Title: Non-Gaussian harmonizable processes as sum of harmonics with random frequencies
Authors: Anastassia Baxevani - University of Cyprus (Cyprus) [presenting]
Krzysztof Podgorski - Lund University (Sweden)
Abstract: A class of strictly stationary non-Gaussian stochastic processes is discussed that allows to independently specify the spectral and marginal first-order distribution of the process. The proposed method models the sinusoidal component frequencies as random variables with a distribution specified by the spectral distribution. This is a key departure from the classical representation of a stationary process by the spectral theorem. While it is known that this class belongs to the non-ergodic Harmonizable Processes (HP), we identify the field of invariant sets, and thus, also the limiting behavior of the time averages of functionals of such a process. The ergodic properties of the derived models are detailed for the class of such processes with G-type marginal distributions.