ATLAS Muon Calibration Framework
Manoj Kumar Jha INFN BolognaOn Behalf of ATLAS Muon Calibration
Group 20th October 2010/CHEP 2010, TaipeiATLAS Muon Calibration
FrameworkOutline20/10/2010Manoj Kumar Jha (INFN Bologna) 2ATLAS
muon spectrometerMDT drift tube chambersResolution
requirementsCalibration strategystreamprocessingLatency
ATLAS Muon SpectrometerStandalone Muon Spectrometry pT / pT <
10% up to 1 TeV Coverage || < 2.7 (|| < 860)Layout
Superconducting air coil toroid Trigger chambers: RPC (Barrel) and
TGC (end-cap) Precision chambers: MDT, CSC (inner end-cap)Muon
Stations
20/10/20103Manoj Kumar Jha (INFN Bologna) MDT Drift Tube
ChambersDrift Tube Ar/CO2 (93/7) @ 3 bar absolute Tube = 30 mm and
Wire = 50 m High Voltage = 3080 Gas gain = 2.0 x 1014 Resolution =
80 m MDT Chambers 2 multi-layers with 3 or 4 layers Surface 0.5 to
11 m2 Monitoring of geometry, temperature and magnetic field.
20/10/20104Manoj Kumar Jha (INFN Bologna) 4Resolution
RequirementsFirst Collisions pT < 100 GeV: Resolution dominated
by multiple scattering and energy loss Required rt precision: O(100
m)Design Parameters pT > 100 GeV: Resolution dominated by
spatial resolution. Required rt precision: O(20 m)
20/10/20105Manoj Kumar Jha (INFN Bologna) Calibration Strategy
t0 determinationStarting point of drift time spectrumNeeded for
equalization of time measurement of all the drift tubes~10K hits
per tube are neededIntegration of drift time integrationGives start
rt with precision O(100 m)Auto calibrationFit segments using input
rtDeviation of rdrift from rsegment gives correction
IterateResolution Fit segment omitting one tube (residual) =
residual drift IterateRequired precision 20 m to be done within 24
hrs after data taking.
20/10/20106Manoj Kumar Jha (INFN Bologna) Correction
FunctionsOne rt relation is fitted by chamber. To compensate for
effects, which are inhomogeneous inside one chamber, rt correction
functions are appliedWire-sagBarrel: Bend tubes so that the wire is
always in middleEnd-cap: Apply rt-correctionTemperature Magnetic
field20/10/20107Manoj Kumar Jha (INFN Bologna) Calibration
StreamMuon triggerRPC-chambersmuon trigger level 1momentumfrom
mdt-chambersmuon triggerlevel 2 event filterlevel-3
triggerCalibration centersoutside CERN
(tier-2 subclusters)
Michigan
Munich (MPI+LMU)
Roma & Napelscalibrationstream~100 kHz~1 kHz~100
Hz1/31/31/3
20/10/20108Manoj Kumar Jha (INFN Bologna) Calibration
Processing: Data FlowCalibration datasets from ATLAS Tier0 are sent
to remote calibration center using distributed data management
system.Calibration Farms at remote center process the files as soon
as they are available in local storage element. 20/10/20109Manoj
Kumar Jha (INFN Bologna) Calibration Processing: Calibration center
Calibration centers are established at ATLAS Tier2 with additional
job and data management components to control the calibration
taskDynamic partitioning of computing resourcesReservation of
storage resourcesJob managementEntry point for grid jobs which
needs a digital certificate to operate and are integrated in Batch
Queue Manager.Recently, job submission tool Ganga has been
integrated for job management and bookkeeping Tasks can be
submitted to different backend like Panda and LCGTasks can be
splitted on basis of number of files, events and subjobs.Better
management of task output.For more information on Ganga , see talk
by Johannes Elmeshuer.20/10/201010Manoj Kumar Jha (INFN Bologna)
Calibration Processing: ArchitectureLocal Calibration Data Splitter
(LCDS) permanently watches for incoming data.As soon as the first
data arrive in local storage element , LCDS starts submitting a set
of jobs to calibration batch queue.Monitoring of calibration task
is performed by monitoring client integrated in LCDS.Relevant data
gets store at the end of calibration task and data could be in
principle released and marked for deletion from Tier2 storage.
20/10/201011Manoj Kumar Jha (INFN Bologna) LatencyCalibration
tasks were submitted using job submission tool Ganga.
Number of subjobs corresponding to a dataset depends upon number
of files in the dataset.
Latency is peaked at around 2 hrs. Most of the datasets are
processed in the calibration centers after 2 hrs of data
taking.
20/10/201012Manoj Kumar Jha (INFN Bologna) 12SummaryGanga as job
submission tool has been integrated into LCDS.Remote center used
for calibration of MDT chambers have been setup and their
functionalities have been tested with real data.Latency is peaked
around 2 hrs.ATLAS muon calibration framework can be extended for
calibration to other sub system of detector. 20/10/201013Manoj
Kumar Jha (INFN Bologna) 13
