sage.plotting.learning_parameter_prior

Filename : learning_parameter_prior.py Description : Short description of the file

Created on 2026-03-21 17:34:06

__author__ = Narenraju Nagarajan __copyright__ = Copyright 2026, ProjectName __license__ = MIT Licence __version__ = 0.0.1 __maintainer__ = Narenraju Nagarajan __affiliation__ = N/A __email__ = N/A __status__ = [‘inProgress’, ‘Archived’, ‘inUsage’, ‘Debugging’]

GitHub Repository: NULL

Documentation: NULL

Functions

plot_learning_parameter_prior(epoch, source_params, ...)

Overlay learned detection efficiency on the source-parameter prior distribution.

Module Contents

plot_learning_parameter_prior(epoch, source_params, pred_stat, labels, export_dir=None, save=True, save_name='stat', bin_width=500, step=10)[source]

Overlay learned detection efficiency on the source-parameter prior distribution.

For each source parameter, shows the prior histogram alongside the detection fraction curve — highlighting where the prior is dense versus where the network actually achieves high detection efficiency. Useful for diagnosing whether efficiency follows the prior or is biased toward high-probability regions.

Parameters:

  • epoch (int or str) – Epoch identifier used in the output subdirectory name.

  • source_params (dict[str, array-like]) – Per-signal parameter arrays.

  • pred_stat (array-like, shape (N_signal,)) – Predicted ranking statistic for signal events.

  • labels (array-like, shape (N,)) – Binary ground-truth labels for the full validation set.

  • export_dir (str or None) – Parent directory; plots saved under LEARN_PARAMS/epoch_{epoch}/.

  • save (bool) – If True, save to disk; otherwise display interactively.

  • save_name (str) – Prefix for saved filenames (default "stat").

  • bin_width (int) – Number of events per parameter bin (default 500).

  • step (float) – Step size for threshold sweep (default 10).