A novel reconstruction framework for an imaging calorimeter for HL-LHC

To sustain the harsher conditions at high luminosity LHC (HL-LHC), the CMS experiment has designed a high granularity calorimeter (HGCAL) for the endcap regions, that uses silicon sensors to achieve radiation tolerance, with the additional benefit of a very high readout granularity. In regions characterised with lower radiation levels, scintillator tiles with individual SiPM readout are employed. A novel reconstruction approach, "The iterative clustering" (TICL) framework, is under developed to fully exploit the HGCAL potential. The inputs to TICL are clusters of energy deposited in individual HGCAL layers delivered by a clustering algorithm based on energy-density, "CLUE", which has recently been developed and tuned. In view of the expected challenges on the computing capacity at the HL-LHC, the algorithms and their data structures are designed with GPUs in mind. Preliminary studies show that significant speed-up can be obtained when running these algorithms on GPUs. Machine learning techniques are being investigated and integrated into the reconstruction framework. This note presents preliminary results on single-particle energy resolution in events with no pileup interactions, using showers reconstructed using the latest TICL developments. No corrections are applied for ECAL rear-leakage (for photons) or for hadronic non-compensation for pions (i.e. software compensation).