Skip to content

Remove finalsink superevent threshold

We've had a run of spammy uploads in ER-15, all with bad far, a low snr in one detector, and a massive snr & chisq in virgo. The issue is an old one, and was addressed in part of !163.

We have postprocessing checks on triggers before upload, including chisq thresholds (0-3, planning to change to 0.3-3) per detector to deal with cases where cohsnr and cmbchisq are dominated or skewed by a glitch in a single detector (complicating the far calculation). These postprocessing checks were only performed if a trigger exceeded an old 'superevent' far threshold.

This change removes that old 'superevent' threshold, requiring all uploads to meet the postprocessing criteria. It should catch all uploads in ER-15 which have slipped through due to glitches in virgo.

This change will be included in a branch off spiir-O4-review-v3 as a patch.

It shouldn't affect zerolags, but we could test it by looking at coincs and making sure that none have the issue seen in ER-15.

Merge request reports