Andrew Stuart (California Institute of Technology, USA)
Lévy(-type) processes arise naturally as models of a (state dependent) jump behaviour in a wide variety of situations in natural sciences and finance. The topic of this talk is induced by the need to measure the distance between the model and the data, e.g. to obtain "goodness-of-fit" tests or to estimate certain parameters of the model. After specifying the problem we fomulate it in terms of noise sensitivity of certain classes of stochastic differential equations.