In many areas of science and engineering, for example, in oceanography, meteorology and target tracking, one uses information from an uncertain model, supplemented by a stream of noisy and incomplete observations, to deduce the current state of a complex system.
I will explain sequential Monte Carlo methods that are currently used in the assimilation of data, show problems that arise as the models become increasingly complex, and how these problems are tackled by the recently developed implicit particle filter. I will present in detail two examples: a stochastic version of the Kuramoto-Sivashinsky equation and a problem that arises in the modeling of the secular variation of the earth's magnetic field.