Synthetic data generation

The NA62 Monte Carlo simulations use BDSim for beam halo overlay. MC Validation, in order to study changes in the model and the impact in comparison to data, requires a lot of overlay statistics, especially since many beam halo particles are "lost" during reconstruction, selections, trigger cuts, etc. We are investigating the use of GANs to generate synthetic data samples based on an original BDSim sample, e.g. events using the BDSim output at a z plane (e.g. CEDAR z=69657 mm). Naively, this would be similar to a parametrisation in x, y, px, py, pz for mu-, mu+, and K+. Such a synthetic "extension" of the BDSim dataset with the use of GANs would allow one to effortlessly study multiple configurations of the model since one could quickly generate the necessary statistics without running full BDSim simulations.

Project parts

This work was started as a summer project (SP1), continued as a semester 1 student project (SP2) and currently the work is continued within the NA62 group.

  1. Project part 1, GANs for mu+, mu-: SP1
  2. Project part 2, GAN fine-tuning: SP2
  3. Various studies, inclusion of K+: P3; first synthetic dataset tests
  4. Current work: P4

Page last updated on March 10, 2022