Particle identification and analysis in GEM TRD using machine learning

Abstract

Particle Identification is one of the major challenges in experimental physics. The identification of a stable particle is done either on the basis of their interaction or by determining their masses. In traditional particle physics experiments, parti cles are identified by the characteristic signature they leave in the detector. Conventionally, particle identification was done using cut based method where a threshold was fixed and if the signature of the particle was more than the threshold value, then it was classified as a signal. One of the major drawbacks of cut based method was that we can only implement one feature at a time, and also the linear cuts can result in loss of signal along with background. With the advancement of superior hardware and smart algorithms, various machine learning and deep learn ing techniques came into existence. Deep learning and Artificial Neural Networks (ANN) have become the most popular tool for research, data driven and prediction based applications. To summarize the research work done in this thesis, an Artificial Neural Network (ANN) based approach is implemented to perform particle identification in GEM (Gas Electron Multiplier) based Transition Radiation Detector (TRD). This GEM based TRD is a part of detector R&D of the upcoming Electron Ion Collider (EIC) experiment at BNL, USA which will enhance the current knowledge of the nucleon structure. Keywords : Gas Electron Multiplier (GEM); Transition Radiation Detector (TRD); Artificial Neural Netowrk (ANN); Specific energy loss; Pion rejection factor