Cargando…
Towards a time-aware global event interpretation at the CMS experiment
To cope with the elevated pileup and luminosity of the HL-LHC, the CMS experiment will introduce sub-detectors able to perform time measurements during collisions. These sub-detectors, the MIP Timing Detector (MTD) and the High Granularity Calorimeter (HGCAL), will provide precise timing information...
Autor principal: | |
---|---|
Lenguaje: | eng |
Publicado: |
2023
|
Materias: | |
Acceso en línea: | http://cds.cern.ch/record/2872557 |
Sumario: | To cope with the elevated pileup and luminosity of the HL-LHC, the CMS experiment will introduce sub-detectors able to perform time measurements during collisions. These sub-detectors, the MIP Timing Detector (MTD) and the High Granularity Calorimeter (HGCAL), will provide precise timing information essential for mitigating the effects of the high pileup, performing particle ID, and 4D vertexing. The efforts to integrate time information into event reconstruction started with the creation of Monte Carlo truth information for MTD clusters. Subsequently, a first attempt to combine the time information from MTD and HGCAL in the global event interpretation has been carried out in the linking between tracks and energy deposits reconstructed in the endcap region with CLUE 3D, called tracksters. An improved version of the linking has been developed making use of the correct time information from the sub-detectors and performing a local time compatibility between tracks and tracksters. This new version of the linking showed an increase in the rejection of fake connections. Another study performed concerns the use of MTD time information on electron isolation to avoid contamination of pileup tracks when computing the isolation. If pileup tracks are not compatible in time with the electron they are removed from the computation. This proved effective in increasing the efficiency with respect to not using any time information in 200 pileup collisions. Furthermore, CMS is upgrading its data acquisition system to handle increased event rates and detector complexity. To achieve this, heterogeneous computing, i.e. the combination of CPUs with accelerators like Graphics Processing Units (GPUs), has been adopted at the High Level Trigger. GPUs accelerate event reconstruction without significantly increasing costs or energy consumption. The performance portability library SYCL/oneAPI has been tested and the portability of a single source code across different architectures has been demonstrated. |
---|