1: \begin{abstract}

2: The semi-implicit Euler-Maruyama (EM) method is investigated to approximate a class of time-changed stochastic differential equations, whose drift coefficient can grow super-linearly and diffusion coefficient obeys the global Lipschitz condition. The strong convergence of the semi-implicit EM is proved and the convergence rate is discussed. When the Bernstein

3: function of the inverse subordinator (time-change)

4: is regularly varying at zero, we establish the mean square polynomial stability of the underlying equations. In addition, the numerical method is proved to be able to preserve

5: such an asymptotic property. Numerical simulations are presented to demonstrate the theoretical results.

6: 

7: \medskip \noindent

8: {\small\bf Key words}: time-changed stochastic differential equations, semi-implicit Euler-Maruyama method, strong convergence, mean square polynomial stability, subordinator.

9: \par \noindent

10: {\small\bf 2010 MSC2010}: 60H10, 65C30, 60J60.

11: \end{abstract}

12: