This method minimizes an approximate log-likelihood function to obtain an estimate of the parameters of a process. It is semi-parametric in that it only assumes the form of the spectral density close to the origin. It does not produce a graphical output.

A more detailed description.


There are several routines associated with the local Whittle estimator.

  1. locwhitt is the basic routine. Has parameters h which is starting H value for minimization, and im which tells program to only use N/im frequencies in minimization.
  2. locwhitt.cont -- as above, but computes estimate for several different values of im, and returns a vector of results.
  3. locwhitt.cont.theory -- as locwhitt.cont, but does it for theoretical spectral density instead of periodogram. Spectral density must be supplied.
  4. All of the above routines but with the suffix 2 use the cumulative periodogram instead of the periodogram to do the same things.