Tutorial for Analyses Based on GWPopulation

A component of the data release for "The Population of Merging Compact Binaries Inferred Using Gravitational Waves Through GWTC-3"

Authors: Bruce Edelman, Amanda Farah, and Jacob Golomb on behalf of the LIGO Scientific Collaboration, Virgo Collaboration and KAGRA Collaboration

This software is provided under license: Creative Commons Attribution 4.0 International (https://creativecommons.org/licenses/by/4.0/legalcode)

This notebook is intented to show readers how to load in the data associated with some of the gwpopulation-based models used in "The population of merging compact binaries inferred using gravitational waves through GWTC-3" (herein "the companion paper"). Here, we will show you how to load in hyperposterior samples, plot probability distributions as a function of source parameters, and make less-polished versions of the plots in the paper. The models highlighed in this particular tutorial are:

These are all included in or built off of the open source python package, gwpopulation, which includes several other parameterized distributions as well, many of which are highlighted in "Population Properties of Compact Objects from the Second LIGO-Virgo Gravitational-Wave Transient Catalog." Note that other parameterized (MultiSource and NS mass models) and non-parametric (Binned Gaussian Process, Flexible Mixtures) models not based on gwpopulation exist and are used in the companion paper. There are separate tutorials for those.

This notebook will not show you how to run the population analysis. If you are intersted in this, please look at the documention for gwpopulation: https://colmtalbot.github.io/gwpopulation/

We will begin by reading in the results of the population inference on our default event list for the BBH population. In this example, we will read in the results for the Powerlaw + Peak model. These results are stored as bilby result files in this data repository. A description of the parameters used in this model can be found in Appendix B1 of the companion paper.

Reading in Hyperposterior Samples

There is a lot of information in the Result object. Lets take a look at some useful attributes.

What you are probably most intersted in are the hyperposterior samples, which are saved as a pandas dataframe.

We can pull 1D marginal distributions from this by picking out a specific column. As an example, lets try the powerlaw slope on m1

We can plot all of these samples with their many correlations using a corner plot: