/// **************************************************** /// How to use the t tbar analysis software /// /// Author: Katja Seidel; 31.07.2012 ///******************************************************* //-------------------------------------------------------- Input: There are 5 sorts of channels: tt, qq, WW, ZZ, WWZ Mostly we have more statistics than needed.That mean the additional evnts can/should be used for training (flavor tagging, background, rejection, final fit) Exspecially for tt there are three sets: 604, 604_secondRound, 604_thirdRound. Second Round ist used as the final data set. Third Round is used for the signal fit in the last step of the analysis. 604 can be used as testing (or can be deleted) //--------------- Overview ------------------------------------- The analysis is devided into 5 parts: 1 - prepare the analysis; cross checks (training) 2 - do analysis up to the tt->bbWW identification using the LCIO software 3 - do Kinematic Fit using the LCIO software 4 - do background rejection in a standalone software using root 5 - do final unbinned maximum likelihood fit Before steps 1-5 can be done the software needs to be setup and installed (step 0) //------------------------------------------------------------------ 0) Setup Marlin software //-------------------------------------------------------------------- a) - Make sure you have to folder: TTBarAnalysis with the following subfolders: BTagging-Input - should be empty Channels - should be empty KinFitOutput - shpuld be empty KinematicFitting - source code for the kinematic fitting Selection-Rejection-Classification - source code for all other Marlin Processors SteeringFiles - scripts and templates and files: the required gear.xml file for the detector model CMakeLists.txt envMarlin.sh.in envSetup_v0XXXX.sh README - installation guide for the Marlin part b) - Do the installion according to the installation guide 1 ) Prepare Analysis ///************************************************* - This step is to do some checks and required tasks before starting the real analysis a) Prepare enviroment: source PATH/TTBarAnalysis/envSetup_v01-14.sh source PATH/TTBarAnalysis/buildv01-14/envMarlin.sh b) Run the MarlinProcessor TTbarLeptonVeto to find the optimal steering values for this processor - To get an estimate on how well the lepton veto performs the processor needs the following input collections: PandoraReconstructedParticles, MCParticles - This are the steering values that should the changed: IsolatedLeptonConeAngle - opening angle of cone around the isolated lepton candidate (default: 10 degree) LeptonEnergyCut - cut on energy of possible isolated lepton candidate (default: 10GeV, can easily be compared with MC info) EnergyCut - cut on energy of the particles, for which it is checked if they are inside the cone (default: 0.5 GeV) - Additional steering values that can the changed: NumberAllowedChargedTracksInCone - there should be no other charged particles in the cone (default: 0) ChargeCut - cut on charged of the particles, for which it is checked if they are inside the cone (default: 0.00001, they have to have a charge) NumberTrack - cut on number of allowed tracks in the cone (default 0) NumberIsolatedLepton - cut on how many isolated leptons we are searching for (default 1) At the end of the processor the performance concerning the applied steering parameters are listed c) Run the flavor tagging training - Go to folder: TTBarAnalysis/SteeringFiles/LCFIVertexBTagging/CrossChecks/ - Use the script createSteeringFile-TrainNN.sh and the template SteeringFileTemplate-TrainNN.xml to create a suitable steering file d) Optional In folder TTBarAnalysis/SteeringFiles/LCFIVertexBTagging/CrossChecks - The jet clustering can be evaluated ( ttbar_Overlay_4JetClusteringCheck.xml, ttbar_Overlay_6JetClusteringCheck.xml ) - The precision of the lepton energy, theta and phi aswell as the jet theta and phi measurement can be evaluated ( use macros/src/JetResolution.C and macros/src/LeptonResolution.C and the output of the Marlin jobs using ttbar_Overlay_4JetClusteringCheck.xml, ttbar_Overlay_6JetClusteringCheck.xml ) - The flavour tagging can be evaluated ( use macros/src/displayEfficiencyVsPurity.C and the output of the PlotProcessor in CrossChecks/ttbar_FlavourTaggingQuality_6Jets.xml and CrossChecks/ttbar_FlavourTaggingQuality_4Jets.xml ) 2. ) Marlin //----------------------------------------------------------------------- a) Prepare enviroment source PATH/TTBarAnalysis/envSetup_v01-14.sh source PATH/TTBarAnalysis/buildv01-14/envMarlin.sh b) - Go to folder: TTBarAnalysis/SteeringFiles/LCFIVertexBTagging/ -> The REAME file explains the usage of the script createSteeringFile.sh and the according template SteeringFileTemplate.xml - Here only variables have to be changed which are idential for all events (signal and background), like the path and name of the gear file - Go to the folder TTBarAnalysis/Channels and choose the subdirectory, e.q. tt - In each subdirectory should the the folders: logs, steer, output - Go to the folder TTBarAnalysis/Channels/. There you find a submit.sh file. This you have to edit for different channels like tt - Run it ./submit.sh - The log file is in log, the root output and the slcio output can be found in output 3) //----------------------------------------------------------------- - Go the the folder KinFitOut - There should be the subfolders steer, logs and the channel folders like tt, qq, .... - Go to steer subfolder and prepare the steering files -> Create two steering files per channel: one for the full-hadronic and one for the semi-leptonic decay branch - The slcio input files are the output files on the folders: Channel/tt/output/X/ and so on - Make shure that the output paths of the created root files are set to e.q. PATH/TTBarAnalysis/KinFitOut/tt/ - The most important output files are a root file called: LikelihoodVariables_FullHadronic-XXX.root LikelihoodVariables_SemiLeptonic-XXX..root since these are the input files for the next steps, which is the Background Rejection !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Depending in the available number of signal and background events these steps should be repeated for training events (e.q. training events are needed for the flavor tagging, background rejection and the final fit) and events which are used to optain the final analysis results !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 4. ) Background Rejection //--------------------------------------------------------------------------- - Make sure there is the folder: BackgroundRejection - Source the root version which is specified in the Makefile (or change the Makefile) - In Main.cc the input files for the training and the final background rejection need to the specified - In the file LikelihoodRejection.cc you can find the cut on the Likelihood for the final event selection (default 0.6) - More information in README - Save output root files for signal only (high statistics), signal+background, only background for the semi-leptonic and full-hadronic analysis branch 5. ) Final Fit //------------------------------------------------------------------------------------ - Make sure you have the folder: FinalFit/unbinnedLikelihoodFit - Check and if needed adjust the Makefile - Source root - Source envSetup.sh - In mtop.cc the input files need to the specified (signal only (high statistics), signal+background, only background) - Also make sure that the input parameter file mtop.dat is available and contains sensible parameters - To run: Just compile by typing make and ./mtop - The output are 4 root macros (.C) and eps files