ACVStoPACS: Computes partial autocorrelations from autocovariances
Description
Given an autocovariance sequence (ACVS) for a stationary process,
computes the corresponding partial autocorrelation sequence (PACS).
Usage
ACVStoPACS(acvs)
Value
a vector with the PACS for lags 1, ..., p.
Arguments
acvs
the ACVS for lags 0, 1, ..., p, where p must be a positive integer.
Details
The PACS
(sometimes called the reflection coefficient sequence)
is computed from the ACVS using the Levinson-Durbin recursions.
Note that the autocorrelation sequence can be used as
input rather than the ACVS since the PACS does not
in fact depend on the variance of the process
(i.e., the ACVS at lag 0).
References
S-Plus 5 Guide to Statistics,Section 24.2.
D. Percival and A. Walden, Spectral Analysis for Physical Applications,
Cambridge University Press, 1993, Section 9.4.