Tool to recover electrons misidentified as photons

advertisement
Tool to recover electrons
misidentified as photons in the
AOD:
EMPhotonToElectron
Fany Dudziak
RD. Schaffer, L. Iconomidou Fayard
Egamma phone meeting November 17 2008
1
Overview
• The tool
Reminder: a fraction of electrons are found in photon
container (in part due to conversion reco) with tracks
in front. Try to recover them
• Test of different trackMatch (standard, new
Thomas’one, our).
• Impact on efficiencies and rejections
• Improvements to come.
2
Reminder : losts og efficiency at low Pt
Loss at low Pt : due to Bremsstrahlung :
- electron is turning more
- P more difficult to estimate
Ideal
Track far from the cluster
 cluster is getting wider
 Track lost when electron radiates too much
 some electrons are reconstructed as photons.
3
EMPhotonToElectron Tool
•
In egammaTools, to be used on AOD:
– converts a photon to an electron if the EMtrackMatch tool finds a track for
the cluster.
– it reclusterizes and recalibrates the particles’cluster as electron (following
Mark Cooke’s Twiki : Instructions for Re-Clustering from the AOD)
– runs
• EMShowerBuilder : creates the shower link
• EMBremsstrahlungBuilder : build brem’s variables
• EMFourMomBuilder : build 4-momentum variables
• EMPIDBuilder : computes electron isEM flags
– Can be « driven » by an Alg and run in ESDtoAOD or AODtoAOD
– #include ‘’ egammaInterfaces/IEMPhotonToElectron.h’’
virtual StatusCode execute(const Analysis::Photon* phot,
Analysis::Electron*& newelec,
const CaloClusterContainer* clusTES)
Returns the StatusCode FAILURE if no track has been found.
4
Impact on efficiency and rejection:
Analysis :
For H4l sample : AOD.025585 130 GeV
matching between e truth and e reco within dR<0.2 :
- first with electron author 1
- second (when no e reco found) with photons, no author asked
- the tool runs on all photons.
-> calculate efficiency w.r.t the truth
For rejection : jf17 sample AOD.105802 (only 50000 events  )
use Oleg Fedine’s Tool egammaMCTruthClassifier to find the
type of the reconstructed e
-> choose the ones with electronType == 4 ie « NonElectron »
-> calculate the fraction of remaining fakes after different
isem cuts
 Need to have more jets!! And minbias
5
Impact on efficiency and rejection:
Tool + Classic EMTrackMatch
• Impact on the efficiency using the standard EMTrackMatch:
Efficiency
Before
After tool
matching
86.5%
-> 93.4%
Isem loose
84.6%
Isem
medium
Isem tight
Rejection
Before
After
-> 91.2%
13.4 %
13.4%
75.1%
-> 78.4%
2.9%
2.8%
64.8%
-> 66.4%
- Not enough - Not enough
(~0.1%)
(~0.1%)
6
Impact on efficiency and rejection:
Tool + Thomas’ EMTrackMatch
•Impact on the efficiency using Thomas’ EMTrackMatch:
-> select preferentialy nonTRT tracks + non E/P cut
Efficiency
Before
After tool
matching
86.5%
-> 93.4%
Isem loose
84.6%
Isem
medium
Isem tight
Rejection
Before
After
-> 91.2%
13.4 %
14.0%
75.1%
-> 79.3%
2.9%
3.17%
64.8%
-> 66.8%
- Not enough
(~0.1%)
idem
7
Impact on efficiency and rejection:
Tool + RD’s TrackMatch (new Δφ)
• Impact on the efficiency using the RD’s EMTrackMatch:
-> new Δφ + Thomas’ emprovements
Efficiency
Before
After tool
matching
86.5%
-> 95.6%
Isem loose
84.6%
Isem
medium
Isem tight
Rejection
Before
After
-> 93.3%
13.4 %
15.1%
75.1%
-> 80.5%
2.9%
3.43%
64.8%
-> 66.8%
- Not enough
(~0.1%)
Idem
8
To do
• Run on more jets (jf17 & minbias)
• Study the contribution of Softe Electrons
• New author for the output electrons?
• Make a TWiki to explain how to use the tool.
9
Download