1. RD. Schaffer, L. Iconomidou Fayard, D. Fournier
Impact of new trackselection, new track extrapolation on the H4e analysis.
Fany Dudziak
RD. Schaffer, L. Iconomidou Fayard, D. Fournier
Higgs WG meeting March 27, 2009

2. Reminder
H->4l (M=130GeV) suffers from low efficiency coming from the Z*->lowPt electrons. Essentially radiating electrons
The track/cluster matching is sensitive to brems which produce a lower reconstructed track momentum
 E.g from the simple formula:
dphi =8.6 mrad *(50GeV/ptelec)*(rclus/rcalo)
..one sees that a 25GeV Pt electron with E/p=2 will be shifted by 20mrad and will be rejected by the standard Δφ cut (20mrad).
 Important losses of efficiency
for low Pt Electrons due to cut on
Δφ(tr-cal) cut
 Rescaling the momentum and extrapolating from the perigee improves the track/cluster matching conserving these electrons

3. Overview
We compare efficiencies/rejections for electrons in releases (dataset used for studies: H4l130):
« Standard reco »  : i.e. rel 14.2.X
Release 15.0 : changes in track match and priority for electrons vis- a-vis conversions.
Here we will use : “Simulated” Rel 15.0 = “Simulated” Rel starting from Rel14 and WITHOUT E/P cut
Release new track extrap : available as option in rel 15.1
We show impact on the H->4e analysis for a set of MH (120,130,150 and 200 GeV)
Reminder : For CSC note comparison:
- Conversions are kept in Electron Container
- Since CSC, isEM has changed:
E.g. Medium now has Δη<0.005 and calo iso cut in.

4. Datasets, release
Mc08 samples :
H4l
mc PythiaH130zz4l.recon.AOD.e352_s462_r604_tid032239/40
mc PythiaH120zz4l.recon.AOD.e352_s462_r604_tid038862/63/039066
mc PythiaH150zz4l.recon.AOD.e352_s462_r604_tid038858
mc PythiaH150zz4l.recon.AOD.e352_s462_r604_tid038859
mc PythiaH200zz4l.recon.AOD.e352_s462_r604_tid038861
mc PythiaH200zz4l.recon.AOD.e352_s462_r604_tid038860
DiJets minbias : mc filtered_minbias6.recon.AOD.e347_s462_r604_tid
Zz4l mc Pythiazz4l_3MultiLeptonFilterElecMu.recon.AOD.e379_s462_r617
Analysis with
EMPhotonToElectron tool used to make the new electrons with a modified EMTrackMatch tool (in rel 15.1)
egammaMCTruthClassifier used to classify the electrons in jet sample

5. TrackMatch definition VS Releases
Rel x:
Extrapolate the track to calo from the last measurement point
Has losses due to :
E/P cut at <10
Choose the best track in φ => sometime took TRT track instead of Si track
Priority given to conversions => electrons end up in photon container
Rel 15 reco:
Recuperate above losses:
Select preferentially tracks with hits in Pixel and SCT
E/P cut removed (finally not, but removed here for studies)
Conversions ignored for electrons
New trackmatch
As in Rel 15 with removed E/p.
Rescale track momentum and extrapolate from perigee
Recuperates effects of bremsstrahlungs

6. From Rel14=>Simulated Rel15+NewExtrapolation: How to make a new (transient) container of electrons
EMPhoton-
-toElectron +
EMTrack-
Match
photons
!= 2
Electrons
author&0x1
New
electrons
2.8% of e rejected
by trackMatch
19.1% of γ remain γ
rejected
NB :The different TrackMatch algorithm are simulated via our EMPhotonToElectronTool

7. Results on electrons

8. The New deltaPhi trackMatch
Signal e
Fake e
deltaPhi(tr-cal)
deltaPhi(tr-cal)
Rel Δφ : asymmetric + large tails
Rel15+newΔφ : more symmetric + narrower
Rel Δφ : peak centered at 0
Rel15+ New Δφ : peak off-centered and wider because of energy rescaling

9. Rejection vs efficiency
CSCnote
for et>20GeV
CSCMedium : MediumRel14 - CaloIsolation-DeltaEta

10. Sim15.0+new extrapolation
Efficiency %
Electron candidates
loose
medium
medium + Δφ
14.0
88.3 ± 0.2
86.0 ± 0.2
73.5 ± 0.2
69.2 ± 0.2
Sim15.0
96.0 ± 0.1
93.6 ±0.1
80.1 ±0.2
74.2 ±0.2
Sim15.0+new extrapolation
98.7 ± 0.1
96.2 ±0.1
81.8 ±0.2
80.3 ±0.2
+11.1%
+8.3%
Simulated new extrapolation:
Improvement by 8.3% (absolute) vs Rel14 and by 1.7% (absolute) vs Rel 15 for mediumId efficiency
With Medium+Δφ cut: small loss vs Medium
(-1.5% in 15.0+New instead of -5.9% in 15.0) . Allows to improve fake rejection

11. Efficiency -- Rel 14.0 eff =73.5% efficiency
electrons eff =80.1%
new extrapolation eff =81.8%
efficiency
Pt(GeV)
efficiency
Gain more events at low Pt and
at large eta η

12. SimRel 15.0+new extrapolation
Rejection
Min bias sample : Pt>8GeV
Electron fake = electron with electrontype==4 in egammaMCTruthClassifier
Rejection = nb_event *0.31 /(nb_cut * 0.057)
loose
medium
medium + Δφ
Rel 14.0
640 ± 9
3638 ± 119
4705 ± 177
SimRel 15.0
± 11
4018 ± 129
4551 ± 156
SimRel 15.0+new extrapolation
611 ± 8
3817 ± 119
5052 ± 181

13. Results on H4e analysis

14. H->ZZ*->4e 130 GeV +35.3% +5.7% Rel 14 (%) Rel 15 (%)
Rel 15 +new
Extrapolation (%)
4e loose, |η|<2.5
61.5 ±0.7
79.4 ±0.5
87.2 ±0.5
2e w/ Pt>20 GeV
4 w/ Pt>7GeV
47.1 ±0.7
62.7 ±0.7
69.1 ±0.6
mZ, mZ* (*)
18.9 ±0.5
25.0 ±0.6
26.8 ±0.6
|m4e-130|<10 GeV
|m4e-130|<4.4 GeV
(2σ)
18.1 ± 0.5
15.0 ±0.5
23.9 ± 0.5
19.2 ±0.5
25.6 ± 0.5
20.3 ± 0.5
+35.3%
+5.7%
To be compared with 12.5% of
efficiency in csc for H4e but with 2 more cuts (impact parameters, track cone ~>2%)
(*) opposite charges, electron medium, |me1e2-mZ|<15 GeV, mZ*>20 GeV

15. H->ZZ*->4e different masses
Efficiencies for cut on mH ± 2σ for different possible masses of the Higgs
Rel 14 (%)
Rel 15 (%)
Rel 15 +new
Extrapolation (%)
H 120 GeV
10.5 ± 0.4
14 ± 0.5
14.7 ± 0.5
H 130 GeV
15.0 ± 0.5
19.2 ± 0.5
20.3 ±0.5
H 150 GeV
18.3 ± 0.5
23.5 ± 0.6
24.3 ± 0.6
H 200 GeV
27.8 ± 0.7
36.1 ± 0.7
37.8 ± 0.5

16. Summary about Higgs efficiencies
---14
---15 (without E/P cut)
---15+new extrapolation
Gain in efficiency:
14-15
15-15+
14-15+
120 GeV
28.4% 5%
34.9%
130 GeV
22.7%
5.4%
29.3%
150 GeV
27.0%
3.4%
31.4%
200 GeV
28.5%
4.7%
34.5%

17. Background to the H4e signal
ZZ4e : for 130 GeV
Rel 14 (%)
Rel 15 (%) simulée
Rel 15 +new
Extrapolation (%)
4e loose, |η|<2.5
57.4 ± 0.4
78 ± 0.4
85.3 ± 0.3
2e w/ Pt>20 GeV
4 w/ Pt>7GeV
50.0 ± 0.4
69.3 ± 0.4
76.1 ± 0.4
mZ, mZ* (*)
28.1 ± 0.4
38.5 ± 0.4
40.8 ± 0.4
|m4e-200|<7.8 GeV
0.28 ± 0.05
0.63 ±0.07
0.7 ± 0.1
X2 the zz4l background as we increase efficiency for electrons

18. Summary about Higgs efficiencies
Gain in efficiency:
14-15
15-15+
14-15+
120 GeV
33.3% 5%
40%
130 GeV
28%
5.7%
29.3%
150 GeV
28.4%
3.4%
32.8%
200 GeV
29.9%
4.7%
36%
---14
---15
---15+new extrapolation
Increase with Pt
Decrease with Pt

19. Summary
Rel 15.0 electron reco improves electron efficiency relative to rel 14 (6.6% in absolute at medium level)
Further improvement is obtained with a rescaled momentum extrapolation from perigee for efficiency and rejection:
For efficiency : 8.3% in absolute for medium
For rejection : 10.4% of gain for 14->15 and 5% for 14->15+ new extrapolation
Application to the H4l analysis shows encouraging results
at 130 GeV efficiency increased by 22.7% for 14->15 and
29.3% for 14->15+ new extrapolation

20. Back-up slides

21. H->ZZ*->4e 120 GeV +34.9% +5% Rel 14 (%) Rel 15 (%) Rel 15 +new
Extrapolation (%)
4e loose, |η|<2.5
60.1 ±0.7
77.5 ±0.5
84.9 ±0.5
2e w/ Pt>20 GeV
4 w/ Pt>7GeV
41.8 ±0.7
55.2 ±0.7
60.5 ±0.7
mZ, mZ* (*)
13.4 ±0.5
18.2 ±0.5
19.3 ±0.6
|m4e-120|<4.4
GeV
10.9 ±0.5
14.0 ±0.5
14.7 ±0.6
+34.9%
+5%
(*) opposite charges, electron medium, |me1e2-mZ|<15 GeV, mZ*>15 GeV

22. H->ZZ*->4e 150 GeV +31.4% +3.4% Rel 14 (%) Rel 15 (%)
Rel 15 +new
Extrapolation (%)
4e loose, |η|<2.5
64.1 ±0.7
83.5 ±0.5
90.0 ±0.4
2e w/ Pt>20 GeV
4 w/ Pt>7GeV
56.0 ±0.7
74.6 ±0.6
81.2 ±0.5
mZ, mZ* (*)
23.6 ±0.6
31.4 ±0.6
33 ±0.6
|m4e-150|<4.6 GeV
18.5 ±0.5
23.5 ±0.5
24.3 ±0.6
+31.4%
+3.4%
(*) opposite charges, electron medium, |me1e2-mZ|<15 GeV, mZ*>30 GeV

23. H->ZZ*->4e 200 GeV +34.5% +4.7% Rel 14 (%) Rel 15 (%)
Rel 15 +new
Extrapolation (%)
4e loose, |η|<2.5
64.9 ±0.8
84.9 ±0.5
91.3 ±0.4
2e w/ Pt>20 GeV
4 w/ Pt>7GeV
60.9 ±0.8
82.2 ±0.7
89.1 ±0.4
mZ, mZ* (*)
34.5 ±0.5
45.2 ±0.7
47.8 ±0.7
|m4e-200|<7.8 GeV
28.1 ±0.5
36.1 ±0.7
37.8 ±0.5
+34.5%
+4.7%
(*) opposite charges, electron medium, |me1e2-mZ|<15 GeV, mZ*>60 GeV