Split a spectrum in parts that all have the same integration and return the x values of the splits.
The first entry splits the spectrum in two halves of equal integration, the two next ones split each half
again, etc. This is the same as looking for the 1/2, then 1/4 and 3/4, then 1/8, 3/8, 5/8 and 7/8 fractions
of the total integration, which is how it is computed.
This describes a spectrum by a small vector and should provide an efficient way to store and search
similar spectra, comparing the vectors with xEqualIntegrationVectorSimilarity.
A split never lands between the peaks: it follows the integration, so it points at the signal itself. The
noise of a long baseline does not drag it either, because a point is never weighted by its own x value.
The data is considered as a piecewise linear density, so the result varies continuously with the y values
and hardly depends on the sampling density.
The window may extend beyond the data, the missing points being considered to have a y value of zero.
Parameters
data: DataXY
object that contains property x (an ordered increasing array) and y (an array).
Split a spectrum in parts that all have the same integration and return the x values of the splits. The first entry splits the spectrum in two halves of equal integration, the two next ones split each half again, etc. This is the same as looking for the 1/2, then 1/4 and 3/4, then 1/8, 3/8, 5/8 and 7/8 fractions of the total integration, which is how it is computed. This describes a spectrum by a small vector and should provide an efficient way to store and search similar spectra, comparing the vectors with
xEqualIntegrationVectorSimilarity. A split never lands between the peaks: it follows the integration, so it points at the signal itself. The noise of a long baseline does not drag it either, because a point is never weighted by its own x value. The data is considered as a piecewise linear density, so the result varies continuously with the y values and hardly depends on the sampling density. The window may extend beyond the data, the missing points being considered to have a y value of zero.