Fast simulation of the ATLAS calorimeter system with Generative Adversarial Networks

The extensive physics program of the ATLAS experiment at the Large Hadron Collider needs large scale and high fidelity simulated samples which forwards the research and development of better fast simulation techniques. Building on the recent success of deep learning algorithms, Generative Adversarial Networks are exploited for modelling the response of the ATLAS detector calorimeter for different particle types. FastCaloGAN is a tool capable of simulating calorimeter showers for photons, electrons and pions over a range of energies (between 256~MeV and 4~TeV) in the full detector $\eta$ range. The properties of synthesised showers are compared to showers simulated from a full detector simulation with \textsc{GEANT4}. This study demonstrates the potential of the algorithm for fast calorimeter simulation for the ATLAS experiment in the future and opens the possibility to improve current simulation techniques.