Transcript Noise Rate Tests - Indico

1
M2-M5 Efficiency
and Timing checks on 7TeV beam data
Alessia, Roberta
R.Santacesaria, April 23rd , 2010- Muon Operation meeting@CERN
2
Efficiency 7TeV data for M2-M5
Method reminder:
- MuonTrackRec standalone reconstruction using 4 stations,
- Look for Clusters in the 5th station within 6s around the
prediction, no time cut
- Assume the found cluster multiplicity distribution is due to
efficiency + poissonian background
- Fit with this assumption
Data samples
TAE
noTAE
HV=2650V
~ 1Mevts
~17Mevts
HV optimized
~ 1Mevts
~ 6Mevts
3
Applied cuts
General cuts:
- require that at least one Tracker track extrapolated to M1 matches
within 1s the M1 cluster belonging to the Muon standalone track
- P(track)>8 GeV
- to discard fake tracks due to combinatorial (in R1 essentially) ,
require that the best matched Tracker track to Muon track on M1,
extrapolated to M5, matches the M5 cluster within 1s.
- for the analysis of M5 P(track)>15GeV, match of the Tracker track
on M3 and M4
Fiducial volume cuts
- 6s fiducial volume cut in the inner border of R1
- 6s fiducial volume cut in the left-right borders of M4R4
- In 2 regions a clear inefficient zone is identified and discarded
to compute the “ pure” efficiency value
Results
4
M2
HV=2650, noTAE
HV Optim, noTAE
M3

M4
M5 Region
FV cuts applied on R1 and M4R4,
bad zones of M3R2 and M4R2 excluded
Statistical errors only, ~0.1 systematic error must be included, due to bg subtraction
5
TAE events at 2650V and effect of bad zones
M2
M3
M4
M5
Region
HV=2650V, TAE FV cuts, bad zones excluded
HV=2650V, TAE bad zones included  ~1% is lost on M3R2, 0.2% on M4R2
6
M3R2 inefficient chamber 16A3 A side
With more statistics clearly visible that 1 FEB is inefficient on the Cathods
in 1/8 of the chamber Estimated efficiency 50% on those cathods
7
M3R2 16A3 A side
Also clearly visible by channel statistics
Beam 2010
TED 2010
Efficiency ~ 50% since a long time
8
M4R2
Low efficiency due to two dead channels
9
Q4M4R2H1
2 Dead channels
10
M4R4
Some FV cuts needed also on M4R4 right-left sides to reject fake tracks
due to hits on left-right borders in M2 and M5 (calorimeters not perfect shielding)
11
M2 M5 X illumination on R4
M2R4
M5R4
X(mm)
Some tracks are faked on M4 with a random noise on M3 and M1(easy)
12
R1 – Effect of Cleaning cuts and FV cuts
M2
M3
M4
M5
Region
HV=optimal, noTAE  cleaning cuts
HV=optimal, noTAE  no cleaning cuts ~1.5% lost on M3R1, ~0.5% on M4R1
HV=optimal, noTAE  no FV cuts ~1.5% lost on M3R1, ~0.5% on M4R1
13
Multiplicities at HV_optimal.vs.2650V settings
M1
M2
M3
M4
M5
Region
Ratio of the average number of Pads (red) and Clusters (blue)
HV_optimal/2650V for noTAE events (time centering for HV_optimal)
~ larger DV corresponds to lower ratio (M2R1,M3R1),
also Clusters change, even if less than Pads
14
Check on timing: FEB timing in inner regions
FEB average time on M2-M5 R1 regions
M3R1
M2R1
4ns
M4R1
M5R1
Inner regions seems to be well aligned, though there are some FEB’s 3-4
bins apart. To be better studied or….wait for huge statistics and align per
channel? It is better to have more statistics anyway
15
Conclusions
-Absolute efficiency everywhere > 99% in 25ns in M2-M5
- Some FV cuts are needed on R1 to correct for limited precision of the
prediction near the beampipe hole
- At 7 TeV some cuts are necessary to reject “combinatorial” tracks
expecially in R1.
- Worth while to investigate the reason for low efficiency zones in M3R2 and
M4R2, they are there since a long time
- Timing needs some touching up, though efficiency is good enough.
we have to decide whether go for channel-wise correction as soon as
statistics available, or do one or more intermediate steps.
To do a refined analysis (cross-talk, choice of the hit to be used, clean
tracks…) O(50Mevts) events needed with a stable setting.
16
Spare slides from the old presentations
17
Efficiency calculation
Number of clusters found within the defined windows in M4
R1
R3
R2
R4
18
Reminder : efficiency calculation on M2-M5
From each histogram of the #Clusters within the tolerance, the efficiency e
and the background probability Pbg is extracted by fitting the following function:
M4
k=# of clusters
If k>0
SUM [(1-e) Pbgk e -Pbg / k! + e Pbgk-1 e -Pbg / k-1!]
If k=0
SUM[ e -Pbg (1-e) ]
Where
SUM = Sum of the entries
e = Efficiency
P bg= Background probability
In this way the event-correlated background is taken into account automatically
since it is estimated in the vicinity of the track