ML4GW/aframe

Investigate SV dip for production curves/apply vetoes

Closed this issue · 3 comments

alecgunny commented

As it stands, the two re-seeded production training runs have noticeable dips in SV at the lowest FARs, indicating possibly anomalously loud background events.

  1. Is this actually the case? Do their foreground distributions match those from earlier runs?
  2. Are these events nearly the same for both models?
  3. What happens when we apply vetoes?
alecgunny commented

@EthanMarx I can tackle (1) if you can tackle (2) and (3)

EthanMarx commented

Sounds good!

EthanMarx commented

It was discovered that this dip was due to a known event contaminating the background in a timeslide.

  • 👍1