ml-gw-pop

This repo contains material for the workshop Scientific Machine Learning for Gravitational Wave Astronomy, 2-6 June 2025.

The slides.pdf introduce three applications of machine learning for population inference of gravitational-wave sources.

There is a Python notebook for each of these applications. You can run with Google Colab (and GPUs!) with the following buttons for each notebook:

• flow-variational-inference.ipynb
• neural-posterior-estimation.ipynb
• simulation-based-prior.ipynb

This will require installing packages and downloading data first.

Alternatively, you can clone the repo and run locally or on a cluster - it is entirely self contained.

This material is focused on normalizing flows, which use neural networks for statistical inference. Below is a bonus notebook for Hamiltonian Monte Carlo:

• hamiltonian-monte-carlo.ipynb