Approximate Maximum Likelihood Estimation of a Step Function Spectral Density.
CARNEGIE-MELLON UNIV PITTSBURGH PA DEPT OF STATISTICS
Pagination or Media Count:
Approximate maximum likelihood estimates are obtained for the ordinates of a step function spectra density in the Gaussian case. The estimates are simply integral averages of the periodogram over the frequency bands in which the density is constant. Whittles form of the approximate likelihood is used. Results are given for scalar and vector processes. Author
- Theoretical Mathematics