CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Complex-valued stochastic process modelling with some physical applications Authors:  Adam Sykulski - Lancaster University (United Kingdom) [presenting]
Abstract: In many applications, bivariate time series are represented as complex-valued time series. This representation is useful for separating series that are circular vs noncircular (sometimes referred to as proper vs improper). We present a framework for the parametric modelling of such signals using stochastic processes. We apply our framework to two applications. The first uses a novel widely-linear autoregressive process to model noncircular seismic signals. The second uses a novel anisotropic Matern process to model time series obtained from particle trajectories seeded in fluid dynamic models of turbulence.