Simulating the time projection chamber responses at the MPD detector using generative adversarial networks

A Maevskiy,A Zinchenko,F Ratnikov,V Riabov

doi:10.1140/epjc/s10052-021-09366-4

Abstract

High energy physics experiments rely heavily on the detailed detector simulation models in many tasks. Running these detailed models typically requires a notable amount of the computing time available to the experiments. In this work, we demonstrate a new approach to speed up the simulation of the Time Projection Chamber tracker of the MPD experiment at the NICA accelerator complex. Our method is based on a Generative Adversarial Network – a deep learning technique allowing for implicit estimation of the population distribution for a given set of objects. This approach lets us learn and then sample from the distribution of raw detector responses, conditioned on the parameters of the charged particle tracks. To evaluate the quality of the proposed model, we integrate a prototype into the MPD software stack and demonstrate that it produces high-quality events similar to the detailed simulator, with a speed-up of at least an order of magnitude. The prototype is trained on the responses from the inner part of the detector and, once expanded to the full detector, should be ready for use in physics tasks.

Highlights

The MPD detector is one of the two experiments at the NICA accelerator complex – a new heavy ion accelerator facility being constructed at the Joint Institute for Nuclear Research and located in Dubna, Russia [9,10]
To make a more precise quantitative evaluation, we introduce a set of metrics that we profile as a function of input variables and compare between the generated and validation data
This means that Geant3 energy deposits in the detector material can be used to scale the predicted integrated amplitude and account for any mismodeling of the amplitude introduced by the Generative Adversarial Network (GAN)

Summary

Introduction

A typical approach to constructing models for fast simulation of particle physics detectors is to use a simplified detector geometry and a simplified model of the interaction of particles with matter [13]. This approach is justified for subsystems with a flat sensitive volume, such as silicon trackers, that measure the two-dimensional coordinate of a passing particle. For systems with a large volume, such as calorimeters or TPC-based trackers, this approach makes it difficult to achieve a reasonable compromise between accuracy and simulation speed

Results

Discussion

Conclusion