Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Slide 23Slide 24Slide 25Slide 26Slide 27Slide 288.882 LHC PhysicsExperimental Methods and MeasurementsEfficiency and Acceptance[Lecture 15, April 1, 2009]C.Paus, LHC Physics: Efficiency and Acceptance 2Organization Project 1●completed – in process of reading and correcting, few comments so far == very good Project 2●due April 9 (the following week Thursday) Project 3●instructions are complete●due May 2 Project 4 and Conference Session●they are considered the final, contents to be definedC.Paus, LHC Physics: Efficiency and Acceptance 3Organization Our little conference●one student one presentation Proposed rough program●“Overview – The LHC Project and Status”●“Interesting Physics at the LHC”●“A Charge Multiplicity Measurement”●“Measurement of the Upsilon Cross Section”●“Measurement of the B lifetime”●“Standard Model Higgs Search: H→ZZ*”●“Standard Model Higgs Search: H→WW*”●“Standard Model Higgs Search: H→ττ”●“Standard Model Higgs Search: H→γγ”●....‘09A Physics and Chemistry joint ColloquiumThursday, April 2 at 4:15 pm in room 10-250George WhitesidesHarvard University "Problems at the Interface between Physics, Chemistry, and Energy"SpringFor a full listing of this semester’s colloquia, please visit our website at web.mit.edu/physicsColloquium SeriesPhysicsC.Paus, LHC Physics: Efficiency and Acceptance 5Lecture Outline Efficiency and Acceptance●introduction●details about the Upsilon data●how where they triggered?●is all data good? goodrun lists!●details of the Upsilon Monte Carlo sample●rough generator description●decaying Upsilons according to phase space●how to derive a relative and absolute efficiency?●some systematic uncertaintiesC.Paus, LHC Physics: Efficiency and Acceptance 6Introduction Acceptance●refers to purely geometric fiducial volume of the detector Efficiency●refers to purely detector effectiveness in finding objects which have passed through the detector In practice: ambiguous●inside acceptance●one leg mostly out of acceptance: efficiency will matter●both legs outside of acceptanceC.Paus, LHC Physics: Efficiency and Acceptance 7Introduction Cross section analysis●cross section is given by Ingredients of the analysis●L – integrated luminosity (provided to you)●Nobserved – various methods exist (usually straight forward)●simple sideband subtraction, binned χ2 or unbinned likelihood fits●a – acceptance from the Monte Carlo●not clear how to get this without storing every event●also must be able to carefully calculate fiducial volume per muon●ε – efficiency again from Monte Carlo●only possible to quote separately if acceptance known●it makes sense to combine a and ε into one number●often people refer to efficiency as the product: a εC.Paus, LHC Physics: Efficiency and Acceptance 8Introduction Our efficiency (a ε) can be subdivided into●trigger efficiency●level1●level2●level3●ntuple●reconstruction●pre-selection efficiency●your analysis●reconstruction and final selection efficiency●deal with efficiencies of trigger and your analysis only●ntuple related efficiency is implicitly taken care of: apply harder analysis requirementsC.Paus, LHC Physics: Efficiency and Acceptance 9CDF Muon Detection System Muon detection starts at the muon chambers CMU●on HCAL●|η| < 0.6 CMP●add steel●|η| < 0.6 CMX●0.6<|η|<1.0 IMU●1.0<|η|<1.5●no triggerCMUCMPCMXC.Paus, LHC Physics: Efficiency and Acceptance 10Trigger Essentials Trigger tables●every event has to follow one or more exactly defined sequences through the level-1/level-2/level-3 system●avoids all volunteers Volunteer (ex. our upsilon sample)●level1 requires CMUP muon●level3 requires CMUP muon, higher quality data●some CMUP muon identified at level3 but not at level1●exact defined path avoids events without level1 CMUP ●if other level1 triggers (ex. track trigger) are considered additional events can show up, efficiency for those events is very difficult to determineC.Paus, LHC Physics: Efficiency and Acceptance 11Trigger Essentials Deadtime●full detector read out takes a finite amount of time●this time is larger then time between beam crossings●this is also true for a pipelined trigger system which is called 'deadtimeless'●during this time no additional events can be accepted●this time is called deadtime●if accept rate too high deadtime can seriously affect data taking: every event receives the same deadtime●rule of thumb: deadtime should be kept well below 10%C.Paus, LHC Physics: Efficiency and Acceptance 12Trigger Essentials Prescales in CDF jargon●too avoid too high accept rates certain triggers get prescaled: this means accepted events get rejected, at a given scale: the prescale●prescale of 2 means: only every second event passing all trigger conditions gets accepted●can be applied at all trigger level (usually level1, level2)●simple prescale (PS): a fixed scale is applied throughout the data taking period to reject events●dynamic prescale (DPS): the value of the prescale gets dynamically adjusted throughout the data taking period●inst. luminosity decreases more bandwidth is available●on a macroscopic timescale bandwidth is saturated●fully reproducible because average prescale per run can be calculatedC.Paus, LHC Physics: Efficiency and Acceptance 13Trigger Essentials Prescales in CDF jargon, continued●űber prescale (UPS): saturates the bandwidth at a microscopic level●in CDF this is done at the first trigger level●level2 trigger has four buffers●on average they are mostly full when running at a given rate with a given trigger table●at the microscopic level (396 ns, beam crossing) there must be instances where more then 1 buffer is free, even up to 4 can be free●űber prescale monitors activity in the buffers and will fill the buffers if there are free slots●problem: it is not possible to determine the effective scale anymore●trigger path with UPS needs to be separated so some analysis can ignore these events (ex. cross section analysis cannot use UPS)C.Paus, LHC Physics: Efficiency and Acceptance 14Trigger Essentials Access to trigger data and Monte Carlo in ntuple●module: TPrereqFast●specify names