The Argo Experiment In The Gamma / Hadron Method For Distinguishing Between Studies

The origin and acceleration mechanism of the cosmic rays are the unsolved problems in cosmic ray physics. The measurement of the characteristics of cosmic rays from certain astrophysical sources can provide experimental information for solving these problems. Charged cosmic ray particles are deflected by the interstellar magnetic field before reaching the earth, losing the direction information of the sources which they originated.γ- rays are not affected by the interstellar magnetic field, so they carry the direction information of the sources which produced them. For this reason, theγrays play a very important rule in the cosmic ray physics study. However,γray detection suffers from the huge background constituted by ordinary cosmic rays, whose flux is 3 orders of magnitude larger than the typical -ray emitter flux. So effectiveγ/hadron separation is crucial inγastronomy studies for ground based experiment.The apparatus of ARGO experiment located at Yangbajing consists of a full coverage array of Resistive Plate Chambers (RPC). It can measure the arrival time and lateral density distributions of the secondary particles produced in an Extensive Air Shower (EAS). These information could be used to separate the showers initiated by primaryγand hadrons.The purpose of this thesis is to investigate the possibility of separating the EAS induced by primaryγrays and hadrons by using the space and time information measured by the ARGO detector array. A large amount of EAS showers initiated byγand proton are generated by a Monte Carlo simulation program. The transportation of the secondary particles in EAS inside the ARGO detector and the detector response are also simulated. The different characteristics ofγ-initiated and hadron-initiated showers are studied by using these generated events. 5 variables which could be used to characterize the difference are gotten. These variables are: the RMS of the EAS particle density distribution along the x and y direction; the weighted mean lateral spread; the minimum mean tree length and the variable that characterize the skewness of the density distribution. Using these variables as input, multi-variant analysis is performed with TMVA (Toolkit for Multivariate Analysis) program to separate the γ-initiated and hadron-initiated showers. Three identification methods are used: the rectangular cut optimization, the projective likelihood method, and the artificial neuron network (ANN) method. From the test, we found that the efficiency for selecting the γ-initiated showers can reach more than 80% while rejecting more than 75% of the hadron-initiated showers. The best quality factor is 2.06.By comparing the three methods, we found that the ANN method is the best method to be used for the γ-induced shower selection.