Dear colleagues,
The Event List from GWOSC records the GPS time of each event, where GPS time is said to be " End time or merger time of the event in GPS seconds" in Event Portal Usage Notes. I wonder what is the definition of “end time or merger time”.
I also wonder how to choose which to be recorded.
Hi yidelee,
I am not sure about the answer to this question, but I do see someone use the following method:
Waiting for more answer.
Dear yidelee,
The GPS time given for the events on the GWOSC webpages is obtained from the search pipelines and is thus the merger time when the template used is a binary black hole waveform model that includes the merger-ringdown (such as SEOBNRv4[_ROM]), and is the end time when the template used is a post-Newtonian waveform (such as TaylorF2 or SpinTaylorT4), which is used in the low-mass portion of the parameter space. See, e.g., Appendix D of the GWTC-3 paper for the description of the searches (in that particular case, the configurations used for O3b).
The minimally modelled coherent WaveBurst pipeline also described in that appendix also detected some of the events given in GWOSC, notably the marginal event 200214_224526. I am not sure how exactly it determines the GPS time of a detection, but presume that the results from the templated searches are given precedence for the cases where they also detect an event. However, I can enquire further about this if you are particularly interested in cases like 200214_224526, where there are also no parameter estimation results.
The merger time provided by the templated searches is indeed the time of largest absolute value of the waveform, with the appropriate definition of that quantity. The official LVK convention is that the merger time should be the time of maximum of the sum of the squares of the absolute values of all the l = 2 spin-weighted spherical harmonic modes of the waveform. (This somewhat complicated definition is chosen to apply to models for binaries with spin precession, which are used for parameter estimation, though not yet for the searches the collaboration carries out.) However, some waveforms use different conventions that are close to the official convention. Regardless, the merger time will be close to the maximum of the sum of the squares of the two polarizations of the waveform.
The end time used for post-Newtonian waveforms is the time at which the frequency becomes (formally) infinite.
As far as I can tell, the specific waveform used to obtain a given GPS time from the searches is not given in GWOSC. However, the parameter estimation results will give posterior samples for the GPS time (called geocent_time) and say exactly which waveforms they use, so those should be used if you need a precise result and/or need to know whether you’re getting the merger time or end time for an event near the boundary between where the searches use binary black hole and post-Newtonian waveforms.
The waveforms used for parameter estimation of binary neutron stars, which do not include the merger phase, similarly would provide an end time. The neutron star-black hole waveforms generally give a time near the peak (so a “merger time”), but give a time well after the peak for some more extreme parameters.
I would point you to technical documents giving details about the conventions used in the various waveform models, but the details all seem just to be in internal LVK documentation.
Dear BarryG,
Thank you for your timely suggestion and sorry for my late reply. I’ve tried your method before posting this question, but I encountered some difficulties when calculating SNR. As a result, I was seeking for a procedure without localizing the waveform. However I appreciate your advice and I think your method can be somehow “self-consisting” in my research. Thanks again.
Dear nkjm,
Thank you very much for your extremely clear and smooth explanation. Sorry for my late reply.
The GPS time given for the events on the GWOSC webpages…
Thank you for providing me the link of pipelines. I can now understand that different pipelines result in specific localization of events.
The minimally modelled coherent WaveBurst pipeline…
Thank you for mentioning the special case while I would not consider it in my research. Thank you for your kindness.
The merger time provided by the templated searches…
This paragraph greatly solved my problem which had been bothering me for a long time. However I couldn’t find the definition “the merger time should be the time of maximum of the sum of the squares of the absolute values of all the l = 2 spin-weighted spherical harmonic modes of the waveform” in documents from GWOSC or relevant published paper. I would like to ask about the reference of this definition.
The end time used for post-Newtonian waveforms…
Thank you for separately mentioning the definitions.
As far as I can tell, the specific waveform…
I also couldn’t find the corresponding waveform in GWOSC, and I’ve overlooked parameter estimation because I was mainly focusing on waveform denoising. I found the parameter geocent_time in the python package bibly. Thanks you for providing the information.
The waveforms used for parameter estimation…
Thank you for also mentioning how we can obtain the end time.
I would point you to technical documents…
I would like to ask if you could provide me the documents for the definitions of merger time and end time.
Dear yidelee,
Apologies for not getting back to you sooner, but I was waiting to hear back from the waveforms chairs to see if there was a public reference for the official LVK merger time convention I could point you to. Unfortunately, there is not (though they hope to have things better documented and standardized in the future), nor is there anything public about the specific definitions used in the individual models, except for the code (all in LALSimulation; see also its documentation). You can also use LALSimulation to evaluate a given waveform for a specific binary–see this tutorial, which shows how to use the PyCBC interface to the LALSimulation waveforms. The merger or end time is at t = 0 in this case. I’m sorry that I can’t be more helpful here.