View
214
Download
0
Category
Preview:
Citation preview
Scalar Events Synchronization
2nd Farm Replay Summary
Jin HuangM.I.T.
For Transversity Analysis MeetingJun 18, 2009 @ JLab
&
Jin Huang <jinhuang@mit.edu> 2
• Why do synchronization?• Technical : One way to impalement - TSelector • Application • Beam Trip Cut• BCM Stability Check w/ Beam Trip Cut• Dead Time w/ Beam Trip Cut
Synchronizing Scalar Insertion
• Coin Trigger: LT/LT Asym with/without beam trip cuts• Other Trigger
Dead Time w/ Beam Trip Cut
• Farm Production Summary• Coincident Timing Update
Other
Transversity Analysis Meeting
Out Line
Jin Huang <jinhuang@mit.edu> 3
Asymmetry Calculation requires two parts of information:◦ L1A Events Counts (with PID and Kinematics Cut)◦ Scalar Sums (bcm … )
If part of data should be cut away (ex. Beam trip), then Events & Scalar should be cut for exactly same period
Transversity Analysis Meeting
Why scalar synchronization?
Jin Huang <jinhuang@mit.edu> 4
3 Forms of scalar data insertion 1. Each Event based (DL.*)
automatically synchronized to L1A events Limited information
major trigger gated/ungated bcm u/d3 clocks
2. Every 100 event based Full data set Non-syncronized to each event
3. Every 2s based Full data set Slower comparing with #2
Transversity Analysis Meeting
Why scalar synchronization?
Jin Huang <jinhuang@mit.edu> 5
Upgrading my old beam trip code producing raw asymmetry to do scalar synchronizationAutomatically generated class base using T->MakeSelector()Automatically loop through whole root file, flexible process: full c++ supportFastest way to go through a root file (backbone for Draw & Scan)Require lots of coding with carefulnessNot for doing simple analysis
Transversity Analysis Meeting
Technical : One way to impalement - TSelector
Jin Huang <jinhuang@mit.edu> 6
~10% of our data is taken during beam trip◦ Fast Drop; Slower climb◦ Even slower PID lock back to equilibrium current
Transversity Analysis Meeting
Application/ Anatomy Beam Trip
Trip
Ram p
Norm
Norm
Lock
Example of an aggressive Trip cutCurrent Calculated Each Scalar Insert
Before Cut
T (min)
Cu
rren
t (u
A)
Scalar Insertion
s
Jin Huang <jinhuang@mit.edu> 7Transversity Analysis Meeting
Application/ Beam Trip Cut
Although there are Bad Days
Average Runs
Jin Huang <jinhuang@mit.edu> 8Transversity Analysis Meeting
Average Beam Current w/Beam Cut
Jin Huang <jinhuang@mit.edu> 9Transversity Analysis Meeting
BCM u3 vs d3 Diff w/Beam Cut
This Drop persists
This fluctuation Persists
Jin Huang <jinhuang@mit.edu> 10
Difference change for different Spin Drift with time?
Transversity Analysis Meeting
Zoom in the fluctuation part
Jin Huang <jinhuang@mit.edu> 11
Dead time asymmetry is addable correction to physical asymmetry ◦ Comparing with polarization asymmetry, which is
multiplied to physical asymmetry, dead time asym is more important
Dead time change from different beam condition
DAQ LiveTime = 1-DeadTime = NEvents/(Ntrigger/Prescale)
Studied with ~70% of our production data
Transversity Analysis Meeting
Dead Time Study w/beam cuts
Jin Huang <jinhuang@mit.edu> 12
Live time -> 84~90% after beam cuts
Transversity Analysis Meeting
DAQ Live Time
Before Cut
After Cut
Jin Huang <jinhuang@mit.edu> 13
Live time is higher during trip and ramp
Transversity Analysis Meeting
DAQ Live Time
Trip Part
Ramp Part
Not very high due to aggressive beam cuts
Jin Huang <jinhuang@mit.edu> 14
Seems stable
Transversity Analysis Meeting
DAQ Live Time Asym, w/trip cut
SSA
DSA
Jin Huang <jinhuang@mit.edu> 15Transversity Analysis Meeting
DAQ Live Time Asym, w/trip cut
Sum 600runs, average ~100ppm
Jin Huang <jinhuang@mit.edu> 16Transversity Analysis Meeting
DAQ Live Time Asym, w/o trip cut
On same level, Fluctuation is larger
Jin Huang <jinhuang@mit.edu> 17Transversity Analysis Meeting
Also for other triggers
T6
T3
Jin Huang <jinhuang@mit.edu> 18
Coin Trigger DT is 84%~90% Beam cut improve stability of live time by a
factor of ~2 Overall SSA LT Asym is ~100ppm (small) Single Run SSA LT Asym ~0.1% DT Correction half-pair by half-pair is doable
but not necessary
Transversity Analysis Meeting
DAQ Live Time/Summary
Jin Huang <jinhuang@mit.edu> 19
the majority of our data are replayed LHRS TOF calibration from Chiranjib (time
walk is not included) final BB MWDC calibration from Xin Scalar variable updates from Kalyan BB Gas Cer variables are added into DL.*
following requests from Joe Target Spin Azimuthal Angle theta_s, phi_s
are calculated
Transversity Analysis Meeting
2nd Farm Production
Jin Huang <jinhuang@mit.edu> 20
600GB increase in disk space/cleaning of old files -> More free space◦ ~670 (100%) 1M event top quality production runs
from my list. ◦ ~230 (50%) flag-15 production runs from database It
took 48hour to replay on farm with a parallel
processing of 60 runs at the same time. Root files were stored at
◦ Temporary Data Disk Directory: /w/halla/transversity/disk3/ProductionROOTfiles
◦ Permanent MSS Tape Directory: /mss/halla/e06010/ProdReplay2
Transversity Analysis Meeting
2nd Farm Production
Jin Huang <jinhuang@mit.edu> 21
Y axis is number of events passing a “have track” cuts
Transversity Analysis Meeting
Run Distribution
Jin Huang <jinhuang@mit.edu> 22
Pion resolution drop from ~400ps -> ~360ps
Limit by BB side (~300- ps)
Transversity Analysis Meeting
Coin Time after LHRS Calibration w/o time walk yet
p
K+
Pi+
Recommended