A lattice algorithm for factoring the spectrum of a moving average process

Triangular decomposition of the semi-infinite covariance matrix of a moving average process can be used as a spectral factorization technique. An efficient lattice algorithm is derived for performing the necessary computations. This technique is a special case of the fast Cholesky decomposition of stationary covariance matrices. The algorithm can be used to factor multichannel spectra to a desired degree of accuracy.