 Research
 Open Access
 Published:
A class of linear dynamical equations for a Banach space on time scales
Advances in Difference Equations volume 2013, Article number: 318 (2013)
Abstract
In this paper we consider the existence and uniqueness of global solutions to linear dynamical equations for a Banach space on time scales from a new point of view. We characterize those linear dynamical equations for a Banach space whose existence and uniqueness of global solutions do not depend on concrete time scales.
MSC:34G10, 34K30, 39A13.
1 Introduction
The calculus of time scales was introduced by Hilger in his PhD dissertation [1] in order to unify continuous and discrete analysis. In 2001, Bohner and Peterson published a comprehensive introduction to time scales [2]. Another book which consolidates further research in the area is [3]. The reader is referred to [1–9] for better understanding of time scales. We begin with introducing briefly some notation and terminologies.
We denote a time scale, which is a closed subset of the real numbers, by the symbol \mathbb{T}. Thus ℝ and ℤ, i.e., the real numbers and the integers, are the examples of time scales. The forward jump operator and backward jump operator \sigma ,\rho :\mathbb{T}\to \mathbb{T} and the graininess \mu :\mathbb{T}\to [0,\mathrm{\infty}) are defined by \sigma (t):=inf\{s\in \mathbb{T}:s>t\}, \rho (t):=sup\{s\in \mathbb{T}:s<t\} and \mu (t):=\sigma (t)t for t\in \mathbb{T}, respectively (supplemented by inf\mathrm{\varnothing}=sup\mathbb{T} and sup\mathrm{\varnothing}=inf\mathbb{T}). A point t\in \mathbb{T} is called rightdense, rightscattered, leftdense, leftscattered, if \sigma (t)=t, \sigma (t)>t, \rho (t)=t, \rho (t)<t hold, respectively. The set {\mathbb{T}}^{k} which is derived from the time scale \mathbb{T} is defined as follows: If \mathbb{T} has a leftscattered maximum m, then {\mathbb{T}}^{k}=\mathbb{T}\{m\}. Otherwise, {\mathbb{T}}^{k}=\mathbb{T}.
Let \mathcal{X} be a complex Banach space, \mathcal{B}(\mathcal{X}) be the class of all bounded linear operators on \mathcal{X}. A delta derivative for a function f:\mathbb{T}\to \mathcal{X} was introduced by Hilger [7].
Let t\in {\mathbb{T}}^{k}. Then we define {f}^{\mathrm{\Delta}}(t) to be the element in \mathcal{X} (provided it exists) with the property that given any \epsilon >0, there is a neighborhood U of t (i.e., U=(t\delta ,t+\delta )\cap \mathbb{T} for some \delta >0) such that \parallel [f(\sigma (t))f(s)]{f}^{\mathrm{\Delta}}(t)[\sigma (t)s]\parallel \le \epsilon \sigma (t)s for all s\in U(t,\delta ). In this case, {f}^{\mathrm{\Delta}}(t) is called the delta (or Hilger) derivative of f at t. Moreover, f is said to be delta (or Hilger) differentiable on {\mathbb{T}}^{k} if {f}^{\mathrm{\Delta}}(t) exists for all t\in {\mathbb{T}}^{k}.
This delta derivative is equal to {f}^{\prime} (the usual derivative) if \mathbb{T}=\mathbb{R}, and it is equal to △f (the usual forward difference) if \mathbb{T}=\mathbb{Z}. So, the study of dynamical equations on time scales allows a simultaneous treatment of differential and difference equations. And this field has attracted many researchers’ attention.
Especially for linear dynamical equations in finitedimensional spaces on time scales, a lot of results have been obtained (see [2–4, 10–15]). While for dynamical equations in Banach spaces on time scales, only a few results have been obtained (see [7, 16, 17]). In [7, 17], Hilger obtained the existence and uniqueness conditions for the global solutions to nonlinear and linear dynamical equations, while the conditions are closely dependent on the concrete time scales. It is natural to ask whether or not there exists a class of equations whose existence and uniqueness of global solutions do not depend on the concrete time scales.
In this paper, we consider the linear dynamical equation of the IVP
where \mathbb{T} is a time scale, the function y is from \mathbb{T} to \mathcal{X}, {y}_{0}\in \mathcal{X}, g:\mathbb{T}\to \mathcal{X} is rdcontinuous and A:\mathbb{T}\to \mathcal{B}(\mathcal{X}) is rdcontinuous in the strong operator topology (SOT).
The purpose of this paper is to investigate those classes \mathcal{A} of operators satisfying that for any time scale \mathbb{T}, if V(A):=\{A(t)\in \mathcal{B}(\mathcal{X}):t\in \mathbb{T}\}\subseteq \mathcal{A}, then equation (1) has exactly one global solution on the whole time scale. In order to state our main result, we first define a class \mathfrak{U} of operators as
where {\sigma}_{1}(A):=\{\lambda \in \mathbb{C}:A\lambda I\text{is not invertible}\} denotes the spectrum of A and {\mathbb{R}}^{}:=(\mathrm{\infty},0).
The main result of this paper is the following theorem.
Theorem 1.1 The class \mathfrak{U} is the largest one of those classes \mathcal{A} ensuring that equation (1) has exactly one global solution for any time scale \mathbb{T} and any rdcontinuous A:\mathbb{T}\to \mathcal{B}(\mathcal{X}) with V(A)\subseteq \mathcal{A}.
From Theorem 1.1 we obtain the largest class of equations whose existence and uniqueness of global solutions do not depend on the concrete time scales. In Section 2, we make some preparations about time scales and show some needed lemmas. In Section 3, we give the proof of Theorem 1.1.
2 Preparations
Firstly, we introduce some needed definitions.
Definition 2.1 [2]
A function f:\mathbb{T}\to \mathcal{X} is said to be rdcontinuous if it is continuous at rightdense points in \mathbb{T} and its leftsided limits exist at leftdense points in \mathbb{T}. The set of all rdcontinuous functions f:\mathbb{T}\to \mathcal{X} will be denoted by {\mathrm{C}}_{\mathrm{rd}}(\mathbb{T},\mathcal{X}).
Similarly, the set of all rdcontinuous functions f:\mathbb{T}\to \mathcal{B}(\mathcal{X}) in the norm topology will be denoted by {\mathrm{C}}_{\mathrm{rd}}(\mathbb{T},\mathcal{B}(\mathcal{X})).
Definition 2.2 [2]
A function A:\mathbb{T}\to \mathcal{B}(\mathcal{X}) is called regressive if I+\mu (t)A(t) is invertible for all t\in {\mathbb{T}}^{k}.
Definition 2.3 [2]
Let \mathbb{T} be a time scale and \mathcal{X} be a Banach space. A function f:\mathbb{T}\times \mathcal{X}\to \mathcal{X} is said to be

(i)
rdcontinuous if g(t)=f(t,x(t)) is rdcontinuous for any continuous function x:\mathbb{T}\to \mathcal{X};

(ii)
regressive at t\in {\mathbb{T}}^{k} if the mapping id+\mu (t)f(t,\cdot ):\mathcal{X}\to \mathcal{X} is invertible (where id is the identity function), and f is called regressive on {\mathbb{T}}^{k} if f is regressive at each t\in {\mathbb{T}}^{k};

(iii)
bounded on a set S\subset \mathbb{T}\times \mathcal{X} if there exists a constant M>0 such that \parallel f(t,x)\parallel \le M for all (t,x)\in S.
Firstly, we introduce the existence and uniqueness theorem for the dynamical equation
where \mathbb{T} is a time scale, the function y is from \mathbb{T} to \mathcal{X}, {y}_{0}\in \mathcal{X}, f(t,y) is a function from \mathbb{T}\times \mathcal{X}\to \mathcal{X}.
Lemma 2.4 [17]
If f(t,y) is regressive and rdcontinuous on {\mathbb{T}}^{k}, then equation (2) has exactly one global solution on \mathbb{T}.
Especially for the linear form of equation (2), we obtain simpler conditions in the following lemma.
Lemma 2.5 Let \mathbb{T} be some time scale and f(t,y)=A(t)y+g(t), where A:\mathbb{T}\to \mathcal{B}(\mathcal{X}) and y,g:\mathbb{T}\to \mathcal{X}. Then the following two statements hold.

(i)
The function f(t,y) is regressive on {\mathbb{T}}^{k} if and only if the function A(t) is regressive on {\mathbb{T}}^{k}.

(ii)
The function f(t,y) is rdcontinuous on {\mathbb{T}}^{k} if and only if the function A(t) is rdcontinuous in the strong operator topology and g(t) is rdcontinuous on {\mathbb{T}}^{k}.
Proof (i) `⟹’. Assume that f(t,y)=A(t)y+g(t) is regressive on {\mathbb{T}}^{k}. Then, for any t\in {\mathbb{T}}^{k}, b(t):=I+\mu (t)f(t,\cdot ):\mathcal{X}\to \mathcal{X} is invertible. So [I+\mu (t)A(t)]=b(t)\mu (t)g(t) is invertible from \mathcal{X} to \mathcal{X} and therefore A(t) is regressive on {\mathbb{T}}^{k}.
`⟸’. If the function A(t) is regressive on {\mathbb{T}}^{k}, then for any t\in {\mathbb{T}}^{k}, I+\mu (t)A(t) is invertible on \mathcal{X}. And then [I+\mu (t)A(t)](\cdot )+\mu (t)g(t) is invertible from \mathcal{X} to \mathcal{X}.
(ii) `⟹’. From Definition 2.3, it is easy to see that g(t) is rdcontinuous on {\mathbb{T}}^{k} if we take the continuous function y(t)\equiv 0. Similarly, for any fixed y\in \mathcal{X}, let the function y(t)\equiv y on \mathbb{T}, then A(t)y+g(t) is also rdcontinuous. Therefore A(t)y=(A(t)y+g(t))g(t) is rdcontinuous on {\mathbb{T}}^{k}. So the function A(t) is rdcontinuous in the strong operator topology.
`⟸’. For any continuous function y(t) on {\mathbb{T}}^{k}, and any fixed rightdense point {t}_{0}\in {\mathbb{T}}^{k}, we have
For A(t) being rdcontinuous in the strong operator topology, we have
So it suffices to show that for any \epsilon >0, there exists \delta >0 such that for any t with t{t}_{0}<\delta, we have \parallel A(t)y(t)A(t)y({t}_{0})\parallel <\epsilon.
If not, then there exist {\epsilon}_{0}>0 and a sequence {\{{t}_{n}\}}_{n=1}^{\mathrm{\infty}} with {lim}_{n\to \mathrm{\infty}}{t}_{n}={t}_{0} such that
For A(t) being rdcontinuous in the strong operator topology, we have
So {sup}_{n\ge 1}\parallel A({t}_{n})y\parallel <\mathrm{\infty} for any y\in \mathcal{X}. By the principle of uniform boundedness [18], there exists M>0 such that {sup}_{n\ge 1}\parallel A({t}_{n})\parallel <M. And then {lim}_{n\to \mathrm{\infty}}\parallel A({t}_{n})(y({t}_{n})y({t}_{0}))\parallel =0 from the continuity of y(t). This is a contradiction to inequality (3).
So f(t)=A(t)y(t) is continuous on rightdense points. Similarly, we can show that for leftdense points, the leftsided limits of f(t)=A(t)y(t) exist. □
It should be noticed that the proof of Lemma 2.5 is independent of special time scale features, except for the notion of rdcontinuity at the end. We can obtain the following corollary easily by Lemma 2.4 and Lemma 2.5.
Corollary 2.6 Let \mathbb{T} be some time scale, f(t,y)=A(t)y+g(t) be the righthand side of equation (2). If the function A(t) is regressive and rdcontinuous in the strong operator topology on {\mathbb{T}}^{k}, g(t) is rdcontinuous on {\mathbb{T}}^{k}, then equation (2) has exactly one global solution on \mathbb{T}.
Remark 2.7 Corollary 2.6 implies the result for finitedimensional spaces (see [[2], Theorem 5.8]) and improves the conclusion for infinitedimensional spaces (see [[7], Section 6]) because the strong operator topology is equal to the norm topology in finitedimensional spaces but weaker than the latter in infinitedimensional spaces.
Therefore the condition ‘strong operator topology’ in (ii) of Lemma 2.5 cannot be replaced by ‘norm topology’. In fact, we have the following counterexample.
Example 2.8 Let {L}^{2}[0,1] be the space of Lebesgue square integrable functions on [0,1], and let \mathcal{B}({L}^{2}[0,1]) be the class of all bounded linear operators on {L}^{2}[0,1]. For t\in [0,1] and x(s)\in {L}^{2}[0,1], define
Then A:[0,1]\to \mathcal{B}({L}^{2}[0,1]) is an operatorvalued function. We first show that A(t) is rdcontinuous in strong operator topology.
Arbitrarily, choose x\in {L}^{2}[0,1]. For any {t}_{0}\in (0,1) being rightdense, we have
This implies that A:[0,1]\to \mathcal{B}({L}^{2}[0,1]) is continuous with respect to the SOT in the range space. On \mathbb{T}=[0,1] the two notions of rdcontinuity and continuity coincide.
On the other hand, it is obvious that for any t,{t}^{\prime}\in [0,1], t>{t}^{\prime}, A(t)A({t}^{\prime}) is an orthogonal projection on {L}^{2}[0,1] with norm 1. Therefore we conclude that A(t) is not rdcontinuous in the norm topology.
3 Proof of the main result
From Lemma 2.4 and Corollary 2.6, we can see that existence and uniqueness of the global solution to equation (1) depend closely on the regressivity of the operatorvalued function A(\cdot ). Now we are going to give the proof of the main theorem.
Proof of Theorem 1.1 We give the proof by two steps.
Step 1. \mathcal{U} is a class of operators which ensures that equation (1) has a unique solution for any time scale \mathbb{T} and any A:\mathbb{T}\to \mathcal{B}(\mathcal{X}) with V(A)\subseteq \mathcal{A}.
We have the graininess \mu :\mathbb{T}\to [0,\mathrm{\infty}). If V(A)\subseteq \mathcal{U}, then I+\mu (t)A(t) is invertible for any t\in {\mathbb{T}}^{k}. So A(\cdot ) is regressive on any time scale. Moreover, we know from equation (1) that g(\cdot ) is rdcontinuous and A(\cdot ) is rdcontinuous in the strong operator topology. From Corollary 2.6, equation (1) has a unique solution on the whole time scale.
Step 2. Suppose that \mathfrak{m} is a class of operators with \mathfrak{M}\u2288\mathfrak{U} satisfying condition (i). We need only to prove that there exist a time scale {\mathbb{T}}_{0} and a function A from {\mathbb{T}}_{0} to \mathcal{B}(\mathcal{X}) with V(A)\subseteq \mathfrak{M} such that equation (1) either has no solution or has at least two solutions on {\mathbb{T}}_{0}.
Take {A}_{0}\in \mathfrak{M}\mathrm{\setminus}\mathfrak{U}, then there exists \lambda >0 such that \lambda \in \sigma ({A}_{0}). We consider the following equation on {\mathbb{T}}_{0}=\frac{1}{\lambda}\mathbb{Z}:
where {y}_{0}\ne 0. Obviously, the solution to equation (4) is
It follows from \lambda \in \sigma ({A}_{0}) that {A}_{0}+\lambda I is not invertible.
If {A}_{0}+\lambda I is not injective, then y({t}_{0}\frac{1}{\lambda}) is not unique.
If {A}_{0}+\lambda I is not surjective, then we can choose {y}_{0}\in \mathcal{X}\mathrm{\setminus}\mathcal{R}(I+\frac{1}{\lambda}{A}_{0}) (where \mathcal{R}(I+\frac{1}{\lambda}{A}_{0}) is the range of (I+\frac{1}{\lambda}{A}_{0})). Hence y({t}_{0}\frac{1}{\lambda}) does not exist. This completes the proof. □
Remark 3.1 For \mathcal{B}(\mathcal{X}) being the matrix algebra {M}_{n}(\mathbb{C}), it is easy to see that the closure of \mathfrak{U} is equal to {M}_{n}(\mathbb{C}). Hence we can deduce that \mathfrak{U} is a very large class of operators in {M}_{n}(\mathbb{C}). We can also consider the closure and interior of the class \mathfrak{U} in a Hilbert space in the future.
References
Hilger, S: Ein maskettenkalkul mit anwendung auf zentrumsmannigfaltigkeiten. Dissertation, PhD, Univ. Wurzburg (1988)
Bohner M, Peterson A: Dynamic Equations on Time Scales. Birkhäuser, Boston; 2001.
Bohner M, Peterson A: Advances in Dynamic Equations on Time Scales. Birkhäuser, Boston; 2003.
Agarwal RP, Bohner M, O’Regan D, Peterson A: Dynamic equations on time scales: a survey. J. Comput. Appl. Math. 2002, 141(12):126. 10.1016/S03770427(01)004320
Akin E, Bohner M, Erbe L, Peterson A: Existence of bounded solutions for second order dynamic equations. J. Differ. Equ. Appl. 2002, 8(4):389401. In honor of Professor Lynn Erbe 10.1080/1026190290017414
Anderson D, Peterson A: Asymptotic properties of solutions of a 2 n thorder differential equation on a time scale. Math. Comput. Model. 2000, 32(56):653660. 10.1016/S08957177(00)00162X
Hilger S: Analysis on measure chains  a unified approach to continuous and discrete calculus. Results Math. 1990, 18(12):1856. 10.1007/BF03323153
Hilger S: Matrix Lie theory and measure chains. J. Comput. Appl. Math. 2002, 141(12):197217. 10.1016/S03770427(01)004460
Peterson AC, Tisdell CC: Boundedness and uniqueness of solutions to dynamic equations on time scales. J. Differ. Equ. Appl. 2004, 10(1315):12951306. 10.1080/10236190410001652793
Agarwal RP, Bohner M, Shakhmurov VB: Linear and nonlinear nonlocal boundary value problems for differentialoperator equations. Appl. Anal. 2006, 85(67):701716. 10.1080/00036810500533153
Bohner M, Saker SH: Oscillation of second order halflinear dynamic equations on discrete time scales. Int. J. Differ. Equ. 2006, 1(2):205218.
DaCunha, JJ: Lyapunov stability and Floquet theory for nonautonomous linear dynamic systems on time scales. Dissertation, PhD, Baylor University (2004)
Došlý O, Hilger S: A necessary and sufficient condition for oscillation of the SturmLiouville dynamic equation on time scales. J. Comput. Appl. Math. 2002, 141(12):147158. 10.1016/S03770427(01)004423
Guseinov GS, Zafer A: Stability criteria for linear periodic impulsive Hamiltonian systems. J. Math. Anal. Appl. 2007, 335(2):11951206. 10.1016/j.jmaa.2007.01.095
Messer, KR: Linear dynamic equations on time scales. Dissertation, PhD, The University of Nebraska, Lincoln (2003)
Feng YL: Acta Math. Sci. Ser. A Chin. Ed.. 2011, 31A(2):439446.
Hilger S: Differential and difference calculus  unified! 30. Proceedings of the Second World Congress of Nonlinear Analysts, Part 5 1997, 26832694.
Conway JB Graduate Texts in Mathematics 96. In A Course in Functional Analysis. 2nd edition. Springer, New York; 1990.
Acknowledgements
The author gratefully acknowledges the invaluable advice and encouragement of Professor You Qing Ji. I would like to express my sincere thanks to the reviewers for their helpful comments and advice. Supported by NNSF of China (11271150) and ‘Twelfth FiveYear Plan’ Science and Technology Research Project of the Education of Jilin Province (2013215, 2012186, 2013373, 2011113).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
About this article
Cite this article
Feng, Y. A class of linear dynamical equations for a Banach space on time scales. Adv Differ Equ 2013, 318 (2013). https://doi.org/10.1186/168718472013318
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/168718472013318
Keywords
 time scales
 linear dynamical equations
 existence
 uniqueness