pygwb.delta_sigma_cut

In general, the noise level in ground-based detectors changes slowly on time-scales of tens of minutes to hours. The variance associated to each segment is an indicator of that level of noise, which typically changes at roughly the percent level from one data segment to the next. However, there are occasional very loud disturbances to the detectors, such as glitches, which violate the Gaussianity of the noise. Auto-gating procedures are in place to remove loud glitches from the data; however the procedure does not remove all non-stationarities. To avoid biases due to these noise events, an automated technique, called delta-sigma cut, to exclude them from the analysis has been developed, which flags specific segments to be cut from the analyzed set.

Examples

As an example, we show how to use delta sigma cut. To this end, we import the relevant packages:

>>> import numpy as np
>>> from pygwb.delta_sigma_cut import dsc_cut

For concreteness, we use some randomly generated data arrays as placeholders for naive_sigma and sliding_sigma:

>>> naive_sigma = np.random.normal(size=10)
>>> sliding_sigma = np.random.normal(size=10)

The dsc_cut method can be called with its default parameters:

>>> dsigma_mask, dsigma = dsc_cut(naive_sigma, sliding_sigma)

The result is a mask containing booleans, which indicates whether or not the segment should be considered in the remainder of the analysis. In addition, the actual value of the difference in sigmas is given as well.

Functions

dsc_cut(naive_sigma, slide_sigma[, dsc, ...])	Function that performs the delta sigma cut, a veto that marks certain GPS times as unusable if the estimation of the PSD in the naive (estimating sigma in bin J) and sliding (estimating sigma in bins J pm 1) differ by more than a certain threshold:
run_dsc(dsc, segment_duration, psd1_naive, ...)	Function that runs the delta sigma cut.