CERN Accelerating science

An innovative approach to particle identification (PID) analyses employing machine learning techniques and its application to a physics case from the fixed-target programme at the LHCb experiment at CERN are presented. In general, a PID classifier is built by combining the response of specialized subdetectors, exploiting different techniques to guarantee redundancy and a wide kinematic coverage. At analysis level, the efficiency of PID selections changes thus as a function of several experimental observables, notably the particle momentum, the collision geometry and the experimental conditions. To precisely model the distribution of the PID classifier overcoming the unavoidable imperfections of the simulation, large samples of calibration channels reconstructed and selected in data are needed. In the presented approach, conceived for all applications where the collection of sufficiently-large-size calibration samples is not possible, the PID classifier is modeled on another high-statistics training sample using a Gaussian Mixture Model whose parameters are determined by Multi Layer Perceptrons. These are fed with the relevant experimental features and the non-trivial dependencies of the PID classifier are learned and predicted for the lower-statistics sample. Thanks to its speed and easy configuration, the presented approach, demonstrated on a proof-of-principle physics case to perform as or better than the detailed simulation, is expected to be employable on a large variety of use cases dealing with experimental observables depending on a sizeable number of experimental features.

The LHCb experiment started in 2015 a pioneering fixed-target program studying the collisions between LHC protons and gas targets injected into the LHC beam pipe through a system called SMOG. The gas was free to spread between a wide area covering the pp luminous region, limiting the injectable gas species to noble ones. Fixed-target data were mostly acquired during special runs of limited duration to avoid interfering with the $pp$ physics program. The upgrade of the SMOG program, SMOG2, will mainly consist in the installation of a confinement cell for the gas covering a 20 cm long region upstream the nominal $pp$ interaction point. This will allow to increase the gas density by up to two orders of magnitude and to potentially extend the gas species to be injected. Moreover, the large separation between the $pp$ and SMOG2 regions opens the scenario of a simultaneous data-taking, provided that the reconstruction performance of both fixed-target and $pp$ events is not significantly affected and the tight time constraints for the online reconstruction and selection are met. This poster contains some preliminary results for the efficiencies of the Run3 tracking sub-detectors: the performance are demonstrated to be comparable between $pp$ and SMOG2 collisions and they do not degrade for $pp$ when adding Helium or Argon gas targets.