David Aristoff's homepage Title: Stochastic sensitivity analysis for continuous time jump Markovian processes

Abstract: Monte Carlo simulation is a valuable tool for sensitivity analysis of high dimensional stochastic systems. The most commonly used approaches for stochastic sensitivity analysis are the finite difference (FD) method, Girsanov transformation (GT) method and pathwise derivative (PD) method. It has been numerically observed that FD and PD methods tend to have lower variance than the GT method. I will provide a theoretical justification for this observation in terms of scaling limit analysis. If time permits, I will also talk about the recently derived path space information bounds for screening insensitive parameters.