Graph Generative Adversarial Networks for Sparse Data Generation in High Energy Physics
This work addresses the problem of generating sparse high-energy physics data for physicists, which is crucial for simulations and detector development.
This paper develops a graph generative adversarial network (GGAN) to generate sparse data, demonstrating its effectiveness on sparse MNIST digits and particle jets from simulated LHC collisions. The model successfully generates data that agrees with real data, quantified by graph-based Fréchet Inception distance for MNIST and 1-Wasserstein distance for particle and jet features.
We develop a graph generative adversarial network to generate sparse data sets like those produced at the CERN Large Hadron Collider (LHC). We demonstrate this approach by training on and generating sparse representations of MNIST handwritten digit images and jets of particles in proton-proton collisions like those at the LHC. We find the model successfully generates sparse MNIST digits and particle jet data. We quantify agreement between real and generated data with a graph-based Fréchet Inception distance, and the particle and jet feature-level 1-Wasserstein distance for the MNIST and jet datasets respectively.