LowPtCaloTagMuon

Introduction

This page provides an explanation of Low pT calo tag muon algorithm.
The target in the algorithm is low pt muon below 5 GeV .
To use this algorithm, information of calo sampling is needed.

Low pT calo tag muon can de used in

• 1< pT[GeV] < 5
• |eta| <1.6

Useful link

Slides in MCP meeting

• Hiroaki's slide v1(about position matching and OLD algorithm)
• Hiroaki's slide v2(about Update of algorithm and performance estimation in data)

Other TWiki page

• MuonEfficiencyCorrectionsSubgroup

Derivation framework

• Merge request of new decoration tool for low pt calo tag

You can add this decorator to your Derivation. (example: SUSY19)

available DAOD files

// official DAOD

SUSY19 p4417

MUON1 p4513

(MUON1 dose not contain data17_13TeV:data17_13TeV.00341294.physics_Main.merge.AOD.r10803_p3630_tid15380666_00 since segmentation faults)

// private DAOD

data

user.hhibi:user.hhibi.MUON1.data18_13TeV.periodC.physics_Main.PhysCont_EXT0

user.hhibi:user.hhibi.MUON1.data18_13TeV.periodK.physics_Main.PhysCont.v5_EXT0

user.hhibi:user.hhibi.MUON1.data18_13TeV.periodO.physics_Main.PhysCont.v5_EXT0

user.gsumi:user.gsumi.MUON1.data17_13TeV.periodE.physics_Main.PhysCont_EXT0

user.gsumi:user.gsumi.MUON1.data17_13TeV.periodB.physics_Main.PhysCont_EXT0

user.gsumi:user.gsumi.MUON1.data17_13TeV.periodI.physics_Main.PhysCont_EXT0

user.hhibi:user.hhibi.MUON1.data16_13TeV.periodB.physics_Main.PhysCont_EXT0

user.hhibi:user.hhibi.MUON1.data16_13TeV.periodG.physics_Main.PhysCont.v2_EXT0

user.hhibi:user.hhibi.MUON1.data16_13TeV.periodE.physics_Main.PhysCont_EXT0

user.hhibi:user.hhibi.MUON1.data16_13TeV.periodB.physics_Main.PhysCont_EXT0

MC

user.hhibi:user.hhibi.mc16_13TeV.424108.Pythia8B_A14_CTEQ6L1_Jpsimu6.merge.r10201_r10210.DAOD.MUON1_EXT0

command example

please check AMIConfig tag in AMIConfig tag TWiki

pathena --trf "Reco_tf.py --inputAODFile %IN --outputDAODFile %OUT.DAOD.pool.root --reductionConf MUON1 --maxEvents -1 --AMIConfig p*****" --inDS NameOfinDS --outDS NameOfoutDS --nFilesPerJob 10

How to use

Input variables

30 inputs are needed to DNN

• track eta
• track pT [MeV]
• deposit energy in each sampling [MeV] × 28 ( sampling definition )

Where is the input variable in AOD?

The track is in the InDetTrackParticles container.
deposit energy is in the CaloCalTopoClusters container.

• CaloCalTopoClustersAux .e_sampl : vector<vector<float> >

Where is the input variable in DAOD?

TrackAssociatedCaloSampleDecorator adds information on position matched sampling to a low pT track.
You can add this decorator. (example: SUSY19)

sampling information is saved as an aux-data for position matched track.

• InDetTrackParticlesAuxDyn .trackCaloClusEta : vector<vector<float> >
• InDetTrackParticlesAuxDyn .trackCaloClusPhi : vector<vector<float> >
• InDetTrackParticlesAuxDyn .trackCaloSampleE : vector<vector<vector<float> > >
• InDetTrackParticlesAuxDyn .trackCaloSampleNumber : vector<vector<vector<int> > >

How to calculate low pt calo tag muon score from input variables

Lwtnn ( link to Github of lwtnn )

The DNN was trained by Keras(python).
lwtnn is used to implement the DNN in Athena (mainly c++).
input type should be store into std::map<std::string, double> .

Table of input name.

example

• implemented into MuonPerformanceAnalysis

Definition of WP

• Hiroaki's target is WP optimized for the EWK SUSY search.
• WP definition with DNN

Progress of calibration development

• Implemented low pt calo tag muon algorithm into private MuonPerformanceAnalysis.(2020/Sep/09)