Thomson's multitaper estimate: the eigenvalue-weighted average of the
periodograms of the signal tapered by the first K Slepian sequences,
sum_k lambda_k |X_k(f)|^2 / (fs sum_k lambda_k), returned as a
one-sided density (every bin except DC, and Nyquist for even nfft,
doubled, as in scipy.signal.periodogram) so that it integrates to the
variance of the signal.
Usage
morie_psdmt(x, fs = 1, nw = 4, n_tapers = NULL, nfft = NULL)
Arguments
- x
Numeric signal.
- fs
Sampling frequency in Hz.
- nw
Time-bandwidth product.
- n_tapers
Number of tapers; 2 nw - 1 by default.
- nfft
FFT length; length(x) by default.
Value
List with value (the integrated power), frequencies,
psd, n_tapers, concentrations.
Examples
x <- sin(2 * pi * 10 * (0:255) / 100)
r <- morie_psdmt(x, fs = 100)
r$frequencies[which.max(r$psd)]
#> [1] 10.15625
