Calculate the minimum variance and corresponding optimal weights of a localized multitaper spectral estimate.
weight_opt = SHMTVarOpt (
var_opt: float, dimension (
- The minimum variance of the multitaper spectral estimate for degree
lusing 1 through
var_unit: float, dimension (
- The variance of the multitaper spectral estimate using equal weights for degree
lusing 1 through
weight_opt: float, dimension (
- The optimal weights (in columns) that minimize the multitaper spectral estimate’s variance using 1 through
- The angular degree to determine the minimum variance and optimal weights.
tapers: float, dimension (
- A matrix of localization functions obtained from
taper_order: integer, dimension (
- The angular order of the windowing coefficients in TAPERS. If this matrix was created using
SHReturnTapersM, then this array must be composed of zeros.
sff: float, dimension (
- The global unwindowed power spectrum of the function to be localized.
lwin: optional, integer, default =
- The spherical harmonic bandwidth of the localizing windows.
kmax: optional, integer, default =
- The maximum number of tapers to be used when calculating the minimum variance and optimal weights.
nocross: optional, integer, default = 0
- If 1, only the diagonal terms of the covariance matrix Fij will be computed. If 0, all terms will be computed.
SHMTVarOpt will determine the minimum variance that can be achieved by a weighted multitaper spectral analysis, as is described by Wieczorek and Simons (2007). The minimum variance is output as a function of the number of tapers utilized, from 1 to a maximum of
kmax, and the corresponding variance using equal weights is output for comparison. The windowing functions are assumed to be solutions to the spherical-cap concentration problem, as determined by a call to
SHReturnTapersM. The minimum variance and weights are dependent upon the form of the global unwindowed power spectrum,
If the optional argument
nocross is set to 1, then only the diagnonal terms of
Fij will be computed.
Wieczorek, M. A. and F. J. Simons, Minimum-variance multitaper spectral estimation on the sphere, J. Fourier Anal. Appl., 13, doi:10.1007/s00041-006-6904-1, 665-692, 2007.