Multichannel time varying autoregressive modeling: a circular lattice-smoothness priors realization

An algorithm for multichannel time varying autoregressive (MCTVAR) modeling of nonstationary covariance time series data is shown. The multichannel modeling is achieved by doing things one channel at a time using only scalar computations. The method exploits the smoothness priors modeling (W. Gersch and G. Kitagawa, 1988) of partial correlation coefficients in a time-varying linear regression model and the `circular lattice-form´ structure (H. Sakai, 1982) for multichannel stationary time series modeling. The circular lattice structure permits the multichannel model to be realized one channel at a time. Smoothness priors permit fitting the MCTVAR model with the explicit computation of only a small number of hyperparameters. An example is shown