Author
Title
Multilevel Quadrature Methods
Abstract
Stochastic sampling methods (such as Monte Carlo methods) are used to simulate physical systems whose model parameters are uncertain. To maintain a given level of accuracy, the spatial fidelity of the physical system being simulated and the number of samples used to estimate the stochastic quantity of interest should be sufficiently high, which could be computationally expensive. Multilevel methods aim to achieve the same overall accuracy as traditional sampling methods but at a much reduced computational cost, by making use of a series of simulation models instead of just one, each with a different level of spatial detail.