Spectral factorization of z-domain polynomials using non-linear least-squares

A new recursive method for spectral factorizing polynomials is given. The method uses non-linear least-squares. The method is simple to implement and has applications in many areas of adaptive filtering, system identification and control. The optimization method is not new but to date has not been applied to spectral factorization. Convergence is much faster than ordinary root finding algorithms. (superlinear versus quadratic).