TauID Algorithm に関する研究 Motivation Tau identification To do for improvement conclusion 中村浩二 Motivation (TauIDの改良) VBF Htau tau is promising channel for Higgs discovery But as we can see the statistics of the signal is limited. Pt distribution of tau jet (MH=125GeV) 40 GeV 20 GeV Invariant mass distribution of VBF Htautau (MH=120GeV) with fast simulation BG Signal Our Pt threshold of tau jet is 40GeV But many signals are in the lower Pt region. Low Pt Tau jet identification and reconstruction are very important to increase statistics. We have developed tau identification optimized low Pt region Tau identification Tau candidates are selected based on track (finding Isolate 1 track or 3 tracks) We also use cells information . Find Pt >5 GeV tracks only 1 track Tau 1prong Tau 3prong QCD Jet We also required isolation there is no more Pt>2GeV track in DR<0.4. 3tracks in DR<0.1 (Find two more Pt>2GeV tracks) Find a seed cell associated with this track, and make Cluster (The seed is defined as a cell with larger than 6s of Noise near the track) reconstruction identification “Narrowness” cut is applied for identification ( I will talk in later slides) Energy deposit on EM calorimeter ATLASのEM calorimeter は、 presampler と、3つのsampling 層 からなる。 Sampling 2が、main component で、DhxDf=0.025x0.025と、 segment が細かい。 phi PreSampler EM1 EM2 EM3 0.4 0.4 eta We use cell based narrowness definition not DR. Cell-based narrowness LArEM Sampling 2 Tile Sampling 0 LArHEC Sampling 0 0.025 f 0.025 0.1 If Seed Cell is not on LArEM 33 7 7 0.1 h EMEt33 Narrowness EMEt77 Seed Cell Narrowness HDEt1x1 HDEt3x3 Narrowness distribution after 1prong/3prong selection We can enhance the signal against the BG by setting narrowness threshold. For example 0.7/0.6(1prong/3prong) 1prong 3prong Signal QCD(17-35GeV) QCD(35-70GeV) Narrowness Narrowness Efficiency & Rejection We plotted efficiency/rejection as For di-jet sample the function of visible Tau Pt. (20GeV-140GeV) Tagging Efficiency 1pr…0.7 3pr…0.6 1prong 3prong 0.7 0.6 Total gluon jet quark jet 1/R 10 50 Visible tau Pt [GeV] Jet Pt [GeV] 50%のefficiency が得られる。Rejectionに関しては、 改善の可能性がある。 To do… Track invariant mass for 3prong Mp±=139.6 [MeV] を仮定して、3本の track の invariant mass を求める。 Signal Di-Jet Track invariant mass for 3prong [GeV] Electron veto Gen Matching Tau purely tau Inclusive Tau Inclusive Tau Electron peak ! Signal MC all tau candidate (tau, electoron,muon,VBFjets) Gen Matching Tau 1.0 EMEt/TrkPt Pt(Pi+-)<Pt(Pi0) 1.0 EMEt/TrkPt ~ 1 && HDET/TrkPt ~ 0 を veto Cut のoptimize をする。 EMEt/TrkPt muon veto Gen Matching Tau Muonは、tau jet Et cut でveto される Inclusive Tau muon EMEt~HDEt~0 EMEt/TrkPt Electron HDEt=0 EMEt~1 QCD Event Topology のcheck Di-Jet event Pt>2GeV tracks phi phi Pt>2GeV track event display ▼ # of Tracks distribution Rec Track (iPad) ~10% becomes 1pr/3pr candidate Mean :6.3 # of Tracks in 1 Jet eta phi phi eta phi eta eta phi eta eta Conclusion VBF process で生成し、tau レプトン対に崩壊するHiggs 粒子 探索で、統計を増やすため、low Ptのtauにoptimizeした、track base のtauIDのアルゴリズムを開発中である。 Efficiency に関しては、50%程度出ているが、Rejectionに関 しては、十分な性能が出ていない。 QCD background をsuppress するため、Rejection factor を、現在の3-5倍よくする必要がある。-> Wjjj, jjjj BG から。 To do (Plan) 3prong のtrack invariant mass cut Lepton veto QCD event topology の再チェック p0 p+- invariant mass すべての cut valuable のoptimization などなど Back Up Signal selection (1) electron Pt>25GeV,muon Pt>20GeV (2) Tau jet with Pt>40GeV DR >0.7(from lepton) (3) Forward Jet : At least 2 Jet , Pt>20GeV(Leading 40GeV) (4) Forward Jet : h1*h2<0,|h1-h2|>4.4 (5) h1+0.7< hl,t <h2-0.7 (6) |cosflh| < 0.9 for missing Et (7) Transverse mass between lepton and missing energy less than 30GeV (8) Missing Et > 30GeV (9) Invariant mass of the tagged forward jets larger than 700GeV (10)Central Jet Veto : There is no Pt>20GeV jet in h1+0.7< h <h2-0.7 Background Signal… Major QCD background o W+3jets : Wjjj -> ltaujj -> depend on mainly TauID o bbjj : bbjj -> l tau jj bbjj -> l j tau j o jjjj : jjjj o WWjj : WWjj -> tau tau jj -> cross section is small o ttbar -> l tau jj last talk of this session : WWbb ->tau tau jj Fake of W->lepton(tau) is also suppress with W transverse Mass cut Z->tt background o Z+njets : Zjj -> tau(l)tau(h)jj Final state is the same as the signal So we can only use “central jet veto” for suppress the BG Introduction(2) … ATLAS ATLAS検出器 LAr HEC Tile HD LAr FCal 飛跡検出器 Pixel シリコンピクセル検出器 SCT シリコンマイクロストリップ検出器 TRT 遷移輻射ストローチューブチェンバー LAr EM Muon System カロリーメータ LAr EM 液体アルゴンカロリーメータ (Pb) Tile HD シンチレータタイルカロリメータ LAr HECハドロンエンドキャップ(Cu) LAr FCal 超前方カロリーメータ (Pb) Muon 検出器 MDTドリフトチューブチェンバー TGC,RPC ,CSCトリガー用チェンバー
© Copyright 2024 ExpyDoc