Selection of Tau Candidates in Muon+Jets Events
5.4 Selection of Tau Candidates in Muon+Jets Events
Starting from the preselected muon+jets sample, we select events in the sample with at least one tau candidate. This section describes the selection and studies of the muon+tau+jets samples.
We require the following criteria for the tau candidates:
µ µ + + ≥ ≥ 1 Jet 1 Jet
Data 38536
QCD 5521 ± 125 (14.3 ± 0.324%)
W 33016 ± 310 (85.7 ± 2500 0.804%) 2500 Events/5 GeV Events/5 GeV
W Transverse Mass (GeV) W Transverse Mass (GeV)
Figure 5.1: Distributions of the transverse mass from the muon+jets sample used to determine the W normalization factor. The filled histograms show the templates used in the fit and have been normalized using the results. Contributions from Z and t¯ t have been subtracted from the data.
• Tau calorimeter cluster E T > 10/5/10 GeV for type 1/2/3. • Tau is required to be reconstructed in the central calorimeter, |η τ
CAL | < 1.0. • The tau’s leading track p T > 7/5/5 GeV for type 1/2/3. In addition, for tau type 3, the total
sum of tau tracks’ p T , Pp τ track T > 7 GeV. • Tau is separated from the muon. We require the muon’s track and tau’s leading track to be
separated by ∆R > 0.5. We also require that the muon’s track and the tau’track are not the same object.
• Tau is coming from the same vertex as the muon, ∆z(µ track, τ leading track) < 1.0 cm. • The tau’s charge is opposite that of the muon’s charge. • Tau neural net output is greater than 0.8 for all types.
For events with more than one tau candidate, we choose the one with the highest tau neural net output. After selecting events with a tau candidate, we recount the jet multiplicity in the
L = 994 pb 8000 -1 8000
DØ Run II Preliminary
L = 994 pb -1
DØ Run II Preliminary
W → µ + light jets
Multijet W → µ + light jets W → µ +c c W → µ +c c
- µ + + + jets b W Z → → µ µ - µ + +b + jets Z b → τ τ + jets t t → lepton + jets
Z W → → µ µ - +b
Z → τ - τ + + jets t t → 5000 lepton + jets 5000
t t → dilepton
t t → dilepton
Number of events 4000 4000
Number of events 2500 2500
Leading jet p T (GeV), ≥
1 jets
Second lead jet p T (GeV), ≥
2 jets
5000 5000 DØ Run II Preliminary
L = 994 pb
DØ Run II Preliminary
W → µ + light jets 4000 4000
W → µ + light jets
W → µ +c c 800
Z W → → µ µ - - µ + +b + jets b 3500 3500
W → µ +c
- µ + +b - b + + jets
+ jets
Z → τ τ + + jets
t t → lepton + jets 3000 3000
t t → → lepton + jets
t t dilepton
t t → dilepton
Number of events 2500 2500
Number of events 500 500
Leading jet η , ≥
1 jets
Second lead jet η , ≥
2 jets
2000 2000 DØ Run II Preliminary
L = 994 pb 1800 1800
L = 994 pb
1000 1000 DØ Run II Preliminary
Multijet W → µ + light jets
DATA
W → µ + light jets
W → µ +c c Z W → → µ µ - µ + +b + jets b W Z → → µ µ - - µ + + +b - + + jets b 1400 1400
t t 1200 1200
t t t → lepton + jets
t → dilepton
t t → → dilepton lepton + jets
Number of events 1000 1000
Number of events 500 500
Leading jet φ (GeV), ≥
1 jets
Second lead jet φ (GeV), ≥
2 jets
Figure 5.2: Distributions of control variables from the muon+jets sample including all contributions (part 1 of 2).
DØ Run II Preliminary
L = 994 pb -1
4000 4000 DØ Run II Preliminary
L = 994 pb -1
Multijet 8000 DATA 8000
Multijet DATA
W → µ + light jets 7000 7000
W → µ + light jets
W → µ +c c W → µ +c c
Z W → → µ µ - µ +b - b - + + jets 6000 6000
W Z → → µ µ - µ + +b + jets b 3000 3000
Z → τ τ + + jets
Z → τ τ + + jets
t t → lepton + jets
t t → lepton + jets
t t → 5000 dilepton 5000 Number of events
t t → dilepton
Number of events
1 jets 2000 2000 DØ Run II Preliminary
Muon p T (GeV), ≥
1 jets
Muon η , ≥
10 10 6 6 DATA L = 994 pb DATA 1800 1800
L = 994 pb -1
DØ Run II Preliminary
W → µ + light jets 1600 1600
W → µ +c c 10 10 5 5 W → µ +c c Z W → → µ µ - µ + +b + jets b Z W → → µ µ - µ + +b + jets b 1400 1400
W → µ + light jets
Z → τ - τ + + jets
τ 4 - 4 Z → τ + + jets
t t → lepton + jets 1200 1200
t t → lepton + jets
10 10 t t → dilepton
t t → dilepton
Number of events 1000 1000
Number of events 10 10 3 3
Number of jets DØ Run II Preliminary
Muon φ , ≥
1 jets
L = 994 pb 5000 -1 5000
L = 994 pb -1
DØ Run II Preliminary
Multijet DATA
W → µ + light jets 4000 4000
W → µ + light jets
W → µ +c c W → µ +c c
W Z → → µ µ - µ + +b + jets b 3500 3500
W Z → → µ µ - µ + +b + jets b 5000 5000
Z → τ - τ + + jets
Z → τ - τ + + jets
t t 3000 3000
t t t t → → lepton + jets dilepton
t t → → dilepton lepton + jets
Number of events 2500 2500
Number of events
m T (muon,MET) (GeV), ≥
1 jets
MET (GeV), ≥
1 jets
Figure 5.3: Distributions of control variables from the muon+jets sample including all contributions (part 2 of 2).
DØ Run II Preliminary
L = 994 pb -1
DØ Run II Preliminary
L = 994 pb -1
10 10 4 4 DATA Multijet
Multijet DATA
W → µ - + jets
W → µ + jets
Z → τ τ + + jets
Z → τ - τ + + jets
Z → µ - µ + + jets
Z → µ - µ + + jets
WW, WZ
WW, WZ
10 10 3 3 t t → dilepton non-
t t → lepton + jets
t t → lepton + jets
t t → dilepton non- µ τ
Number of events
Number of events 400 400
0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.5 0.6 0.6 0.7 0.7 0.8 0.8 0.9 0.9 1 1 0.5 0.5 0 0 0.55 0.55 0.6 0.6 0.65 0.65 0.7 0.7 0.75 0.75 0.8 0.8 0.85 0.85 0.9 0.9 0.95 0.95 1 1 Tau neural network output (all type)
Tau neural network output (all type, 0.5 ≤ NN_ τ ≤ 1.0))
Figure 5.4: Distributions of tau neural net output (N N τ ) in the preselected muon+tau+jets sample before N N τ cut. Left figure is for all values of N N τ in logarithmic scale, right figure is for values of N N τ greater than 0.5 in linear scale.
event. We check if one of the selected jets is the same object as the selected tau candidate. A jet is considered to be the same object as the tau candidate if it is separated from the tau candidate by a distance less than 0.5 in the η − φ plane
(5.8) If a jet is matched to the tau candidate, it is discarded from the list of selected jets. We require all
∆R(τ, j) < 0.5.
selected jets in the event to be not matched to the tau. We reapply the jet criteria (e.g. leading jet p T > 30 GeV) with the revised jet list.
Before the tau neural net cut is applied, we cross-check the validity of tau neural net output in data and Monte Carlo. Figure 5.4 shows the distributions of tau neural net output in the above sample before the application of tau neural net cut. We found good agreement between data and Monte Carlo in the prediction of tau neural net output. As expected, W +jets sample which has fake taus peaks near zero, while Z → ττ sample which has real taus peaks near one.
We divide the lepton+tau+jets events sample into two disjoint samples. The first sample contains muon-tau pairs with opposite charge sign (OS). This sample contains the signal events, as well as various background contributions from Z+jets, W +jets, diboson, and multijet processes. The second sample contains the muon-tau pairs with have same-sign charge (SS). This sample is dominated by multijet and W +jets events. We use this sample to estimate contributions from multijet processes to the OS sample, as will be discussed in Section 5.5.1.
DØ Run II Preliminary 4 4 L = 994 pb -1
DØ Run II Preliminary 4 4 L = 994 pb -1
DATA Multijet
Z → τ - - τ + + jets
W → µ + jets
Z → τ + - τ + + jets
W → µ + jets
+ jets 3 3 Z → µ - µ + 10 + jets 10 WW, WZ
10 10 WW, WZ t t
t t lepton + jets
t t → → lepton + jets dilepton non-
dilepton non-
Number of events 10 10 2 2
Number of events 10 10 2 2
NN τ (OS, all type)
NN τ (SS, all type)
400 400 DØ Run II Preliminary
L = 994 pb
100 100 DØ Run II Preliminary
DATA Multijet
90 90 Multijet W → µ + jets
Z → τ - - τ + + + jets 300 300
Z → τ - - τ + + + jets
W → µ + jets
WW, WZ Z → µ µ + jets
80 80 Z WW, WZ → µ µ + jets
t t → lepton + jets
70 70 t t → t lepton + jets t → dilepton non- µ τ t t → dilepton non- µ τ
60 60 t t → µ τ
Number of events 200 200
Number of events 50 50
NN τ (OS, all type, 0.5 ≤ NN τ ≤ 1.0)
NN τ (SS, all type, 0.5 ≤ NN τ ≤ 1.0)
Figure 5.5: Distributions of tau neural net output (N N τ ) in the preselected muon+tau+jets sample before N N τ cut. Left figure is for all values of N N τ in logarithmic scale, right figure is for values of N N τ greater than 0.5 in linear scale.
5.4.1 Monte Carlo to data correction factor for jets faking taus
A significant number of events passing our tau criteria arise from having a jet fake a tau. This results in non-negligible contributions from W +jets, multijet and t¯ t → ℓ+jets. Since the W and t¯ t components are estimated from MC it is important to account for any differences between data and MC with respect to jets faking taus.
We use the muon+jets data sample from Section 5.3 and W +jets Monte Carlo to derive the Monte Carlo to data correction factor of jets faking taus. In addition to the muon+jets selection criteria, we add a cut on the minimum transverse mass of 40 GeV to reduce the multijet component. We also subtract the estimated contribution of Z → ττ, Z → µµ, and t¯t to the data sample to
Table 5.4: Number of events in data and Monte Carlo W +jets enriched sample. The Monte Carlo sample is normalized to the generator cross-section.
Sample
Number of events without tau requirement with tau requirement
Number of events
W +jets Monte Carlo
increase the sample purity. We cross-check that the samples used in this study are indeed dominated by W +jets events. Figure 5.6 shows the W transverse mass plots used in this calculation, before and after tau selection. Within statistical limitations (especially after tau selection), we found that the shapes of the transverse mass distributions; before and after tau selection; are consistent with samples which are enriched in W +jets events.
The correction factor is obtained by taking the ratio of tau fake rate in data and Monte Carlo. This will ensure that the correction factor is independent of Monte Carlo normalization. Table
5.4 lists the estimated number of W +jets events in data and Monte Carlo without and with tau requirement. The correction factor is defined as:
#of data w/ tau
f #of data =
#of MC w/ tau = 6789.0 84.3 (5.9)
#of MC
(5.10) This value is then applied to W +jets and t¯ t → ℓ+jets samples in the rest of this analysis.