Predicts 3D halo distributions from dark matter simulations using a physically motivated Wasserstein mapping network
The network architecture, training methodology and results are detailed in:
"Painting halos from 3D dark matter fields using Wasserstein mapping networks,"
Doogesh Kodi Ramanah, Tom Charnock, Guilhem Lavaux [arXiv:1903.10524]
- The notebook
wasserstein_halo_painting_network.ipynbcontains an in-depth and stepwise description of the network implementation and training;
- Please cite the above paper if you make use of our code;
network_learning_movie.mp4depicts the network predictions for a given thick slice of dark matter density field, as a function of weight updates, and provides a visualization of the network learning progress.