Specifying data
A crucial ingredient to any population analysis is the data. There are two main forms of data that are used in population analyses:
posterior samples for each of the observed events. We expect that these are provided in the
PESummaryformat.`simulated signals for quantifying the search sensitivity (sensitivity injections).
Posterior samples
These are specified via the --sample-regex argument which should be a dictionary
where the keys identify subsets of the events, e.g., grouped by observing run, and the values
are glob patterns for a set of result files.
Sometimes we wish to omit specific events from the analysis. This can be done by specifying
a list of events to omit via the --ignore option. Each entry in this list is used as
a glob pattern against the event file names.
Each PESummary result file often contains multiple results with different analysis choices.
To specify which result to use, you can provide a list with a precedence order of
--preferred-labels. The first matching label is used to extract the posterior samples.
If there are no matches inside the file, the first set of samples that contains the desired parameters
is used.
Finally, the run time of the analysis may scale with the number of posterior samples used for each event.
To limit the number of samples used, you can specify a maximum number of samples to use via the
--samples-per-posterior option.
Sensitivity injections
These must be provided in one or more hdf5 files via the --vt-file option.
If you have multiple files, you can specify a glob pattern to match all of the files.
These files are expected to match the format of sensitivity injections produced by
the LIGO-Virgo-Kagra collaboration.