cogwheel.postprocessing.EventdirPostprocessor

class cogwheel.postprocessing.EventdirPostprocessor(eventdir, reference_rundir=None, tolerance_params=None)

Bases: object

Class to gather information from multiple runs of an event and exporting summary to pdf file.

The method .postprocess_eventdir executes all the functionality of the class. It is suggested to use the top-level function postprocess_eventdir for simple usage.

Parameters:

eventdiros.PathLike

Path to directory containing rundirs.

reference_rundiros.PathLike, optional

Path to reference run directory. Defaults to the first (by name) rundir in eventdir.

tolerance_paramsdict

Items to update defaults from DEFAULT_TOLERANCE_PARAMS. Values higher than their tolerance are highlighted in the table. Keys include:

• ‘asd_drift_dlnl_std’

  Tolerable standard deviation of log likelihood fluctuations due to choice of reference waveform for ASD-drift.

• ‘asd_drift_dlnl_max’

  Tolerable maximum log likelihood fluctuation due to choice of reference waveform for ASD-drift.

• ‘lnl_max_exceeds_lnl_0’

  Tolerable amount by which the log likelihood of the best sample may exceed that of the reference waveform.

• ‘lnl_0_exceeds_lnl_max’

  Tolerable amount by which the log likelihood of the reference waveform may exceed that of the best sample.

• ‘relative_binning_dlnl_std’

  Tolerable standard deviation of log likelihood fluctuations due to the relative binning approximation.

• ‘relative_binning_dlnl_max’

  Tolerable maximum log likelihood fluctuation due to the relative binning approximation.

Methods

get_rundirs	Return a list of rundirs in self.eventdir for which sampling has completed.
make_table	Return a pandas DataFrame with a table that summarizes the different runs in rundirs.
postprocess_eventdir	Make diagnostics plots aggregating multiple runs of an event and save them to pdf format in {eventdir}/{DIAGNOSTICS_FILENAME}.

Attributes

DEFAULT_TOLERANCE_PARAMS
DIAGNOSTICS_FILENAME

get_rundirs()

Return a list of rundirs in self.eventdir for which sampling has completed. Ignores incomplete runs, printing a warning.

make_table(rundirs=None)

Return a pandas DataFrame with a table that summarizes the different runs in rundirs. The columns report differences in the samplers’ run_kwargs, plus the runtime and number of samples of each run.

Parameters:

rundirssequence of pathlib.Path

Run directories.

postprocess_eventdir(outfile=None)

Make diagnostics plots aggregating multiple runs of an event and save them to pdf format in {eventdir}/{DIAGNOSTICS_FILENAME}. These include a summary table of the parameters of the runs, number of samples vs time to completion, and corner plots comparing each run to a reference one.