Computational aspects on the application of the Gaussian state space model

by Ibrahim Amrani & Abdelhalim Skalli.

Abstract: The goal of this article is to show the possibility of combining the Rosenberg algorithm with the diffuse method of Koopman, in order to calculate the initial conditions of a state vector which comprises stochastic and fixed components. Under these conditions, and by applying EM algorithm, the covariance matrix of the disturbance vector of the transition equation is estimated while being diagonal. The unknown parameters of the transition matrix are estimated by applying selection matrices in the equation of EM algorithm.

Key Words: : EM algorithm; Rosenberg algorithm; Diffuse initial conditions; State space model

Authors:
Ibrahim Amrani, ibrahim_amrani@yahoo.fr
Abedelhalim Skalli, skalli@emi.ac.ma

Editor: Ke, Weiming, Weiming.Ke@sdstate.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (128087 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 1988 times since OCTOBER 14, 2008.


Return to the InterStat Home Page.