R/pspectrum2mvspectrum.R

In ForeCA: Forecastable Component Analysis

# # @rdname mvspectrum
# # 
# # @description
# # \code{.pspectrum2mvspectrum} converts the output of \code{\link[psd]{pspectrum}} to a 3D array with
# # frequencies in the first and the spectral density matrix at a given frequency in the latter
# # two dimensions.
# # 
# # @details
# # \code{.pspectrum2mvspectrum} is typically not called by the user, but by
# # \code{\link{mvspectrum}}.
# # 
# # @param pspectrum.output an object of class \code{"amt"} from \code{\link[psd]{pspectrum}}
# # @keywords ts manip

.pspectrum2mvspectrum <- function(pspectrum.output) {
  if (is.null(pspectrum.output$pspec)) {
    f.lambda <- pspectrum.output$spec
    f.lambda[pspectrum.output$freq > 0]
    # Turn into a 3D array (N x 1 x 1) for consistency with API.
    dim(f.lambda) <- c(length(f.lambda), 1, 1)
  } else {
    f.lambda <- pspectrum.output$pspec
    # remove frequency 0
    f.lambda <- f.lambda[pspectrum.output$freq > 0,,]
  }
  # Divide by 2 to match all other normalizations for other method arguments.
  f.lambda <- f.lambda / 2.
  attr(f.lambda, "frequency") <- pspectrum.output$freq[pspectrum.output$freq > 0] * pi
  return(f.lambda)
}