Skip to content

Advertisement

Open Access

Exponential dichotomy on time scales and admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\)

Advances in Difference Equations20152015:69

https://doi.org/10.1186/s13662-015-0409-7

Received: 15 December 2014

Accepted: 9 February 2015

Published: 1 March 2015

Abstract

In this paper, we study a relation between exponential dichotomy on time scales and admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) for an evolution family on time scales. We establish a sufficient criterion for the existence of exponential dichotomy on time scales in terms of the admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) for the evolution family. Conversely, with the help of exponential dichotomy on time scales, we give the admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) for an input-output equation on time scales.

Keywords

time scalesexponential dichotomyadmissibility

MSC

34N0534D09

1 Introduction

The concept of exponential dichotomies was first introduced by Perron in 1930 [1] to study the conditional stability of the linear differential equations and the existence of bounded solutions of the nonlinear differential equations. Then Li [2] established the corresponding analogous concept for the linear difference equations. The theory of exponential dichotomies has been playing an important role in the study of the theory of differential equations and difference equations (see [35]). An interesting and challenging problem is to establish necessary and sufficient criteria for the existence of exponential dichotomies. Among the many methods and tools, the admissibility techniques or input-output methods have been extensively applying to study the existence of exponential dichotomies for differential equations and difference equations [615].

It is well known that the theory of dynamic equations on time scales provides a unifying structure for the study of differential equations in the continuous case and difference equations in the discrete case and has tremendous potential for applications in mathematical models of real processes and phenomena [1619]. In recent years, the theory of exponential dichotomies on time scales for the linear dynamic equations on time scales extends the idea of hyperbolicity from autonomous dynamic equations on time scales to explicitly nonautonomous ones and has progressed greatly [2030]. In view of the important role of the admissibility techniques or input-output methods in the study of the exponential dichotomy on differential equations and difference equations, it is natural for us to ask whether the admissibility techniques or input-output methods can be applied to deal with problems of exponential dichotomies on time scales for an evolution family on time scales.

Motivated by the results of admissibility and exponential dichotomy for differential equations and difference equations in [615], in this paper, we establish a relation between exponential dichotomy on time scales and admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) for an evolution family on time scales. The paper is organized as follows. In the next section, we present some basic information concerning exponential dichotomies on time scales and admissibility for an evolution family. In Section 3, we construct an equivalent relation between exponential dichotomy on time scales and the admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) for the evolution family on time scales. Our result extends related results known for differential equations and difference equations on the half-line to more general time scales.

2 Preliminaries and basic definitions

In this section, we first introduce the following concepts related to the notion of time scales, which can be found in [16, 17, 30]. A time scale \(\mathbb{T}\) is defined as a nonempty closed subset of the real numbers. Define the forward jump operator \(\sigma :\mathbb{T}\rightarrow \mathbb{T}\) and the graininess function \(\mu(t)=\sigma (t)-t\) for any \(t\in\mathbb{T}\). In the following discussion, the time scale \(\mathbb{T}\) is assumed to be unbounded above and below. Let \(\mathrm{C}_{\mathrm{rd}}(\mathbb{T},{\mathbb {R}})\) be the set of rd-continuous functions \(g:\mathbb{T}\to {\mathbb {R}}\) and \(\mathcal{R}^{+}(\mathbb{T},{\mathbb {R}}):= \{g\in\mathrm{C}_{\mathrm{rd}}(\mathbb{T},{\mathbb {R}}):1+\mu(t)g(t)>0, t\in\mathbb{T}\}\) be the space of positively regressive functions. We define the exponential function on time scales by
$$e_{\varphi }(t,s) =\exp\biggl\{ \int_{s}^{t} \zeta_{\mu(\tau)} \bigl(\varphi (\tau) \bigr)\Delta \tau\biggr\} \quad\mbox{with } \zeta_{h}(z)= \begin{cases} z &\mbox{if } h=0,\\ \operatorname{Log}(1+hz)/h&\mbox{if } h\neq0, \end{cases} $$
for any \(\varphi \in\mathcal{R}^{+}(\mathbb{T},{\mathbb {R}})\) and \(s,t \in\mathbb{T}\), where Log is the principal logarithm. Define
$$\begin{aligned}& (\varphi \oplus\psi) (t): =\varphi (t)+\psi(t)+\mu(t)\varphi (t)\psi(t), \\& \ominus \varphi : =-\frac{\varphi (t)}{1+\mu(t)\varphi (t)}, \\& (\omega\odot \varphi ) (t): =\lim_{h\searrow\mu(t)}\frac{(1+h\varphi (t))^{\omega}-1}{h} \end{aligned}$$
for a given \(\omega \in {\mathbb {R}}^{+}\) and for any \(t\in\mathbb{T}\), \(\varphi ,\psi\in\mathcal{R}^{+}(\mathbb{T},{\mathbb {R}})\). Let
$$\begin{aligned}& \mathbb{T}^{+}=\mathbb{T}\cap[0,+\infty), \quad \kappa =\min\bigl\{ t\in\mathbb{T}^{+} \bigr\} ,\quad\quad [t,s]_{\mathbb{T}^{+}}=[t,s] \cap\mathbb{T}^{+},\quad t,s\in\mathbb{T}^{+}, \\& {}[\varphi ]^{*}:=\sup_{t\in\mathbb{T}^{+}} \bigl(\varphi (t) \bigr),\quad\quad [\varphi ]_{*}:=\inf _{t\in\mathbb{T} ^{+}} \bigl(\varphi (t) \bigr) \end{aligned}$$
for any bounded function \(\varphi \in\mathrm{C}_{\mathrm{rd}}(\mathbb{T}^{+},{\mathbb {R}})\).

Let \((X, \|\cdot\|)\) be a Banach space and \({\mathcal {B}}(X)\) be the space of bounded linear operators defined on X. Now we give some definitions on time scales.

Definition 2.1

\(\{U(t,s)\}_{t\geq s}\subset {\mathcal {B}}(X)\) is said to be an evolution family on a time scale \(\mathbb{T}^{+}\) if
  1. (i)

    \(U(t,t)=\mathrm{id}\) for every \(t\in\mathbb{T}^{+}\) and \(U(t,\tau)U(\tau,s)=U(t,s)\) for any \(t\geq\tau\geq s\geq \kappa \);

     
  2. (ii)

    for each \(s\in\mathbb{T}^{+}\) and any \(x\in X\), \(U(\cdot,s)x\) is rd-continuous on \([s,\infty )_{\mathbb{T}^{+}}\) for the first variable and \(U(s,\cdot)x\) is rd-continuous on \([\kappa ,s]_{\mathbb{T}^{+}}\) for the second variable.

     

Remark 2.1

In the general case, an evolution family \(\{U(t,s)\}_{t\geq s}\) is related to an evolution operator of a linear dynamic equation on time scales.

Definition 2.2

An evolution family \(\{U(t,s)\}_{t,s\in\mathbb{T}^{+}}\) is said to be exponential growth on a time scale \(\mathbb{T}^{+}\) if there exist positive constants L and ρ such that
$$ \bigl\| U(t,s) \bigr\| \leq Le_{\rho}(t,s),\quad t\geq s, t,s\in\mathbb{T}^{+}. $$
(2.1)

Definition 2.3

An evolution family \(\{U(t,s)\}_{t,s\in\mathbb{T}^{+}}\) is said to admit an exponential dichotomy on a time scale \(\mathbb{T}^{+}\) if there exist projections \(\{P(t)\}_{t\in\mathbb{T}^{+}}\) such that \(U(t,s)P(s)=P(t)U(t,s)\) for any \(t\geq s\geq \kappa \) and \(U(t,s)|_{\operatorname {Ker}P(s)}:\operatorname {Ker}P(s)\rightarrow \operatorname {Ker}P(t)\) is an isomorphism for any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\) and there exist a constant \(K>0\) and \(\alpha \in\mathrm{C}_{\mathrm{rd}}(\mathbb{T}^{+},{\mathbb {R}})\) with \([\alpha ]_{*}>0\) such that
  1. (i)

    \(\|U(t,s)x \|\leq Ke_{\ominus \alpha }(t,s) \|x \|\) for all \(x\in \operatorname {Range}P(s)\) and any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\);

     
  2. (ii)

    \(\|U(t,s)y \|\geq\frac{1}{K}e_{\alpha }(t,s) \|y \|\) for all \(y\in \operatorname {Ker}P(s)\) and any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\).

     

Remark 2.2

The exponential function on time scales can display different forms when we choose different time scales. For example, when \(\mathbb{T}={\mathbb {R}}\) or \(\mathbb{T}={\mathbb {Z}}\), we have \(e_{\ominus \alpha }(t,s)=e^{-\alpha (t-s)}\) or \(e_{\ominus \alpha }(t,s)=(1/(1+\alpha ))^{t-s}\) if α is a constant. Let \(\mathbb{T}=q^{{\mathbb {N}}_{0}}\), \(q>1\), then \(e_{\ominus \alpha }(t,s)=\prod_{\tau\in[s,t)}[1/ (1+(q-1)\alpha \tau)]\). More examples for the exponential function on different time scales can be found in [16]. This shows that the exponential dichotomy on time scales is more general and unifies the notions of exponential dichotomies on the continuous and discrete case. On the other hand, we have
$$ e_{\alpha }(t,s)\leq e^{\alpha (t-s)},\quad\quad e^{-\alpha (t-s)}\leq e_{\ominus \alpha }(t,s) $$
(2.2)
for any \(t\geq s\) and any time scale \(\mathbb{T}\) (see (3.3) in [29]).
We let
$$\mathrm{C}_{\mathrm{rd}}^{b} \bigl(\mathbb{T}^{+},X \bigr):= \Bigl\{ u\in\mathrm{C}_{\mathrm{rd}}\bigl(\mathbb{T}^{+},X \bigr)| \|u \|_{\infty }:=\sup_{t\in\mathbb{T}^{+}} \bigl\| u(t) \bigr\| < \infty \Bigr\} $$
and
$$\begin{aligned} L^{p} \bigl(\mathbb{T}^{+},X \bigr) :=& \biggl\{ f\Big|f:\mathbb{T}^{+}\rightarrow X \mbox{ is a Bochner measurable function with} \\ & \|f \|_{p}:= \biggl(\int _{\kappa }^{\infty }\bigl\| f(t) \bigr\| ^{p}\Delta t \biggr)^{1/p}< \infty \biggr\} \end{aligned}$$
for \(p>1\). It is not difficult to show that \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X), \|\cdot\|_{\infty })\) and \((L^{p}(\mathbb{T}^{+},X), \|\cdot\|_{p})\) are both Banach spaces (see [31]). We consider the integral equation on time scales
$$ u(t)=U(t,s)u(s)+\int^{t}_{s}U(t, \tau)f(\tau)\Delta\tau,\quad t\geq s, t,s\in\mathbb{T}^{+}, $$
(2.3)
where \(f\in L^{p}(\mathbb{T}^{+},X)\) and \(u\in\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X)\).

Definition 2.4

The pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X),L^{p}(\mathbb{T}^{+},X))\) is said to be admissible for an evolution family \(\{U(t,s)\}_{t,s\in\mathbb{T}^{+}}\) if for every \(f\in L^{p}(\mathbb{T}^{+},X)\) there exists a function \(u\in\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X)\) such that the pair \((u,f)\) satisfies (2.3). We say that \(L^{p}(\mathbb{T}^{+},X)\) is the input space and \(\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X)\) is the output space.

We easily show that a pair \((u,f)\) satisfies (2.3) if and only if \((u,f)\) satisfies
$$ u(t)=U(t,\kappa )u(\kappa )+\int^{t}_{\kappa }U(t, \tau)f(\tau)\Delta\tau, \quad t\geq \kappa , t\in\mathbb{T}^{+}. $$
(2.4)
In fact, if (2.4) holds, then for each \(s\geq \kappa \)
$$u(s)=U(s,\kappa )u(\kappa )+\int^{s}_{\kappa }U(s,\tau)f( \tau)\Delta\tau $$
and
$$\begin{aligned} U(t,s)u(s) =&U(t,s)U(s,\kappa )u(\kappa )+\int^{s}_{\kappa }U(t,s)U(s, \tau)f(\tau)\Delta\tau \\ =&U(t,\kappa )u(\kappa )+\int^{s}_{\kappa }U(t,\tau)f(\tau) \Delta\tau \\ =&u(t)-\int^{t}_{s}U(t,\tau)f(\tau)\Delta\tau \end{aligned}$$
for any \(t\geq s\), \(t\in\mathbb{T}^{+}\).

3 Main result

In this section, we establish a relation between exponential dichotomy on time scales and admissibility of the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X),L^{p}(\mathbb{T}^{+},X))\) for an evolution family on time scales. Let the linear subspace \(E_{\kappa }:=\{x\in X|U(\cdot, \kappa )x\in\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X)\}\). Now we state our main result.

Theorem 3.1

Assume that an evolution family \(U(t,s)_{t\geq s}\) admits an exponential growth on a time scale \(\mathbb{T}^{+}\) with \([u]^{*}<\infty \). Then the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) is admissible for the evolution family \(U(t,s)_{t\geq s}\) on the time scale \(\mathbb{T}^{+}\) and \(E_{\kappa }\) is closed and complemented in X if and only if \(U(t,s)_{t\geq s}\) admits an exponential dichotomy on the time scale \(\mathbb{T}^{+}\).

The proof of Theorem 3.1 is nontrivial, we shall divide it into several steps and assume that the conditions in Theorem 3.1 are always satisfied. We first establish some auxiliary results. If \(E_{\kappa }\) is closed and complemented in X, then there is a closed linear subspace \(F_{\kappa }\) such that \(X=E_{\kappa }\oplus F_{\kappa }\). We define the linear subspace \(\mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X):=\{u\in\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X):u(\kappa )\in F_{\kappa }\}\). Using similar arguments to that of Lemma 2.1 in [15], we conclude that if the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) is admissible, then for every \(f\in L^{p}(\mathbb{T}^{+},X)\) there exists a unique function \(\bar{u}\in\mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X)\) such that the pair \((\bar{u}, f)\) satisfies (2.3). Therefore, we can define the input-output operator \(J:L^{p}(\mathbb{T}^{+},X)\rightarrow \mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X)\) by \(J(f)=\bar{u}\), where the pair \((\bar{u},f)\) satisfies (2.3).

Lemma 3.1

The operator J is bounded.

Proof

According to the closed graph theorem, we only need to prove that J is closed. We assume that \(\{f_{n}\}_{n\in {\mathbb {N}}}\subset L^{p}(\mathbb{T}^{+},X)\), \(f\in L^{p}(\mathbb{T}^{+},X)\), and \(f_{n}\rightarrow f\) in \(L^{p}(\mathbb{T}^{+},X)\) as \(n\rightarrow \infty \) and there exists a function \(\bar{u}\in\mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X)\) such that \(\bar{u}_{n}=J(f_{n})\rightarrow \bar{u}\) in \(\mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X)\) as \(n\rightarrow \infty \). It follows from (2.4) that
$$ \bar{u}_{n}(t)=U(t,\kappa )\bar{u}_{n}(\kappa )+ \int_{\kappa }^{t}U(t,\tau)f_{n}(\tau)\Delta \tau,\quad t\in\mathbb{T}^{+}. $$
(3.1)
On the other hand, by (2.1) and the Hölder inequality on time scales, we have
$$\begin{aligned} \biggl\Vert \int_{\kappa }^{t}U(t,\tau) \bigl(f_{n}(\tau)-f(\tau) \bigr)\Delta \tau\biggr\Vert \leq&\int _{\kappa }^{t} \bigl\| U(t,\tau) \bigr\| \bigl\| f_{n}(\tau)-f( \tau) \bigr\| \Delta \tau \\ \leq& L\int_{\kappa }^{t}e_{\rho}(t,\tau) \bigl\| f_{n}(\tau)-f(\tau) \bigr\| \Delta \tau \\ \leq& L \biggl(\int_{\kappa }^{t}e_{q\odot\rho}(t, \tau)\Delta \tau\biggr)^{1/q} \biggl(\int_{\kappa }^{t} \bigl\| f_{n}(\tau)-f(\tau) \bigr\| ^{p}\Delta \tau\biggr)^{1/p} \\ =&L \biggl(\int_{\kappa }^{t} \bigl[1/\ominus(q\odot \rho) \bigr]e^{\Delta }_{\ominus (q\odot \rho)}(\tau,t)\Delta \tau\biggr)^{1/q} \|f_{n}-f \|_{p} \\ \leq&\frac{1+[(q\odot\rho)\mu]^{*}}{[q\odot\rho]_{*}}e_{\rho}(t,\kappa ) \| f_{n}-f \|_{p} \end{aligned}$$
for each \(t\in\mathbb{T}^{+}\), where \(1/q+1/p=1\). Then \(\int_{\kappa }^{t}U(t,\tau)f_{n}(\tau)\Delta \tau \rightarrow \int_{\kappa }^{t}U(t,\tau)f(\tau)\Delta \tau\) since \(f_{n}\rightarrow f\) in \(L^{p}(\mathbb{T}^{+},X)\) as \(n\rightarrow \infty \). Combining with (3.1) gives
$$\bar{u}(t)=U(t,\kappa )\bar{u}(\kappa )+\int_{\kappa }^{t}U(t, \tau)f(\tau)\Delta \tau $$
since \(\bar{u}_{n}\rightarrow \bar{u}\) in \(\mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X)\) as \(n\rightarrow \infty \). This implies that \(J(f)=\bar{u}\). The proof is completed. □
For each given \(s\in\mathbb{T}^{+}\), we let
$$ E_{s}:= \Bigl\{ x\in X:\sup_{t\geq s} \bigl\| U(t,s)x \bigr\| < \infty \Bigr\} ,\quad\quad F_{s}:=U(s,\kappa )F_{\kappa }. $$
(3.2)

Lemma 3.2

If the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) is admissible for the evolution family \(U(t,s)_{t\geq s}\) on the time scale \(\mathbb{T}^{+}\), then the subspace \(E_{s}\) is closed for every \(s\in\mathbb{T}^{+}\) and there is a positive constant \(K_{1}\) and \(\alpha \in\mathrm{C}_{\mathrm{rd}}(\mathbb{T}^{+},{\mathbb {R}})\) with \([\alpha ]_{*}>0\) such that
$$ \bigl\| U(t,s)x \bigr\| \leq K_{1}e_{\ominus \alpha }(t,s) \|x \| $$
(3.3)
for any \(x\in E_{s}\) and \(t\geq s\).

Proof

Let
$$ \gamma :=\frac{1}{ \|J \|} \biggl(\frac{[p\odot \beta ]_{*}}{1+[(p\odot \beta )\mu ]^{*}} \biggr)^{1/p},\quad 0< \beta <\gamma ,\quad\quad \alpha =\gamma \ominus \beta , $$
(3.4)
where J is defined in Lemma 3.1. A direct calculation gives \([\gamma \ominus \beta ]_{*}>0\). For each given \(s\in\mathbb{T}^{+}\) and any \(x\in E_{s}\setminus\{0\}\), we let
$$d_{s,x}=\sup\bigl\{ t\in\mathbb{T}^{+}| U(t,s)x\neq0, t\geq s \bigr\} $$
and
$$ \eta_{s}=\inf\bigl\{ t\in\mathbb{T}^{+}|t\geq s+1 \bigr\} . $$
(3.5)
Next we consider two different cases.
The first case is \(d_{s,x}>\eta_{s}\). We let \(f_{t}:\mathbb{T}^{+}\rightarrow X\) by
$$f_{t}(r)= \chi_{[s,t)_{\mathbb{T}^{+}}}(r)e_{\beta }(r,s) \frac{U(r,s)x}{ \|U(r,s)x \|} $$
and \(u_{t}:\mathbb{T}^{+}\rightarrow X\) by
$$u_{t}(r)=\int^{r}_{\kappa } \frac{\chi_{[s,t)_{\mathbb{T}^{+}}}(\tau)}{ \|U(\tau ,s)x \|}e_{\beta }(\tau,s)\Delta\tau U(r,s)x $$
for every \(t\in(s,d_{s,x})_{\mathbb{T}^{+}}\). Then
$$\begin{aligned} \|f_{t} \|_{p} =& \biggl(\int^{\infty }_{\kappa } \bigl\| f_{t}(r) \bigr\| ^{p}\Delta r \biggr)^{1/p} = \biggl(\int ^{t}_{s}e_{p\odot \beta }(r,s)\Delta r \biggr)^{1/p} \\ < & \bigl(1/[p\odot \beta ]_{*} \bigr)^{1/p}e_{\beta }(t,s)<\infty \end{aligned}$$
(3.6)
and \(\sup_{r\in[t,\infty)_{\mathbb{T}^{+}}} \|u_{t}(r) \|<\infty\) since \(u_{t}(r)=\int^{t}_{s}\frac{e_{\beta }(\tau,s)}{ \|U(\tau,s)x \|}\Delta \tau U(r,s)x\) for \(r\geq t\) and \(\sup_{r\in[s,\infty )_{\mathbb{T}^{+}}} \|U(r,s)x \|<\infty\) for \(x\in E_{s}\). This implies that \(f_{t}\in L^{p}(\mathbb{T}^{+},X)\) and \(u_{t}\in\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X)\) for every \(t\in\mathbb{T}^{+}\). Direct calculation shows that the pair \((u_{t},f_{t})\) satisfies (2.3). Therefore, we have \(u_{t}=J(f_{t})\) and
$$ \|u_{t} \|_{\infty }\leq\|J \| \|f_{t} \|_{p} $$
(3.7)
since \(u_{t}(\kappa )=0\in F_{\kappa }\). Noting that \(t\in(s,d_{s,x})_{\mathbb{T}^{+}}\) is arbitrary, by (3.7), (3.6) and (3.4), we have
$$ \int_{s}^{t}\frac{e_{\beta }(\tau,s)}{ \|U(\tau,s)x \|} \Delta\tau\leq\frac{ \| J \|}{([p\odot \beta ]_{*})^{1/p}}\frac{e_{\beta }(t,s)}{ \|U(t,s)x \|} \leq\frac {1}{\gamma } \frac{e_{\beta }(t,s)}{ \|U(t,s)x \|} $$
(3.8)
for any \(t\in(s,d_{s,x})_{\mathbb{T}^{+}}\) since \(\|u_{t}(t) \|\leq\|u_{t} \|_{\infty}\). On the other hand, it follows from (3.8) that
$$\begin{aligned}& \biggl(e_{\ominus \gamma }(t,\kappa )\int^{t}_{s} \frac{e_{\beta }(\tau ,s)}{ \| U(\tau,s)x \|}\Delta\tau\biggr)^{\Delta} \\& \quad=e_{\ominus \gamma }^{\Delta}(t,\kappa )\int^{t}_{s} \frac{e_{\beta }(\tau ,s)}{ \|U(\tau,s)x \|}\Delta\tau+ e_{\ominus \gamma } \bigl(\sigma(t),\kappa \bigr) \biggl( \int^{t}_{s}\frac{e_{\beta }(\tau ,s)}{ \|U(\tau,s)x \|}\Delta\tau \biggr)^{\Delta} \\& \quad =(\ominus \gamma )e_{\ominus \gamma }(t,\kappa )\int^{t}_{s} \frac{e_{\beta }(\tau ,s)}{ \|U(\tau,s)x \|}\Delta\tau+ e_{\ominus \gamma }(t,\kappa ) \bigl(1+\mu(t) (\ominus \gamma ) \bigr)\frac{e_{\beta }(t,s)}{ \| U(t,s)x \|} \\& \quad =-\frac{\gamma e_{\ominus \gamma }(t,\kappa )}{1+\mu(t)\gamma }\int^{t}_{s} \frac{e_{\beta }(\tau ,s)}{ \|U(\tau,s)x \|}\Delta\tau+ \biggl(\frac{e_{\ominus \gamma }(t,\kappa )}{1+\mu(t)\gamma } \biggr)\frac {e_{\beta }(t,s)}{ \|U(t,s)x \|} \\& \quad =\frac{e_{\ominus \gamma }(t,\kappa )}{1+\mu(t)\gamma } \biggl(-\gamma \int^{t}_{s} \frac{e_{\beta }(\tau,s)}{ \|U(\tau,s)x \|}\Delta\tau+ \frac{e_{\beta }(t,s)}{ \|U(t,s)x \|} \biggr) \\& \quad\geq 0 \end{aligned}$$
for any \(t\in(s,d_{s,x})_{\mathbb{T}^{+}}\), which implies that \(e_{\ominus \gamma }(t,\kappa )\int^{t}_{s}\frac{e_{\beta }(\tau,s)}{ \| U(\tau ,s)x \|}\Delta\tau\) is nondecreasing on \((s,d_{s,x})_{\mathbb{T}^{+}}\). Then
$$ e_{\ominus \gamma }(t,\kappa )\int^{t}_{s} \frac{e_{\beta }(\tau,s)}{ \| U(\tau ,s)x \|}\Delta\tau\geq e_{\ominus \gamma }(\eta_{s},\kappa )\int ^{\eta_{s}}_{s}\frac{e_{\beta }(\tau ,s)}{ \|U(\tau,s)x \|}\Delta\tau $$
(3.9)
for any \(t\in[\eta_{s},d_{s,x})_{\mathbb{T}^{+}}\). By (2.1), we have
$$ \bigl\| U(t,s)x \bigr\| \leq Le_{\rho}(t,s) \|x \|\leq Le_{\rho}( \eta_{s},s) \|x \| $$
(3.10)
for any \(t\in[s,\eta_{s}]_{\mathbb{T}^{+}}\). It follows from (3.8), (3.9), and (3.10) that
$$\begin{aligned} \frac{1}{Le_{\rho}(\eta_{s},s) \|x \|} \leq&\int^{\eta_{s}}_{s} \frac{1}{Le_{\rho}(\eta_{s},s) \|x \|}\Delta\tau\leq\int^{\eta _{s}}_{s} \frac{e_{\beta }(\tau,s)}{ \|U(\tau,s)x \|}\Delta\tau \\ \leq& e_{\ominus \gamma }(t,\eta_{s})\int^{t}_{s} \frac{e_{\beta }(\tau,s)}{ \| U(\tau ,s)x \|}\Delta\tau\leq\frac{e_{\ominus \gamma }(t,\eta_{s})}{\gamma }\frac{e_{\beta }(t,s)}{ \| U(t,s)x \|} \end{aligned}$$
for any \(t\in[\eta_{s},d_{s,x})_{\mathbb{T}^{+}}\). Then
$$\begin{aligned} \bigl\| U(t,s)x \bigr\| \leq&\frac{L}{\gamma }e_{\rho}(\eta_{s},s)e_{\ominus \gamma }(t, \eta_{s}) e_{\beta }(t,s) \|x \| \\ =&\frac{L}{\gamma }e_{\rho}(\eta_{s},s)e_{\ominus \gamma }(s, \eta_{s})e_{\beta \ominus \gamma }(t,s) \|x \| \\ =&\frac{L}{\gamma }e_{\rho\oplus \gamma }(\eta_{s},s)e_{\ominus \alpha }(t,s) \| x \| \end{aligned}$$
for any \(t\in[\eta_{s},d_{s,x})_{\mathbb{T}^{+}}\). To obtain the conclusion, we need to show that \(\delta(s):=e_{\rho\oplus \gamma }(\eta_{s},s)\) is bounded for any \(s\in\mathbb{T}^{+}\). For the definition of \(\eta_{s}\) (see (3.5)), there are the following three different cases:
  1. Case 1.

    \(s+1\in\mathbb{T}^{+}\). We have \(\eta_{s}=\inf\{t\in \mathbb{T}^{+}|t\geq s+1\}=s+1< s+1+[\mu]^{*}\).

     
  2. Case 2.

    \(s+1\notin\mathbb{T}^{+}\) and \((s,s+1]\cap\mathbb{T}^{+}\neq\emptyset\). Let \(t^{*}=\max\{t\in[s,s+1]_{\mathbb{T}^{+}}\}\). We have \(\sigma (t^{*})>t^{*}\). In fact, if \(\sigma (t^{*})=t^{*}\), then \(t^{*}\) is a right-dense point, which implies that there is a point \(t^{**}>t^{*}\) and \(t^{**}\in[s,s+1]_{\mathbb{T}^{+}}\). This is a contradiction. By the definition of \(t^{*}\), we get \(\eta_{s}=\sigma (t^{*})\) and \(\eta_{s}\leq s+1+\sigma (t^{*})-t^{*}\leq s+1+[\mu]^{*}\).

     
  3. Case 3.

    \((s,s+1]\cap\mathbb{T}^{+}=\emptyset\). We have \(\eta_{s}=\sigma (s)>s\) and \(\eta_{s}\leq s+\sigma (s)-s\leq s+1+[\mu]^{*}\).

     
In view of the above discussion and (2.2), we have
$$\delta(s)\leq e_{\rho}(\eta_{s},s)e_{\gamma }( \eta_{s},s)\leq e^{\rho(\eta_{s}-s)}e^{\gamma (\eta_{s}-s)}\leq e^{(\rho+\gamma )(\eta_{s}-s)} \leq e^{(\rho+\gamma )(1+[\mu]^{*})}:=L_{1} $$
for any \(s\in\mathbb{T}^{+}\). Then
$$ \bigl\| U(t,s)x \bigr\| \leq(LL_{1}/\gamma )e_{\ominus \alpha }(t,s) \|x \| $$
(3.11)
for all \(t\in[\eta_{s},d_{s,x})_{\mathbb{T}^{+}}\). Moreover, by (2.1), we get
$$\begin{aligned} \bigl\| U(t,s)x \bigr\| \leq& Le_{\rho}(t,s) \|x \|=L e_{\rho}(t,s)e_{\alpha }(t,s)e_{\ominus \alpha }(t,s) \|x \| \\ \leq& Le_{\rho}(t,s)e_{\gamma }(t,s)e_{\ominus \alpha }(t,s) \|x \| \leq Le_{\rho}(\eta_{s},s)e_{\gamma }( \eta_{s},s)e_{\ominus \alpha }(t,s) \|x \| \\ \leq& LL_{1}e_{\ominus \alpha }(t,s) \|x \| \end{aligned}$$
(3.12)
for all \(t\in[s,\eta_{s}]_{\mathbb{T}^{+}}\). It follows from (3.11) and (3.12) that
$$ \bigl\| U(t,s)x \bigr\| \leq K_{1}e_{\ominus \alpha }(t,s) \|x \| $$
(3.13)
for all \(t\in[s,d_{s,x})_{\mathbb{T}^{+}}\), where \(K_{1}=\max\{LL_{1},(LL_{1}/\gamma )\}\).
The second case is \(s\leq d_{s,x}\leq\eta_{s}\). It follows from (2.1) and (3.13) that
$$ \bigl\| U(t,s)x \bigr\| \leq Le_{\rho}(t,s) \|x \|\leq Le_{\rho}( \eta_{s},s)e_{\gamma }(\eta_{s},s)e_{\ominus \alpha }(t,s) \|x \|\leq K_{1}e_{\ominus \alpha }(t,s) \|x \| $$
(3.14)
for all \(t\in[s,d_{s,x}]_{\mathbb{T}^{+}}\).

Based on (3.13), (3.14), and the definition of \(d_{s,x}\), we conclude that (3.3) holds. Let \(s\in\mathbb{T}^{+}\) and \(\{x_{n}\}_{n\in {\mathbb {N}}}\subset E_{s} \) with \(x_{n}\rightarrow x\) as \(n\rightarrow \infty \). Combining with (3.3) gives \(\|U(t,s)x_{n} \|\leq K_{1} \|x_{n} \|\) for any \(n\in {\mathbb {N}}\) and any \(t\geq s\). Thus, we get \(\|U(t,s)x \|\leq K_{1} \|x \|\) for any \(t\geq s\). This implies that \(x\in E_{s}\) and \(E_{s}\) is closed. The proof is completed. □

Lemma 3.3

If the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) is admissible for the evolution family \(U(t,s)_{t\geq s}\) on the time scale \(\mathbb{T}^{+}\), then the subspace \(F_{s}\) is closed for every \(s\in\mathbb{T}^{+}\) and there is a positive constant \(K_{2}\) and \(\alpha \in\mathrm{C}_{\mathrm{rd}}(\mathbb{T}^{+},{\mathbb {R}})\) with \([\alpha ]_{*}>0\) such that
$$ K_{2} \bigl\| U(t,s)y \bigr\| \geq e_{\alpha }(t,s) \|y \| $$
(3.15)
for any \(y\in F_{s}\) and \(t\geq s\). Moreover, \(U(t,s)|_{F_{s}}:F_{s}\rightarrow F_{t}\) is an isomorphism for any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\).

Proof

Let β, γ be positive constants and α be a rd-continuous function defined in (3.4). For \(y\in F_{\kappa }\setminus\{0\}\), we have \(U(t,\kappa )y\neq0\) for any \(t\in\mathbb{T}^{+}\). In fact, if there is \(\bar{t}\in\mathbb{T}^{+}\) such that \(U(\bar{t},\kappa )y=0\), then \(U(t,\kappa )y=U(t,\bar{t})U(\bar{t},\kappa )y=0\) for any \(t\geq\bar{t}\) and \(y\in E_{\kappa }\). This means that \(y\in E_{\kappa }\cap F_{\kappa }\) and \(y=0\). This is a contradiction to \(y\in F_{\kappa }\setminus\{0\}\). For each \(t\in\mathbb{T}^{+}\), we choose \(\{\tau_{n}^{t}\}_{n\in {\mathbb {N}}}\subset\mathbb{T}^{+}\) such that \(t< \tau_{1}^{t}<\tau_{2}^{t}<\cdots<\tau_{n}^{t}<\cdots\) and \(\tau_{n}^{t}\rightarrow \infty \) as \(n\rightarrow \infty \). We define \(f_{\tau_{n}^{t}}:\mathbb{T}^{+}\rightarrow X\) by
$$f_{\tau_{n}^{t}}(s)=-\chi_{[t,\tau_{n}^{t}]_{\mathbb{T}^{+}}}e_{\ominus \beta }(s,\kappa ) \frac{U(s,\kappa )y}{ \|U(s,\kappa )y \|} $$
and \(u_{\tau_{n}^{t}}:\mathbb{T}^{+}\rightarrow X\) by
$$u_{\tau^{t}_{n}}(s)=\int^{\infty}_{s} \frac{\chi_{[t,\tau ^{t}_{n}]_{\mathbb{T} ^{+}}}(\tau)e_{\ominus \beta }(\tau, \kappa )}{ \|U(\tau, \kappa )y \|}\Delta\tau U(s,\kappa )y. $$
It follows that
$$ \|f_{\tau_{n}^{t}} \|_{p}\leq\biggl(\int _{t}^{\infty }e_{\ominus (p\odot \beta )}(s,\kappa )\Delta s \biggr)^{1/p}\leq\biggl(\frac{1+[(p\odot \beta )\mu]^{*}}{[p\odot \beta ]_{*}} \biggr)^{1/p}e_{\ominus \beta }(t, \kappa )< \infty . $$
(3.16)
Moreover, \(u_{\tau^{t}_{n}}\) is rd-continuous,
$$u_{\tau^{t}_{n}}(\kappa )= \biggl(\int^{\tau^{t}_{n}}_{t} \frac {e_{\ominus \beta }(\tau, \kappa )}{ \|U(\tau, \kappa )y \|}\Delta\tau\biggr) y\in F_{\kappa }$$
and \(u_{\tau^{t}_{n}}(s)=0\) for \(s\geq\tau^{t}_{n}\), which implies that \(u_{\tau^{t}_{n}}\in\mathrm{C}_{\mathrm{rd}}^{b,F_{\kappa }}(\mathbb{T}^{+},X)\). Then \(u_{\tau^{t}_{n}}=J(f_{\tau^{t}_{n}})\) and \(\|u_{\tau^{t}_{n}} \|_{\infty }\leq\|J \| \|f_{\tau^{t}_{n}} \|_{p}\) for any \(n\in {\mathbb {N}}\) since it is easy to show that the pair \((u_{\tau^{t}_{n}}, f_{\tau^{t}_{n}})\) satisfies (2.3). It follows from \(\|u_{\tau^{t}_{n}}(t) \|\leq \|u_{\tau^{t}_{n}} \|_{\infty}\) and (3.16) that
$$ \int^{\tau^{t}_{n}}_{t}\frac{e_{\ominus \beta }(\tau, \kappa )}{ \|U(\tau ,\kappa )y \|}\Delta\tau\bigl\| U(t, \kappa )y \bigr\| \leq \|J \| \biggl(\frac {1+[(p\odot \beta )\mu]^{*}}{[p\odot \beta ]_{*}} \biggr)^{1/p} e_{\ominus \beta }(t,k) = \frac{1}{\gamma }e_{\ominus \beta }(t,\kappa ) $$
for any \(n\in {\mathbb {N}}\). This also reads
$$ \gamma \int^{\infty }_{t}\frac{e_{\ominus \beta }(\tau, \kappa )}{ \|U(\tau ,\kappa )y \|} \Delta\tau\leq\frac{e_{\ominus \beta }(t,\kappa )}{ \|U(t,\kappa )y \|} $$
(3.17)
as \(n\rightarrow \infty \). It follows from (3.17) that
$$\begin{aligned}& \biggl(e_{\gamma }(t,\kappa )\int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau ,\kappa )}{ \|u(\tau, \kappa )y \|}\Delta\tau\biggr)^{\Delta }\\& \quad=e^{\Delta}_{\gamma }(t,\kappa )\int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau, \kappa )}{ \|u(\tau, \kappa )y \|}\Delta\tau+e_{\gamma } \bigl(\sigma (t),\kappa \bigr) \biggl( \int_{t}^{\infty }\frac{e_{\ominus \beta }(\tau ,\kappa )}{ \|u(\tau, \kappa )y \|}\Delta\tau \biggr)^{\Delta} \\& \quad=\gamma e_{\gamma }(t,\kappa )\int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau, \kappa )}{ \| u(\tau, \kappa )y \|}\Delta\tau- \bigl(1+\mu(t)\gamma \bigr)e_{\gamma }(t,\kappa ) \frac{e_{\ominus \beta }(t,\kappa )}{ \| U(t,\kappa )y \|} \\& \quad\leq e_{\gamma }(t,\kappa ) \biggl(\gamma \int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau ,\kappa )}{ \|u(\tau, \kappa )y \|}\Delta\tau-\frac{e_{\ominus \beta }(t,\kappa )}{ \|U(t,\kappa )y \|} \biggr)\leq0. \end{aligned}$$
Thus, we get
$$ e_{\gamma }(t,\kappa )\int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau, \kappa )}{ \| u(\tau, \kappa )y \|}\Delta\tau\leq e_{\gamma }(s,\kappa )\int ^{\infty}_{s}\frac{e_{\ominus \beta }(\tau, \kappa )}{ \| U(\tau, \kappa )y \|}\Delta\tau $$
(3.18)
for any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\). Combining with (3.17) gives
$$ \gamma e_{\gamma }(t,s)\int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau, \kappa )}{ \| u(\tau, \kappa )y \|}\Delta\tau\leq\frac{e_{\ominus \beta }(s,\kappa )}{ \| U(s,\kappa )y \|} $$
(3.19)
for any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\). On the other hand, due to (2.1), it is sufficient to have
$$\bigl\| U(\tau, \kappa )y \bigr\| = \bigl\| U(\tau,t)U(t,\kappa )y \bigr\| \leq Le_{\rho}(\tau,t) \bigl\| U(t, \kappa )y \bigr\| $$
for any \(\tau\geq t\), \(\tau,t\in\mathbb{T}^{+}\). This implies that
$$\begin{aligned} \int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau, \kappa )}{ \|u(\tau, \kappa )y \| }\Delta\tau =&e_{\ominus \beta }(t,\kappa ) \int^{\infty}_{t} \frac{e_{\ominus \beta }(\tau,t)}{ \|u(\tau, \kappa )y \| }\Delta\tau \geq \frac{e_{\ominus \beta }(t,\kappa )}{L \|U(t,\kappa )y \|}\int^{\infty }_{t}e_{\ominus(\beta \oplus\rho)}( \tau,t)\Delta\tau \\ \geq&\frac{(1+[(\beta \oplus\rho)\mu]_{*})e_{\ominus \beta }(t,\kappa )}{L[\beta \oplus\rho]^{*} \|U(t,\kappa )y \|}. \end{aligned}$$
(3.20)
By (3.19) and (3.20), we have
$$ e_{\gamma \ominus \beta }(t,s) \bigl\| U(s,\kappa )y \bigr\| \leq\frac{L[\beta \oplus\rho ]^{*}}{\gamma (1+[(\beta \oplus\rho)\mu]_{*})} \bigl\| U(t,\kappa )y \bigr\| $$
(3.21)
for any \(t\geq s\), \(t,s\in\mathbb{T}^{+}\). Together with \(F_{s}=U(s,\kappa )F_{\kappa }\), \(K_{2} \|U(t,s)y \|\geq e_{\alpha }(t,s) \|y \|\) holds for any \(y\in F_{s}\) and \(t\geq s\), where \(K_{2}=(L[\beta \oplus\rho]^{*}/\gamma (1+[(\beta \oplus\rho)\mu]_{*}))\).

We easily conclude that the subspace \(F_{s}\) is closed for every \(s\in\mathbb{T}^{+}\) since \(F_{s}=U(s,\kappa )F_{\kappa }\) and \(F_{\kappa }\) is closed. It follows from \(F_{t}= U(t,\kappa )F_{\kappa }= U(t,s)F_{s}\) and (3.15) that \(U(t,s)|_{F_{s}}:F_{s}\rightarrow F_{t}\) is well defined and bijection \(t\geq s\), \(t,s\in\mathbb{T}^{+}\). The proof is completed. □

We are now at the right position to establish Theorem 3.1.

Proof of Theorem 3.1

(Sufficiency). If \(U(t,s)_{t\geq s}\) admits an exponential growth and an exponential dichotomy on the time scale \(\mathbb{T}^{+}\), then
$$ \|x+y \|\geq\frac{1}{L}e_{\ominus \rho}(t,s) \bigl\| U(t,s) (x+y) \bigr\| \geq \frac {1}{L}e_{\ominus \rho}(t,s) \biggl(\frac{1}{K}e_{\alpha }(t,s)-Ke_{\ominus \alpha }(t,s) \biggr) $$
for any \(t\geq s\) and \(x\in \operatorname {Range}P(s)\), \(y\in \operatorname {Ker}P(s)\) with \(\|x \|= \|y \|=1\). This shows that there is a positive constant \(\hat{c}\) such that
$$\begin{aligned} \hat{c} \leq&\inf_{s\in\mathbb{T}^{+}} \bigl\{ \|x+y \||x\in \operatorname {Range}P(s),y\in \operatorname {Ker}P(s), \|x \|=1, \|y \|=1 \bigr\} \\ \leq&\biggl\Vert \frac{P(s)z}{ \|P(s)z \|}+\frac{\mathrm{id}-P(s)z}{ \|\mathrm{id}-P(s)z \| } \biggr\Vert \leq \frac{2 \|z \|}{ \|P(s)z \|} \end{aligned}$$
for any \(z\in X\), which implies that \(\|P(s) \|\leq2/\hat{c}:=c\) for any \(s\in\mathbb{T}^{+}\). For every \(f\in L^{p}(\mathbb{T}^{+},X)\), we let
$$u(t)=\int_{\kappa }^{t}U(t,\tau)P(\tau)f(\tau)\Delta \tau- \int_{t}^{\infty }U(t,\tau) \bigl(\mathrm{id}-P( \tau) \bigr)f(\tau)\Delta \tau. $$
It follows from (i) and (ii) in Definition 2.3 that
$$\begin{aligned} \bigl\| u(t) \bigr\| \leq& Kc\int_{\kappa }^{t}e_{\ominus \alpha }(t, \tau) \bigl\| f(\tau) \bigr\| \Delta \tau+K(1+c)\int_{t}^{\infty }e_{\ominus \alpha }(\tau,t) \bigl\| f(\tau) \bigr\| \Delta \tau \\ \leq&\biggl(\frac{Kc}{[q\odot \alpha ]_{*}} \biggr)^{1/q} \|f \|_{p}+ \biggl(\frac {1+[(q\odot \alpha )\mu]^{*}}{[q\odot \alpha ]_{*}} \biggr)^{1/q} \|f \|_{p} \end{aligned}$$
for any \(t\in\mathbb{T}^{+}\), where \(1/q+1/p=1\). Then \(u\in\mathrm{C}_{\mathrm{rd}}(\mathbb{T}^{+},{\mathbb {R}})\). A direct calculation gives the pair \((u,f)\) that satisfies (2.4). Thus, the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) is admissible for the evolution family \(U(t,s)_{t\geq s}\) on the time scale \(\mathbb{T}^{+}\). In view of (i) and (ii) in Definition 2.3, for any \(x\in E_{\kappa }\), we have \(\sup_{t\in\mathbb{T}^{+}} \|U(t,\kappa )x \|<\infty \) and
$$\begin{aligned} \frac{e_{\alpha }(t,\kappa )}{K} \bigl\| \bigl(\mathrm{id}-P(\kappa ) \bigr)x \bigr\| \leq&\bigl\| U(t, \kappa ) \bigl(\mathrm{id} -P(\kappa ) \bigr)x \bigr\| \\ \leq&\sup_{t\in\mathbb{T}^{+}} \bigl\| U(t,\kappa )x \bigr\| +Ke_{\ominus \alpha }(t,\kappa ) \bigl\| P( \kappa )x \bigr\| \\ \leq&\sup_{t\in\mathbb{T}^{+}} \bigl\| U(t,\kappa )x \bigr\| +Kc \|x \|< \infty \end{aligned}$$
for any \(t\in\mathbb{T}^{+}\). Therefore, \((\mathrm{id}-P(\kappa ))x=0\) and \(x\in \operatorname {Range}P(\kappa )\). On the other hand, it is clear that \(\operatorname {Range}P(\kappa )\subset E_{\kappa }\). This means that \(E_{\kappa }=\operatorname {Range}P(\kappa )\) is closed and complemented in X.
(Necessity). By Lemmas 3.2 and 3.3, if the pair \((\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+}, X),L^{p}(\mathbb{T}^{+},X))\) is admissible for the evolution family \(U(t,s)_{t\geq s}\) on the time scale \(\mathbb{T}^{+}\), then \(E_{s}\) and \(F_{s}\) (see (3.2)) are both closed linear subspaces for every \(s\in\mathbb{T}^{+}\). Let \(P(s)\) be the projection satisfying \(P(s)(X)=E_{s}\) for every \(s\in\mathbb{T}^{+}\). To obtain the conclusions, we need to prove \((I-P(s))(X)=F_{s}\). If \(z\in E_{s}\cap F_{s}\) for every \(s\in\mathbb{T}^{+}\), then there is \(\hat{z}\in F_{\kappa }\) such that \(U(s,\kappa )\hat{z}=z\). By \(U(t,\kappa )\hat{z}=U(t,s)U(s,\kappa )\hat{z}=U(t,s)x\in\mathrm{C}_{\mathrm{rd}}^{b}(\mathbb{T}^{+},X)\), we get \(\hat{z}\in E_{\kappa }\cap F_{\kappa }=\{0\}\) and \(z=U(s,\kappa )\hat{z}=0\). Thus, \(E_{s}\cap F_{s}=\{0\}\). For any \(z\in X\), we have \(f(t):=\chi_{[s,\eta_{s})_{\mathbb{T}^{+}}}U(t,s)z\in L^{p}(\mathbb{T}^{+},X)\) and there exists \(u\in\mathrm{C}_{\mathrm{rd}}^{b.F_{\kappa }}(\mathbb{T}^{+},X)\) such that
$$\begin{aligned} u(t) =&J(f)=U(t,s)u(s)+\int_{s}^{t}U(t,\tau)f( \tau)\Delta \tau \\ \geq& U(t,s)u(s)+\int_{s}^{\eta_{s}}U(t,\tau)f(\tau)\Delta \tau \\ \geq& U(t,s) \bigl(u(s)+z \bigr) \end{aligned}$$
for any \(t\geq\eta_{s}\), where \(\eta_{s}\) can be found in (3.5). Then we get \(u(s)+z\in E_{s}\). This implies together with the fact that \(u(s)\in F_{s}\) since \(u(\kappa )\in F_{\kappa }\) that \(z=u(s)+z-u(s)\in E_{s}+F_{s}\). Combining with \(E_{s}\cap F_{s}=\{0\}\) gives \(X=E_{s}\oplus F_{s}\). This means that \((I-P(s))(X)=F_{s}\) is well defined. Hence, we have \(U(t, s)P(s)=P(t)U(t,s)\), \(\operatorname {Range}P(s)=E_{s}\) and \(\operatorname {Ker}P(s)=F(s)\). It follows from Lemma 3.2 and Lemma 3.3 that \(U(t,s)_{t\geq s}\) admits an exponential dichotomy on the time scale \(\mathbb{T}^{+}\), where \(K=\max\{K_{1},K_{2}\}\) and β, γ, α can be found in (3.4). □

Remark 3.1

Our result extends related results known for differential equations [15] and difference equations [12] on the half-line to more general time scales.

Declarations

Acknowledgements

This research is supported by National Natural Science Foundation of China (No. 11201128 and No. 11426077), China Postdoctoral Science Foundation, Natural Science Foundation of Heilongjiang Province-A201414, Science and Technology Innovation Team in Higher Education Institutions of Heilongjiang Province (No. 2014TD005), the Heilongjiang University Fund for Distinguished Young Scholars (JCL201203) and the Fund of Heilongjiang Education Committee (No. 12541127).

Open Access This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.

Authors’ Affiliations

(1)
School of Mathematical Sciences, Heilongjiang University, Harbin, P.R. China
(2)
School of Applied Sciences, Harbin University of Science and Technology, Harbin, P.R. China

References

  1. Perron, O: Die Stabilitätsfrage bei Differentialgleichungen. Math. Z. 32, 703-728 (1930) View ArticleMATHMathSciNetGoogle Scholar
  2. Li, T: Die Stabilitäsfrage bei Differenzengleichungen. Acta Math. 63, 99-141 (1934) View ArticleMathSciNetGoogle Scholar
  3. Coppel, WA: Dichotomies in Stability Theory. Lecture Notes in Mathematics, vol. 629. Springer, Berlin (1978) MATHGoogle Scholar
  4. Chicone, C, Latushkin, Y: Evolution Semigroups in Dynamical Systems and Differential Equations. Mathematical Surveys and Monographs, vol. 70. Amer. Math. Soc., Providence (1999) MATHGoogle Scholar
  5. Massera, J, Schäffer, J: Linear Differential Equations and Function Spaces. Pure and Applied Mathematics, vol. 21. Academic Press, New York (1966) MATHGoogle Scholar
  6. Van Minh, N, Räbiger, F, Schnaubelt, R: Exponential stability, exponential expansiveness and exponential dichotomy of evolution equations on the half line. Integral Equ. Oper. Theory 32, 332-353 (1998) View ArticleMATHGoogle Scholar
  7. Van Minh, N, Huy, NT: Characterizations of dichotomies of evolution equations on the half-line. J. Math. Anal. Appl. 261, 28-44 (2001) View ArticleMATHMathSciNetGoogle Scholar
  8. Megan, M, Sasu, AL, Sasu, B: Discrete admissibility and exponential dichotomy for evolution families. Discrete Contin. Dyn. Syst. 9, 383-397 (2003) MATHMathSciNetGoogle Scholar
  9. Huy, NT: Exponential dichotomy of evolution equations and admissibility of function spaces on a half-line. J. Funct. Anal. 235, 330-354 (2006) View ArticleMATHMathSciNetGoogle Scholar
  10. Sasu, AL, Sasu, B: Exponential dichotomy on the real line and admissibility of function spaces. Integral Equ. Oper. Theory 54, 113-130 (2006) View ArticleMATHMathSciNetGoogle Scholar
  11. Sasu, B, Sasu, AL: Exponential trichotomy and p-admissibility for evolution families on the real line. Math. Z. 253, 515-536 (2006) View ArticleMATHMathSciNetGoogle Scholar
  12. Sasu, B: Uniform dichotomy and exponential dichotomy of evolution families on the half-line. J. Math. Anal. Appl. 323, 1465-1478 (2006) View ArticleMATHMathSciNetGoogle Scholar
  13. Sasu, B, Sasu, AL: Exponential dichotomy and \((l^{p},l^{q})\)-admissibility on the half-line. J. Math. Anal. Appl. 316, 397-408 (2006) View ArticleMATHMathSciNetGoogle Scholar
  14. Sasu, AL, Sasu, B: Integral equations, dichotomy of evolution families on the half-line and applications. Integral Equ. Oper. Theory 66, 113-140 (2010) View ArticleMATHMathSciNetGoogle Scholar
  15. Sasu, AL, Babuţia, MG, Sasu, B: Admissibility and nonuniform exponential dichotomy on the half-line. Bull. Sci. Math. 137, 466-484 (2013) View ArticleMATHMathSciNetGoogle Scholar
  16. Bohner, M, Peterson, A: Dynamic Equations on Time Scales: An Introduction with Applications. Birkhäuser, Boston (2001) View ArticleGoogle Scholar
  17. Hilger, S: Analysis on measure chains - a unified approach to continuous and discrete calculus. Results Math. 18, 18-56 (1990) View ArticleMATHMathSciNetGoogle Scholar
  18. Agarwal, RP, Bohner, M, O’Regan, D, Peterson, A: Dynamic equations on time scales: a survey. J. Comput. Appl. Math. 141, 1-26 (2002) View ArticleMATHMathSciNetGoogle Scholar
  19. Agarwal, RP, Bohner, M: Basic calculus on time scales and some of its applications. Results Math. 35, 3-22 (1998) View ArticleMathSciNetGoogle Scholar
  20. Bohner, M, Lutz, DA: Asymptotic behavior of dynamic equations on time scales. J. Differ. Equ. Appl. 7, 21-50 (2001) View ArticleMATHMathSciNetGoogle Scholar
  21. Hamza, AE, Oraby, KM: Stability of abstract dynamic equations on time scales. Adv. Differ. Equ. 2012, 143 (2012) View ArticleMathSciNetGoogle Scholar
  22. Li, YK, Wang, C: Almost periodic functions on time scales and applications. Discrete Dyn. Nat. Soc. 2011, Article ID 727068 (2011) Google Scholar
  23. Li, YK, Wang, C: Pseudo almost periodic functions and pseudo almost periodic solutions to dynamic equations on time scales. Adv. Differ. Equ. 2012, 77 (2012) View ArticleGoogle Scholar
  24. Lizama, C, Mesquita, JG: Almost automorphic solutions of dynamic equations on time scales. J. Funct. Anal. 265, 2267-2311 (2013) View ArticleMATHMathSciNetGoogle Scholar
  25. Siegmund, S: A spectral notion for dynamic equations on time scales. J. Comput. Appl. Math. 141, 255-265 (2002) View ArticleMATHMathSciNetGoogle Scholar
  26. Wang, C: Almost periodic solutions of impulsive BAM neural networks with variable delays on time scales. Commun. Nonlinear Sci. Numer. Simul. 19, 2828-2842 (2014) View ArticleMathSciNetGoogle Scholar
  27. Xia, YH, Li, J, Wong, PJY: On the topological classification of dynamic equations on time scales. Nonlinear Anal., Real World Appl. 14, 2231-2248 (2013) View ArticleMATHMathSciNetGoogle Scholar
  28. Zhang, JM, Fan, M, Zhu, HP: Existence and roughness of exponential dichotomies of linear dynamic equations on time scales. Comput. Math. Appl. 59, 2658-2675 (2010) View ArticleMATHMathSciNetGoogle Scholar
  29. Zhang, JM, Fan, M, Zhu, HP: Necessary and sufficient criteria for the existence of exponential dichotomy on time scales. Comput. Math. Appl. 60, 2387-2398 (2010) View ArticleMATHMathSciNetGoogle Scholar
  30. Zhang, JM, Song, YJ, Zhao, ZT: General exponential dichotomies on time scales and parameter dependence of roughness. Adv. Differ. Equ. 2013, 339 (2013) View ArticleMathSciNetGoogle Scholar
  31. Rynne, BP: \(L^{2}\) spaces and boundary value problems on time-scales. J. Math. Anal. Appl. 328, 1217-1236 (2007) View ArticleMATHMathSciNetGoogle Scholar

Copyright

© Yang et al.; licensee Springer. 2015

Advertisement