Approximate likelihood inference for discretely observed rough differential equations
Theodore Papamarkou (University of Glasgow)
Thursday 23rd November, 2017 16:00-17:00 Seminar room 116
This talk is split in three parts. The first part will provide a brief introduction to basic concepts from rough path theory. The second part will introduce a new iterative algorithm for approximating the increments of the latent driver of a rough differential equation (RDE) given discrete-time observations of the RDE’s response signal. The increments of the driver induce a nested sequence of linear paths, which are assumed to converge in the underlying p-variation topology to the p-rough path driving the RDE. In the third part, it will be explained how the increments of the driver can be used for constructing an approximate likelihood of the discretely observed RDE response to estimate the RDE parameters.
This is joint work with Anastasia Papavasiliou and Yang Zhao.