Software

Creation of input ROOT files from original ROOT files

All the software used until now reside in subdirectories of:

/lhome/ific/j/julio

On the one hand, we have the subdirectory TopNtupleAnalysis where we perform the first part of the analysis generating ntuples derived from the original ones. Those original ntuples are created by members of the Physics Group ttbar-1l-2l. The analysis proceeds according to 2 steps, though previously we were analyzing the original ntuples, thus missing some information.

Within the directory TopNtupleAnalysis we create the subdirectories build, source and run following the guidelines for ATLAS software. In the subdirectory source we find the packages required for the first step:

• TopNtupleAnalysis^?
  • This is the package where we can find the files the determine selection cuts and modify the structure of the final ntuple. It was obtained from a Gitlab repository at CERN
• • ssh://git@gitlab.cern.ch:7999/atlas-phys/exot/hqt/R21-ttbar-1lep/TopNtupleAnalysis.git

• HQTTtResonancesTools^?
  • ssh://git@gitlab.cern.ch:7999/atlas-phys/exot/hqt/R21-ttbar-1lep/HQTTtResonancesTools

• TtResonancesTools^?
  • ssh://git@gitlab.cern.ch:7999/atlas-phys/exot/hqt/R21-ttbar-1lep/TtResonancesTools

The important files that have to be changed are the following:

• TtresEventSaverFlatNtuple^?.h under source/HQQTtResonancesTools/HQQTtResonancesTools/: to add m_chi2 variable to the output TTree
• selections.py under source/TopNtupleAnalysis/python/: to modiy the cut on max_chi2
• observables.py under source/TopNtupleAnalysis/python/: to select those variables that end up in the TTree of the final output file (e.g. MC matching and Chi2 matching related variables)

First of all one has to create an updated TopNtupleAnalysis/share/sumOfWeights_R21.txt. The easiest way is to modify the already existing Create_SumOfWeights.sh script in the run subdirectory. Check the input files needed and create them for your initial ROOT files. Beware the options: -m w to write the file and -m a to append to an already existing file. The execution of the script generates a file also in the build area.

Process_TNA.sh generates the resolved and boosted output files for both electrons and muons according to objects and cuts specified within the script. Currently it adds the contributions of electrons and muons in the resolved regime and deletes the ROOT files for the boosted regime. The output files are of the type Zp{ MZprime }_TNA.root.

The ROOT files are stored under /lustre/ific.uv.es/grid/atlas/t3/julio/ttbar-ml/ttbar_Input/TNA/ and they are the inputs from which we extract and process the variables that end up in the csv file that is read by the ML algorithm.

Generation of csv files

The relevant directory now is /lhome/ific/j/julio/ttbar-ml. The subdirectory Ntuple_Reading contains the software to read the input ROOT files and build the csv files that are processed in the next step of the analysis.

Application of ML algorithms to jet classification

The code for the processing of the csv files resides under /lhome/ific/j/julio/ttbar-ml/ML_Analysis

-- JulioLozano - 22 Sep 2021