1: \begin{abstract}

2:           {\rm\small  

3:           This article is concerned with the

4:           multilevel Monte Carlo (MLMC) methods for approximating expectations of some functions of the solution to the Heston 3/2-model 

5:           from mathematical finance, 

6:           which takes values in $(0, \infty)$

7:           and possesses superlinearly growing drift and diffusion coefficients. 

8:           To discretize the SDE model, a new Milstein-type scheme is proposed to produce independent sample paths.

9:           The proposed scheme can be explicitly solved and is positivity-preserving unconditionally, i.e., for any time step-size $h >0$. This  positivity-preserving property for large discretization time steps is particularly desirable in the MLMC setting.

10:           Furthermore, a mean-square convergence rate of

11:           order one is proved in the non-globally Lipschitz regime, which is not trivial, as the diffusion coefficient grows super-linearly. 

12:           The obtained order-one convergence in turn promises the desired relevant variance of the multilevel estimator

13:         and justifies the optimal complexity $\mathcal{O}(\epsilon^{-2})$ for the MLMC approach, where

14:         $\epsilon > 0$ is the required target accuracy. 

15:         Numerical experiments are finally reported to 

16:         confirm the theoretical findings.

17:           %covering SDEs  with super-linearly growing diffusion coefficients, such as the popular 3/2-model .

18:            } \\

19: 

20: \textbf{AMS subject classification: } {\rm\small 60H35, 60H15, 65C30.}\\

21: %\textbf{PACS: 02.60.Lj}

22: 

23: \textbf{Key Words: }{\rm\small}  Multilevel Monte Carlo methods, Heston 3/2-model, 

24: positivity-preserving methods, strong convergence rate

25: \end{abstract}

26: