Central limit theorem for the multilevel Monte Carlo Euler method (Q2258530)

The multilevel Monte Carlo method for the computation of quantities \(\mathbb E[f(X_T)]\), where \((X_t)_{0\leq t \leq T}\) is a diffusion process and \(f\) some function (e.g. to price options), yields a remarkable reduction of the complexity in contrast to crude Monte Carlo algorithms. In this paper, the authors derive a central limit theorem of Lindeberg-Feller type for the multilevel Monte Carlo Euler scheme. The main tool is a stable law convergence theorem for the Euler scheme error of two consecutive levels of the multilevel algorithm. This result gives a precise description for the choice of the parameters to run the multilevel Monte Carlo Euler scheme. The explicitly characterized limiting variance in the central limit theorem allows to construct more accurate confidence intervals. A complexity analysis confirms the efficiency of the multilevel method. Moreover, a Berry-Esseen type bound is established.