|
|
|
**Agenda**
|
|
|
|
# Agenda
|
|
|
|
|
|
|
|
The pp-tests and fiducial injections are all passing; the dev team is happy with how they look. That said, we have just released Bilby-v0.5.4, and are currently re-running all pp-tests and fiducial injections. There are no substantial changes to the code in this version, so we expect these to pass. We will look at these in detail at the next review call (August 7). The agenda today is therefore to look at the known-event comparisons. Again, this is with a slightly old version of the code; we will re-run these with 0.5.4 this week, but we do not expect anything to change.
|
|
|
|
|
|
|
|
* GW150914: [pesummary page](https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/home.html) |
|
|
\ No newline at end of file |
|
|
|
* GW150914: [pesummary page](https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/home.html)
|
|
|
|
|
|
|
|
|
|
|
|
## Minutes
|
|
|
|
|
|
|
|
Greg apologies for not keeping better minutes as we went. Here are my "recalled" minutes so I may have missed somethings.
|
|
|
|
|
|
|
|
Greg: Bilby 0.5.4 released, reruns ongoing and results to be shared later in the week (**ACTION ITEM**: GA to share results)
|
|
|
|
Greg: Charlie to give a walk through of the latest GW150914 comparison (using <0.5.4 bilby, rerun ongoing)
|
|
|
|
Charlie: In general things look good, issues in the waveform plotting for L.I. Publication plots look great.
|
|
|
|
|
|
|
|
Minor issues noticed as we went through the 1d histograms:
|
|
|
|
* Simon notes that KS tests and JS divergences are not "perfect", but quantify the "by eye" agreement
|
|
|
|
* JS Divergence agreed on as a better statistic
|
|
|
|
* Issues noted in spin angles, could be due to the waveforms interface
|
|
|
|
* Known issue in phi_jl due to the waveform interface
|
|
|
|
* Network MF SNR strongly disagrees: likely a bug in pesummary (**ACTION ITEM**: CH to fix)
|
|
|
|
* Sylvia notes discussion of evidences (fixed on master and in rerun)
|
|
|
|
* Spin magnitudes show disagreement
|
|
|
|
* MP wonders if this is due to bugs in the nest/MCMC L.I.
|
|
|
|
* **ACTION ITEM ** CH and SB to dig up separate nest and MCMC runs so comparisons can be made against these
|
|
|
|
* GA: if the disagreement is still there, the next candidate is the waveforms interface
|
|
|
|
|
|
|
|
Moritz: 15 D Gaussian a combined run with 100+ individual runs has finished and looks equal demonstraing the difference in heights between the two peaks is not a systematic issue, but a convergence issue
|
|
|
|
** ACTION ITEM ** Moritz to clean up the wiki page and MP to review
|
|
|
|
|
|
|
|
|
|
|
|
## Chat:
|
|
|
|
|
|
|
|
> <08:02:09> "Greg Ashton [h]": https://git.ligo.org/lscsoft/bilby_pipe/wikis/O3-review/minutes/190730
|
|
|
|
> <08:03:09> "Matt Pitkin [h]": sounds good
|
|
|
|
> <08:03:14> "Simon Stevenson": ypu
|
|
|
|
> <08:03:43> "Greg Ashton [h]": https://git.ligo.org/lscsoft/bilby_pipe/wikis/O3-review/known-events
|
|
|
|
> <08:04:13> "Greg Ashton [h]": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/home.html
|
|
|
|
> <08:04:50> "Charlie Hoy": In the version tab, this is just the version of pesummary I am afraid
|
|
|
|
> <08:05:57> "Charlie Hoy": I can go through them if you would like?
|
|
|
|
> <08:07:46> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/Publication.html
|
|
|
|
> <08:09:02> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/Comparison.html
|
|
|
|
> <08:09:32> "Greg Ashton [h]": Hand up
|
|
|
|
> <08:10:01> "Simon Stevenson": Yeah this is fine
|
|
|
|
> <08:10:14> "Greg Ashton [h]": Sorry for the interuption!
|
|
|
|
> <08:10:47> "Matt Pitkin [h]": hand up
|
|
|
|
> <08:10:57> "Simon Stevenson": BTW low KS = bad not good
|
|
|
|
> <08:12:46> "Greg Ashton [h]": Hand up on these statistics
|
|
|
|
> <08:13:10> "Isobel Romero-Shaw": By the way, it doesn't look like these runs were with the phi_jl fix https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/Comparison_phi_jl.html
|
|
|
|
> <08:13:35> "Charlie Hoy": yes I think so
|
|
|
|
> <08:13:43> "Simon Stevenson": Yeah
|
|
|
|
> <08:14:48> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/Comparison_phi_2.html
|
|
|
|
> <08:15:42> "Greg Ashton [h]": Hand up
|
|
|
|
> <08:16:05> "Charlie Hoy": Colm sent around an O3 comparison
|
|
|
|
> <08:16:12> "Greg Ashton [h]": Ah yes
|
|
|
|
> <08:16:24> "Charlie Hoy": It could be...
|
|
|
|
> <08:16:31> "Charlie Hoy": Most probably becaus eof phi_jl
|
|
|
|
> <08:16:50> "Greg Ashton [h]": https://ldas-jobs.ligo.caltech.edu/~colm.talbot/bilby_review/0.5.3/190521r/EXP0/comparison/home.html
|
|
|
|
> <08:17:09> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~colm.talbot/bilby_review/0.5.3/190521r/EXP0/comparison/html/Comparison_phi_jl.html
|
|
|
|
> <08:18:18> "Sylvia Biscoveanu": Hand up
|
|
|
|
> <08:21:03> "Charlie Hoy": hand up
|
|
|
|
> <08:21:21> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/Comparison_network_matched_filter_snr.html
|
|
|
|
> <08:22:09> "Charlie Hoy": I will take a look in PESummary
|
|
|
|
> <08:22:12> "Charlie Hoy": it is probably my fault!
|
|
|
|
> <08:22:22> "Sylvia Biscoveanu": Are the individual IFOs ok?
|
|
|
|
> <08:22:36> "Charlie Hoy": optimal snrs are perfect
|
|
|
|
> <08:23:13> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/bilby_bilby_network_matched_filter_snr.html
|
|
|
|
> <08:23:37> "Charlie Hoy": hand up
|
|
|
|
> <08:24:34> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/bilby_bilby_H1_matched_filter_abs_snr.html
|
|
|
|
> <08:24:43> "Charlie Hoy": https://ldas-jobs.ligo.caltech.edu/~cbc/pe/bilby_lalinference_comparison/GW150914/html/bilby_bilby_L1_matched_filter_abs_snr.html
|
|
|
|
> <08:25:11> "Simon Stevenson": chi_p, a1 and a2 are a bit different. This may again may be related to waveform interface
|
|
|
|
> <08:26:47> "Sylvia Biscoveanu": Mass ratio is also slightly different
|
|
|
|
> <08:28:00> "Greg Ashton [h]": https://ldas-jobs.ligo.caltech.edu/~colm.talbot/bilby_review/0.5.3/190521r/EXP0/comparison/html/Comparison_a_1.html
|
|
|
|
> <08:28:28> "Simon Stevenson": Can we do comparisons against LALInference MCMC and NEST posteriors seperately?
|
|
|
|
> <08:29:05> "Charlie Hoy": I can run some LALInference runs is you need volunteers
|
|
|
|
> <08:29:42> "Sylvia Biscoveanu": I can volunteer to do that for next time
|
|
|
|
> <08:30:28> "Simon Stevenson": Yeah just for this one is fine. And those runs should already exist (I hope)
|
|
|
|
> <08:30:52> "Charlie Hoy": Ed F-J ran the O2 re-run so I can see if he still has the data
|
|
|
|
> <08:30:53> "Charlie Hoy": tomorrow
|
|
|
|
> <08:30:54> "Sylvia Biscoveanu": They do exist, and I'm happy to track them down without rerunning
|
|
|
|
> <08:31:00> "Sylvia Biscoveanu": I really don't think we need to rerun
|
|
|
|
> <08:31:10> "Charlie Hoy": I think Ed has the data on ARCCA/CIT
|
|
|
|
> <08:33:29> "Charlie Hoy": Ed has sent me the links to the mccm and nest runs for 150914
|
|
|
|
> <08:33:34> "Charlie Hoy": so I can make the plots tomorrow
|
|
|
|
> <08:33:40> "Matt Pitkin [h]": Excellent
|
|
|
|
> <08:33:42> "Sylvia Biscoveanu": Awesome
|
|
|
|
> <08:33:47> "Charlie Hoy": sure no problem
|
|
|
|
> <08:33:55> "Moritz Hübner": 15d gaussian
|
|
|
|
> <08:34:49> "Matt Pitkin [h]": Cool, thanks
|
|
|
|
> <08:35:19> "Moritz Hübner": sure
|
|
|
|
> <08:36:17> "Sylvia Biscoveanu": Thanks, bye!
|
|
|
|
> <08:36:17> "Matt Pitkin [h]": Thank, bye
|
|
|
|
> <08:36:18> "Charlie Hoy": bye> "Matt Pitkin [h]": Excellent
|
|
|
|
> <08:33:42> "Sylvia Biscoveanu": Awesome
|
|
|
|
> <08:33:47> "Charlie Hoy": sure no problem
|
|
|
|
> <08:33:55> "Moritz Hübner": 15d gaussian
|
|
|
|
> <08:34:49> "Matt Pitkin [h]": Cool, thanks
|
|
|
|
> <08:35:19> "Moritz Hübner": sure
|
|
|
|
> <08:36:17> "Sylvia Biscoveanu": Thanks, bye!
|
|
|
|
> <08:36:17> "Matt Pitkin [h]": Thank, bye
|
|
|
|
> <08:36:18> "Charlie Hoy": bye |
|
|
\ No newline at end of file |
