1. Monitoring SCT Efficiency and Noise
SCT Monitoring Meeting
1st July 2008
Richard Batley
(Cambridge)
SCT efficiency :
do we really need all those Tools ?
SCT noise :
can it really be measured with SpacePoints ?
compare measured and true efficiency and noise in MC

2. SCT Hit Efficiency Algorithm
For every SCT hit fitted to a reconstructed track :
extrapolate track fit to opposite side of module
(using simple intersection of a straight line with a plane)
look for nearest cluster (if any) to extrapolated point
declare opposite side to be “inefficient” if no cluster found within 1mm in local-x
Note :
no extrapolator tools, no track refitting, no HoleSearch
the nearest opposite-side cluster will usually already be included in the track fit

3. Testing the efficiency algorithm
Use single-muon and multihadron (ZZ→ttmm) MC :
simulated with (m) and 12.0.X (ZZ)
digitised and reconstructed with
(SCT_Digitization )
study SCT barrel only for now
True MC inefficiency :
default SCT digitisation has strip-level inefficiency of 1%
(1% of strips above threshold are randomly rejected)
cluster-level inefficiency may be different
(due to multi-strip clusters, secondary particles, …)
define “true inefficiency” as no cluster within 1mm of true track position

4. Distance to nearest cluster on opposite side
In ZZ events :
o/flow includes case where there is no cluster at all on the other side

5. Extrapolated track position for inefficient hits
(no cluster on opposite side of module within 1mm of extrapolated track)
(N.B. without track fit info, this could look like same-side noise)

6. Distance to nearest cluster on opposite side
Require track extrap to be away from edges and bond-gap :
o/flow includes case where there is no cluster at all on the other side

7. Measured and true inefficiencies: single muons
SCT barrel
(all track extrapolations)
(track extrap within active region)
(true inefficiency)
measured efficiency consistent with true value

8. Measured and true inefficiencies: ZZ events
SCT barrel
(track extrap within active region)
(true inefficiency, from single-m events)
measured efficiency consistent with true value

9. Monitoring SCT Noise via Spacepoints
SpacePoint algorithm (in SCT_Monitoring package) :
estimates noise from rate of clusters not in SP’s
doesn’t need reconstructed tracks etc.
(just data from each SCT module alone)
(aims to give approx. noise for monitoring purposes, not final best noise value)
Study using single-m and ZZ→ttmm MC :
compare SP estimate with true noise cluster rate
true noise cluster :
no contribution from any true particle in any strip of cluster
default noise rate per strip in MC is only ~ 5 x 10-6
(same for all module types)
(harder to measure noise in MC than in real data)

10. Width-improved Spacepoint algorithm
Find SP alg doesn’t work for multihadron events :
try to improve using cluster width information
(nstrip >= 2 is a good signal for a track-induced, non-noise cluster)
N.B. “noise” means random, uncorrelated, single-strip noise
Combine width and SP info :
extrapolate into nstrip = 1:
normalise to nstrip = 2 :
count nstrip = 1and nstrip = 2 clusters in and out of SP’s

11. Cluster and noise rates: single m events
all clusters
SP noise alg
true noise clusters
width-improved
short middle
barrel
outer
middle
inner
SP noise alg gives reasonable estimate of true noise
Improved by width info for barrel

12. Cluster and noise rates: ZZ events
all clusters
SP noise alg
true noise clusters
width-improved
short middle
barrel
outer
middle
inner

13. Cluster and noise rates: ZZ events
Same plot as previous slide, with expanded vertical scale :
SP noise alg
barrel
true noise clusters
width-improved
SP noise alg overestimates noise by large factor
Width-improved SP alg gives big improvement in barrel

14. Summary SCT hit efficiency :
can be reliably determined using a simple straight-line extrapolation to opposite side of module
SCT noise :
SP alg badly overestimates noise in busy events
much improved (in barrel) using cluster width info

15. BACKUP SLIDES