Computing the likelihood and its dierivatives for a gaussian ARMA model

Robert Kohn, Craig F. Ansley · Journal of Statistical Computation and Simulation · 1985

A fast algorithm is obtained for computing the likelihood of a Gaussian ARMA (p,q) model and its derivatives by combining the algorithms of Pearlman (1980) and Ansley (1979). Only p + 3 q + 2 operations are required per observation. In seasonal models structural zeros are exploited to speed up the computation.

Read the paper · More papers on PaperTik