On the convergence of solutions to dynamic equations on time scales (Q2628043)

Continuous dependence of solutions to dynamic equations with respect to the choice of the time scale has been the subject of numerous papers (see, e.g., [\textit{P. E. Kloeden}, in: Nonlinear analysis and applications: To V. Lakshmikantham on his 80th birthday. Vol. 2. Dordrecht: Kluwer Academic Publishers. 645--659 (2003; Zbl 1071.34058); \textit{B. M. Garay} et al., in: New progress in difference equations. Proceedings of the 6th international conference on difference equations, Augsburg, Germany July 30--August 3, 2001. Boca Raton, FL: CRC Press. 279--287 (2004; Zbl 1065.39030); \textit{L. Adamec}, J. Difference Equ. Appl. 17, No. 5, 647--656 (2011; Zbl 1227.34095); \textit{M. Cichoń} and \textit{A. Yantir}, Appl. Math. Comput. 252, 473--483 (2015; Zbl 1338.34173)]. Let \(\{\mathbb T_n\}_{n=1}^\infty\) be a sequence of time scales such that \(\mathbb T_n\to\mathbb T\) in some sense. Moreover, assume that all the time scales \(\mathbb T_n\) contain a point \(t_0\in\mathbb R\). Now, consider a function \(f:\hat\mathbb T\times\mathbb R^d\to\mathbb R^d\), where \(\hat\mathbb T=\overline{\mathbb T\cup\bigcup_{n=1}^\infty\mathbb T_n}\). The problem is to find sufficient conditions guaranteeing that solutions of the initial-value problems \[ x_n^\Delta(t)=f(t,x_n(t)),\quad t\in\mathbb T_n,\quad x_n(t_0)=x_0,\quad n\in\mathbb N,\tag{1} \] are in some sense convergent to a solution of the initial-value problem \[ x^\Delta(t)=f(t,x(t)),\quad t\in\mathbb T,\quad x(t_0)=x_0. \tag{2} \] Since \(x_n\) and \(x\) are defined on different domains (whose intersection might be empty), it is necessary to describe the precise meaning of the convergence of \(x_n\) to \(x\). The authors of the present article suppose that \(\mathbb T_n\to\mathbb T\) in the Hausdorff distance, all the time scales \(\mathbb T_n\) have bounded graininess, \(f\) is continuous on its domain and Lipschitz-continuous in the second variable. For each \(t\in\mathbb T\) and \(n\in\mathbb N\), they let \(t_n^*\) be the point of \(\mathbb T_n\) that is closest to \(t\) (if there are two such points, choose the smaller number). Then for each \(T>t_0\), the solutions \(x_n\) of Eq. (1) on \([t_0,T]_{\mathbb T_n}\) approach the solution of Eq. (2) on \([t_0,T]_{\mathbb T}\) in the following sense: \[ \lim_{n\to\infty}\sup_{t\in[t_0,T]_\mathbb T}\|x(t)-x_n(t^*_n)\|=0.\tag{3} \] The result is illustrated on the example of the logistic equation \[ x^\Delta(t)=x(t)(1-4x(t)),\quad t\in\mathbb T. \] Although the corresponding function \(f\) is not Lipschitz-continuous, the authors remark that their convergence result remains valid if \(f\) is locally Lipschitz-continuous and \(\{x_n\}_{n=1}^\infty\) is uniformly bounded. The second main result of the paper is concerned with the rate of convergence in (3). It says that if \(f\) is Lipschitz-continuous in both variables, then \(\|x(t)-x_n(t^*_n)\|\) can be estimated by a constant multiple of the Hausdorff distance between \(\mathbb T_n\) and \(\mathbb T\) (the constant depends on the choice of \(T\)). In the conclusion, the authors mention an open problem of studying continuous dependence of solutions to nabla dynamic equations \(x^\nabla(t)=f(t,x(t))\) with respect to the choice of the time scale. Before the present paper was published, they have already managed to solve the problem in their paper [Dyn. Syst. Appl. 24, No. 4, 451--466 (2015; Zbl 1336.34132)].