Numerical analysis of multilevel Monte Carlo path simulation using the Milstein discretisation

The multilevel Monte Carlo path simulation method introduced by Giles (Operations Research, 56(3):607-617, 2008) exploits strong convergence properties to improve the computational complexity by combining simulations with different levels of resolution. Previous research has analysed its efficiency when using the Euler-Maruyama discretisation, and also demonstrated its improved efficiency using the Milstein discretisation with its improved strong convergence. In this paper we analyse its efficiency for scalar SDEs using the Milstein discretisation, bounding the order of convergence of the variance of the multilevel estimator, and hence determining the computational complexity of the method.