Author: Aldana Grichener, agrichener@arizona.edu

The data underlying this project was generated using Modules for Experiments in Stellar Astrophysics (https://github.com/MESAHub/mesa), 
bbq (https://github.com/rjfarmer/bbq) and PyTorch Lightning (https://github.com/Lightning-AI/pytorch-lightning).

To use this data, cite the manuscript "Nuclear Neural Networks: Emulating Late Burning Stages in Supernova Progenitors" by Grichener A., Renzo, M., Kerzendorf W., et al. 2025.  
 
Detailed documentation is provided in a guide found in this link: https://sites.google.com/view/aldanagrichener/nuclear-neural-networks?authuser=0 

Subfolders:

Mesa_models: contains two mesa runs of a 20Mo stars with large nuclear reaction networks and two other publicly available supernova progenitors. 
The stellar models were used to determine the parameter space for training the NNNs (section 2)

training_sets: contains the datasets on which the NNNs were trained (section 2).

test_datasets: contains the bbq outputs used to evaluate the NNNs performance (section 3).

trained_NNN_models: contains all NNN trained models produced in this study.

python_scripts_for_analysis: includes the python script used to generate the training sets, train the NNNs and evaluate their performance. 
We also provide the scripts used to create the figures. 

The complete bbq runs we generated for this project include additional timesteps for training the NNNs and datasets for extended density ranges, comprising a total of 4TB, 
and will be shared upon request from the corresponding author.