Pass 8 is a comprehensive revision of the entire Fermi-LAT analysis chain from reconstruction to the event-level particle identification. At the reconstruction level, Pass 8 will provide significant improvement in all areas of LAT performance including effective area, point-spread function, and energy resolution. The event-level analysis is the final stage of the analysis chain in which high-level variables such as energy, direction, and photon class are assigned. To date, the Pass 8 effort has focused primarily on improving the event reconstruction. However that work is nearing completion, and we have started developing the event-level analysis for Pass 8.
We use the various components of the existing (Pass 7) event-level analysis as a starting point: Pass 7 uses Classification Tree (CT) based multivariate analyses to choose the best estimates of photon energy and direction, and creates estimators of the quality of the energy and direction reconstruction as well as the probability that a given event is a photon as opposed to cosmic-ray background. We discuss a framework used for training elements of the Pass 8 event-level analysis using TMVA, a ROOT-based multivariate classification package. We show "starting point" performance estimates for Pass 8 based on naively porting the Pass 7 analysis to Pass 8 without consideration of the reconstruction improvements, and finally consider how the reconstruction improvements in Pass 8 will allow us to further improve on the performance.