Dear all:
My experience in using hosking.sim to generate random time series is
that it requires an n-length autocorrelation function (ACF; order 0 to
n-1) to generate n-length random vectors. I have attempted to utilize
shorter than n-length (i.e. truncated) ACF's with hosking.sim to
generate n-length random vectors, but the function will not operate for
n-length time series without the full-length ACF series.
The help available for this function via R's online capability only
gives a generic description of the nature of the function, with no
details.
I am aware that, when modeling time series with AR models, generally
low-order AR processes are used, and that there are objective criteria
(e.g. BIC and AIC) to optimally limit the order of the AR model. In
light of this, I would like to understand why hosking.sim appears to
require the full (0 to n-1) ACF to generate n-length random time series
vectors.
Any information in this regard would be helpful.
Sincerely, Gene
*******************************
Dr. Eugene R. Wahl
Asst. Professor of Environmental Studies
Alfred University
[[alternative HTML version deleted]]
