Electron and Photon HLT alley M. Witek K. Senderowska, A. Żurański

Calorimeter meeting - M. Witek2 Ecal Alley - Introduction  Two lines ready: γ -line and e-line γ -line  B s →φγ optimized using offline selection by Lesya e-lines  B + → K + e + e - offline selection for trigger (Hans)  B s →J/ψ(ee)φ lifetime unbiased selection (Adlene) π 0 -line  Work not started yet

Calorimeter meeting - M. Witek3 L0 decision statistics  L0 „crosstalk” Signal photons do not always produce photon L0 candidate (same for electrons and π 0 )  Specific feature of Ecal Alley. Need to be taken into account in HLT Example: start photon line for L0Photons and L0Electrons Bs→φγBs→φγB s →J/ψ(ee)φ e γ π 0 γ e π 0 e γ not e π 0 not e, not γ γ e not γ π 0 not γ, not e L0 Candidates from signal that trigger

Calorimeter meeting - M. Witek4 γ - line

Calorimeter meeting - M. Witek5 γ -line  B s →φγ φ and γ are two independent elements:  High p T γ (off-line cut p T >2.8 GeV)  φ→KK, kaons correlated, relatively low p T  Strategy: 1. Start line if L0DecPhoton or L0DecElectron L0 electron: anti-confirmation (no track in T) → proceed as L0 Photons L0 photon: confirmation (π 0 – merged removal) 2. Find 2 high IP tracks with a bit of p T

Calorimeter meeting - M. Witek6 L0 photon – π 0 merged removal  85 % of L0 photons that trigger min bias are due to π 0  Check quality of L0 photon – remove π 0 to reduce mbias rate Inspired by method of Sergey Barsuk – „ γ/π 0 separation at high E T ”   ECAL reconstruction needed to calculate shower shape variables. At this early stage of HLT1 it is has to be a fast version of Calo reconstruction. Provided by Victor Egorychev   Fast reconstruction in the area around a given seed cell  Treat 3 ECAL regions separately (different cell size) γ from signal π 0 merged selected from min bias events (two photons contributing to one cluster)

Calorimeter meeting - M. Witek7 L0 photon – shower shape variables γ shape γ asymγ kappa π 0 shape π 0 kappaπ 0 asym

Calorimeter meeting - M. Witek8 L0 photon – π 0 merged removal Signal eff = 90 % and 50 % reduction of mbias events Cut here Input variables: 1.E γ 2.Shape 3.Tails/Core 4.Asymmetry 5.Kappa

Calorimeter meeting - M. Witek9 Hadron part - φ→KK 3D IP 2D IP Correlation in PT distributions of kaons Large fraction – both PT<1500 MeV Loss in 2D => 3D HLT phase of VELO track reconstruction ET1 vs ET2 IP1 vs IP2 ET vs IP 2D 3D IP Not available at 1 MHz

Calorimeter meeting - M. Witek10 Electron lines

Calorimeter meeting - M. Witek11 Electron lines – standard selection  Two lines (correlated) Single electron Di-electron  B + → K + e + e -. Focus on e + e -, IP cut used.  Both starting from L0 Electron e + vs e - distributions for L0Electron = 1 L0: p T >2.6 GeV IP 2D loss can be recovered if L0 e from signal

Calorimeter meeting - M. Witek12 Structure of electron lines L0 electron Second electron 1. T confirmation 2. Velo – T matching (2D followed by 3D) 3. IP > 0.15 mm PT>3 GeV: single electron decision 4. Velo 3D 5. e+e- vertex 6. Forward reco 7. Calo confirmation- fast reco Optional (sufficient MB reduction without this step) PT>1 GeV: di-electron decision

Calorimeter meeting - M. Witek13 Di-electron line – lifetime unbiased selection  B s →J/ψ(ee)φ  Difference wrt standard selection No cut on IP Invariant mass of e + e - around J/ψ mass Adjust PT cut φ not used → essentialy J/ψ selection  Electron momentum correction for radiation used offline Same needed in HLT First implementation is ready

Calorimeter meeting - M. Witek14 ECAL alley performance  Tuning still going on (feedback Hlt1 – off-line selections) Photon line  working point seems to be close to optimal  adding L0Elelectrons crutial to get good overall L0*HLT1 efficiency Electron lines  standard selection (IP cut)  Lifetime unbiased selection to be improved HLT1 decisionSignal efficiencyMin bias rate γ - line 75 %~5 kHz Single electron70 %~2 kHz Di electron - standard78 %~3 kHz Di electron – no IP~60 %~few kHz Example of performance numbers

Calorimeter meeting - M. Witek15 Summary  ECAL alley status. γ -line: close to optimal. e-lines: implemented,  tuning still going on – start with L0 π 0 local,  improve lifetime unbiased selection π 0 - line: work to be started

Calorimeter meeting - M. Witek16 Backup slides

Calorimeter meeting - M. Witek17 L0 photon – π 0 merged removal  Check quality of L0 photon – remove π 0 to reduce mbias rate Inspired by method of Sergey Barsuk – „ γ/π 0 separation at high E T ”  Shower Shape Tails/Core Asymmetry Kappa Variables for π 0 /γ resolution