M. D. Ruiz Medina, R. Salmerón Gómez
The autoregressive Hilbertian time series (ARH(p) series) framework (see Bosq, 2000; Bosq, 2010; Bosq and Blanke, 2007) provides a new perspective for deriving models and tools in the statistical analysis of spatiotemporal data displaying weak-dependence in time (see Salmerón y Ruiz-Medina, 2009). Moment-based parameter estimation has been addressed in Bosq (2000). These results are alternatively formulated in the framework of projection into a general orthogonal basis in Bosq and Blanke (2007). The maximum-likelihood parameter estimation problem is investigated in Ruiz-Medina and Salmerón (2010). In this paper, we extend the asymptotic study initiated in Ruiz-Medina and Salmerón (2011) on the large-sample variance properties of these estimators, in terms of the SEM (Structural Expectation Maximization) algorithm, to the derivation of their consistency and asymptotic distributional properties.
Palabras clave: ARH(p) processes, functional limit results, Hilbert-valued processes, maximum likelihood estimation
Programado
VB7 Probabilidad, convergencias y teoremas límite
20 de abril de 2012 10:30
Sala Roma II
