cogwheel.postprocessing.RundirPostprocessor¶

class cogwheel.postprocessing.RundirPostprocessor(rundir, relative_binning_boost: int = 4)¶

Bases: object

Postprocess posterior samples from a single run.

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

Methods

`compute_lnl_aux`	Add columns self._lnl_aux_cols to self.samples with log likelihood computed by detector, at high relative binning resolution, with no ASD-drift correction applied.
`get_lnl_aux_cols`	Return names of auxiliary log likelihood columns.
`process_samples`	Call the various methods of the class sequentially, then save the results.
`save_tests_and_samples`	Save `.tests` and `.samples` in `.rundir`.
`test_asd_drift`	Compute typical and worse-case log likelihood differences arising from the choice of somewhat-parameter-dependent asd_drift correction.
`test_relative_binning`	Compute typical and worst-case errors in log likelihood due to relative binning.

Attributes

LNL_COL

compute_lnl_aux()¶: Add columns self._lnl_aux_cols to self.samples with log likelihood computed by detector, at high relative binning resolution, with no ASD-drift correction applied.

static get_lnl_aux_cols(detector_names)¶: Return names of auxiliary log likelihood columns.

process_samples()¶

Call the various methods of the class sequentially, then save the results. This computes:

Columns for standard parameters
Column for log likelihood
Auxiliary columns for log likelihood (by detector, at high relative binning resolution and with no ASD-drift correction applied)
Tests for log likelihood differences arising from reference waveform choice for setting ASD-drift
Tests for log likelihood differences arising from relative binning accuracy.

save_tests_and_samples()¶: Save .tests and .samples in .rundir.

test_asd_drift()¶

Compute typical and worse-case log likelihood differences arising from the choice of somewhat-parameter-dependent asd_drift correction.

Store in .tests['asd_drift'].

test_relative_binning()¶

Compute typical and worst-case errors in log likelihood due to relative binning.

Store in .tests['relative_binning']. If the samples are weighted, the weights are considered in the standard deviation of the errors but ignored in the maximum.