changes delta sigma cut
Correct delta sigma cut. The delta sigma cut consist on comparing the sigma from a segment with that from adjacent segments and if there exists a difference larger than a certain threshold (we choose 0.2) then the segment is flagged as "bad":
Numerator: Sliding_sigma_correct_alpha * sliding_bias_factor - naive_sigma_correct_alpha * naive_bias_factor
Denominator: Sliding_sigma_correct_alpha * sliding_bias_factor
Segment removed if: abs(Numerator)/denominator >= dsc
The sliding and naive sigmas are calculated starting with the function calculate_point_estimate_sigma_spectra
. This function calculates S_alpha
, given by
This S_alpha is then used to compute sigma^2
as
The inverse of sigma^2
is then computed to get the naive and sliding sensitivity integrands:
Finally, performing the square root of the inverse of this result is our final sliding or naive sigma:
The bias factors are computed with calc_bias
from util.py