madness-deblender 1.0.0

Galaxy deblender from variational autoencoders

MADNESS Deblender

Maximum A posteriori with Deep NEural networks for Source Separation arXiv:2408.15236

This repository contains the code for the MADNESS project under development within the LSST Dark Energy Science Collaboration (LSST DESC). MADNESS obtains the MAP solution to deblend galaxies in a blended scene by performing gradient descent in a VAE latent space.

Installation

For testing the deblender, the package can be installed directly from GitHub

 pip install madness-deblender

For contributing (further instructions to be added soon):

git clone https://github.com/LSSTDESC/madness/
cd madness
pip install -e .[dev]

Contributors ✨

Thanks goes to these wonderful people (emoji key):

Biswajit Biswas 💻 🤔 ⚠️ 🚧

Junpeng Lao 💻 👀 🤔

Alexandre Boucaud 👀 🤔

Eric Aubourg 🤔

Cécile Roucelle 🤔

Axel Guinot 🤔

This project follows the all-contributors specification. Contributions of any kind welcome!

License

MIT