Fast Simulation of a High Granularity Calorimeter by Generative Adversarial Networks

TL;DR

This paper introduces 3DGAN, a generative adversarial network that efficiently simulates high granularity calorimeter data with high accuracy, transfer learning capabilities, and practical application in physics analysis, achieving significant speedups over traditional methods.

Contribution

The paper demonstrates the successful application of GANs for 3D calorimeter data simulation, including transfer learning across particle types and energies, and validates the generated data's utility in physics analysis.

Findings

GAN-generated showers are within 10% accuracy of Monte Carlo data.

The method achieves three orders of magnitude speedup in data generation.

Transfer learning enables simulation across multiple particle types and energy ranges.

Abstract

We present the 3DGAN for the simulation of a future high granularity calorimeter output as three-dimensional images. We prove the efficacy of Generative Adversarial Networks (GANs) for generating scientific data while retaining a high level of accuracy for diverse metrics across a large range of input variables. We demonstrate a successful application of the transfer learning concept: we train the network to simulate showers for electrons from a reduced range of primary energies, we then train further for a five times larger range (the model could not train for the larger range directly). The same concept is extended to generate showers for other particles (photons and neutral pions) depositing most of their energies in electromagnetic interactions. In addition, the generation of charged pion showers is also explored, a more accurate effort would require additional data from other detectors not included in the scope of the current work. Our further contribution is a demonstration of using GAN-generated data for a practical application. We train a third-party network using GAN-generated data and prove that the response is similar to a network trained with data from the Monte Carlo simulation. The showers generated by GAN present accuracy within $10\%$ of Monte Carlo for a diverse range of physics features, with three orders of magnitude speedup. The speedup for both the training and inference can be further enhanced by distributed training.