: Razumiknin method for impulsive functional differential equations of neutral type. Chaos Solitons Fractals 101, 41–49 (2017)
Li, X., Ding, Y.: Razumiknin-type theorems for time-delay systems with persistent impulses. Syst. Control Lett. 107, 22–27 (2017)
Acknowledgements
The author wishes to express his sincere gratitude to the editors and reviewers for their useful suggestions, which helped to improve the paper. This work is supported by National Natural Science Foundation of China (11601269, 61503214), Natural Science Foundation of Shandong Province (ZR2017MA048), and A Project of Shandong Province Higher Educational Science and Technology Program (J15LI02). The paper has not been presented at any conference.
Availability of data and materials
Not applicable.
Authors’ information
Qiang Xi received the B.S. and M.S. degrees in applied mathematics from Liaocheng University, Liaocheng, China, in 2003 and Shandong Normal University, Ji’nan, China, in 2006, respectively, and the Ph.D. degree in fundamental mathematics from Shandong University, Ji’nan, China, in 2014. He is currently an associate professor with School of Mathematics and Quantitative Economics, Shandong University of Finance and Economics. His current research interests include stability theory, delay systems, impulsive control theory, artificial neural networks, and applied mathematics
Funding
National Natural Science Foundation of China (11601269); National Natural Science Foundation of China (61503214); Natural Science Foundation of Shandong Province (ZR2017MA048); the Project of Shandong Province Higher Educational Science and Technology Program (J15LI02).
Author information
Authors and Affiliations
Contributions
The main idea of this paper was proposed by QX. QX prepared the manuscript initially and performed all the steps of the proofs in this research. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Ethics approval and consent to participate
Not applicable.
Competing interests
The author declares that there is no competing interest regarding the publication of this paper.
Consent for publication
Not applicable.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Xi, Q. Razumikhin-type theorems for impulsive differential equations with piecewise constant argument of generalized type. Adv Differ Equ 2018, 267 (2018). https://doi.org/10.1186/s13662-018-1725-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13662-018-1725-5