Photon+Jets+MET Analysis

Mailing list archive

SUSY Photon Channel Meetings

SVN

SUSY Working Group HN

Introduction

This page provides useful information on the single photon + jets + Etmiss analysis being performed within the SUSYWG (2013).

In the context of GGM models, ATLAS has searched for SUSY production in the diphoton+MET, γ+l+MET and γ+bb+MET channels, covering the bino-like, wino-like and higgsino admixture (μ<0), respectively. The search described here is intended to fill the missing scenario in the GGM phase space.

Motivation

In models of General Gauge Mediation, the LSP is the gravitino, and the NLSP can be a neutralino admixture of higgsino and bino eigenstates. (This scenario is typical for |μ|≈M1.) This specific admixture means, provided μ>0, the NLSP decays to γ + gravitino or to Z + gravitino. With the Z boson decaying hadronically, this resulting signature is γ+jets+MET.

General decay chain of GGM neutralino in strong production events.

• Formats: PNG, PDF, EPS

Documents

Document	COM note	INT note
Supporting Note	SVN CDS	SVN CDS

Contact Information

Editors

The editors are Martin Tripiana and Francisco Alonso . Please feel free to contact them with any questions, comments, or concerns.

Mailing Lists

Please be certain to sign up for the relevant mailing lists for this analysis:

• SUSY Photons+MET Mailing List : atlas-phys-susy-photon@cern.ch - Discussion list for the SUSY searches including photons.

NB: be sure to remove the SPAMNOT from the address!

People

Martin Tripiana (UNLP)
Francisco Alonso (UNLP)
Tere Dova (UNLP)

(Please add/remove your name a piacere)

References and Bibliography

Experimental References

• CMS PAS : Search for Supersymmetry in Events with Photons, Jets and Missing Energy, CMS-PAS-SUS-11-009 (2011).
• CMS PAS : Search for supersymmetry in events with photons and missing energy, CMS-PAS-SUS-12-018 (2012).

Theoretical references

• P. Meade, N. Seiberg, and D. Shih, General Gauge Mediation, Prog. Theor. Phys. Suppl. 177 (2009) 143–158.
• M. Buican, P. Meade, N. Seiberg, and D. Shih, Exploring General Gauge Mediation, JHEP 03 (2009) 016.

Event & Object Selection

• Following at the moment all the SUSY WG recommendations detailed in SusyObjectDefinitionsr178TeV.
• Based on SUSYTOOLS-00-02-09-01

Event Preselection
LAr Hole	Smart Veto Remove events only if the estimated energy missed in the LAr hole contributes significantly to MET using the variable: Events are rejected if Ep > E_min and Ep > frac*MET where E_min and frac are two parameters and MET is the transverse missing energy. Default configuration is E_min=10GeV and frac=0.1. Check FakeMetEstimator for details.
GRL [Data only]	data12_8TeV.periodAllYear_DetStatus-v58-pro14-01_DQDefects-00-00-33_PHYS_StandardGRL_All_Good.xml
Bad/Corrupted events	Following DP recommendations: * veto LAr noise burst and data corruption. if (larError==2) { // reject event (D3PD)} * veto Tile corrupted events. if (tileError==2) { // reject event (D3PD)} * veto Tile corrupted events passing other quality criteria in periods G-J. TTileTripReader::checkEvent(unsigned int run, unsigned int lbn, unsigned int event) Instructions and more details can be found here: TileTripReader * veto incomplete events: In 2012 data-taking the TTC restart was developed to recover certain detector busy conditions without a run-restart. In the lumi-block after a TTC restart there can be events with incomplete events (where some detector information is missing from the event). if ((coreFlags&0x40000) != 0) { // This is an incomplete event remove from analysis }
Trigger	The current trigger configuration in data / MC12 is SMK: 325, L1PSK: 142, HLTPSK: 266. Among all the single photon triggers available, this analysis makes use of EF_g120_loose which has been the lowest-pt unprescaled trigger during the whole 2012. Please check whether this problem is pertinent to your data.
Vertex	The reconstructed primary vertex of the event has more than 4 associated tracks. D3PD : vx_nTracks[0]>4 .
Hot Tile [Data]	Only for periods B1-B2 Remove events where a jet points to the eta,phi region close to LBC28 (eta = -0.15 & phi = 2.7), if this jet has its highest energy fraction in the Tile second layer (Layer-ID=13) and this fraction is large (Elayer/Ejet>0.6). As described in HowToCleanJets2012
MET cleaning	BadJet Veto Reject events where at least one AntiKt4LCTopo jet (after overlap removal) with pT > 20 GeV is characterized as BAD by the ``Looser'' (aka LooseMinus aka Very Loose) criteria (see below). Cosmic Veto The event has at least one selected muon with a |z0| ≥ 1 mm or |d0| ≥ 0.2 mm, where the values are calculated with respect to the primary vertex.

Physics Object Baseline Selection
Photons Reference Link
Overlap Removal	See OverlapRemoval below.
γ-ID	Tight selection. Recomputed at D3PD level with PhotonIDTool , tune tight(2012) OBS: TightARegammaPID::PhotonTightAR not usable yet (until next AODFix)
	Ambiguity Resolver (AR) : To reduce electron fake. bool isPhotonAR = true; if ((ph_isEM & 0x800000) != 0) isPhotonAR = false;
Kinematics	ET > 125 GeV
	|ηcluster| < 2.37 [excluding the crack region]
ObjectQuality	OQ has no bits in BADCLUSPHOTON set : MaskedCellCore(bit 10), MissingFEBCellCore(bit 5), MissingFEBCellEdgeS1(bit 7), MissingFEBCellEdgeS2(bit 8), DeadHVS1S2S3Core(bit 1), DeadHVS1S2S3Edge(bit 2), BadS1Core(bit 15); Code snippet on D3PDs... Hide if ((ph_OQ& 34214)==0) { cout << "this is a good OQ photon" << endl; }
PhotonCleaning	Remove clusters with large amount of energy from bad cells and very narrow fake candidates (ph_OQ LArCleaning bit (27)), and out-of-time candidates (ph_OQ OutTime bit (26)). Code snippet on D3PDs... Hide if( !( (ph_OQ&134217728)!=0 && (ph_reta>0.98||ph_rphi>1.0||(ph_OQ&67108864)!=0) ) ){ cout << "this a good photon wrt the new photon cleaning" << endl; } After this cleaning, some bad clusters may remain. Photons with LArCleaning > 0.8 are checked by hand at the end of the analysis (event displays + LAr data quality info). Code snippet on D3PDs... Hide if((ph_OQ&134217728)!=0) cout << "this photon has a LArCleaning > 0.8" << endl; cout << "This can be a bad photon, especially if reta/rphi are close to 1 or if the timing is large (~ 10ns)" << endl; }
Isolation	ph_topoEtcone40_corrected < 5 GeV Corrections applied for out-of-core signal energy leakage and for UE/pileup, using the CaloIsolationCorrections tool. (See below).
Overlap Removal	See OverlapRemoval below.
Electrons Reference Link
Algorithm	egammaParameters::AuthorElectron ( el_author==1 or 3)
e-ID	mediumPP . Recomputed at D3PD level using macro provided as part of the egammaAnalysisUtils package.
Kinematics	ET > 25 GeV (calculated using EM cluster energy and matched track direction)
	|ηcluster| < 2.47 [including the crack region]
ObjectQuality	OQ has no bits in BADCLUSELECTRON set : MaskedCellCore(bit 10), MissingFEBCellCore(bit 5), MissingFEBCellEdgeS1(bit 7), MissingFEBCellEdgeS2(bit 8), DeadHVS1S2S3Core(bit 1), DeadHVS1S2S3Edge(bit 2); and cluster is not out of time (bit 26): |t| ≤ (10 + 2/|Eclus|) ns. Code snippet on D3PDs... Hide if ((el_OQ&1446)==0 && (el_OQ&egammaPID::OutTime)==0) { cout << "this is a good electron" << endl; }
Isolation	el_topoEtcone20_corrected < 5 GeV Corrections applied for out-of-core signal energy leakage and for UE/pileup, using the CaloIsolationCorrections tool. (See below).
Overlap Removal	See OverlapRemoval below.
Muons
μ-ID	Loose Combined or Segment-Tagged Staco Muons following the recommendations of the MCP group. In D3PDs: Select author either 6 or 7.
Kinematics	pT > 10 GeV, |η| < 2.4
ID Hits	- ( ! expectBLayerHit) || (numberOfBLayerHits > 0) - Number of pixel hits+number of crossed dead pixel sensors > 0 (was 1 in 2011). - Number of SCT hits+number of crossed dead SCT sensors > 4 (was 5 in 2011). - Number of pixel holes + number of SCT holes < 3. - TRT extension : If the ID track is within the TRT acceptance a TRT extension is required: Let n = nTRThits + nTRTouts, where nTRThits (nTRTouts) denote the number of TRT hits (outliers) of the muon track, then: Case 1: 0.1 < |η| < 1.9. Require n > 5 and nTRTouts < 0.9 n. Case 2: |η|≤ 0.1 or |η| ≥ 1.9. If n > 5, then require nTRTouts < 0.9 n. Show D3PD variables relevant for above cuts... Hide mu_expectBLayerHit mu_nBLHits mu_nPixHits mu_nPixelDeadSensors mu_nSCTHits mu_nSCTDeadSensors mu_nPixHoles mu_nSCTHoles mu_nTRTHits mu_nTRTOutliers NB: Certain cuts have been loosened compared to 2011 without increasing the misidentification rate (see page 8 of this talk).
Cosmic Cut	|d0| < 1 mm (where d0 is relative to the primary vertex, mu_staco_d0_exPV) and |Z0 - Zvtx| < 1 cm (at the beamline, mu_staco_z0_exPV)
Jets
Algorithm	AntiKt4LCTopo. See JetCollections
Jet Cleaning	Reject jets which are characterized as BAD according to the Looser (aka LooseMinus aka Very Loose) criteria definition
Kinematics	pT > 20 GeV and |η| < 2.5
Pileup removal	Require |!JetVertexFraction| > 0.75 (wrt the primary vertex). Jets w/o associated tracks have JVF=-1 and pass this requirement.

Selection Variables Definition

• H_T : Total hadronic transverse energy

It is defined as the E_T sum of all good jets in the event, according to the criteria above.

• R_T² :

It is defined as the ratio of the two leading jet E_T sum to H_T.

• Δφ(jet, MET) :

It is defined as the minimum Δφ distance from the selected photon to the two leading jets.

Corrections [Scaling + Smearing + Reweighting]

SHOW ME! OK enough...

What	Apply to data?	Apply to MC?	Tool info
Event Reweighting
b-tag scale factor	no	yes	CalibrationDataInterface-00-01-02 with the calibration file for MC11b final release. Weight each good jet by results of either getScaleFactor or getInefficiencyScaleFactor , depending on whether or not the jet was tagged. Multiply all jet weights within an event.
Pileup Reweighting	no	yes	PileupReweighting-00-02-05 (see this Twiki for details.) Code snippet on D3PDs... Hide After adding the root files with the histograms, use: m_tPileUp->SetUnrepresentedDataAction(2); // "just carry on" m_tPileUp->UsePeriodConfig("MC12a"); m_tPileUp->SetDataScaleFactors(1./1.11); //mandatory : to put *before* the call to AddLumiCalcFile(...) m_tPileUp->AddConfigFile(m_mcRootFileName); m_tPileUp->AddConfigFile(m_mcRootFileName); m_tPileUp->Initialize(); Reweight each event with: m_tPU->GetCombinedWeight(RunNumber, mc_channel_number, averageIntPerXing); A good description of minimum bias vertex multiplicity (with ≥2 tracks) with the Pythia8 tuning used in MC12 is obtained by scaling the <μ> in MC by 11±0.03 (details)
MC generator weights
MC@ NLO samples	no	yes	use mcevt_weight[0] (in D3PDs: ((*mcevt_weight)[0])[0])
Photon related
Energy scale	yes	no	Use egammaAnalysisUtils-00-03-24 EnergyRescaler er; er.useDefaultCalibConstants("2012"); new_E = scale.applyEnergyCorrectionMeV (ph_cl_eta, ph_cl_phi, ph_E, ph_Et, SYST_FLAG, PHO_TYPE); //energies in MeV PHO_TYPE set to 'UNCONVERTED_PHOTON', 'CONVERTED_PHOTON' accordingly SYST_FLAG is 0=NOMINAL, 1=ERR_DOWN, 2=ERR_UP (one sigma variations). Reference
Energy resolution	no	yes	Use egammaAnalysisUtils-00-03-24 EnergyRescaler er; er.useDefaultCalibConstants("2012"); smear.SetRandomSeed (EventNumber + 1001 * iph); double smearedEnergy = oldEnergy * er.getSmearingCorrectionMeV(cl_eta, oldEnergy, SYST_FLAG, CT_FLAG); iph is the index of the photon within its collection (before any selections). SYST_FLAG is 0 for nominal scale, 1 or 2 for 1-sigma variations. CT_FLAG should be false for all MC12 physics samples (produced with zero constant term for the LAr EM resolution). Reference
ID Efficiencies	no	yes	use the tag FudgeMCTool-00-00-16 to apply the FudgeFactors with pre-selection 13. (EGamma recommendation)
Electron related
Energy & direction	yes	yes	Code snippet on D3PDs... Hide //4-momentum variables if ((nSCT + nPix) < 4){ // basically TRT stand-alone tracks eta = el_cl_eta [el_index]; phi = el_cl_phi [el_index]; Et = el_cl_et [el_index]; E = el_cl_E [el_index]; } else{ eta = el_tracketa [el_index]; phi = el_trackphi [el_index]; Et = el_cl_E [el_index] / cosh(el_tracketa [el_index]); E = el_cl_E [el_index]; } //**WARNING** Use always cluster (eta,phi) for calorimeter related cuts/bins (crack removal, efficiency binning, OTX maps, energy scale corrections) eta = el_cl_eta [el_index]; phi = el_cl_phi [el_index]; Reference
Energy scale	yes	no	Use egammaAnalysisUtils-00-03-24 EnergyRescaler er; er.useDefaultCalibConstants("2012"); new_E = scale.applyEnergyCorrectionMeV (e_cl_eta, e_cl_phi, e_E, e_Et, SYST_FLAG, "ELECTRON"); //energies in MeV SYST_FLAG is 0=NOMINAL, 1=ERR_DOWN, 2=ERR_UP (one sigma variations). Reference
Energy resolution	no	yes	Use egammaAnalysisUtils-00-03-24 EnergyRescaler er; er.useDefaultCalibConstants("2012"); smear.SetRandomSeed (EventNumber + 1001 * iele); double smearedEnergy = oldEnergy * er.getSmearingCorrectionMeV(cl_eta, oldEnergy, SYST_FLAG, CT_FLAG); iele is the index of the electron within its collection (before any selections). SYST_FLAG is 0 for nominal scale, 1 or 2 for 1-sigma variations. CT_FLAG should be false for all MC12 physics samples (produced with zero constant term for the LAr EM resolution). Reference
ID Efficiencies	no	yes	egammaSFclass, egammaAnalysisUtils-00-03-24 egamamSFclass egammaSF; egammaSF.scaleFactor (el_cl_eta, el_pt, 6, 0, 8, True); // rel = 8 : release 17.2 for 2012 data and G4 FullSim MC12a, "ICHEP2012 recommendations" // set = 6 : Medium++ Reference
Reco Efficiencies	no	yes	egammaSFclass, egammaAnalysisUtils-00-03-24 egamamSFclass egammaSF; egammaSF.scaleFactor (el_cl_eta, el_pt, 4, 0, 8, True); // rel = 8 : release 17.2 for 2012 data and G4 FullSim MC12a, "ICHEP2012 recommendations" // set = 4 : Reco + track quality requirements Reference
Muon related
pT smearing	no	yes	MuonMomentumCorrections-00-06-03, (SmearingClass) Code snippet on D3PDs... Hide MuonSmear::SmearingClass mcp_smear ("Data12", "staco", "q_pT", "Rel17.2_preliminary", MCP_DIR); mcp_smear.UseScale(1); //do muon scale correction as well mcp_smear.SetSeed (EventNumber, imu); if muon is combined: mcp_smear.Event (mu_MS_pt, mu_ID_pt, mu_pt, mu_eta, mu_charge); mu_pt = mcp_smear.pTCB(); if mcp_smear.ChargeFlipCB() < 0: mu_charge *= -1; else: mcp_smear.Event (mu_pt, mu_eta, "ID", mu_charge); mu_pt = mcp_smear.pTCB(); if mcp_smear.ChargeFlipID() < 0: mu_charge *= -1; imu is the index of the muon within its collection (before selections). MCP_DIR is the directory containing the MuonMomentumCorrections data files. Reference
ID Efficiency (SF)	no	yes	Using MuonEfficiencyCorrections-02-00-03+ Code snippet on D3PDs... Hide std::string directory=""; //directory with the SF file. Empty string to use files provided with the package, std::string file="SF.txt" //name of the file std::string unit("MeV"); // for MeV; for GeV use "GeV" Analysis::AnalysisMuonConfigurableScaleFactors::Configuration config=Analysis::AnalysisMuonConfigurableScaleFactors::Default //Default is the configuration for SFs which are not binned per run or per period //If the SFs are provided per Period (as for 2011 SFs), the configuration can be PerPeriod (to get the SF per single period) // or AverageOverPeriods (to get an average over the periods) //If the SFs are provided per Run (as for 2012 SFs), the configuration can be PerRun (to get the SF per single period) // or AverageOverRuns (to get an average over the runs) Analysis::AnalysisMuonConfigurableScaleFactors m_MCPsf(directory,file,unit,config); m_MCPsf.Initialise(); // getting the scale factors... // With the Default, AverageOverPeriods, AverageOverRuns configuration m_MCPsf.scaleFactor(p); // scale factor of a muon with fourmomentum (TLorentzVector) p // With the PerRun configuration m_MCPsf.scaleFactor(p,run); // scale factor of a muon with fourmomentum (TLorentzVector) p for a given (int) run // With the PerPeriod configuration m_MCPsf.scaleFactor(p,period); // scale factor of a muon with fourmomentum (TLorentzVector) p for a given (TString) period Reference
Jet related
LAr hole veto	yes	periods E-H	Veto events with any selected jets in the LAr hole with pT above threshold. LAr hole is defined by -0.1 < eta < 1.5 and -0.9 < phi < -0.1. pT threshold is 40 GeV in Monte Carlo. In data, threshold is 40 GeV * (1-BCH_CORR_JET) / (1-BCH_CORR_CELL).

Overlap removal strategy

• no with DR(e,) < 0.01
• no jet with DR(jet,/e) < 0.2
• no /e with 0.2 < DR(jet,/e) < 0.4
• no with DR(jet,) < 0.4

Cut flow tables

MC

• Benchmark : mc12_8TeV.173417.Herwigpp_UEEE3CTEQ6L1_GGM_gl_neut_800_600.merge.NTUP_SUSY.e1577_s1499_s1504_r3658_r3549_p1328 [ No rw . No pT/PV smearing ]

La Plata

DATA

• Benchmark : group.phys-susy.data12_8TeV.periodG.physics_Egamma.PhysCont.NTUP_SUSYSKIM.repro14_v01_p1328_p1329 PeriodG in runQuery

La Plata

SusyD3PDs

• The SusyD3PD ntuples generated by the SUSYD3PDMaker package are centrally produced since the beginning of the 2011 data taking period.

• Useful links:
  • SusyD3PDs = description on the ntuple content and production details.
  • Sofware coordinators : Chris Young and Borge Gjelsten
  • SUSYD3PD hypernews [atlas-phys-susy-d3pd] = Discussions related to SUSYD3PDMaker, SUSYD3PD production and SUSYTools package.

2013 Production

For both data and MC, D3PDs are produced with the AtlasPhysics-17.2.2.1.2 cache, SUSYD3PDMaker-00-12-10

Data Samples

• production tag for 2012 data (data12_8TeV): p1328_p1329

ContNumEvts12

• a private production of skimmed D3PDs (passing EF_g120_loose + some variables' slimming) is available on the grid
  • latest tag is 'UNLP7' (May 2013)
  • they all follow this naming scheme:

• a private production of skimmed D3PDs (passing EF_e60_medium1 + some variables' slimming) is available on the grid
  • latest tag is 'UNLP7_ELTRIG' (May 2013)
  • they all follow this naming scheme:

Data Periods

Period	Run range	Luminosity [pb-1]	runQuery details
Total	200804 - 215643	20344.37
A	200804 - 201556	795.91	link
B	202660 - 205113	5113.61	link
C	206248 - 207397	1409.06	link
D	207447 - 209025	3297.54	link
E	209074 - 210308	2534.11	link
G	211522 - 212272	1279.54	link
H	212619 - 213359	1449.04	link
I	213431 - 213819	1018.45	link
J	213900 - 215091	2605.48	link
L	215414 - 215643	841.634	link

Integrated luminosity used in this analysis. For each data taking period the run range and the integrated luminosity are given.

MC Datasets

• production tag for 2012 MC (mc12_8TeV): p1032

• private skimmed D3PDs (passing EF_g120_loose + some variables' slimming)

Signal . GGM bino-higgsino grid

In models of General Gauge Mediation, a higgsino-like neutralino decays to Z+gravitino or h+gravitino. We consider a GGM model where the lightest neutralino is a mixture of bino and higgsino, and we fix μ to be positive so that the branching ratio to Z boson id greater than the branching ratio to the lowest-mass supersymmetric Higgs boson.

With these considerations, the final state signature contains (most of the times) one photon, jets and missing energy.

Representative mass spectra for GGM signal samples (m_gl=800GeV, mχ₁₀=600GeV)

The SLHA files are defined using SUSPECT 2.41, SDECAY 1.3b, and HDECAY 3.4. We set the following global parameters for all signal points:

• M₂ = 2.5 TeV
• tan β = 1.5
• cτ < 1 mm
• M_A = 2 TeV
• m_h = 126 GeV

• BR(γG) = 0.51
• BR(ZG) = 0.48
• BR(hG) = 0.01

• Details of first gluino-neutralino grid proposal can be checked in this page

Decay chains

These are the dominant decay chains for this channel. The relative contribution in different regions of phase space have driven the optimization of the signal regions for this analysis.

2-body gluino decay to chi_10	2-body gluino decay to chi_20	2-body gluino decay to chi_30
3-body gluino decay to chi_1_+-	Relative contribution to total BR in phase space.

The particles expected in the final state are highlighted in red.

Background

Event Generation Details (cross sections, etc)

The details of the PhotonMetSusy8TeV datasets (e.g. xsections, k-factors and filter efficiencies) are given this file.

The SUSY::CrossSectionDB class (provided inside the SUSYTools package) is used to decode the cross-sections' values in the text file. For a how-to example please see the README file there.

• Useful links:
  • AtlasMcProductionMC12 - generator information
  • AtlasProductionGroupMC12a - production lists, MC12a high and normal priority, production tags

Major updates:
-- MartinTripiana - 12-Jun-2012

Responsible MartinTripiana
Review Never reviewed