guest ::
| Home > Publications database > Performance of a soft secondary vertex tagger usingproton-proton collisions collected at 13.6 TeV withthe ATLAS detector |
| Home > Publications database > Performance of a soft secondary vertex tagger usingproton-proton collisions collected at 13.6 TeV withthe ATLAS detector |
| Master Thesis | PUBDB-2025-04944 |
; ;
2025
Abstract: This thesis evaluates a b-tagging algorithm optimized to identify low pT (soft) b-hadrons in theATLAS experiment at the LHC. The algorithm, called the NewVrtSecInclusiveTool, reconstructssoft secondary vertices (SSVs), which can be connected to the decay of a soft b-hadron. Theanalysis evaluates the properties of these soft secondary vertices and compares them with theproperties of b-hadrons using a dileptonic t¯t sample from the Monte Carlo campaign denotedas mc23a which corresponds to the ATLAS data taking in 2022. An acceptance definition isintroduced to specifically test the identification ability of the algorithm outside of jets. Furthermore,a ΔR matching procedure is developed to assess if a soft secondary vertex can be associated tothe decay of a b-hadron. This procedure divides the SSVs into true SSVs and fake SSVs. Acomparison of these objects is done and an explanation for the origin of the fake SSVs is given.The b-hadrons and SSVs in acceptance, the matched and fake SSVs and the matched b-hadronsare used to develop an efficiency definition for the algorithm. Moreover, the average number offake SSVs nF is introduced to assess how often the algorithm makes a wrong tagging decision.The efficiency and the average number of fake SSVs are then analysed as a function of b-hadronproperties and event variables. Additionally, they are used to evaluate the 3 working points ofthe algorithm. The overall efficiency is in the order of 0.2 to 0.25 depending on the working point.The overall average number of fake SSVs is between 0.01 and 0.04 depending on the workingpoint. Furthermore, regions which are enhanced in matched and fake SSVs are constructed and afurther splitting of these regions is discussed in an effort to enable a calibration of the algorithmin the future. Finally, Monte Carlo to data comparisons are performed using data from 2022 and2023 corresponding to the Monte Carlo campaigns denoted as mc23a and mc23d respective
|
The record appears in these collections: |
PDF
PDF (PDFA)