CMStatistics 2016: Start Registration
View Submission - CFE
A1007
Title: Multivariate all-pass time series models: Modelling and estimation strategies Authors:  Bernd Funovits - University of Helsinki (Finland)
Juho Nyholm - University of Helsinki (Finland) [presenting]
Abstract: Multivariate all-pass time series models are analyzed, i.e. rational matrix functions $Q(z)$ for which $Q(z)Q(z^{-1})' = I$ holds. In particular, the poles of the determinant of all-pass matrix functions are the reciprocals of its roots. Multivariate all-pass models are important for analysing non-causal and non-invertible time series models as recently proposed. For i.i.d white noise input, all-pass models generate uncorrelated (white noise) processes; however, these processes are not independent in the non-Gaussian case. We use the theory of rational matrix factorization to gain insight into the structure of all-pass models. Furthermore, we propose estimation and identification strategies.