Eur.Phys.J.CmanuscriptNo. (willbeinsertedbytheeditor) Lepton identification at particle flow oriented detector for the future e+e− Higgs factories DanYu1,2,ManqiRuana,1,VincentBoudry2,HenriVideau2 1IHEP,China 2LLR,EcolePolytechnique,France 7 1 0 2 Received:date/Accepted:date b e F Abstract Theleptonidentificationisessentialforthephysics In terms of Higgs measurements, the electron-positron 7 programs at high-energy frontier, especially for the precise colliders play a role complementary to the hadron collid- 1 measurementoftheHiggsboson.Forthispurpose,aToolkit erswithdistinguishableadvantages.Manyelectron-positron for Multivariate Data Analysis (TMVA) based lepton iden- Higgs factories have been proposed, including the Interna- ] t tification (LICH1) has been developed for detectors using tional Linear Collider (ILC), the Compact LInear Collider e d highgranularitycalorimeters.Usingtheconceptualdetector (CLIC),theFuturee+e-CircularCollider(FCC-ee)andthe - geometryfortheCircularElectron-PositronCollider(CEPC) CEPC[1][5][6].Theseproposedelectron-positronHiggsfac- s n andsinglechargedparticlesampleswithenergylargerthan tories pick and reconstruct Higgs events with an efficiency i 2 GeV, LICH identifies electrons/muons with efficiencies closeto100%,anddeterminetheabsolutevalueoftheHiggs . s higherthan99.5%andcontrolsthemis-identificationrateof couplings.ComparedtotheLHC,thesefacilitieshavemuch c i hadron to muons/electrons to better than 1%/0.5%. Reduc- betteraccuracyontheHiggstotalwidthmeasurementsand s y ingthecalorimetergranularityby1-2ordersofmagnitude, Higgs exotic decay searches, in addition the accuracies of h the lepton identification performance is stable for particles Higgs measurements are dominated by statistic errors. For p withE>2GeV.AppliedtofullysimulatedeeH/µµHevents, example, the circular electron-positron collider (CEPC) is [ the lepton identification performance is consistent with the expectedtodeliver1millionHiggsbosonsinitsHiggsop- 2 singleparticlecase:theefficiencyofidentifyingallthehigh eration,withwhichtheHiggscouplingswillbemeasuredto v energyleptonsinanevent,is95.5-98.5%. percentorevenpermillelevelaccuracy[6]. 2 TheleptonidentificationisessentialtothepreciseHiggs 4 5 measurements.TheStandardModelHiggsbosonhasroughly 7 1 Introduction 10%chancetodecayintofinalstateswithleptons,forexam- 0 ple, H→ WW* →llvv/lvqq, H→ZZ*→llqq, H→ττ, H→ . After the Higgs discovery, the precise determination of the 1 µµ,etc.TheSMHiggsalsohasabranchingratioBr(H→bb) 0 Higgsbosonpropertiesbecomesthefocusofparticlephysics = 58%, while the lepton identification provides an impor- 7 experiments.Phenomenologicalstudiesshowthatthephysics tant input for the jet flavor tagging and the jet charge mea- 1 atTeVscalewouldberevealediftheHiggscouplingscould : surement.Ontopofthat,theHiggsbosonhasasignificant v reachthepercentlevelmeasurementaccuracy[1][2]. chancetobegeneratedtogetherwithleptons.Forexample, i X TheLHCisapowerfulHiggsfactory.However,thepre- in the ZH events, the leading Higgs generation process at cisionofHiggsmeasurementsattheLHCislimitedbythe r 240-250 GeV electron-positron collisions, about 7% of the a hugeQCDbackground,thelargetheoreticalandsystemati- Higgs bosons are generated together with a pair of leptons caluncertainties.Inaddition,theHiggssignalattheLHCis ( Br(Z→ee) and Br(Z→ µµ) = 3.36% ). At the electron- usually tagged by the Higgs decay products, making those positron collider, ZH events with Z decaying into a pair of measurementsalwaysmodeldependent.Therefore,thepre- leptonsisregardedasthegoldenchannelfortheHZZcou- cisionofHiggscouplingsattheHL-LHCistypicallylimited pling and Higgs mass measurement[7]. Furthermore, lep- to5-10%leveldependingontheoreticalassumptions[3][4]. tons are intensively used as a trigger signal for the proton ae-mail:[email protected] colliderstopickupthephysicseventsfromthehugeQCD 1LeptonIdentificationforCalorimeterwithHighgranularity backgrounds. The Particle Flow Algorithm (PFA) becomes 2 theparadigmofdetectordesignforthehighenergyfrontier[8, tracking system and a calorimeter systems with extremely 9,6,12].Thekeyideaistoreconstructeveryfinalstatepar- highgranularity. ticle in the most suited sub-detectors, and reconstruct all InthisCEPCconceptualdetectordesign,theforwardre- the physics objects on top of the final state particles. The gion,andtheyokethicknesshavebeenadjustedtotheCEPC PFA oriented detectors have high efficiency in reconstruct- collisionenvironmentwithrespecttotheILDdetector.The ingphysicsobjectssuchasleptons,jets,andmissingenergy. core part of this detector is a large solenoid of 3.5 Tesla. The PFA also significantly improves the jet energy resolu- Thesolenoidsystemhasaninnerradiusof3.4metersanda tion, since the charged particles, which contribute the ma- lengthof8.05meters,insidewhichbothtrackerandcalori- jority of jet energy, are usually measured with much better metersystemareinstalled.Thetrackingsystemiscomposed accuraciesinthetrackersthaninthecalorimeters[14,9,10, ofaTPCasthemaintracker,avertexsystem,andthesili- 11,13]. contrackingdevices.Theamountofmaterialinfrontofthe Toreconstructeveryfinalstateparticle,thePFArequires calorimeter is kept to ∼ 5% radiation length. Both ECAL excellentseparationbyemployinghighly-granularcalorime- andHCALusesamplingstructuresandhaveextremelyhigh ters. In the detector designs of the International Large De- granularity. The ECAL uses tungsten as the absorber and tector (ILD) or the Silicon Detector (SiD) [1,15], the total silicon for the sensor. In depth, the ECAL is divided into numberofreadoutchannelsincalorimetersreachesthe108 30 layers and in the transverse direction, each layer is di- level. In addition to cluster separation, detailed spatial, en- videdinto5by5mm2 cells.TheHCALusesstainlesssteel ergyandeventimeinformationontheshowerdevelopments absorberandGRPC(GlassResistivePlateChamber)sensor is provided. An accurate interpretation of this recorded in- layers.Ituses10by10mm2cellsandhas48layersintotal. formationwillenhancethephysicsperformanceofthefull As a Higgs factory, the CEPC will be operated at 240- detector[16]. 250GeVcenterofmassenergy.Tostudytheadequatelep- Using the information recorded in the high granularity tonidentificationperformance,wesimulatedsingleparticle calorimeterandthedE/dxinformationrecordedinthetracker, samples(pion+,muon-,andelectron-)overanenergyrange LICH(LeptonIdentificationinCalorimeterwithHighgran- of1-120GeV(1,2,3,5,7,10,20,30,40,50,70,120GeV). ularity),adedicatedleptonidentificationalgorithmforHiggs Ateachenergypoint,100keventsaresimulatedforeachpar- factories has been developed. Using CEPC conceptual de- ticle type. These samples follow a flat distribution in theta tectorgeometry[6](basedonILD)andtheArbor[14]recon- andphioverthe4π solidangle. structionpackage,itsperformanceistestedonsingleparti- ThesesamplesarereconstructedwithArbor(version3.3). clesandphysicsevents.Forthesingleparticleswithenergy To disentangle the lepton identification performance from higher than 2 GeV, LICH reaches an efficiency better than the effect of PFA reconstruction and geometry defects, we 99.5%inidentifyingthemuonsandtheelectrons,and98% select those events where only one charged particle is re- forpions.Itsperformanceonphysicsevents(eeH/µµH)and constructed. The total number of these events is recorded the final efficiency agrees with the efficiency at the single asN ,andthenumberoftheseeventsidentifiedwith 1Particle particlelevel. correctparticletypesisrecordedasN .Theperfor- 1Particle,T This paper is organized as follows. The detector geom- manceofleptonidentificationisthenexpressedasamigra- etry and the samples are presented in section 2. In section tion matrix in Table 2, its diagonal elements εi refer to the i 3,thediscriminantvariablesmeasuredfromchargedrecon- identification efficiencies (defined as N /N ), 1Particle,T 1Particle structed particles are summarized and the algorithm archi- andtheoffdiagonalelementPi representtheprobabilityof j tectureispresented.Insection4,theLICHperformanceon atypeiparticletobemis-identifiedastype j. single particle events is presented. In section 5, the corre- lationsbetweenLICHperformanceandthecalorimeterge- Table1 MigrationMatrix ometryareexplored.Insection6,theLICHperformanceon ZHeventswhereZdecaysintoeeorµµ pairsisstudied,the e−like µ−like π+like undefined resultsarethencomparedwiththatofsingleparticleevents. e− εe Pe Pe Pe e µ π und In section 7, the results are summarized and the impact of µ− Peµ εµµ Pπµ Puµnd calorimetergranularityisdiscussed. π+ Pπ Pπ επ Pπ e µ π und 2 Detectorgeometryandsample Inthispaper,thereferencegeometryistheCEPCconceptual 3 Discriminantvariablesandtheoutputlikelihoods detector[6],whichisdevelopedfromtheILDgeometry[1]. ILDisaPFAorienteddetectormeanttobeusedforcentreof LICHtakesindividualreconstructedchargedparticlesasin- massenergiesupto1TeV.Itisequippedwithalowmaterial put, extracts 24 discriminant variables for the lepton iden- 3 tification,andcalculatesthecorrespondinglikelihoodtobe anelectronoramuon.Thesediscriminantvariablescanbe characterizedintofivedifferentclasses: Electron 800 – dE/dx Muon ForatrackintheTPC,thedistributionofenergylossper 600 Pion unit distance follows a Landau distribution. The dE/dx estimatorusedhereistheaverageofthisvaluebutafter 400 cuttingtailsatthetwoedgesoftheLandaudistribution (first7%andlast30%).ThedE/dxhasastrongdiscrim- 200 inantpowertodistinguishelectrontracksfromothersat lowenergy(under10GeV)(Figure1). 0 0 0.2 0.4 0.6 0.8 1 FD_all · 10-6 x 0.5 Fig.2 FractaldimensionusingbothECALandHCALfore−,µ−and d 90 E/ π+at40GeV d 80 0.4 70 60 energydepositedinacylinderaroundtheincidentdirec- 0.3 50 tionwitharadiusof1and1.5Moliereradius. 40 30 – HitsInformation 0.2 20 Hits information refers to the number of hits in ECAL 10 and HCAL and some other information obtained from 0.1 0 0 0.2 0.4 0.6 0.8 1 1.2 hits,suchasthenumberofECAL(HCAL)layershitby log10(TrkEn) theshower,numberofhitsinthefirst10layersofECAL. Fig.1 dE/dxfore−,µ−andπ+,forelectronsitisstablearound2.4× 10−7,formuonandpionitissmalleratenergylowerthan10GeVand – ShowerShape,SpatialInformation afterthattheystartmixingwithelectron Thespatialvariablesincludethemaximumdistancebe- tweenahitandtheextrapolatedtrack,themaximumdis- tanceandaveragedistancebetweenshowerhitsandthe – FractalDimension axis of the shower (defined by the innermost point and The fractal dimension (FD) of a shower is used to de- the center of gravity of the shower), the depth (perpen- scribe the self-similar behavior of shower spatial con- diculartothedetectorlayers)ofthecenterofgravity,and figurations,followingtheoriginaldefinitionin[16],the thedepthoftheshowerdefinedasthedepthbetweenthe fractal dimension is directly linked to the compactness innermosthitandtheoutermosthit. oftheparticleshower. Atafixedenergy,theEMshowersaremuchmorecom- Thecorrelationofthosevariablesatenergy40GeVare pactthanthemuonorhadronshower,leadingtoalarge summarized in Figure 3, the definitions of all the variables FD.Themuonshowerusuallytakestheconfigurationof are listed in Appendix A. It is clear that the dE/dx, mea- a1-dimensionalMIP(MinimumIonizingParticle)track, sured from tracks, does not correlate with any other vari- thereforehasaFDclosetozero.TheFDofthehadronic ables which are measured from calorimeters. Some of the shower usually lays between the EM and MIP tracks, variablesarehighlycorrelated,suchasFD_ECAL(FDcal- sinceitcontainsbothEMandMIPcomponents.Atyp- culated from ECAL hits) and EcalNHit (number of ECAL icaldistributionofFDfor40GeVshowersispresented hits).Howeverallthesevariablesarekeptbecausetheircor- inFigure2, relationschangewithenergyandpolarangle. Foranycalorimetercluster,LICHcalculates5different LICH uses TMVA[17] methods to summarize these in- FDvalues:fromitsECALhits,HCALhits,hitsin10or put variables into two likelihoods, corresponding to elec- 20firstlayersofECAL,andallthecalorimeterhits. tronsandmuons.MultipleTMVAmethodshavebeentested – EnergyDistribution andtheBoostedDecisionTreeswithGradientboosting(BDTG) LICH builds variables out of the shower energy infor- methodischosenforitsbetterperformance.Thee-likeness mation,includingtheproportionofenergydepositedin (L )and µ-likeness(L )fordifferentparticlesina40GeV e µ the first 10 layers in ECAL to the entire ECAL, or the sampleareshowninFigure4. 4 – barrel1:middleofbarrel(|cosθ|<0.3), – barrel2:edgeofbarrel(0.3<|cosθ|<0.7), 100 dEdx NH_ECALF10 – overlap:overlapregionofbarrelandendcap(0.7<|cosθ| FD_ECALL20 80 <0.8), FD_ECALF10 aAvL__NEHcHal 60 – endcap:(0.8<|cosθ|<0.98). rms_Hcal EEClu_r 40 EEClu_R Take the sample of 40 GeV charged particle as an ex- EEClu_L10 20 FD_HCAL ample,themigrationmatrixisshowninTable2.Comparing FD_FEDC_AaLll 0 thistabletotheresultofALEPHforenergetictaus[18],the MaxDisHel minDepth -20 efficiencies are improved, and the mis-identification rates cluDepth graDepth -40 fromhadronstoleptonsaresignificantlyreduced. EcalEn avDisHtoL -60 maxDisHtoL NLHcal NLEcal -80 Table2 MigrationMatrixat40GeV(%) HcalNHit EcalNHit -100 Type e−like µ−like π+like Fig.3 Thecorrelationmatrixofallthevariables e− 99.71±0.08 <0.07 0.21±0.07 µ− <0.07 99.87±0.08 0.05±0.05 π+ 0.14±0.05 0.35±0.08 99.26±0.12 Theleptonidentificationefficiencies(diagonaltermsof the migration matrix) at different energies are presented in Figure5forthedifferentregions.Theidentificationefficien- ciessaturateat99.9%forparticleswithenergyhigherthan 2GeV.Forthosewithenergylowerthan2GeV,theperfor- mancedropssignificantly,especiallyinbarrel2andoverlap regions.Fortheoverlapregion,thecomplexgeometrylim- its the performance; while for the barrel2 region, charged particles with Pt < 0.97 GeV cannot reach the barrel, they willeventuallyhittheendcapsatlargeincidentangle,hence theirsignalismoredifficulttocatalog. Concerningtheoff-diagonaltermsofthemigrationma- Fig.4 Thee-likelinessandµ-likenessofe−,µ− andπ+ at40GeV, trix,thechancesofelectronstobemis-identifiedasmuons greylinesarethecutsfordifferentcatalogsinnextsection and pions are negligible (Pe,Pe <10−3), the crosstalk rate µ π Pµ is observed at even lower level. However, the chances e 4 Performanceonsingleparticleevents of pions to be mis-identified as leptons (Pπ, Pπ) are of the e µ order of 1% and are energy dependent. In fact, these mis- Thephasespacespannedbythelepton-likelihoods(L and e identificationsaremainlyinducedbytheirreduciblephysics Lµ)canbeseparatedintodifferentdomains,corresponding effects: pion decay and π0 generation via π-nucleon colli- to different catalogs of particles. The domains for particles sion.Meanwhile,themuonsalsohaveasmallchancetobe of different types can be adjusted according to physics re- mis-identifiedaspionsatenergysmallerthan2GeV.Figure quirements. In this paper, we demonstrate the lepton iden- 6 shows the significant crosstalk items (Pπ, Pπand Pµ) as e µ π tification performance on single particle samples using the a function of the particle energy in the endcap region. The followingcatalogs: green shaded band indicates the probability of pion decay – Muon:L >0.5 beforereachingthecalorimeter,whichisroughlycompara- µ – Electron:Le>0.5 blewithPµπ. – Pion:1-(L +L )>0.5 µ e – Undefined:L <0.5&L <0.5&1-(L +L )<0.5 µ e µ e 5 Leptonidentificationperformanceonsingleparticle Theprobabilitiesofundefinedparticlesareverylow(<10−3)eventsfordifferentgeometries atsingleparticlesampleswiththeabovecatalog. polaSrinanceglteheoEfcdHalticNNhsaHtlLeNNriEitLHmibcHniatuaalcixvtatDDEiiloaicisgsanlHHrlEacttpoDloonuLaLmefDpriMentttphihDatcFheexDlpDFse_tDiehasF_(HlDlEθvEe_ClaE)HEArC,CEiLEltauCAEh_brlLCuLmel_al1esuRv0T_s_A_HrNMLdcFH_aDeElVFH_pcDEAaNe_lCHEndAC_iEdsLEdAsFCxLt1AorL0aLn2i0Fn1t0ehde Tetheerspyoswteemrcsocnasluemwpitthiotnheannduemlebcetrroonfircecaodsotuotfcthhaenncaellos.riImt’s- independentlyonfoursubsets: importanttoevaluatethephysicsperformancefordifferent 5 barrel1 barrel2 number less than 20, marginal performance degradation is %)102 %)102 observed: the efficiency of identifying muons degrades by eff(100 eff(100 1-2% for low energy particles (E ≤ 2 GeV), and the iden- 98 98 tification efficiency of pion degrades slightly over the full 96 electron 96 electron energyrange,seeFigure7. muon muon 94 94 pion pion 92 92 hcal cellsize 60mm hcal 20 layers 90 90 1 10 102 1 10 102 %)102 %)102 overlap Energy endcap Energy tagged eff(19080 tagged eff(19080 %)102 %)102 96 muon 96 muon eff(100 eff(100 94 pion 94 pion 98 98 92 92 96 electron 96 electron 90 90 muon muon 1 10 Ener1g02y 1 10 Ener1g02y 94 94 pion pion 92 92 Fig.7 Theefficiencyofleptonidentificationfortwodifferentgeome- 90 90 tries 1 10 102 1 10 102 Energy Energy Fig. 5 The efficiency of lepton identification for e−, µ− and π+ as functionofparticleenergyinthefourregions 6 Performanceonphysicsevents endcap The Higgs boson is mainly generated through the Higgsst- %)6 eff( Ppm rbaohslounngfupsirooncepsrsoc(ZesHse)saantdemlecotrreonm-parogsiintraolnlyHthigrogusgfahcvtoerciteosr. Pp 4 em AsignificantpartoftheHiggsbosonswillbegeneratedto- Pp pion decay rate gether with a pair of leptons (electrons and muons). These leptonsaregeneratedfromtheZbosondecayoftheZHpro- 2 cess.Forthe electrons,theycanalsobegenerated together withHiggsbosonintheZbosonfusionsevents,seeFigure 8. At the CEPC, 3.6×104 µµH events and 3.9×104 eeH 0 eventsareexpectedatanintegratedluminosityof5ab−1.In 1 10 Ene1r0g2y theseevents,theparticlesareratherisolated. Fig.6 Themis-identificationratesofleptonidentificationfor µ and π in ∼ 5000 events for the endcap region; Pion decay rate band (to accountforthepolaranglespread)isindicatedforcomparison calorimetergranularities,atwhichtheLICHperformanceis analyzed. The performance is scanned over certain ranges of the followingparameters: Fig.8 FeynmandiagramsofmajorHiggsproductionwithleptonsat CEPC:theHiggsstrahlungandZZfusionprocesses. – the number of layers in ECAL, taking the value of 20, 26,30; TheeeHandµµHeventsprovideanexcellentaccessto – thenumberoflayersinHCAL:20,30,40,48; themodel-independentmeasurementtotheHiggsbosonus- – the ECAL cell size = 5×5 mm2, 10×10 mm2, 20×20 ingtherecoilmassmethod[7].Therecoilmassspectrumof mm2,40×40mm2 eeHandµµHeventsisshowninFigure9,whichexhibitsa – HCAL cell size = 10×10 mm2, 20×20 mm2, 40×40 highenergytailinducedbytheradiationeffects(ISR,FSR, mm2,60×60mm2,80×80mm2 bremsstrahlung,beamstrahlung,etc),whileinCEPCthebe- In general, the lepton identification performance is ex- amstrahlungeffectisnegligible.Thebremsstrahlungeffects tremely stable over the scanned parameter space. Only for forthemuonsaresignificantlysmallerthanthatfortheelec- HCAL cell size larger than 60×60 mm2 or HCAL layer trons,therefore,ithasahighermaximumandasmallertail. 6 thelastrowofTable3.Theeventidentificationefficiencies is roughly the square of the identification efficiency of the initialleptons.Comparingtheperformanceofbothgeome- tries,itisshownthatwhenthenumberofreadoutchannelsis reducedby4,theeventreconstructionefficiencyisdegraded by1.3%and1.7%,forµµHandeeHeventsrespectively. Fig. 9 The recoil mass spectrum of ee/µµ, low energy peak in eeH correspondstotheZfusionevents 7 Conclusion Figure 10 shows the energy spectrum for all the recon- Thehighgranularitycalorimeterisapromisingtechnology structedchargedparticlesin10keeH/µµHevents.Thelep- fordetectorsincolliderfacilitiesoftheHighEnergyFron- tons could be classified into 2 classes, the initial leptons tiers.Itprovidesgoodseparationbetweendifferentfinalstate (those generated together with the Higgs boson) and those particles, which is essential for the PFA reconstructions. It generatedfromtheHiggsbosondecaycascade.FortheeeH alsorecordstheshowerspatialdevelopmentandenergypro- events, the energy spectrum of the initial electron exhibits filetoanunprecedentedlevelofdetails,whichcanbeused a small peak at low energy, corresponding to the Z fusion fortheenergymeasurementandparticleidentifications. events. The precise identification of these initial leptons is To exploit the capability of lepton identification with the key physics objective for the lepton identification per- high granularity calorimeters and also to provide a viable formanceofthedetector. toolkitforthefutureHiggsfactories,LICH,aTMVAbased leptonidentificationpackagededicatedtohighgranularca- lorimeter,hasbeendeveloped.Usingmostlytheshowerde- 101404104 OElreIMOEincglrtuieirtincogioataninrnll o famefnrrlloe uoefcmmorlteonr c HmoHtnri igoHggngisgsgs 104 IMnuitoianl fmroumon Higgs scription variables extracted from the high granularity ca- 101303103 MHauEHMHdolaareundodcontrrnrooonnn 103 EHlaedctrroonn lorimeter and also the dE/dx information measured from 102 101202 102 tracker, LICH calculates the e-likeness and µ-likeness for 10 1010 10 each individually reconstructed charged particle. Based on 1 11 0 20 40 MC60 Energ8y0 (GeV1)00 1 these output likelihoods, the leptons can be identified ac- 00 2020 4040 6060 8080 101000 0 20 40 60 80 100 MMC CE nEenregryg (yG (GeVe)V) MC Energy (GeV) cordingtodifferentphysicsrequirement. Fig.10 EnergySpectrumofchargedparticlesineeHeventat250GeV Applied to single particle samples simulated with the centerofmassenergy CEPC_v1 detector geometry, the typical identification effi- ciencyforelectronandmuonishigherthan99.5%forener- Since the lepton identification performance depends on gieshigherthan2GeV.Forpions,theefficiencyisreaching the particle energy, and most of the initial leptons have an 98%.Theseefficienciesarecomparabletotheperformance energyhigherthan20GeV,wefocusedontheperformance reached by ALEPH, while the mis-identification rates are studyofleptonidentificationonthesehighenergyparticles significantlyimproved.Ultimately,theperformancesarelim- atdetectorswithtwodifferentsetsofcalorimetercellsizes. itedbytheirreducibleconfusions,inthesensethatthechance The µ-likeliness and e-likeliness of electrons, muons, for muon to be mis-identified as electron and vice versa is and pions, for eeH events and µµH events are shown in negligible, the mis-identification of pion to muon is domi- Figure11andFigure12.Table3summarizesthedefinition natedbythepiondecay. of leptons and the corresponding performance at different The tested geometry uses a ultra-high granularity calo- conditions.Theidentificationefficienciesfortheinitiallep- rimeter: the cell size is 1 by 1 cm2 and the layer number tons is degraded by 1-2% with respect to the single parti- ofECAL/HCALis30/48.Inordertoreducethetotalchan- cle case. This degradation is mainly caused by the shower nelnumber,LICHisappliedtoamuchmoremodestgran- overlap,andit’smuchmoresignificantforelectronsaselec- ularity,itisfoundthattheleptonidentificationperformance tronshowersaremuchwiderthanthatofmuon,leadingtoa degrades only at particle energies lower than 2 GeV for an larger chance of overlapping. The electrons in µµH events HCALcellsizebiggerthan60×60mm2 orwithanHCAL andviceversa,aregeneratedintheHiggsdecay.Theiriden- layernumberlessthan20. tification efficiency and purity still remains at a reasonable The lepton identification performance of LICH is also level. For charged leptons with energy lower than 20 GeV, tested on the most important physics events at CEPC. In theperformancedegradesbyabout10%becauseofthehigh theseevents,multiplefinalstateparticlescouldbeproduced statistics of background and the cluster overlap. The event inasinglecollision,theparticleidentificationperformance identification efficiency, which is defined as the chance of willpotentiallybedegradedbytheoverlapbetweennearby successfullyidentifyingbothinitialleptons,ispresentedin particles.TheleptonidentificationoneeH/µµHeventat250 7 Table3 µµH/eeHeventsleptonidentificationefficiency Geom1(ECALandHCALCellSize10×10mm2) Geom2(ECALandHCALCellSize20×20mm2) µµH eeH µµH eeH µdefinition Lµ>0.1 Lµ>0.1 Lµ>0.1 Lµ>0.1 edefinition Le>0.01Lµ<0.1 Le>0.001Lµ<0.1 Le>0.01Lµ<0.1 Le>0.001Lµ<0.1 εe 93.41±0.92 98.64±0.08 91.60±1.02 97.89±0.11 ηe 92.02±1.00 99.74±0.04 89.89±1.10 99.67±0.04 εµ 99.54±0.05 95.53±0.76 99.19±0.06 86.48±1.26 ηµ 99.60±0.04 96.31±0.70 99.83±0.03 95.38±0.81 εevent 98.53±0.13 97.06±0.19 97.24±0.18 95.40±0.24 Fig.11 e-likelihoodandµ-likelihoodofchargedparticleswithE>20GeVineeHevent Fig.12 e-likelihoodandµ-likelihoodofchargedparticleswithE>20GeVinµµHevent GeVcollisionenergyhasbeenchecked.Theefficiencyfora Acknowledgements ThisstudywassupportedbyNationalKeyPro- singleleptonidentificationisconsistentwiththesinglepar- grammeforS&TResearchandDevelopment(GrantNO.:2016YFA0400400), the Hundred Talent programs of Chinese Academy of Science No. ticleresults.Theefficiencyoffindingtwoleptonsdecreases Y3515540U1,andAIDA2020. by1∼2%whenthecellsizedoubles,whichmeansthatthe detectorneeds2∼4%morestatisticsintherunning.IneeH events,theperformancedegradesbecausetheclusteringal- AppendixA: Appendixsection gorithmstillneedstobeoptimized. ListandmeaningofvariablesusedintheTMVAwhichare Toconclude,ultra-highgranularitycalorimeterdesigned notmentionedinthetext: for ILC provides excellent lepton identification ability, for operationclosetoZHthreshold.Itmaybeaslightoverkill – NH_ECALF10:Numberofhitsinthefirst10layersof forCEPCandaslightlyreducedgranularitycanreachabet- ECAL tercompromise.AndLICH,thededicatedleptonidentifica- – FD_ECALL20: FD calculated using hits in the last 20 tionforfuturee+e-Higgsfactory,isprepared. layersofECAL 8 – FD_ECALF10: FD calculated using hits in the first 10 4. CMScollaboration,ProjectedPerformanceofanUpgradedCMS layersofECAL DetectorattheLHCandHL-LHC:ContributiontotheSnowmass Process[J].arXivpreprintarXiv:1307.7135,2013. – AL_ECAL: Number of ECAL layer groups (each five 5. CLICCDR,Amulti-TeVlinearcolliderbasedonCLICtechnol- layersformsagroup)withhits ogy:CLICConceptualDesignReport[J].editedbyM.Aicheler,P. – av_NHH: Average number of hits in each HCAL layer Burrows,M.Draper,T.Garvey,P.Lebrun,K.Peach,N.Phinney, groups(eachfivelayersformsagroup) H.Schmickler,D.SchulteandN.Toge,CERN-2012-007,2012. – rms_Hcal:TheRMSofhitsineachHCALlayergroups 6. M. Ahmad et al (The CEPC-SPPC Study Group), CEPC- SppCPreliminaryConceptualDesignReport:PhysicsandDetec- (eachfivelayersformsagroup) tor,http://cepc.ihep.ac.cn/preCDR/mainpreCDR.pdf,retrieved4th – EEClu_r:Energydepositedinacylinderaroundthein- May2015 cidentdirectionwitharadiusof1Moliereradius 7. Z. Chen,Y. Yang, M. Ruan, et al, Study of Higgsstrahlung Cross – EEClu_R:Energydepositedinacylinderaroundthein- Section and Higgs Mass Measurement Precisions with ZH (Z→ cidentdirectionwitharadiusof1.5Moliereradius µ+µ−)eventsatCEPC[J].arXivpreprintarXiv:1601.05352,2016. 8. CMScollaboration,Technicalproposalforthephase-IIupgradeof – EEClu_L10: Energy deposited in the first 10 layers of the CMS detector[J]. CERN, CERN-LHCC-2015-010. LHCC-P- ECAL 008,2015. – MaxDisHel: Maximum distance between a hit and the 9. M.A.Thomson,ParticleflowcalorimetryandthePandoraPFAal- helix gorithm[J].NuclearInstrumentsandMethodsinPhysicsResearch SectionA:Accelerators,Spectrometers,DetectorsandAssociated – minDepth:Depthoftheinnermosthit Equipment,2009,611(1):25-40. – cluDepth:Depthoftheclusterposition 10. F.Beaudette,TheCMSParticleFlowAlgorithm[J].arXivpreprint – graDepth:Depthoftheclustergravitycenter arXiv:1401.8155,2014. – EcalEn:EnergydepositedinECAL 11. J.C. Brient, Improving the Jet Reconstruction with the Particle – avDisHtoL: Average distance between a hit to the axis FlowMethod;anIntroduction[J].arXivpreprintphysics/0412149, 2004. fromtheinnermosthitandthegravitycenter 12. J.C.Brient,H.Videau,Thecalorimetryatthefuturee+e-linear – maxDisHtoL: Maximum distance between a hit to the collider[J].arXivpreprinthep-ex/0202004,2002. axisfromtheinnermosthitandthegravitycenter 13. H. Videau, Energy flow or Particle flow-The technique of en- – NLHcal:NumberofHCALlayerswithhits ergy flow for pedestrians[C]//International Conference on Linear – NLEcal:NumberofECALlayerswithhits Colliders-LCWS04. Ecole Polytechnique Palaiseau, 2004: 105- 120. – HcalNHit:NumberofHCALhits 14. M.Ruan,Arbor,anewapproachoftheParticleFlowAlgorithm. – EcalNHit:NumberofECALhits arXiv:1403.4784(2014). 15. T. Abe, ILD Concept Group-Linear Collider Collaboration. The International Large Detector: Letter of Intent, 2010[J]. arXiv References preprintarXiv:1006.3396,4(10). 16. M.Ruan,D.Jeans,V.Boudry,J.C.Brient,&H.Videau,(2014), 1. T.Behnke,J.E.Brau,P.N.Burrows,etal,TheInternationalLinear FractalDimensionofParticleShowersMeasuredinaHighlyGran- Collider Technical Design Report-Volume 4: Detectors[J]. arXiv ularCalorimeter,Physicalreviewletters,112(1),012001. preprintarXiv:1306.6329,2013. 17. A.Hoecker,P.Speckmayer,J.Stelzer,J.Therhaag,E.vonToerne, 2. M.E.Peskin,Physicsgoalsofthelinearcollider[J].arXivpreprint H.Voss,...&D.Dannheim(2007),TMVA-Toolkitformultivariate hep-ph/9910521,1999. dataanalysis.arXivpreprintphysics:0703039. 3. ATLAScollaboration,PhysicsataHigh-LuminosityLHCwithAT- 18. Aleph Collaboration, Measurement of the Tau Polarisation at LAS[J].arXivpreprintarXiv:1307.7292,2013. LEP[J].arXivpreprinthep-ex/0104038,2001.