Pass 6: Event Classes The final chapter Working Definitions TRANSIENT: E > 100 MeV ~ 2 Hz. Maximize Aeff at low energy SOURCE: E> 100 MeV ~ .4 Hz. Maximize Aeff 500 MeV - 10 GeV DIFFUSE: E> 100 MeV ~ .1 Hz. Minimize High energy tail > 10 GeV Will try to achieve a greater differentiation between Classes then was the case in Pass 5 1 Basic Losses CPFGamProb > 0 & CALSeal > 0 Total Thin Thick More optimal CPFGamProb > max(.05, .55- .3* (CTBBestLogEnergy - 1.)) & CALSeal > 0 Total Thin Thick The 2 TKR Topoplogy Filters designed to kill sneak-through conversions, Heavy Ions, and back-side entering range-outs TKRShower > 0. Total Thin Thick TKRShower > 0. & TKRHeavies > 0. Total Thin Thick 2 What the other PreFilters do CALGamProb > 0. Total Thin Thick The large losses in these Filters can Be compensated for by taking the OR TKRGamProb > 0. || CALGamProb > 0 TKRGamProb > 0. Total Thin Thick Total Thin Thick 3 Default CPF Seal and Special Event Filter (TKRGamProb > 0 | CALGamProb > 0) & TKRHeavies > 0. & TKRShower > 0. & CPFGamProb > max(.05, .55- .3* (CTBBestLogEnergy - 1.)) & CALSeal > 0. Default Losses Total Thin Thick This will be the applied to Source & Diffuse The large loss below 100 MeV not appropriate for Transient 4 Combining TKRGamProb with CALGamProb Pass 5 Pass 6 Pass 5 Cut variable: (2.*max(0.,TKRGamProb)/CTBBestLogEnergy + CALGamProb ) / (1 - .08*(CTBBestLogEnergy-1.)) Pass 6 Cut variable : 3.*(TKRGamProb + CALGamProb)/CTBBestLogEnergy 5 ALL Prob This is another way of combining the Shower & Topology Information by way of a CT. This doesn't use the combined CALGamProb & TKRGamProb 6 This CT uses CalTkrComboCut and this is the one that will be used! CTs agree that the correlation between CALGamProb & TKRGamProb is a good discriminator 7 Where the Signal and Backgrounds are in the 2 GamProbs Scan CPFGamProb Increase cut on CPFGamProb CPFGamProb > .2 Leaves 153781 Bkg (10.6 Hz) Scan CALGamProb When events fail the TKR PreFilters (TKRGamProb < 0) set CALGamProb > .2 8 When events fail the CAL PreFilters (CALGamProb < 0) Scan TKRGamProb Set TKRGamProb > .2 When events Pass Both PreFilters (CALGamProb > 0 & TKRGamProb > 0) Set (CALGamProb + TKRGamProb) > 1. 9 When events Fail Both PreFilters (CALGamProb < 0 & TKRGamProb < 0) Nothing to rescue Pass 6 Transient Class CPFGamProb > .2 & CALSeal > 0 & ((TKRGamProb < 0 & CALGamProb > .1) | (TKRGamProb > .1 & CALGamProb < 0) | (TKRGamProb+CALGamProb > .5)) Results: Bkg. Left = 29226 = 2.02 Hz Bkg. Above 100 MeV = 14676 = 1.02 Hz 10 Losses: ~ 25% below 100 MeV Gamma and Background Spectra Total Thin Thick For the next 2 Classes, only the CT based combination of TKR and CAL GamProbs will be used One of the PreFilters Vetoes 11 Source Class Req. Transient Class AllProb Knob Use CT with CalTkrComboCut .4 Hz - AllProb > .10 (!) 12 Source Class Remaining Background: All = 7400 (.51 Hz) E > 100 MeV = 4837 (.33 Hz) Gamma and Background Spectra 13 Diffuse Class (By far the hardest due to high energy residual backgrounds) AllProb > .4 Leaves: All = 1860 (.13 Hz) E > 100 MeV = 1627 (.11 Hz) 14 But…. E > 3000 MeV : 87 event (6.0 mHz) Raising AllProb cut to .55 yields 40 events 26 events have McTkr1DirErr > .01 10 Fail CPFGamProb > .08*CTBBestLogEnergy 6 have (AcdActiveDist3D > 0 & AcdActDistTileEnergy > 1.7) 1 has (AcdActiveDist3D > 0 & AcdActDistTileEnergy > .2 & Tkr1SSDVeto == 0) 15 High Energy Residual Backgrounds Applying these cuts results in AllProb > .4 Leaves: All = 978 (.068 Hz) E > 100 MeV = 934 (.065 Hz) No. Events > 3 GeV: 29 Events (2.0 mHz) AllProb > .4 + Cuts For Pass 5 this was 3 mHz. This needs improvement! Event Class Comparison 16 Summary Pass 6 improves on Pass 5 The basics for optimize Event Class definitions are available Improvements needed to compensate for mis-tracking at high energy 17