Multilevel Monte Carlo (MLMC) methods are a general Monte Carlo approach combining simulations with different levels of accuracy, at correspondingly different costs. The overall aim is to obtain an estimate for the expected value of an output quantity with a root-mean-square accuracy of ε at a cost which is O(ε-2).
The first half of the talk will give an overview of the key ideas, which are very simple. After briefly outlining the wide range of applications, the second half will cover some of my current MLMC research with various collaborators.
http://people.maths.ox.ac.uk/gilesm/ -- homepage
http://people.maths.ox.ac.uk/gilesm/mlmc.html -- my MLMC publications
http://people.maths.ox.ac.uk/gilesm/mlmc_community.html -- global MLMC
http://people.maths.ox.ac.uk/gilesm/slides.html -- my MC/MLMC talks