On backward periodic autoregressive processes

摘要: It has been shown that there is a one-to-one correspondence between multivariate autoregress ive (AR) process and scalar periodic AR process. So we can analyze processes by the theory of vice versa. Multivariate have widely studied Levinson–Whittle–Wiggins–Robinson algorithm well known for obtaining predictor coeffic ient matrices. In addition, circular Levinson derived coefficients processes. In this paper, construct backward from auxiliary used in algorithm. A numerical example also presented statistical properties estimated based on sample finite size are derived.