On irregular functionals of SDEs and the Euler scheme (Q964680): Difference between revisions

The main result of the paper contains a general principle that gives a sharp upper bound for the functional \({E}| g(X)-g(\hat{X})| ^p\) in terms of moments of \(X-\hat{X}\). Here \(X\) and \(\hat{X}\) are random variables and \(g\) is a function of bounded variation, e.g. the payoff of the binary option. The convergence result for the functions automatically gives a convergence rate for the functional, and it is arbitrary close to the original rate. The results are applied to the approximation of a solution to a stochastic differential equation by the Euler scheme, and it is demonstrated that the approximation of the payoff of the binary option has asymptotically sharp strong convergence rate 1/2. This result, in turn, is applied to the multilevel Monte Carlo method, and the improvement in the mean square error of the multilevel estimator is obtained.