-
Notifications
You must be signed in to change notification settings - Fork 356
Multiifo Ranking Statistic Development
Gareth Davies edited this page Jul 23, 2019
·
21 revisions
Name of problem/issue | Status | Testing/Verification |
---|---|---|
Total noise rate for each type of coinc | Done | predicted rate vs timeslide rate |
Relate total number of single triggers to thresholded fit | Not yet, set up plotting script | see distribution and fitting of below-threshold triggers |
convert predicted noise rate into statistic | Done, exp_fit_sg_coinc_rate statistic in stat.py. See #2689 | coincident trigger scatter plot, and comparison histograms (LIGO login required). |
Proper combination of different types of coinc via statistic value (not FAR/trials factor) | Done, pycbc_multiifo_add_statmap see #2811 & #2834 | compare unclustered event FAR assignments from trials factor vs adding histograms. stat vs effective trials factor, Note that the prediction doesn't exactly line up for a known, very fiddly reason |
Approximation of network sensitivity from sigma^2 | Done. See #2843 | plot sensitivity vs time /mass/etc. compare to standard plots. |
Add in network sensitivity into statistic | In review #2850 | Newstat vs previousstat as function of time/coinc type. Stat with and without sigma: difference vs time and comparison of statistic values |
Signal extent over time difference, include in signal rate part of statistic | In coinc_rate module | compare to noise trigger extent allowed time differences for noise and signal triggers |
Include P/T/A consistency checks | Being prepared: #2830 | Compare background trigger statistic histograms |