TRT Software Activities - 08/14/2009 SPLASH EVENT IN THE TRT I will mainly focus on the activities where the Duke group is more directly involved 1
TRT SW Offline Duke Group heavily involved in several aspects of the TRT Offline since the end of TRT construction/assembly (about 2006) The TRT Offline encompasses a wide range of activities Bytestream converter Rt and T0 calibration Alignment Monitoring and Run Time Tester (RTT) Simulation and Digitalization Conditions and Database Data/MC Validation Trigger... 2
TRT SW Offline Duke Group heavily involved in several aspects of the TRT Offline since the end of TRT construction/assembly (about 2006) The TRT Offline encompasses a wide range of activities Bytestream converter Rt and T0 calibration (Bocci, NBI, Penn) Alignment (Bocci, Kruse, Penn) Monitoring and Run Time Tester (RTT) (Klinkby,Benjamin, Penn, Yale) Simulation and Digitalization (Klinkby, NBI) Conditions and Database (Bocci, Klinkby, Indiana) Data/MC Validation (Duke, All) Trigger Duke Group Direct Involvement... 3
TRT SW Offline The TRT Offline encompasses a wide range of activities Bytestream converter Rt and T0 calibration Alignment Monitoring and Run Time Tester (RTT) Simulation and Digitalization Conditions and Database Data/MC Validation Trigger Other tasks for the Coordinator and the Deputy Supervision of all the activities Provide adequate coverage of all critical area Prioritization of the work Coordination with the ID (ATLAS) SW Management of the code/releases/tags etc.. 4
TRT SW Offline The TRT Offline encompasses a wide range of activities Bytestream converter Rt and T0 calibration Alignment Monitoring and Run Time Tester (RTT) Simulation and Digitalization Conditions and Database Data/MC Validation Trigger Here I will focus only on activities correlated with the TRT commissioning In the last two years ATLAS engaged in a rapid transition from a set of semiindependent sub-detectors to global combined experiment We had to because we were told to be ready for data by Sep. 2008!! Duke group has been a major player to make this transition successful 5
Commissioning in SR1 Commissioning started just after the assembly in surface, in building SR1 Cosmics triggered by scintillators 6
Commissioning in SR1 Commissioning started just after the assembly in surface, in building SR1 Cosmics triggered by scintillators First Performance Analysis with Built Detector Duke (Bocci, Kruse, student) active on studies of efficiency, noise cross talk, alignment, module deformation 7
Efficiency with SR1 See Esben talk for more on efficiency 2 mm ~ 0% (noise) ~ 97% less 8
Efficiency with SR1 2 mm ~ 0% (noise) ~ 97% less 9
Cross-talk in SR1 Cosmics Data MC Cross-talk (correlated noise) from straws belonging to the same HV pad 10
Cross-talk in SR1 Cosmics Data MC Cross-talk (correlated noise) from straws belonging to the same HV pad Similar work on new cosmics data ongoing (Klinkby and CERN summer student) It will be revisited with collision data for electrons for which higher cross-talk is expected 11
Alignment with SR1 data and beyond TRT alignment framework started to be built for the SR1 data taking One of the major Duke contribution to offline (AB co-author of the code) Subsequently expanded to full barrel, endcap, collision data and entire ID First measurement of the misalignment between the SCT and TRT (Level 1) L1 TRT (w.r.t Silicon) 12
Alignment with SR1 data and beyond TRT alignment framework started to be built for the SR1 data taking Subsequently expanded to full barrel, endcap, collision data and entire ID One of the major Duke contribution to offline (AB co-author of the code) First measurement of the misalignment between the SCT and TRT (Level 1) TRT Aligned by module (Level 2) First study for deformation with SR1 data 13
Commissioning with Milestone Cosmics Runs Once in the pit the TRT commissioned as long as the RODs integration proceeded Dec 2007 - M4 14
Commissioning with Milestone Cosmics Runs Once in the pit the TRT commissioned as long as the RODs integration proceeded Mar 2008 - M6 15
Commissioning with Milestone Cosmics Runs Once in the pit the TRT commissioned as long as the RODs integration proceeded Aug 2008 - M8 16
Commissioning with Milestone Cosmics Runs Once in the pit the TRT commissioned as long as the RODs integration proceeded Mar 2008 - M6 17
Commissioning with Milestone Cosmics Runs Once in the pit the TRT commissioned as long as the RODs integration proceeded Mar 2008 - M6 Alignment constants from M6 runs was taken as baseline for the first collision data in Sep. 2008 (sic!) 18
Alignment with CSC MC Physics events were simulated with a misaligned geometry to emulate an as built detector. For the alignment group the challenge was to restore the perfect geometry 19
Alignment with CSC MC Physics events were simulated with a misaligned geometry to emulate an as built detector. For the alignment group the challenge was to restore the perfect geometry Movement of the order of order of mm (mainly radial) Before alignment 20
Alignment with CSC MC Physics events were simulated with a misaligned geometry to emulate an as built detector. For the alignment group the challenge was to restore the perfect geometry Very challenging but huge improvement in the development of alignment techniques Movement of the order of order of mm (mainly radial) Before alignment After alignment 21
Alignment with FDR Final Dress Rehearsal exercised the software of the complete ATLAS data taking chain, from the trigger sub farm output (SFO) to the physics analysis at Tier 2 22
Alignment with FDR Final Dress Rehearsal exercised the software of the complete ATLAS data taking chain, from the trigger sub farm output (SFO) to the physics analysis at Tier 2 For the alignment (and calibration) it meant to exercise the 24 hours operation scheme 23
Alignment with FDR For the alignment (and calibration) it meant to exercise the 24 hours operation scheme 3 days of data taking with event simulated with the CSC geometry 24
Alignment with FDR For the alignment (and calibration) it meant to exercise the 24 hours operation scheme 3 days of data taking with event simulated with the CSC geometry Bocci coordinated the Inner Detector alignment and the TRT calibration part 25
Condition Database We (Bocci, Klinkby) involved also in the definition and maintenance of the condition database that includes, besides alignment and calibration constants, dead/hot straw list Conditions are determined in the 24-hour scheme before the data are sent to Tiers 1 26
Condition Database We (Bocci, Klinkby) involved also in the definition and maintenance of the condition database that includes, besides alignment and calibration constants, dead/hot straw list Conditions are determined in the 24-hour scheme before the data are sent to Tier 1's Permanently dead (unrecoverable) straws are kept separated to be masked also in simulation 27
Combined Cosmic Runs After Sep 19 accident, ATLAS undertook two major combined data taken periods 2009 Runs 2008 Runs First time with solenoid ON and Xe gas mixture for the TRT 28
Combined Cosmic Runs After Sep 19 accident, ATLAS undertook two major combined data taken periods 29
Cosmic Runs 30
Conclusions Duke group involvement in the offline operation of the TRT has been very strong from the beginning Leadership roles in recognition of our expertise 31
Conclusions Duke group involvement in the offline operation of the TRT has been very strong from the beginning Leadership roles in recognition of our expertise TRT offline is in excellent shape to cope with the challenge of an efficient data taking under the very demanding LHC conditions Still a lot to understand on the performance though 32
Conclusions Duke group involvement in the offline operation of the TRT has been very strong from the beginning TRT offline is in excellent shape to cope with the challenge of an efficient data taking under the very demanding LHC conditions Leadership roles in recognition of our expertise Still a lot to understand on the performance though We plan to continue to play an important role for the TRT commissioning with collision data Continuous operation, high data volume, commission/performance with electrons, etc... 33
Conclusions Duke group involvement in the offline operation of the TRT has been very strong from the beginning TRT offline is in excellent shape to cope with the challenge of an efficient data taking under the very demanding LHC conditions Still a lot to understand on the performance though We plan to continue to play an important role for the TRT commissioning with collision data Leadership roles in recognition of our expertise Continuous operation, high data volume, commission/performance with electrons, etc... Our expertise with the TRT detector may be exploited to study physics signature where the TRT can provide extra discriminating power Double charged particle search (TR goes as q4 ) SUSY particle with low β (slow particle leaves longer signal in TRT) 34
First Observation of Electrons 35
Backup 36
Residuals RESIDUAL: Distance between the measured and the extrapolated hit from the track fit Misalignment of a detector element translates in a systematic shift of the residual mean distribution This is what the alignment business is: Minimize: 37
First Elecrons 38
Residual Is Not All Except for big structure misalignment (L1), residuals does not change so much This is an example of TRT L2 alignment YES IT IS!! 39
Weak Modes Weak Modes: Detector configurations that are poorly constrained by tracks Alignment errors (stat. and/or syst.) not negligible w.r.t intrinsic resolution Consequences: poor convergence, biases, etc.. How we identify them? How we deal with them? How will affect the physics? This is the real alignment problem (Don't expect it to be solved anytime soon...) Step 1: Recognize the enemy 40
Weak Modes (2) Step 2: Assess the damages Created four global systematic ID misaligned geometry and align them 41
Weak Modes (2) Step 2: Assess the damages See the effects on physics after the alignment (if converged) Example: Curl Mode Momentum Bias After Alignment (Blue) 42
First Data Misalignment (and its impact on physics) How much the (residual) misalignment will affect my analysis at the beginning? Of course we cannot answer this question before looking at collision data But we can make a educated guess using the cosmics experience Method: introduce random misalignment to match the residuals width with cosmics Assumption 1: Sensitivity of collision data same as of cosmics (conservative) Assumption 2: Only random displacement (optimistic) Two set of constants: Day 1 and Day 100 43
First Data Misalignment (and its impact on physics) Impact on Z reconstruction 44