Subject: Spin convention change between GWTC-1 and O3a
...
...
@@ -31,3 +30,100 @@ Thanks a lot for your help!
Best regards,
Krishnendu, Saleem, Abhirup, Anuradha, Archisman.
From: Harald Pfeiffer <harald.pfeiffer@aei.mpg.de>
Date: Wed, 24 Jun 2020 at 13:41
Subject: Re: Spin convention change between GWTC-1 and O3a
To: Archisman Ghosh <archisman.ghosh@lorentz.leidenuniv.nl>
Cc: Riccardo Sturani <riccardo@iip.ufrn.br>, Frank Ohme <frank.ohme@aei.mpg.de>, Sebastian Khan <KhanS22@cardiff.ac.uk>, Maria Haney <mhaney@physik.uzh.ch>, Abhirup Ghosh <abhirup.ghosh@aei.mpg.de>, Anuradha Gupta <anuradha1bhu@gmail.com>, Saleem Muhammed <saleem.muhammed.c@gmail.com>, Krishnendu Naderi Varium <krnava@aei.mpg.de>
Hi Archisman,
indeed, there are a few differences, the most noticeable ones marked by a star:
* phi_{JL} is shifted by pi/2 The global max of cos(iota) is pushed to the opposite end phi_{12} seems to be shifted by pi theta_{JN} flips the global maximum from face-on to face-off* h1_cplx_snr_amp is different
The waveforms changed conventions of the frame in which the spin components are specified: the in-plane spin components get rotated by the orbital phase. See https://dcc.ligo.org/LIGO-T1800226 for a summary, in particular the bottom left box.
I can't imagine a mechanism how the spin conventions would give those differences. Specifically, I would expect that if a rotation of the in-plane spins is the problem, then \vec{J} would change by a small amount, and so angles would change by a small amount.
However, it almost seems like the new run changes the **preference of face-on vs. face-off**: That would simultaneously affect theta_JN, iota, and perhaps it could also flip phi_{12} through how its definition is tied to the line-of-sight. I don't know how pi_{JL} is defined, put perhaps it, too, shifts by pi/2 between face-on and face-off?
I can't even imagine why the snr should differ.
Hopefully somebody else has better ideas.
Cheers,Harald
From: Frank Ohme <frank.ohme@aei.mpg.de>
Date: Thu, 25 Jun 2020 at 12:19
Subject: Re: Spin convention change between GWTC-1 and O3a
To: Archisman Ghosh <archisman.ghosh@lorentz.leidenuniv.nl>, Harald Pfeiffer <harald.pfeiffer@aei.mpg.de>
Cc: Riccardo Sturani <riccardo@iip.ufrn.br>, Sebastian Khan <KhanS22@cardiff.ac.uk>, Maria Haney <mhaney@physik.uzh.ch>, Abhirup Ghosh <abhirup.ghosh@aei.mpg.de>, Anuradha Gupta <anuradha1bhu@gmail.com>, Saleem Muhammed <saleem.muhammed.c@gmail.com>, Krishnendu Naderi Varium <krnava@aei.mpg.de>
Hi All,
When PhenomPv1/2 were updated with the new interface, I remember that
(mainly) Alejandro Bohé and I looked closely at the definition of
polarization angle and angles between J and L, and found some
inconsistencies (based differences between the LAL triad and the "Arun
et al" triad [1]). Alejandro fixed those [2]. Among other things, it led
to an extra re-shuffeling of polarizations [3].
Since the differences you find manifest themselves in an apparent
opposite orientation of J, but also in a shift of polarization angle,
Subject: Re: Spin convention change between GWTC-1 and O3a
To: Riccardo Sturani <riccardosturani@gmail.com>, Frank Ohme <frank.ohme@aei.mpg.de>
Cc: Archisman Ghosh <archisman.ghosh@lorentz.leidenuniv.nl>, Harald Pfeiffer <harald.pfeiffer@aei.mpg.de>, Sebastian Khan <KhanS22@cardiff.ac.uk>, Maria Haney <mhaney@physik.uzh.ch>, Abhirup Ghosh <abhirup.ghosh@aei.mpg.de>, Anuradha Gupta <anuradha1bhu@gmail.com>, Saleem Muhammed <saleem.muhammed.c@gmail.com>
Dear Riccardo, All,
Thank you so much for all your comments. We have a much better understanding of the issue now. On June 25, 2020 at 3:30 PM Riccardo Sturani <riccardosturani@gmail.com> wrote:
Hi All,
there has been indeed a convention change, but I need to know more to
be able to help.What are the lalsuite git hash of the "old" and "new" runs?The 'new' runs here refer to the BBH runs using a LALSuite branch which includes changes coming from spin-induced quadrupole moments. On the other hand, the 'old' runs are the ones from the PE sample release repo: https://git.ligo.org/pe/lvc_pe_sample_release/-/tree/master/
Our branch is rebased to a released version of the master with the following git hash: cde08c79 [https://git.ligo.org/lscsoft/lalsuite/-/tags/lalsuite-v6.67]
I am not sure about the LALSuite git hash used for the runs listed in the PE release repo. From the PE page (here https://www.lsc-group.phys.uwm.edu/ligovirgo/cbcnote/ParameterEstimationModelSelection/O2_PE/O2/1186741861p5268#C02_cleaned_runs), I think this was the reviewed version of lalinference_o2 branch which is used for these runs.
