IFR Simulation Status

1
IFR Simulation Status
Thomas Moore University of Massachusetts,
Amherst July 12, 2004

• Data/MC Comparisons
• SP6 vs Converted SP5
• Current Studies

2
Data/MC Comparison

• Data is taken from BPCMuon skim in R14
• MC is SP-3981 (µµ? kk2f generator)
• Unbiased muon tracks are selected using
  BtamumugammaSample in BetaPidCalib
• Analysis-21 (CM2) in cache mode
• Comparisons of most micro variables (and a few
  more) for all CONDALIAS Runs 1-4 can be found
  athttp//www.slac.stanford.edu/BFROOT/www/Detect
  or/IFR/tbmoore/R14_DataMC/index.html

3
SP6 vs Converted SP5

• In R14, the IFR reconstruction began using a
  Kalman filter to build charged clusters.
• In the converted SP5, the IFR reconstruction
  was re-run using the Kalman filter.
• In SP6, however,the RPC efficiencies are taken
  from data µµ events rather than cosmics (SP5).
• Ive used Moose in 14.4.2a to generate some µµ?
  events with July 2000 conditions for comparison
  with the converted SP5.

4
SP6 vs Converted SP5 (2)
Data Converted SP5 SP6
July 2000 Conditions
5
SP6 vs Converted SP5 (3)
Data Converted SP5 SP6
July 2000 Conditions
6
Strip Multiplicity

• The distribution of number of strips fired per
  hit (1DCluster) is measured in data and applied
  to the MC.
• Distributions are stored in the CondDB per month.
• We are currently updating those measurements.
• Technique (locate quality hits)
• Use the same sample of muons from BetaPidCalib.
• Require the track to hit at least 5 layers.
• Locate hits with no other hits in that view and
  require a hit in the opposite view.
• Histogram the number of strips in the hit for
  each sector/layer/view.

7
Strip Multiplicity(2)
Data SP6 New multi.
July 2000 Conditions
New hits appear to be too large?
8
View Efficiency

• Problem
• The RPC efficiency measurement considers the RPC
  efficient if theres a hit in either view.
• So in the simulation when a track crosses an RPC,
  if random lt eRPC a hit is placed in both views.
• Result is that MC tracks tend to have an even
  number of hits.
• This can affect µ IDthrough strips/layer and
  fit ?2.

Data MC
9
View Efficiency (2)

• Possible Solution
• Measure view efficiency in data (given a streamer
  in the RPC, whats the probability of a signal in
  each view).
• In practice, we can use good µ tracks and measure
  the fraction of hits in each view given a hit in
  the opposite view (for each sector/layer/view).
  These tend to be 98-99 but with a significant
  tail.
• Logically eRPC es (e0 e1) we can solve
  for es.
• Modify the simulation, if ( ran lt es )
  if ( ran0 lt e0 ) put a hit in view 0 if (
  ran1 lt e1 ) put a hit in view 1

10
View Efficiency (3)
Data New multi. New multi.vEff
11
View Efficiency(4)
Data New multi. New multi.vEff
12
View Efficiency(5)
Data New multi. New multi.vEff
veryLoose
veryTight
none
loose
tight
13
Conclusions

• R14 Data/MC comparisons have been produced for
  all conditions aliases.
• We see significant improvement in SP6 over
  converted SP5 (at least for the conditions we
  checked).
• Studies of strip multiplicity and view efficiency
  look promising for SP7.
• More work is needed to understand discrepancy in
  Kalman ?2.