Oberseminar Numerische Mathematik / Scientific Computing
Des Higham
University of Strathclyde, UK
The Multi-level Monte Carlo Method
Abstract:
In 2008 Mike Giles introduced and analysed the multi-level Monte Carlo method. This method can dramatically improve computational complexity when we are sampling from the approximate solution of a stochastic differential equation. The performance of the method depends on both the weak and strong convergence properties of the underlying timestepping scheme. I will review the multi-level method and then focus on a new extension to the case of estimating the mean hitting time.
| Datum: | 24.01.11 | Zeit: | 17:00 Uhr | Ort: | FU Berlin, Institut für Mathematik, Arnimallee 6, 14195 Berlin. | Raum: | 031 im Erdgeschoss |