Return a series of least-squares transforms of a complex time series via a divide and conquer algorithm. Each transform is minimized independently at each frequency, starting from maxfreq and descending over ncoeff frequencies for each of noctave octaves.
For each result, the complex result for a given frequency o defines the real and imaginary sinusoids which have the least distance to the data set: for a + bi, the matching sinusoids are a cos (ot) + b i sin (ot).
See also: lscomplex, fastlsreal.
Package: lssa