Fits a multivariable autoregressive model to a time-series.
[Af,Ab,Sf,Sb] = destmar(v,d)
This function fits a multivariable autoregressive model to a time-series v(k). The model-structure is
in which ef(k) and eb(k) are innovation sequences with covariance matrices Sf and Sb respectively. The fitting is performed according to .
v is the time-series, a N x l matrix for a signal having N samples and which is l-dimensional.
d is the desired order d of the AR model.
Af,Ab are the coefficient matrices Af and Ab of the causal and anticausal model.
Sf,Sb are the covariance matrices Sf and Sb of the causal and anticausal innovations.
A direct Hankel-matrix based estimation of the AR model is performed according to .
This is a top-level function that is used directly by the user.
 B. Davis, Parameter Estimation in Nonlinear Dynamical Systems with Correlated Noise. PhD thesis, Universite Catholique de Louvain-La-Neuve, Belgium, 2001.