This repository contains code to accompany the paper Importance nested sampling with normalising flows.
This repo contains git submodules, so it should be cloned using:
git clone --recurse-submodules git@github.com:mj-will/nessai-ins-paper.git
or equivalently with HTTPS URL.
The environment.yml and env-spec-file contain details of all the dependencies to construct a conda environment that contains the relevant Python dependencies. However, custom forks of nessai, bilby and bilby_pipe were used. These are all included in packages directory and can be installed with the following commands:
cd packages/<name>
pip install .
where <name> should be substituted for nessai, bilby and bilby_pipe.
The analysis is split into three directories:
toy_example: contains the code for the toy example in Section 3experiments: contains the analytical likelihood and Rosenbrock testsgw: contains all of the gravitational-wave tests
Each directory contains instructions for how to run the experiments.