Analysis of metrics for trackster characterization in the High-Granularity Calorimeter

The future upgrade of the forward calorimeter at the CMS(Compact Muon Solenoid) experiment to the High-Granularity Calorimeter(HGCAL) will give unprecedented precision. This granularity is utilized by the HGCAL reconstruction to reconstruct particles at great detail into so-called tracksters. Still, some problems occur during certain situations during reconstruction. The aim of this project is to analyze these problems and develop metrics that can classify the reconstructed showers as correct or incorrect. Three different problems were analyzed: Missing layer, Low energy incomplete and Merging. Several different metrics were analyzed both centrality-based and shower shape based. A centrality profile metric was found to be the most optimal for the missing layer case. For low energy and merging the $dR_{EW}$ and $\overline{z}$ shower shapes were found to be most optimal with a very basic linear classification accuracy of 92.7\% for the low energy dataset. Finally, future work was discussed with the main points being a general algorithm and investigating hadronic showers.