Skip to content

Changes

Page history
Update PESummary Review authored by Vivien Raymond's avatar Vivien Raymond
Show whitespace changes
Inline Side-by-side
PESummary-Review.md
View page @ 74196f85
......@@ -39,9 +39,9 @@
# Review Telecons
# Final Review Statement
PESummary is now reviewed. We have run our tests on 0.3.2 version, git hash: [4f8b36c3](https://git.ligo.org/lscsoft/pesummary/-/tree/4f8b36c3896bfbe2512a156b1f5bda8f34f23ff7) .
PESummary is now reviewed in its standard configuration. We have run our tests on git hash: [4f8b36c3](https://git.ligo.org/lscsoft/pesummary/-/tree/4f8b36c3896bfbe2512a156b1f5bda8f34f23ff7) .
PESummary is able to post process posterior samples using the correct
parameter conversions and the correct NR fits, in agreement with `cbcBayesProc` [1]. PESummary ’s skymaps are also in agreement with `Ligo.skymap` [2] .
parameter conversions, in agreement with `cbcBayesProc` [1]. PESummary ’s skymaps are also in agreement with `Ligo.skymap` [2].
The metafile in which PESummary stores the processed data, does not alter the input data [3]. Where downsampling applies, this does not introduce any difference in the probability densities recover.
We have also noticed that the metafile is correctly downloadable from the web and can be handled by any of the most common packages (pandas, numpy, etc.) [4].
......@@ -54,7 +54,12 @@ https://git.ligo.org/lscsoft/pesummary/-/wikis/PESummary-Review .
If people want to see any particular additional point to review, we are happy
to address it with a follow up update.
We stress again, that at the current status PESummary is now reviewed and we confirm it can correctly process posterior samples to provide 1D and 2D plots as well as credible intervals required for extrinsic and intrinsic parameters.
We stress again, that at the current status PESummary is now reviewed and we confirm it can correctly process posterior samples to provide 1D and 2D plots as well as credible intervals required for extrinsic and intrinsic parameters. It passes all review tests covering cbcBayesPostProc's usage, and as such should be used as a replacement for post-processing of PE samples and sample releases.
While not preventing the review and usage, there are some additional extended features that need to be considered in future reviews:
* remnant fits, review on-going
* Calibration Marginalisation plots: to be reviewed alongside the codes doing the calibration marginalisation

Thank you to everyone involved!
[1] https://git.ligo.org/lscsoft/pesummary/-/wikis/SummaryReview-script
......