Event Classification and Estimation of Low Mass Diffraction at the TOTEM Experiment at the LHC

The TOTEM experiment is a forward physics experiment at the Large Hadron Collider (LHC) at CERN. The goal of the experiment is to study the structure of the proton and proton interactions at high energies. This includes a precise measurement of the total proton-proton cross section and the measurement of the cross sections for individual processes like elastic, non-diffractive minimum bias and single and double diffractive. The experiment comprises three different types of detectors: Roman Pot silicon detectors for detecting protons scattered by very small angles (few micro radians) and T1 and T2 forward tracking telescopes for detecting charged particles from inelastic proton-proton scattering. Each detector has a digital readout. The detectors for the T2 forward tracking telescope consist of three planes of gas electron multipliers. The radial coordinate is provided by the strip readout and the azimuthal coordinate and trigger are provided by the pad readout. Due to cross-talk and electronics noise, it is possible for the one particle to turn on many adjacent pads in a detector plane. Two different methods of clusterization of the digitized pad readout for the T2 detector are compared in order to determine the optimal way to recognize clusters corresponding to the ionization signal from one particle. An event classifier is used to distinguish which process is leading to a specific inelastic event. This event classification is done with the charged particle tracks reconstructed from the T1 and T2 telescopes and the training of the classifier is done with simulated data, where the original process leading to the event is known. The classification efficiency and purity is studied using simulated data. Diffractive events can be distinguished with good efficiency and purity from non-diffractive in events that have reconstructed particles in both hemispheres. Events with reconstructed particles in only one hemisphere are mainly diffractive and do not classify well, which can be understood from simple physics arguments. Not all diffractive events are detected because sometimes all final state particles are produced at pseudorapidities larger than the acceptance of T1 and T2. In order to determine the inelastic rate needed for the total cross section measurement, it is important to estimate the total number of inelastic events taking also into account the fraction of events that are not detected. These low diffractive mass events are mainly single diffractive, but a small fraction of double diffractive events also contribute. The number of missed inelastic events can be estimated by extrapolating the diffractive mass distribution in some suitable inverse power of the diffractive mass or by means of selecting events with only one "elastic-like" final state proton and no reconstructed particles in the forward tracking telescopes T1 and T2. The extrapolation is valid if the distribution is approximately linear, which is the case of events generated with PYTHIA for an extrapolation in the inverse mass squared. Adding the estimated number of events from such an extrapolation to the number of detected events, the ratio between the calculated and simulated cross section is measured to be 0.99±0.03.