pk_parabola2#

peaklets.pk_parabola2(Nt)[source]#
Convex parabolic peaklets. The scale is FWHM of the parabola,

which is the distance between roots divided by np.sqrt(2). In this version, the roots of the parabola are placed at integer distances, k_n, from the center. The log-k spacing is coarse at first, but it quickly converges to about 11.5 steps per decade in scale. This avoids the problem of informationless scales when the kernel is small, but delivers high resolution when the scale is larger. The sequence used for the parabola half-width is:

k_n = k_{n-3} + k_{n-4}

Input:

Nt, the length of the time series to be transformed.

Output:

sc, a 1D numpy integer array of scales. pk, a list of 1D numpy float arrays, containing the peaklet

functions associated with each element of sc. Note that len(pk[i]) = 1+sc[i].