Speaker
Description
In high-energy physics experiments, calorimeter simulation involves a wealth of physical interaction processes, and the resulting particle showers encapsulate critical physical information. However, calorimeter simulation also incurs enormous computational resource overhead. Previous studies have addressed the growing computational pressure through machine learning, while quantum machine learning offers potential advantages over classical machine learning in simulation applications. This study explores a fast simulation scheme based on quantum generative adversarial networks (Quantum GAN), focusing on a one-dimensional 8-pixel energy deposition distribution. The scheme adopts a hybrid quantum-classical architecture, employing a parameterized quantum circuit as the generator and a classical neural network as the discriminator for adversarial training. Experiments on an ideal simulator demonstrate that the quantum generator can faithfully reproduce the pixel-wise energy distribution, correlation matrix, and average energy distribution of Geant4 reference data. The model effectively approaches the true distribution within a small number of training epochs, validating the feasibility and effectiveness of this Quantum GAN approach for fast calorimeter simulation.
| 请选择分会 | 粒子物理实验技术 |
|---|
