... | ... | @@ -32,17 +32,7 @@ The comparison plots and numbers are produced using scripts [here](https://git.l |
|
|
|
|
|
### PESummary input
|
|
|
|
|
|
We also test that the user can provide PESummary metafiles as input. Below is an example where we recreate the S190814bv example above. First we consider the case where the inspiral and postinspiral analyses are located in different PESummary metafiles. We used the following command line:
|
|
|
|
|
|
```bash
|
|
|
summarytgr --webdir /home/charlie.hoy/public_html/projects/pesummary_review/summarytgr_review/metafile_input \
|
|
|
--samples /home/apratim.ganguly/public_html/O3/S190814bv/C01/20200609_IMRPhenomPv3HM/powerlaw/inspiral/samples/posterior_samples.h5 /home/apratim.ganguly/public_html/O3/S190814bv/C01/20200609_IMRPhenomPv3HM/powerlaw/post-inspiral/samples/posterior_samples.h5 \
|
|
|
--labels Inspiral:inspiral Merger-Ringdown:postinspiral \
|
|
|
--disable_pe_page_generation \
|
|
|
--imrct_kwargs N_bins:401 final_mass_deviation_lim:2 multi_process:16 --test imrct
|
|
|
```
|
|
|
|
|
|
The output can be seen [here](https://ldas-jobs.ligo.caltech.edu/~charlie.hoy/projects/pesummary_review/summarytgr_review/metafile_input/home.html).
|
|
|
We also test that the user can provide PESummary metafiles as input. Below is an example where we recreate the S190814bv example above. First we consider the case where the inspiral and postinspiral analyses are located in different PESummary metafiles, giving [this output](https://ldas-jobs.ligo.caltech.edu/~charlie.hoy/projects/pesummary_review/summarytgr_review/metafile_input/home.html).
|
|
|
|
|
|
Next we consider the case where the inspiral and postinspiral analyses are located in the same metafile. We used the following command line to combine the two files previously used:
|
|
|
|
... | ... | @@ -52,19 +42,22 @@ summarycombine --webdir ./combined \ |
|
|
--gw
|
|
|
```
|
|
|
|
|
|
We then run the following `summarytgr` command line:
|
|
|
We then obtain [this output](https://ldas-jobs.ligo.caltech.edu/~charlie.hoy/projects/pesummary_review/summarytgr_review/single_metafile_input/home.html).
|
|
|
|
|
|
```bash
|
|
|
summarytgr --webdir /home/charlie.hoy/public_html/projects/pesummary_review/summarytgr_review/single_metafile_input \
|
|
|
--samples /home/charlie.hoy/projects/pesummary_review/summarytgr_review/metafile_input/combined/samples/posterior_samples.h5 \
|
|
|
--labels Inspiral:inspiral Merger-Ringdown:postinspiral \
|
|
|
--disable_pe_page_generation \
|
|
|
--imrct_kwargs N_bins:401 final_mass_deviation_lim:2 multi_process:16 \
|
|
|
--test imrct
|
|
|
```
|
|
|
### Review statement
|
|
|
|
|
|
The output can be seen [here](https://ldas-jobs.ligo.caltech.edu/~charlie.hoy/projects/pesummary_review/summarytgr_review/single_metafile_input/home.html).
|
|
|
I have reviewed the new `summarytgr` code and the implementation of the IMR consistency test in it, checking all parts that will be used for the application of the IMR consistency test in the O3b TGR paper, and sign off on it for that application with Git hash [7ec5b734191feb40325fda244da017ac641eb38d](https://git.ligo.org/lscsoft/pesummary/-/commit/7ec5b734191feb40325fda244da017ac641eb38d). Specifically, we have checked that the new implementation reproduces the results obtained with the [LAL implementation](https://git.ligo.org/lscsoft/lalsuite/-/blob/master/lalinference/bin/imrtgr_imr_consistency_test.py) exactly in a number of different cases, including different settings for the parallelization and when reading in LALInference and parallel Bilby results, as illustrated above.
|
|
|
|
|
|
### Review statement
|
|
|
The new implementation outputs the 2d posterior on the deviation quantities just as the LAL implementation did, though in a single HDF5 file that also contains the GR quantile and other information, including the input samples, rather than separate files. However, it does *not* output the marginalized 1d posteriors on the final mass and final spin deviation, since these are trivial to recompute from the 2d posterior—see the instructions [here](https://docs.ligo.org/lscsoft/pesummary/unstable_docs/gw/tgr_file.html#imrct-data).
|
|
|
|
|
|
There are some options for this implementation of the IMR consistency test that have not been reviewed, viz., `use_kde = True`, `vectorize = True`, and anything other than the default of `interp_method = scipy.interpolate.interp2d`. None of these are anticipated to be used for the O3b TGR paper.
|
|
|
|
|
|
Additionally, there is a [bug in `summarycombine`](https://git.ligo.org/lscsoft/pesummary/-/issues/264) that causes the config data not to be copied over to the new file so, e.g., the cutoff frequencies cannot be read in from the file. This will be corrected in a separate merge request, but is not expected to impact the application for the O3b TGR paper.
|
|
|
|
|
|
Here are a few notes for users of the code:
|
|
|
|
|
|
* The code does not check that the `approximant` names and cutoff frequencies passed on the command line correspond to the settings used for the PE, so be sure to check this carefully for documentation purposes if using these options. It is better to avoid using those options if at all possible, which should be in most use cases, where the code will automatically read in these settings from the files passed to it (at least for Bilby results for the approximant).
|
|
|
|
|
|
* The IMR consistency test will be applied to the evolved spin samples if these are present in the PE samples passed to `summarytgr`, while the standard application of the test (e.g., in all LVC TGR papers to date) does not use the spin evolution.
|
|
|
|
|
|
TBA |
|
|
\ No newline at end of file |
|
|
* If one uses an even value for `N_bins` in this code and the LAL implementation, then one will not obtain exactly the same result for the GR quantile, since the LAL implementation computes this using the nearest grid point to the origin and with an even value for `N_bins` the origin is not on the grid, while PESummary always uses the origin. |
|
|
\ No newline at end of file |