Electrostatic field dataset - Latent Field Discovery in Interacting Dynamical Systems with Neural Fields

This repository contains the "Electrostatic field" dataset from the paper

Latent Field Discovery in Interacting Dynamical Systems with Neural Fields
Miltiadis Kofinas, Erik J Bekkers, Naveen Shankar Nagaraja, Efstratios Gavves
NeurIPS 2023
https://arxiv.org/abs/2310.20679
https://github.com/mkofinas/aether

It contains simulations of trajectories of 5 charged particles in 2 dimensions, interacting via Coulomb forces.

Particles move under the influence of 20 immovable and unknown sources, which are shared in the whole dataset.

There are 50,000 simulations for training, 10,000 for validation, and 10,000 for testing. Simulations last for 49 timesteps.

The features comprise positions and velocities of particles, while edges describe the product of pairwise charges. The dataset also contains the positions of the field sources, meant to be used for visualization.