guest ::
| Home > Publications database > Tracking in Dense Environments for theATLAS ITk Strip Detector |
| Home > Publications database > Tracking in Dense Environments for theATLAS ITk Strip Detector |
| Master Thesis | PUBDB-2024-07874 |
; ; ;
2022
Abstract: The Phase II upgrade of the ATLAS detector will prepare it to handle the 2.5–fold increase in instantaneous luminosity of the High Luminosity LHC project starting in 2028. The upgrade will also involve the tracking sub-system of the ATLAS detector to be replaced by the Inner Tracker (ITk), which is made up of a closer-to-beam-line Pixel and acircumscribing Strip detector. A particular challenge in such a scenario is the reconstruction of highly collimated particle trajectories that overlap and share hits in the ITk. These Dense Environments are ubiquitous since they occur in the cores of high pT jets. The reconstruction of such high pT jets is important to identify boosted objects like highly energetic top quarks or Higgs bosons. These objects are of key importance, for example, in searches for di-Higgs production, the observation of which is one of the central goals of HL-LHC.In the ATLAS software for the current Inner Detector (ID), merged clusters in the pixel detector, resulting from energy deposits of several close-by particles are identified and split using deep neural nets. No such splitting procedure is used for the strip part of the current ID. The necessity of such methods is unclear for the ITk given its superior resolution compared to the current ID. Previous results show that for the ID, the reconstruction efficiency decreases significantly in jet cores with increase of jet pT .In this thesis, first, the performance of cluster and track reconstruction in dense environments is studied for the ITk Strip detector. The goal is to determine if a cluster splitting procedure is needed. The relevant variables sensitive to cluster merging such as cluster size, track merging rates, residual and impact parameter resolution, etc. are studied. It is found that tracks in high-pT jet cores have a high merging rate and suffer from a drop in track reconstruction efficiency as the jet pT increases, suggesting a need for improvement of the track reconstructionsoftware for dense environments.Second, a truth-based cluster splitting, wherein, truth information about the particles is used to identify the positions and their uncertainties of the sub-clusters belonging to the individual particles contributing to a merged cluster, is being implemented. This will allow for the evaluation of the reconstruction performance of the ATLAS tracking software in case of a perfect splitting of multi-particle clusters and hence an estimation of the potential gain from implementing a machine learning based splitting at the reconstruction level. Based onthe outcome of the idealized splitting, a procedure similar to the one implemented for the ID Pixel detector can be adopted in the future.
|
The record appears in these collections: |
PDF
PDF (PDFA)