190820

Agenda

Attendance: Charlie, Ethan, Greg, Isobel, Matt, Nikhil, Paul, Shanika, Sharan, Simon Minute taker: Paul

Tidal stuff has been added to the analytic waveforms test by Nikhil Sarin (https://git.ligo.org/lscsoft/bilby_pipe/wikis/O3-review/analytic-waveforms)

update from last week. Things looked mainly fine, but a little weird. GA emailed the PE list; when there is zero noise, everything looks absolutely fine (https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/bilby_review/bilby0.5.4_bilby_pipe0.2.2/outdir_fiducial_bbh_128s_ROQ_zero-noise/result/fiducial_bbh_128s_ROQ_zero-noise_190801_0435_combined_1d/). Christopher Berry emailed on the list saying that this must therefore just be an artifact of the noise realisation.
SS.: Yep, looks fine.
MP.: Looks good also
Ticked!!!

170814: Was ready last week, but only had LI MCMC.
- IRS: Difference between LI and Bilby is no different than difference between two LI runs for other events (ie MCMC and Nest)
- GA: We don’t expect the spins to match up because this is an O2 run.
- PL and SS: LI has the distance bug, so don’t expect that to match up, either
- MP. “Looks good to me”
- SS: “Yup”
- Ticked!!!
170104 (https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/bilby_pe_event_samples/GW170104/comparison/html/Publication.html). Thanks to Sharan for digging into this.
- distance bug exists. \theta_jn looks slightly different, but note that same difference exists between LI MCMC and Nest for GW151226 as well.
- After investigation, the LI plots are done with SEOBNRv3, so it seems like differences are down to waveform systematics!

Review scope discussion:

MP: The final paper will have comparisons. So it’ll be nice to do these at some point, but don’t necessarily have to have all done for the end of the review.
PL: Yes, but doesn’t have to have all direct comparisons
PL: Need to do GW170817
SS: Yes, agree
SS: Would also like to check some subset of commonly used waveforms. eg aligned spin, … No reason to expect trouble with this, but would nice to have it shown for a couple of events.
GA: Already have set of aligned waveforms for 150914 (https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/waveforms_project/GW150914_calibration/comparison/html/Publication.html). This is done with IMRPhenomD, IMRPhenomC, SEOBNRv4_ROM at the moment, and looks pretty good. Is this sufficient?
SS: This looks good. PhenomD and SEOBNR are the workhorses, so they seem sufficient.
GA: Will put these pages into the review so they can be signed off.
GA: Started GW170817 run yesterday. CH was assigned this, but has recently been assigned an NSBH. Just need to find comparisons.
CH: There’s a git repo for all samples for this event; https://git.ligo.org/pe_event_samples/GW170817
GA: Summary of discussion. Reviewers are happy to have seen ~3-4 events with things agreeing. Happy to take it to next stage once we’ve got GW170817 and calibration sorted. Is that a good summary?
MP: Yes, good summary.
SS: +1

Action: GA looking at 170817. Reviewee’s will then write a review statement and get that signed of

Calibration:

GA: Summary of where we’re at: LI and Bilby both return priors, but priors are slightly different. Action was on Sylvia, Carl, Salvo, to understand this. Sylvia has looked briefly, but is busy and now on holiday. What is the path forward? If the events agree, what is the status of calibration?
MP: Fairly in agreement. Calibration is included in known events, so this suggests calibration is either unimportant or doing the same thing between the two codes.
MP: Did the zero-noise realisations have calibration uncertainty.
GA: No
MP: Are there any tests that compare LI and Bilby without calibration uncertainties, then MP would be happy with that and the current comparison of known events. Hopefully would be quick to run.
GA: Easy enough to do, except not 100% sure how to turn off calibration for LI.
GA: Going to put out a call for volunteers to do this.

AOB: None.

Please register or sign in to add a comment.