1. Fixing Tau HLT (Part 1.5/2) ‏ M.Bachtis

2. 2 L1 Seeding Fix L1: Seeding with High Et Jet paths to increase efficiency with High Et Still using corrected Thresholds (but after tunig them to uncorrected the HLT input will be more or less the same)‏ DoubleTau (was DoubleTauJet40): DoubleTauJet_40 OR DoubleTauJet100 SingleTau (was SingleTauJet80)‏ SingleTauJet_80 OR SingleTauJet150 Tau+MET(was L1SingleTau30_MET30)‏ SingleTauJet30_MET30 OR SingleTauJet150 Muon+Tau(was L1_Mu5_TauJet20)‏ L1_Mu5_Jet20 Will not be useful for me but at least it should be fixed if it exists on the menu Need to fix this

3. 3 L2 Operation Point Threshold Tracking studies show that we cannot reconstruct leading Track for a τ Jet of Et<20(next slides)‏ QCD also dominates under 20. Let’s pick 15 GeV for X Channels, 20GeV for DoubleTau Use Cluster ΔR RMS It goes up with Et without relaxing it It is a low pass filter so cuts a lot of the low QCD Also use ECAL Isolation Et Continues to cut over 50GeV where Cluster ΔR RMS vanishes Make it relax with Et over 100-150 GeV

4. 4 L2 Cuts QCD Z  ττ QCD Z  ττ QCD Z  ττ Jet Et>15 ECALIsolEt<5+0.013Et+0.000052Et 2 Cluster ΔR RMS<0.015 ALL 43.6% of QCD Passes L2

5. 5 Tracking L2.5(Lead track RECO)‏ Reconstruct the pixel Seed Silicon Tracks with Pixel Tracks in a small cone 0.2 around the jet center (at vertex)‏ Find the Leading Track with a matching cone of 0.15 (was 0.1) and lead Track pt >5 GeV L3(Tracker isolation)‏ Reconstruct the rest of the tracks in a cone of 0.5 around the jet Apply new isolation Cut : Calculate the pt weighted direction in the jet Calculate the track Pt weighted ΔR RMS around it Similar to L2

6. 6 Leading Track Finding Cuts QCD Z  ττ QCD Z  ττ

7. 7 Isolation Cuts OLD:#Tracks In Annulus(75% of QCD In the first bin)‏ NEW:#Track Delta R RMS (Reduces QCD by a factor of 2) QCD Z  ττ QCD Z  ττ

8. 8 Tracking Performance Isolation:TrkDRRMS <0.002 Matching Cone=0.15, lead Track Pt>5 GeV 24% of QCD Passes L2

9. 9 Summary and next(final) step Total QCD Efficiency per candidate = 11% Fo the double tau that has a rate problem the efficiency is squared Signal : 67% Per candidate with reference to MC Lead Track Finding costs Proposed improvement (need tracking experts)‏ Do reverse tracking: Start from the calorimeter and propagate towards the beam pipe... Needs unpacking of only the outer clusters etc... Next steps: Rate+Time wise optimization Will check rates and timing and if possible make the tracking cuts more loose Just managed to run OpenHLT