plot_posterior_corner

The plot_posterior_corner executable generates “corner plots” (also known as triangle plots) to visualize the posterior probability distributions of intrinsic parameters.

Overview

This tool takes posterior samples (typically produced by util_ConstructIntrinsicPosterior_GenericCoordinates) or intermediate grid data (composite files) and creates multi-panel figures showing: 1. 1D Marginal Distributions: Histograms/KDEs on the diagonal. 2. 2D Joint Distributions: Contour plots or scatter plots on the off-diagonal.

It is used to assess parameter correlations and the constraints placed on the physical properties of the gravitational wave source.

Usage

Basic Command Line

plot_posterior_corner.py [options]

Primary Options

Input Data

  • --posterior-file: Filename of the posterior samples *.dat file. Can be specified multiple times to compare different runs.

  • --posterior-label: Label for each corresponding posterior file.

  • --composite-file: Filename of an intermediate grid file (composite).

  • --truth-file: File containing the true parameters to be plotted as reference lines.

Parameter Selection

  • --parameter: List of parameters to include in the plot (e.g., mc, eta, lambda1).

  • --parameter-log-scale: Parameters to be plotted on a $log_{10}$ scale.

  • --change-parameter-label: Custom label for a parameter (format name=string).

Plotting Controls

  • --use-legend: Display a legend for multiple posterior files.

  • --use-title: Add a title to the figure.

  • --pdf: Export the plot to PDF instead of PNG.

  • --publication: Use specific figure sizes optimized for journals (e.g., PRD).

  • --use-smooth-1d: Enable smoothed 1D distributions.

  • --plot-1d-extra: Generate separate 1D PDF and CDF plots for each parameter.

Bounds and Cuts

  • --bind-param & --param-bound: Impose specific bounds on parameters for the plot.

  • --lnL-cut: Cut points with log-likelihood below a certain threshold.

  • --sigma-cut: Eliminate points with high fit error from the plot.

Functional Logic

  1. Data Loading: Loads the specified posterior or composite files. It handles various formats, including tidal and eccentricity-based files.

  2. Coordinate Transformation:

    • Expands samples to include derived quantities (e.g., $chi_{eff}$ from component spins).

    • Applies periodicity (e.g., $psi pmod pi$).

    • Handles log-scaling if requested.

  3. Range Determination: Automatically calculates the plotting range for each parameter based on the sample distribution or predefined special_param_ranges.

  4. Corner Plot Generation:

    • Uses the corner library to generate the multi-panel figure.

    • For multiple files, it iterates through the lists, assigning colors and linestyles.

    • If composite files are used, it can color-code the points by their log-likelihood.

  5. Reference Lines: If a truth file is provided, it draws vertical/horizontal lines at the true parameter values.

  6. Export: Saves the resulting figure to a file named corner_<params>.png (or .pdf).

Output Details

The output is a high-resolution image file. The diagonal panels show the 1D marginalized posterior for each parameter, and the off-diagonal panels show the 2D joint posterior for pairs of parameters.