Electron ID for susy..
Download
Report
Transcript Electron ID for susy..
What pT cut on the electrons?
Important to decide to what sample we normalize efficiency to
Related to electron trigger we use: em25i or 2em5i
Precut for electrons: pT>25 GeV
Also previous E/gamma group result for ET=25 (GeV) single electron MC
Would be nice to somehow emulate L1 cuts before estimating performance
Various detector regions are instrumented differently: investigate
Detector regions relevant for electron ID
Pixel (check if there is eta dependence of the number of hits, why do we need to cut on this? Check how the
track isolation variables, E/P, Ntrack narrow cone and spatial track match)
B-layer
||<2.5
Barrel layers
||<1.7
End-cap disks
1.7<||<2.5
SCT (check if there is eta dependence of the number of hit, why do we need to cut on this?)
Barrel layers
||<1.4
End-cap wheels
1.4<||<2.5
TRT (we do see dependence!)
Axial barrel straws
Radial end-cap straws
Presampler (used at all?)
Barrel
End Cap
||<0.7
0.7<||<2.5
||<1.52
1.55 <||<1.80
ECAL, regions with high strip granularity
LB
||<1.4 (full ECAL goes to 1.475)
There is a transition b/w two different lead thickness at 0.8: some variables change with it
End Cap
1.5<||<2.4 (full ECAL in 1.375<||<3.2)
NO FINE STRIP COVERAGE b/w 1.4<||<1.5
3 different strip granularities in End-caps
HCAL
LB
||<1
EB
0.8<||<1.7
ITC
0.8<||<1.7 (calo plug+scintillators)
Foward(Larg)
1.5<||<3.2
Investigate several options
One single eta region
3 |eta| regions
Lhood, Fisher, ANN
Performance estimate
Compute electron efficiency and rejection
Rejection computed as
R = Njet(before trigger)/Njet(passing MVD cut)
Signal efficiency computed as:
Eff = Nelectrons(passing MVD cut)/N(electrons before trigger)
Cut flow
Precuts:
pT>25 GeV (trigger emulation!)
Found as an electron candidate by the E/gamma group algorithm
NTRThits>0
Variables that are input to discriminant have not default values (-999 or -999999 ...)
Cut on discriminant
Single region discriminants
e237/e277
NTRThits(>thres)/NTRThits
MVA>0.73
MVA>0.96
(pT/ehad1
Is weak discriminant)
=0.70
=0.54
R=500
R=2500
Rejection and efficiency
Inspire definitions on ”physics TDR” and note com-phys-2004-074
”Electron/jet separation with DC1 Monte Carlo”
Jet rejection is computed w.r.t # reconstructed jets with pT>17GeV (was
ATLFAST reconstructed, now fully recoed)
Electron efficiency is computed w.r.t. # real electrons with true pT>25 GeV
(LVL1 threshold for low lumi).
Marginal
Signal
Background R
Comment
Precut
0.74
10
Electron: include
to find e/ cluster
MVA>0.73
0.70
2560
Cumulated
0.52
25600
Rejection and efficiency (cont’d)
El_author=1
NTRThits>0
i) Real electrons, non isolated, from jets
- isolation from identified jets
- isolation from hadronic energy
- impact parameter
- track isolation
iv) Photon conversion
- IP
- ntracks in narrow cone
- E/p
- spatial track match
ii) Direct pi0 production
- shower shape in strips
- track E/p
- quality of spatial track match
v) Single pi+/- TRT
- shower shape
- ratio of hadr/em energy
- track match quality
iii) Direct photon with random track
- require track match
- E/p
- quality of spatial track match
vi) Jets with high EM content, leading
pi0
- track isolation (sum of track pT)
- shower shape
- fhadron/fem
1
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
El_author
Algorithm used to generate electron candidate (=1 for E/gamma group)
El_ethad
Et in the HCAL behind EM cluster
El_ethad1
Et in the 1st sampling of HCAL behind EM cluster
El_etcone
Et in a DR=0.45 cone around shower (shower energy not included)
El_etcone20
Et in a DR=0.20 cone ”
”
El_etcone30
Et in a DR=0.30 cone ”
”
El_etcone40
Et in a DR=0.40 cone ”
”
El_emins1
E of strip with min E
El_emaxs1
E of strip with max E
El_wtots1
Total width in 20 strips
El_f1
fraction of energy in the 1st sampling
El_f1core
e131/(e033+e1153+e335) so in 1st sampling
El_f3core
e333/(e033+e1153+e335) so in 3rd sampling
El_pos7diff. b/w shower cell and predicted track in +/- 7 cells
El_iso
ratio of energy in 3x3/3x7
El_weta1
corrected lateral width with 3 strips
El_weta2
corrected lateral width in sample 2
El_widths2
uncorrected width in sample 2
El_e2ts1
energy in group of 3 adjacent strips, this 3 strip cluster must be the 2nd most energetic
one
El_e2tsts1
energy in 2nd most energetic strip
El_fracs1
fraction of energy outside core in S1
El_widths1
width with 5 strips
El_NTRTHits
number of TRT hits
El_NHighThresTRTHits
number of TRT hits above high threshold
Track_El_eta, momentaX/Y/Z, p_T, phi, qOverP : fitted track parameters
Track_El_d0
distance of closest approach (xy), wrt to PV or (0,0,0)? CHECK
Track_El_z0
distance to the PV (z), wrt to PV or (0,0,0) ? CHECK
Track_El_ij
Track error matrix
2
i) Real electrons, non isolated, from jets
- isolation from identified jets→ DR to closest ID jet*
- isolation from had energy → El_ethad, El_ethad1
- impact parameter
→ Track_El_d0 (check definition)
- track isolation
→ Sum of track pT in a 0.4 cone? Or alt.?
ii) Direct pi0 production
- shower shape in strips
→ El_emins1, El_emaxs1, El_wtots1,
El_f1, El_f1core, El_weta1, El_e2tsts1, El_e2ts1, El_fracs1, El_widths1
- track E/p
→ El_EoverP
- quality of spatial track match→ chi2 of track match:
2= (d/)^2 + (dz/z)^2
2= (d/)^2 + (d/)^2
iii) Direct photon with random track
- require track match
→ All track matched in ntuple
used here!
- E/p
→ El_EoverP
- quality of spatial track match→ chi2 of track match
4
iv) Photon conversion
- impact parameter
- ntracks in narrow cone
- track E/p
- spatial track match
→ Track_El_d0 (check definition)
→ missing
→ El_EoverP
→ missing
v) Single pi+/- TRT
→ El_NTRTHits, El_NHighThresTRTHits
- shower shape
→ strip info, El_iso, El_f3core,
El_weta2, El_widths2, El_fracs1, other variables?
- ratio of hadr/em energy
→ El_ethad, El_ethad1
- track E/p
→ El_EoverP
- spatial track match
→ missing
vi) Jets with high EM content, leading pi0
- track isolation (sum of track pT) → missing
- shower shape
→ see above
- ratio of hadr/em energy
→ El_ethad, El_ethad1
5