A Graph Neural Network (GNN) based data-driven simulator to build surrogate models for lagrangian fluid simulations.
Based on https://doi.org/10.48550/arXiv.2002.09405 (https://github.com/deepmind/deepmind-research/tree/master/learning_to_simulate) and https://doi.org/10.1016/j.compgeo.2023.106015 (https://github.com/geoelements/gns/)