Generative machine learning for fast silicon detector simulation

Simulation of physics processes and detector response is not only a vital part of high energy physics research, but also represents a large fraction of computing cost. Generative machine learning is successfully complementing full (Geant4-based) simulation as part of fast simulation setups improving the performance compared to classical approaches. A lot of attention has been given to calorimeters being the slowest part of the full simulation, but their simulation speed becomes comparable to that of silicon semiconductor detectors when fast simulation is used. This makes silicon detectors the next candidate for optimisation, especially with the growing number of channels in future detectors. This work explores the use of transformer architectures for fast simulation of silicon tracking detectors. The Open Data Detector is used as a benchmark detector. Physics performance is estimated comparing tracks reconstructed using the ACTS tracking framework obtained from the full simulation and the machine learning-based one.