WZ,and ZZProduction - Harvard University · 2016. 7. 29. · ATLAS ATLAS-CONF-2013-021 WZ → ℓνℓℓ [13 fb-1] 8 TeV ATLAS-CONF-2013-020 ZZ → ... 00 50 100 150 200 250 300

WW, WZ, and ZZ Production

Masahiro Morii, Harvard University

SM@LHC 2013, Freiburg im Breisgau9–12 April 2013

Masahiro Morii, Harvard WW, WZ, and ZZ

Diboson productionLO diagrams for WW, WZ, and ZZ production are

n Cross sections are calculated to NLOn Gluon-‐gluon enters at NLO. Less than 10% of the cross section

Allows access to triple gauge couplings (TGCs)n New physics may show up as anomalous (= non-‐SM) TGCs

Important background to Higgs and beyond-‐SM searchesn Precise knowledge of cross sections and kinematical distributions are important

2

W

q′

q

Z

W

TGC

s-‐channel

W /Z /γ

W /Z

W /Z q′ Z

q W

t-‐channel

W /Z

W /Zq′ Z

q W

u-‐channel

W /Z

W /Z


Leptonic final states (ℓ = e, μ and ν = νe, νμ, ντ)n High-‐pT isolated leptons + missing ET gives good S/B ratios

n Price: Br(W → ℓν) = 22%, Br(Z → ℓℓ) = 7%, Br(Z → νν) = 20%

Semi-‐leptonic final statesn Br(W → qq) = 68%, Br(Z → qq) = 70% ➔ More signal, but huge background

n b-‐jet identification can enhance Z → bb signal

Measurements

3

W W

+ ′ −

ν ν

W Z

+ ′ −

ν ′ +

Z

+

−

Z

′ −

′ +

Z

ν

ν

Z

−

+

W

+

ν

V

jet

jet

V

jet

jet

Z

+

−

Z

ν

ν

V

jet

jet

V = W or Z


References

4

CDF CDF note 10957 ZZ → ℓℓℓℓ, ℓℓνν [9.7 fb-1]

CDF note 10973 WW + WZ → ℓνjj [8.9 fb-1]

CDF note 10968 VV → MET + jj [9.1 fb-1]

CDF note 10864 ZW + ZZ → ℓℓjj [8.9 fb-1]

PRD 86 (2012) 031104 WZ → ℓνℓℓ [7.1 fb-1]

PRL 104 (2010) 201801 WW → ℓνℓν [3.6 fb-1]

DØ arXiv:1301.1243 WW → ℓνℓν [9.7 fb-1]

PLB 718 (2012) 451 TGC limits based on the following three analyses

PRL 108 (2012) 181803 WW + WZ → ℓνjj [4.3 fb-1]

PRD 85 (2012) 112005 WZ → ℓνℓℓ, ZZ → ℓℓνν [8.6 fb-1]

PRD 84 (2011) 011103 ZZ → ℓℓℓℓ [6.4 fb-1]

arXiv:1204.4496 WZ → ℓνbb [7.5 fb-1], ZZ → ννbb [8.4 fb-1], and ZZ → ℓℓbb [7.5 fb-1]

ATLAS ATLAS-CONF-2013-021 WZ → ℓνℓℓ [13 fb-1] 8 TeVATLAS-CONF-2013-020 ZZ → ℓℓℓℓ [20 fb-1] 8 TeVATLAS-CONF-2012-157 WW + WZ → ℓνjj [4.7 fb-1] 7 TeV

arXiv:1211.6096 ZZ → ℓℓℓℓ, ℓℓνν [4.6 fb-1] 7 TeV

arXiv:1210.2979 WW → ℓνℓν [4.6 fb-1] 7 TeV

EPJC 72 (2012) 2173 WZ → ℓνℓℓ [4.6 fb-1] 7 TeV

CMS CMS-PAS-FSQ-12-010 γγ → WW → eνµν [5.05 fb-1] 7 TeV

arXiv:1301.4698 WW → ℓνℓν [3.5 fb-1] and ZZ → ℓℓℓℓ [5.3 fb-1] 8 TeVJHEP 1301 (2013) 063 ZZ → ℓℓℓℓ [5.0 fb-1] 7 TeV

EPJC 73 (2013) 2283 WW + WZ → ℓνjj [5.0 fb-1] 7 TeV

CMS-PAS-SMP-12-005 WW → ℓνℓν [4.9 fb-1] 7 TeV

CMS-PAS-EWK-11-010 WZ → ℓνℓℓ [1.1 fb-1] 7 TeV


ZZ → ℓℓℓℓZZ → ℓℓℓℓ is the cleanest of all channelsn Two e+e− or μ+μ− pairs in loose Z mass windowsn Very small background from WZ and Z + jets

n CMS allows Z → τ+τ− for one Z

n

5

Subleading Lepton-Pair Mass [GeV]20 40 60 80 100 120 140 160 180 200 220

Lead

ing

Lep

ton

-Pair

Mass [

GeV

]

20

40

60

80

100

120

140

160

180

200

220

Data

lllll→ZZ

-1 L dt = 20 fb∫= 8 TeVs

PreliminaryATLAS

Figure 2: Invariant mass of the leading Z candidate versus the invariant mass of the subleading Z candidate. Theevents observed in the data are shown as solid circles and the signal prediction from simulation as pink boxes.The red box indicates the region defined by the ZZ fiducial cuts on the Z candidate masses. The blue dashed linesindicate the regions where one of the leptons pairs is in the mass window 66 < mll < 116 GeV. Contributions fromevents with one or both Z bosons outside of this mass window are also seen.

(lower pT) lepton pair. Events are required to contain two Z candidates with invariant masses satisfying66 < m`+`� < 116 GeV, resulting in 305 observed events.

4 Signal Acceptance

The ZZ selection requirements described above are applied to Monte Carlo simulation (PowhegBox andgg2zz), and corrections are applied to account for di↵erences in lepton reconstruction and identificatione�ciency, lepton energy scale and resolution, and trigger e�ciencies between data and simulation. Thecorrections are measured in data using samples of single Z ! `+`� events, supplemented by samplesof J/ ! e+e� events for electron e�ciency measurements at low pT. The overall e�ciency correctionfactor is 0.92±0.06 for the e+e�e+e� channel, 0.97±0.03 for the µ+µ�µ+µ� channel and 0.94±0.03 forthe e+e�µ+µ� channel, where the errors are systematic. A smearing and scale correction is added to themuon pT in the simulation [38] so that the Z ! µ+µ� invariant mass distribution in data is reproduced.Similarly, corrections are applied to the electron calorimeter energy resolution in the simulation and tothe electron calorimeter energy scale in the data [39]. A correction parameterized by the number of

5

ATLAS 8 TeV10 7 Summary

[GeV]2l2lm200 400 600 800

Even

ts /

20 G

eV

0

10

20

DATA ZZ WZ/Z + jets

CMS -1 = 8 TeV, L = 5.3 fbs

a)

[GeV]visτ2l2m

100 200 300 400 500 600

Even

ts /

25 G

eV

0

2

4

6 DATA ZZ WZ/Z + jets

CMS -1 = 8 TeV, L = 5.3 fbs

b)

[GeV]ll →Z1 m60 80 100 120

[GeV

]ll

→Z2

m

60

80

100

1202l2l → ZZ → pp

DATA 2eµ 4

µ 2e2

CMS -1 = 8 TeV, L = 5.3 fbs

c)

[GeV]ll →Z1 m60 80 100 120

[GeV

]vi

sττ

→Z2

m

20

40

60

80

100

120τ22l → ZZ → pp

DATA

CMS -1 = 8 TeV, L = 5.3 fbs

d)

Figure 2: Distributions for ZZ candidate events of (a) the four-lepton reconstructed mass for thesum of the 4e, 4µ, and 2e2µ channels and (b) the sum of the 2`2t channels. Points represent thedata, and shaded histograms represent the expected ZZ signal and the reducible background.The shapes of the signal and background are taken from the MC simulation, with each com-ponent normalized to the corresponding estimated value from Table 2. The distributions (c)and (d) demonstrate the relationship between the reconstructed Z1 and Z2 masses. Differentsymbols are used to present different decay channels.

CMS 8 TeV

ATLAS

-CO

NF-2013-020arX

iv:1301.4698

CMS 8 TeV

6

) [GeV/c]1

(lT

p0 50 100 150 200 250 300 350 400 450 5000

1

2

3

4

5

6

7DataSM ZZFakes

CDF Run II Preliminary -1 L dt = 9.7 fb∫

(a)pT (l1).

) [GeV/c]2

(lT

p0 50 100 150 200 250 300 350 400 450 5000

1

2

3

4

5

6

DataSM ZZFakes


(b)pT (l2).

) [GeV/c]3

(lT

p0 50 100 150 200 250 300 350 400 450 5000

2

4

6

8

10 DataSM ZZFakes


(c)pT (l3).

) [GeV/c]4

(lT

p0 50 100 150 200 250 300 350 400 450 5000

1

2

3

4

5

6

7

8

DataSM ZZFakes


(d)pT (l4).

R(ll)]Δmin[0 0.5 1 1.5 2 2.5 30

0.5

1

1.5

2

2.5

3

3.5

4

4.5

DataSM ZZFakes


(e)min(∆R(!!)).

R(ll)Δ0 0.5 1 1.5 2 2.5 3 3.5 40

1

2

3

4

5

6

7 Data

SM ZZ

Fakes


(f)∆R(!!) of both lepton pair.

]2Z [GeV/cT

leading PllM20 40 60 80 100 120 140 160

]2Z

[GeV

/cT

sub

lead

ing

Pll

M

20

40

60

80

100

120

140

160

DataSM ZZFakes


(g)M!! of the reconstructed Zs.

]2 [GeV/cllm0 50 100 150 200 2500

2

4

6

8

10

12 DataSM ZZFakes


(h)m!! of both lepton pair.

(Z) [GeV/c]T

p0 50 100 150 200 250

Even

ts/(2

5 G

eV/c

)

0

1

2

3

4

5

6

7

8

9

10

DataSM ZZFakes


(i)pT (!!) of both lepton pair.

]2 [GeV/c4lm100 200 300 400 500 600

Even

ts/(2

5 G

eV)

0

1

2

3

4

5

6

DataSM ZZFakes


(j)M4!.

(ZZ) [GeV/c]T

p0 20 40 60 80 100 120 140 160 180 200

Even

ts/(2

0 G

eV)

0

1

2

3

4

5

6

7

8

9

10

DataSM ZZFakes


(k)pT (ZZ).

Z decay angle-3 -2 -1 0 1 2 30

0.5

1

1.5

2

2.5

3

3.5

4

4.5

DataSM ZZFakes


(l)Z decay angle.

FIG. 1: Comparison of the MC simulation prediction and observation for some kinematic variables of the events inthe selected !!!′!′ collected sample.

CDF 9.7 fb-1

CD

F note 10957


ZZ → ℓℓννZZ → ℓℓνν looks for missing ET recoiling against a Z → ℓℓn Define “axial missing ET” =

n CDF and DØ use neural-‐net discriminators to separate signal from background

6

−ET ⋅pT pT

[GeV]missTAxial-E

-50 0 50 100 150

Even

ts /

5 G

eV

-1101

10210310410510610710810910

1010ATLAS

0 -1Ldt = 4.6 fb = 7 TeVs

DataZ+XW+X

*a/WaWTopWZ

WW-l+l-l+lAZZii

-l+lAZZTotal Uncertainty

ii-l+lAZZ

ATLAS 7 TeV

arXiv:1211.6096

8

The composition of the sample of events passing these requirements is summarized in TABLE III, including ex-pectations for other minor backgrounds. FIG. 3 shows some kinematic variable comparison between data and MCsimulation for events passing the Signal Region requirements.

) [GeV], Z axisTEφ Δ cos(TE--50 0 50 100 150 200

Even

ts /

5 G

eV

-210

-110

1

10

210

310

410

510

610

710

810

) [GeV], Z axisTEφ Δ cos(TE--50 0 50 100 150 200

Even

ts /

5 G

eV

-210

-110

1

10

210

310

410

510

610

710

810CDF Run II preliminary -1 L = 9.7 fb∫

ttγW

W+jetsDY

WZWW

ZZCDF Data

FIG. 2: E/TAx

distribution for the selected events (before

applying the E/TAx

requirement, comparing data to sim-ulation. The red line over impressed represents the ZZsignal shape, magnified ×5 with respect to the propernormalization, while all the solid process contributionare stacked.

∫

L = 9.7 fb−1

Process candidate eventsDY 317 ± 51.3tt 11.9 ± 2.2W+jets 69.5 ± 18.5Wγ 17.3 ± 2.2WW 114 ± 10.6WZ 37.5 ± 5.3Total Background 567 ± 24.4ZZ 63 ± 11Data 618

ZZ → !!νν Signal Region

TABLE III: Expected and observed number of eventspassing the kinematic requirements defining the SignalRegion.

1. Background modeling test

The modeling of the main background contribution in the defined Signal Region (Drell-Yan, WW ) have been testedin orthogonal sample of collected data with similar kinematic properties to the Signal Region. We test the Drell-Yanbackground modeling comparing data and simulation in a sample of events passing all the requirements applied forthe Signal Region definition but having E/T

Ax≤25 GeV, where the ZZ signal contribution is negligible. The WW

production modeling is tested comparing data to simulation in a sample of e±µ∓ events passing the same requirementsapplied for the Signal Region but 40 ≤ Meµ ≤ 140 GeV/c2, which has a negligible ZZ contribution and a small residualDrell-Yan contamination due to Z → ττ decays. The discrepancies between data and Monte Carlo simulation thataffects in particular the Drell-Yan modeling will be taken into account as a systematic uncertainties.

D. Neural Network separation

In order to improve the signal-to-background ratio further, we use an artificial neural network relying on thesimulated samples of signal and background events. This self-learning machine exploits kinematic information abouta given process (input variables) to produce an output value close to a target value (usually +1 for signal-like and-1 for background-like events). A NeuroBayes c© neural network (NN) [13] is trained using seven event kinematicvariables: the leading lepton transverse momentum (pT (#1)), the E/T significance (E/T /

√

∑

ET [20]), the dileptoninvariant mass (M!!), the dilepton system transverse momentum (p!!T ), the opening angles between the two leptons inthe transverse plane (∆φ(##)), the number of reconstructed jets in the events (Njets), and the angle in the detectortransverse plane between the E/T evaluated as the unbalance in the calorimeter towers and the P/T representing themissing transverse pT from the tracks reconstructed in the tracking system. These variables are the most sensitivefor signal-to-background separation since they exploit the unique features of ZZ production and are shown in Figure3, comparing data to simulations. Figure 4 shows the resulting NN output distributions for data and expected signaland background events in the Signal Region, in which ZZ signal events tend toward higher values and backgroundtoward lower values.

Exploiting the good separation of the signal from the background, we measure the ZZ cross section from a binnedmaximum likelihood fit of the NN output distribution. The likelihood function in the fit is the product of the Poisson

CD

F note 10957

CDF 9.7 fb-1

12

(GeV)’Tp10 20 30 40 50 60 70 80 90 100

Entri

es-110

1

10

210

310

410 )-e+Data (eZZ sig. Z bgd.WW bgd. Other bgd.

-1(a)D0 8.6 fb

(GeV)’Tp10 20 30 40 50 60 70 80 90 100

Entri

es

-110

1

10

210

310

410 )-µ+µData (

ZZ sig. Z bgd.

WW bgd. Other bgd.

-1(b)D0 8.6 fb

(GeV)’Tp10 20 30 40 50 60 70 80 90 100

Entri

es

0

10

20

30

40

50 )µData (e

ZZ sig. Z bgd.

WW bgd. Other bgd.

-1(c)D0 8.6 fb

(GeV)llM40 60 80 100 120 140 160 180 200 220 240

Entri

es

0

5

10

15

20

25

30


-1(d)D0 8.6 fb

(GeV)llM40 60 80 100 120 140 160 180 200 220 240

Entri

es

0

5

10

15

20

25

30 )-µ+µData (

ZZ sig. Z bgd.

WW bgd. Other bgd.

-1(e)D0 8.6 fb

(GeV)llM40 60 80 100 120 140 160 180 200 220 240

Entri

es

0

10

20

30

40

50

60 )µData (e

ZZ sig. Z bgd.

WW bgd. Other bgd.

-1(f)D0 8.6 fb

Neural network output-3 -2 -1 0 1 2 3

Entri

es

0

5

10

15

20

25

30

35


-1(g)D0 8.6 fb


Entri

es

0

5

10

15

20

25

30 )-µ+µData (

ZZ sig. Z bgd.

WW bgd. Other bgd.

-1(h)D0 8.6 fb


Entri

es

0

5

10

15

20

25

30

35

40

45 )µData (e

ZZ sig. Z bgd.

WW bgd. Other bgd.

-1(i)D0 8.6 fb

FIG. 6: (a-c) The p/′T distribution of the ZZ → !+!−νν candidate events before imposing the p/′T requirement. (d-f) The M!!

distribution of the ZZ → !+!−νν candidate events before imposing the M!! requirement. (g-i) The neural network outputdistribution of the accepted ZZ → !+!−νν candidate events. For the e±µ∓ channel, the neural network trained in the e+e−

channel is shown. The vertical dashed lines indicate the requirements on p/′T and M!!. The signal normalization is as describedin Section IV.

TABLE III: Table of predicted signal and background yields for the ZZ → e+e−νν signal and control regions. The systematicuncertainties are provided for the predictions.

Rejected by requirement onProcess Accepted p/′T a/′T Mll Extra lep. Charge Jets

Z → e+e− 0.6 ± 0.3 11666 ± 1665 0 ± 1 0.3 ± 0.2 3 ± 2 0.0 ± 0.0 0.1 ± 0.1

Z → τ+τ− 0.1 ± 0.1 8 ± 2 1.4 ± 0.2 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0WW → !+ν!−ν 35 ± 1 35 ± 1 1.7 ± 0.1 33 ± 1 9 ± 5 0.3 ± 0.1 0.1 ± 0.1WZ → !ν!+!− 2.3 ± 0.1 1.9 ± 0.1 0.2 ± 0.0 0.2 ± 0.1 14 ± 2 0.2 ± 0.1 0.0 ± 0.0W → eν 6 ± 2 13 ± 2 0.3 ± 0.1 5 ± 1 2 ± 1 4 ± 1 0.0 ± 0.0Wγ → eνγ 3.3 ± 0.3 5.5 ± 0.5 0.0 ± 0.1 2.8 ± 0.5 0.6 ± 0.5 3.3 ± 0.4 0.0 ± 0.0ZZ → !+!−!+!− 0.0 ± 0.0 0.1 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 1.3 ± 0.2 0.0 ± 0.0 0.0 ± 0.0tt → !+!−ννbb 1.0 ± 0.2 1.4 ± 0.2 0.4 ± 0.1 1.2 ± 0.1 7 ± 1 0.0 ± 0.0 0.2 ± 0.1Predicted background 48 ± 2 11749 ± 1668 4 ± 1 43 ± 2 37 ± 11 8 ± 1 0.4 ± 0.2

ZZ → !+!−νν 13.6 ± 0.4 7.4 ± 0.2 1.3 ± 0.1 0.6 ± 0.0 4 ± 2 0.2 ± 0.0 0.1 ± 0.0Predicted total 62 ± 3 11756 ± 1668 6 ± 1 43 ± 2 41 ± 13 8 ± 1 0.4 ± 0.2

Observed 61 10560 12 50 63 12 1

DØ 8.6 fb-1

eeννDØ 8.6 fb-1

µµνν

PR

D 85 (2012) 112005


ZZ → ℓℓℓℓ and ℓℓνν

Measured cross sections agreewith the SM predictionn NB: “total” cross section depends on the Z mass window

Measurement precisions arestatistics-‐limitedn Leading systematic errors is the leptonidentification efficiency

7

s L dt∫ Measured cross section (pb) Theory (pb)

CDF 1.96 TeV 9.7 fb−1 1.38 ± 0.19(stat)−0.19+0.20(sys) 1.4 ± 0.1 CDF note 10957

DØ 1.96 TeV 6.4–8.6 fb−1 1.40−0.37+0.43(stat) ± 0.14(sys) 1.4 ± 0.1 PRD 85 (2012) 112005

ATLAS 7 TeV 4.6 fb−1 6.7 ± 0.7(stat)−0.3+0.4(sys) ± 0.3(lumi) 5.89−0.18

+0.22 arXiv:1211.6096

CMS 7 TeV 5.0 fb−1 6.24−0.80+0.86(stat)−0.32

+0.41(sys) ± 0.14(lumi) 6.3 ± 0.4 EPJC 73 (2013) 2283

ATLAS 8 TeV 20 fb−1 7.1−0.4+0.5(stat) ± 0.3(sys) ± 0.2(lumi) 7.2−0.2

+0.3 ATLAS-CONF-2013-020

CMS 8 TeV 5.3 fb−1 8.4 ±1.0(stat) ± 0.7(sys) ± 0.4(lumi) 7.7 ± 0.4 arXiv:1301.4698

[TeV] s0 2 4 6 8 10 12 14

[pb]

ZZto

tal

σ

1

10

<116 GeV)ll) (66<mpZZ (p<116 GeV)llZZ (pp) (66<m

=8 TeV)sLHC Data 2012 (


=1.96 TeV)sTevatron (

-1<116 GeV) L=20 fbll llll (66<m→ATLAS ZZ-1<120 GeV) L=5.3 fbll llll (60<m→CMS ZZ

-1<116 GeV) L=4.6 fbll) (66<mνν ll(ll/→ATLAS ZZ-1<120 GeV) L=5.0 fbll llll (60<m→CMS ZZ

-1<120 GeV) L=8.6 fbll) (60<mνν ll(ll/→D0 ZZ-1) (on-shell) L=6.0 fbνν ll(ll/→CDF ZZ

PreliminaryATLAS NLO QCD (MCFM, CT10.0)

Figure 4: Comparison of experimental measurements and theoretical predictions of the total ZZ produc-tion cross section as a function of centre-of-mass energy

ps. Shown are experimental measurements

from CDF [14] and D0 [15] in pp collisions at the Tevatron atp

s = 1.96 TeV, and experimental mea-surements from ATLAS [5] and CMS [6, 8] in pp collisions at the LHC at

ps = 7 TeV and

ps = 8 TeV.

The blue dashed line shows the theoretical prediction for the ZZ production cross section in pp collisions,calculated at NLO in QCD using MCFM with PDF set CT10. The solid red line shows the theoreticalprediction for the ZZ production cross section in pp collisions, calculated in the same way. The theo-retical curves are calculated using the natural width of the Z boson in the mass range 66 to 116 GeV.Atp

s = 8 TeV, the theoretical prediction using the zero-width approximation is 5% higher than theprediction using the natural width, restricted to the mass range 66 to 116 GeV.

12


[GeV]llM60 70 80 90 100 110 120

Even

ts /

2 G

eV

0

50

100

150

200

250

300

350 dataWZZZ

γW/Z+W+jetZ+jetTop

ATLAS Preliminary-1 L dt = 13 fb∫ = 8 TeV, s

[GeV]WZT

p0 50 100 150 200 250

Even

ts /

10 G

eV

020406080

100120140160180200220 data

WZZZ

γW/Z+W+jetZ+jetTop


) [GeV]missT

(l,ETM20 40 60 80 100 120 140 160

Even

ts /

5 G

eV

0

20

40

60

80

100

120 dataWZZZ

γW/Z+W+jetZ+jetTop


[GeV]WZM0 100 200 300 400 500 600 700 800

Even

ts /

20 G

eV

020406080

100120140160180 data

WZZZ

γW/Z+W+jetZ+jetTop


Figure 4: Distributions of: the invariant mass of the Z candidate (top left), the WZ system transverse momentum(top right), the W transverse mass (bottom left) and the WZ invariant mass (bottom right). All these distributionsare shown after applying all of the WZ selection criteria in the four channels except the Z candidate mass. Thestatistical uncertainty is shown by shaded bands. For top (tt and single top) and Z+jets, the expected shape istaken from simulation but the normalization is taken from the data-driven estimates. The rightmost bins includeoverflow.

8

[GeV]llM60 70 80 90 100 110 120

Even

ts /

2 G

eV

0

50

100

150

200

250

300

350 dataWZZZ

γW/Z+W+jetZ+jetTop


[GeV]WZT

p0 50 100 150 200 250

Even

ts /

10 G

eV

020406080

100120140160180200220 data

WZZZ

γW/Z+W+jetZ+jetTop


) [GeV]missT

(l,ETM20 40 60 80 100 120 140 160

Even

ts /

5 G

eV

0

20

40

60

80

100

120 dataWZZZ

γW/Z+W+jetZ+jetTop


[GeV]WZM0 100 200 300 400 500 600 700 800

Even

ts /

20 G

eV

020406080

100120140160180 data

WZZZ

γW/Z+W+jetZ+jetTop


Figure 4: Distributions of: the invariant mass of the Z candidate (top left), the WZ system transverse momentum(top right), the W transverse mass (bottom left) and the WZ invariant mass (bottom right). All these distributionsare shown after applying all of the WZ selection criteria in the four channels except the Z candidate mass. Thestatistical uncertainty is shown by shaded bands. For top (tt and single top) and Z+jets, the expected shape istaken from simulation but the normalization is taken from the data-driven estimates. The rightmost bins includeoverflow.

8

WZ → ℓνℓℓWZ → ℓνℓℓ is quite clean: S/B ~ 4n e+e− or μ+μ− pair in a tight Z mass window, plus one isolated lepton and missing ET

n Loose cut on mT of W candidate

n Background from Z + jets and ZZ

Enough statistics to measure kinematicaldistributions, e.g. pT(Z) and m(WZ)

n ATLAS “unfolded” (= corrected for experimental resolutions) pT(Z) and m(WZ) distributions with 7 TeV data

8

ATLAS

-CO

NF-2013-021

11

Table 7 Normalized fiducial cross-sections and uncertainties in bins of pZT.

pZT [GeV] [0, 30] [30, 60] [60, 90] [90, 120] [120, 150] [150, 180] [180, 2000]

∆σfidWZ

(pZT)/σfidWZ

0.231 0.350 0.230 0.065 0.045 0.042 0.038Uncertainty 0.034 0.039 0.033 0.019 0.015 0.014 0.013

(a)

[GeV]TZp

0 30 60 90 120 150 180 2000Data

/MC

0.51

1.5

fid WZ

σ/fid W

ZσΔ

0.1

0.2

0.3

0.4

0.5

0.6ATLAS

= 7 TeV)sData 2011 (-1 L dt = 4.6 fb∫

Monte Carlo (MC@NLO)Data Total Uncertainty

(b)

[GeV]WZm170 270 405 2500Da

ta/M

C

0.51

1.5

fid WZ

σ/fid W

ZσΔ

0.2

0.4

0.6

0.8

1ATLAS

= 7 TeV)sData 2011 (-1 L dt = 4.6 fb∫

Monte Carlo (MC@NLO)Data Total Uncertainty

Fig. 7 Normalized fiducial cross-sections ∆σfidWZ

/σfidWZ

in binsof (a) pZT and (b) mWZ compared with the SM prediction.The total uncertainty contains statistical and systematic un-certainties added in quadrature.

for each source of the uncertainty, and combining theresulting changes in the unfolded spectrum. Because ofthe normalization, the results are affected only by theuncertainties that depend on pZT or mWZ . The stabilityof the unfolding procedure is tested in two ways: firstlyby comparing the unfolded spectra after two and af-ter three iterations, and secondly by checking that thetrue variable distribution is correctly reproduced froma simulated sample generated with non-zero anomalouscouplings.

7 Conclusion

Measurements of W±Z production in proton-protoncollisions at

√s = 7 TeV have been presented using a

Table 8 Normalized fiducial cross-sections and uncertaintiesin bins of mWZ .

mWZ [GeV] [170, 270] [270, 405] [405, 2500]

∆σfidWZ

(mWZ)/σfidWZ

0.568 0.283 0.149Uncertainty 0.038 0.030 0.027

data sample with an integrated luminosity of 4.6 fb−1,collected with the ATLAS detector at the LHC. Thecandidate W±Z events were selected in the fully lep-tonic final states with electrons, muons, and large miss-ing transverse momentum. In total, 317 candidates wereobserved with a background expectation of 68 ± 10events. The fiducial and total cross-sections are deter-mined to be

σfidWZ = 92+7

−6(stat.)± 4(syst.)± 2(lumi.) fb,

and

σtotWZ = 19.0+1.4

−1.3(stat.)± 0.9(syst.)± 0.4(lumi.) pb,

respectively, where the fiducial cross-section is definedby pµ,eT > 15GeV for the leptons from the decay ofthe Z bosons, pµ,eT > 20GeV for the leptons from thedecay of the W± bosons, |ηµ,e| < 2.5, pνT > 25GeV,|m"" − mZ| < 10GeV, MW

T > 20GeV, and ∆R > 0.3between the two leptons of all possible pairings of thethree leptons. These results are significantly more pre-cise than the earlier ATLAS measurement [6] whichthis paper supersedes. The total cross-section is con-sistent with the SM expectation of 17.6+1.1

−1.0 pb. Limitson anomalous triple gauge couplings have been derivedbased on the observed pZT distribution. The 95% confi-dence intervals are

∆gZ1 ∈ [−0.057, 0.093]

∆κZ ∈ [−0.37, 0.57]

λZ ∈ [−0.046, 0.047]

without a form factor. The limits are again more strin-gent than the earlier ATLAS measurement. Normalizedfiducial cross-sections have also been presented in binsof pZT and mWZ , and are in good agreement with SMpredictions.

EP

JC 72 (2012) 2173

ATLAS 7 TeVunfolded

ATLAS 8 TeV

ATLAS 8 TeV

9

(GeV)’TE0 20 40 60 80 100 120 140

Entri

es

0

10

20

30

40

50

60Data WZ sig.Z bgd.Other bgd.

-1(a) D0 8.6 fb

(GeV)llM40 60 80 100 120 140 160 180 200 220 240

Entri

es

0

10

20

30

40

50

60 Data WZ sig.Z bgd.Other bgd.

-1(b) D0 8.6 fb

(GeV)WTM

0 20 40 60 80 100 120 140 160 180 200

Entri

es

0

5

10

15

20


-1(c) D0 8.6 fb

(GeV)T

Leading lepton p0 50 100 150 200 250

Entri

es

0

5

10

15

20

25

30


-1(d) D0 8.6 fb

(GeV)T

Subleading lepton p0 20 40 60 80 100 120 140

Entri

es

0

5

10

15

20

25

30

35


-1(e) D0 8.6 fb

(GeV)T

W daughter lepton p20 40 60 80 100 120 140 160 180 200

Entri

es

0

5

10

15

20

25

30


-1(f) D0 8.6 fb

(GeV)llT

p0 50 100 150 200 250 300

Entri

es

0

5

10

15

20

25

30

35


-1(g) D0 8.6 fb

(GeV)WT

p0 50 100 150 200 250 300

Entri

es

0

10

20

30

40

50Data WZ sig.Z bgd.Other bgd.

-1(h) D0 8.6 fb

FIG. 3: Kinematic distributions for the WZ → !ν!+!− signal candidates after combining the different sub-channels. Thefollowing variables are shown: (a) the E/ ′

T ; (b) the invariant mass of the Z → !+!− decay; (c) the W transverse mass; thetransverse momenta of the (d) leading and (e) subleading leptons from the Z → !+!− decay and (f) the charged lepton fromthe W decay; the transverse momenta of the reconstructed (g) Z → !+!− and (h) W → lν decays. The vertical dashed linesindicate the requirements on E/ ′

T and M!!. The signal normalization is as described in Section IV.

llTa

llLa

(2)T

p (1)T

p

llT

p

φΔ

t

Hadronic recoil

FIG. 4: Illustration of the decomposition of the dilepton pTinto aT and aL components.

allowed for the upward variation and protects againstelectrons for which the calorimeter has severely under-measured the energy. The amount by which, e.g., a!!T isreduced, is denoted aδT . These quantities are defined insuch a way that they always carry a negative sign.

B. Calorimeter recoil

Two estimates of the calorimeter recoil are made, fromthe reconstructed jets and from the reconstructed E/T .Jets are reconstructed using the D0 mid-point cone al-gorithm [29] with a cone size of ∆R = 0.5. They mustbe separated from the leptons by at least ∆R > 0.3 andsatisfy pT > 15 GeV. The pT , aT , and aL components

PR

D 85 (2012) 112005

DØ 8.6 fb-1


WZ → ℓνℓℓ

Measured cross sections agreewith the SM predictionn NB: “total” cross section depends on the Z mass window

Measurement precisions arestatistics-‐limitedn Exception: ATLAS 8 TeV, 13 m-‐1

n Leading systematic error is Z + jetsbackground estimate

9


CDF 1.96 TeV 7.1 fb−1 3.93−0.53+0.60(stat)−0.46

+0.59(sys) 3.50 ± 0.21 PRD 86 (2012) 031104

DØ 1.96 TeV 8.6 fb−1 4.50 ± 0.61(stat)−0.25+0.16(sys) 3.21± 0.19 PRD 85 (2012) 112005

ATLAS 7 TeV 4.6 fb−1 19.0−1.3+1.4(stat) ± 0.9(sys) ± 0.4(lumi) 17.6−1.0

+1.1 EPJC 72 (2012) 2173

CMS 7 TeV 1.1 fb−1 17.0 ± 2.4(stat) ±1.1(sys) ±1.0(lumi) (19.8 ± 0.1) CMS-PAS-EWK-11-010

ATLAS 8 TeV 13 fb−1 20.3−0.7+0.8(stat)−1.1

+1.2(sys)−0.6+0.7(lumi) 20.3 ± 0.8 ATLAS-CONF-2013-021

[TeV] s0 2 4 6 8 10 12 14

[pb]

to

tal

WZ

σ

1

10 <116 GeV)ll

)(66<mpWZ (p<116 GeV)

llWZ (pp)(66<m



=1.96 TeV)sTevatron (

-1<116 GeV) L=13 fbllll (66<mν l→ATLAS WZ

-1<116 GeV) L=4.6 fbllll (66<mν l→ATLAS WZ

-1<120 GeV) L=8.6 fbllll (60<mν l→D0 WZ-1ll L=7.1 fbν l→CDF WZ

PreliminaryATLAS

NLO QCD (MCFM, CT10)

Figure 5: Measurements and theoretical predictions of the total W±Z production cross section as a functionof center-of-mass energy. Experimental measurements from CDF and D0 in proton antiproton collisions at theTevatron at

ps = 1.96 TeV, and experimental measurements from ATLAS in proton-proton collisions at the LHC

atp

s = 7 TeV andp

s = 8 TeV are shown. The blue dashed line shows the theoretical prediction for the W±Zproduction cross section in proton anti-proton collisions, calculated at NLO using MCFM with PDF set CT10. Thesolid red line shows the theoretical prediction for the W±Z production cross section in proton-proton collisions,calculated in the same way. The ATLAS results at 8 TeV define the total cross section with a Z boson with massbetween 66 GeV and 116 GeV. The results from CDF define the total cross section assuming zero-width for the Zboson and neglecting the �⇤ contribution. The results from D0 define the total cross section with a Z boson withmass between 60 GeV and 120 GeV.

12


WW → ℓνℓνWW → ℓνℓν needs a tight event selection to fight backgroundn Two isolated opposite-‐sign leptonsn Jet veto to suppress top background

▶ DØ uses 0-‐jet and 1-‐jet samplesn More kinematical cuts to suppress W/Z/γ* + jets

▶ ATLAS/CMS/CDF use a hard missing ET cut for e+e− and μ+μ− to suppress Z/γ* + jets

▶ DØ uses a BDT for e+e− or μ+μ−, mTmin and mT2 for e±μ∓

n ATLAS also unfolded the leading lepton pT distribution

10

8 6 Results

[GeV]maxT

p0 50 100 150

Even

ts /

5 G

eV

0

50

100

150

200 DATA WW VV Z + jets W + jets top

syst⊕ stat

CMS -1 = 8 TeV, L = 3.5 fbs

a)

[GeV]minT

p20 40 60 80

Even

ts /

5 G

eV

0

100

200

300

400 DATA WW VV Z + jets W + jets top

syst⊕ stat

CMS -1 = 8 TeV, L = 3.5 fbs

b)

[GeV]llT

p40 60 80 100 120

Even

ts /

5 G

eV0

100

200

DATA WW VV Z + jets W + jets top

syst⊕ stat

CMS -1 = 8 TeV, L = 3.5 fbs

c)

[GeV]llm0 50 100 150 200

Even

ts /

5 G

eV

0

50

100

150

DATA WW VV Z + jets W + jets top

syst⊕ stat

CMS -1 = 8 TeV, L = 3.5 fbs

d)

Figure 1: Distributions for W+W� candidate events of (a) the leading lepton transverse mo-mentum pmax

T , (b) the trailing lepton transverse momentum pminT , (c) the dilepton transverse

momentum p``T , and (d) the dilepton invariant mass m``. Points represent the data, and shadedhistograms represent the W+W� signal and the background processes. The last bin includesthe overflow. The W+W� signal is scaled to the measured cross section, and the backgroundprocesses are normalized to the corresponding estimated values in Table 1.

arXiv:1301.4698

CMS 8 TeV

1jet WW Discriminant0 0.2 0.4 0.6 0.8 1

Even

ts/0

.02

0

5

10

15

20

25

30

35

40

45 dataWWZ+jetsWZ,ZZW+jetsMultijetttbarBkg. syst.

TE + µ, e-1DØ, 9.7 fb

0jet WW Discriminant0 0.2 0.4 0.6 0.8 1

Even

ts/0

.02

0

20

40

60

80

100 dataWWZ+jetsWZ,ZZW+jetsMultijetttbarBkg. syst.

TE + µ, e-1DØ, 9.7 fb

DØ 9.7 fb-1

1-jetDØ 9.7 fb-1

0-jet

arXiv:1301.1243

12

[GeV]T

Leading lepton p40 60 80 100 120 140 160 180

Data

/MC

0.51

1.5

25 40 60 80 100 120 140 350

]-1 [G

eVT

/dp

WW

fid σ d×

WW

fid σ1/

0.005

0.01

0.015

0.02

0.025

0.03 ATLAS=7 TeV)sData 2011 (

-1 L dt = 4.6 fb∫

Monte Carlo (MC@NLO)DataStat. UncertaintyFull Uncertainty

FIG. 7: The normalized di↵erential WW fiducial cross section as a function of the leading lepton pT compared to the SMprediction.

Leading lepton pT [GeV] [25,40] [40,60] [60,80] [80,100] [100,120] [120,140] [140, 350]Weighted bin center [GeV] 33.6 50.2 70.2 89.1 107.1 127.5 180.41/�fid

WW ⇥ d�

fidWW /dpT [GeV�1] 2.0⇥ 10�2 2.1⇥ 10�2 8.2⇥ 10�3 2.7⇥ 10�3 2.2⇥ 10�3 9.5⇥ 10�4 6.2⇥ 10�5

Relative uncertainty 6.7% 4.8% 8.2% 17.0% 17.1% 25.5% 41.0%Correlation 1 �0.43 �0.33 �0.27 �0.27 �0.13 �0.29

1 �0.29 �0.29 �0.23 �0.30 �0.151 �0.01 �0.04 0.02 0.03

1 0.21 0.11 0.141 0.23 0.11

1 0.271

TABLE VII: Normalized fiducial cross section together with the overall uncertainty in bins of the leading lepton pT. Theweighted bin center is calculated as the cross-section-weighted average of the leading lepton pT in each bin derived frommc@nlo and gg2WW. The correlation coe�cients between di↵erent leading lepton pT bins are also shown. Only half of thesymmetric correlation matrix is presented.

A reweighting method is applied to SM WW eventsgenerated with mc@nlo and processed through the fulldetector simulation to obtain the leading lepton p

T

dis-tribution with anomalous couplings. The reweightingmethod uses an event weight to predict the rate withwhich a given event would be generated if anomalouscouplings were present. The event weight is the ratio ofthe squared matrix elements with and without anomalouscouplings i.e., |M|2/|M|2

SM

, where |M|2 is the matrixelement squared in the presence of anomalous couplingsand |M|2

SM

is the matrix element squared in the SM.The event generator bho [48] is used for the calculationof the two matrix elements. Generator-level comparisonsof WW production between mc@nlo and bho with allanomalous couplings set to zero are performed and con-sistent results are obtained. Samples with di↵erent setsof anomalous couplings are generated and the ratio of

the leading lepton pT

distribution to the SM predictionis parameterized as a function of the input anomalouscoupling parameters. This function is then used to inter-polate the leading lepton p

T

distribution for any givenanomalous couplings. To verify the reweighting method,the event weights for a given set of anomalous couplingsare calculated and applied to events generated with bho

assuming no anomalous couplings. The reweighted dis-tributions are compared to those predicted by the bho

generator, and good agreement is observed for the inclu-sive cross section and for the kinematic distributions asshown in Fig. 8(a).

Figure 8(b) compares the reconstructed leading lep-ton p

T

spectrum in data with that from the sum of ex-pected signal and background contributions. The pre-dicted leading lepton p

T

distributions for three di↵erentanomalous TGC values are also shown. Events at high

arXiv:1210.2979

ATLAS 7 TeVunfolded


WW → ℓνℓν

Cross sections at the LHC are slightly larger than the SM predictionn Significances are small (+1.4σ, +1.0σ, +1.7σ) but starting to draw attention

▶ Are the NLO calculations sufficiently precise?▶ Could this be a subtle sign of new physics?

Measurement precisions are systematics-‐limitedn Leading source of systematics is the jet veto

▶ Experimental: jet-‐energy scale and resolution affects jet pT threshold▶ Theoretical: number of jets in WW + jets

We still really want to see results from the full 8 TeV data

11


CDF 1.96 TeV 3.6 fb−1 12.1± 0.9(stat)−1.4+1.6(sys) 11.7 ± 0.7 PRL 104 (2010) 201801

DØ 1.96 TeV 9.7 fb−1 11.6 ± 0.4(stat) ± 0.6(sys) 11.3 ± 0.7 arXiv:1301.1243

ATLAS 7 TeV 4.6 fb−1 51.9 ± 2.0(stat) ± 3.9(sys) ± 2.0(lumi) 44.7−1.9+2.1 arXiv:1210.2979

CMS 7 TeV 4.9 fb−1 52.4 ± 2.0(stat) ± 4.5(sys) ±1.2(lumi) 47.0 ± 2.0 CMS-PAS-SMP-12-005

CMS 8 TeV 3.5 fb−1 69.9 ± 2.8(stat) ± 5.6(sys) ± 3.1(lumi) 57.3−1.6+2.4 arXiv:1301.4698


WW/WZ → ℓνjjWW/WZ → ℓνjj is a tour de force of SM measurementsn Reconstruct a W candidate in one lepton + missing ETn Two jets with pT > 25–30 GeV

n Fit the di-‐jet invariant mass distribution

n Background must be exquisitely modeled

Signal is established by all experimentsn Precisions limited by both statistics and background modeling

12

=29.42χndof= 28

Data

W+X

Diboson

W/Z+jets

Top

QCD

]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/10

GeV

/c

0

0.2

0.4

0.6

0.8

1310×

-1CDF Run II Preliminary, L = 8.9 fb

]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/10

GeV

/c

-50

0

50

100

150Data-SM (no Diboson)

W+X

Diboson

FIG. 11: Fit to the dijet invariant mass distribution in the electron sample. All corrections

described in text have been applied here. Bottom figure shows data with all backgrounds (except

the diboson contribution) subtracted.

=69.42χndof= 56

Data

Diboson

W/Z+jets

Top

QCD

]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/5 G

eV/c

0100200300400500600700800


]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/5 G

eV/c

-50

0

50

100

150 Data-SM (no Diboson)

Diboson

FIG. 12: Fit to the dijet invariant mass distribution similar to [4]. The corrections described in

text have been applied here. Bottom figure shows data with all backgrounds (except the diboson

contribution) subtracted.

IX. ORTHOGONAL SAMPLES

Similar analyses are performed in two orthogonal samples. Same fit in invariant mass is

done as described in Sec. V and same kinematic cuts are applied as in Sec. II. The only

differences are in the objects selected.

20

=29.42χndof= 28

Data

W+X

Diboson

W/Z+jets

Top

QCD

]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/10

GeV

/c

0

0.2

0.4

0.6

0.8

1310×


]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/10

GeV

/c

-50

0

50

100

150Data-SM (no Diboson)

W+X

Diboson

FIG. 11: Fit to the dijet invariant mass distribution in the electron sample. All corrections

described in text have been applied here. Bottom figure shows data with all backgrounds (except

the diboson contribution) subtracted.

=69.42χndof= 56

Data

Diboson

W/Z+jets

Top

QCD

]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/5 G

eV/c

0100200300400500600700800


]2Dijet-Mass [GeV/c50 100 150 200 250 300

2Ev

ents

/5 G

eV/c

-50

0

50

100

150 Data-SM (no Diboson)

Diboson

FIG. 12: Fit to the dijet invariant mass distribution similar to [4]. The corrections described in

text have been applied here. Bottom figure shows data with all backgrounds (except the diboson

contribution) subtracted.

IX. ORTHOGONAL SAMPLES

Similar analyses are performed in two orthogonal samples. Same fit in invariant mass is

done as described in Sec. V and same kinematic cuts are applied as in Sec. II. The only

differences are in the objects selected.

20

CD

F note 10973

CDF 8.9 fb-14

(GeV)jjm50 100 150

Even

ts /

GeV

0

500

1000

1500

WW/WZW+jetstopQCDZ+jetsdata

TeV = 7s, -1fb dt = 5.0L∫CMS,

(a)

(GeV)jjm50 100 150

Even

ts /

GeV

0

50

100WW/WZdataUncertainty


(b)

(GeV)jjm50 100 150

pull

-4

-2

0

2

4


(c)

Figure 1: (a) Distribution of the dijet invariant mass in data, with the binning chosen based onthe resolution and fit projections of the relevant components overlaid. (b) The dijet invariantmass after subtraction of all components except the electroweak WW+WZ processes. The er-ror bars represent the statistical uncertainties and the hatched bands represent the systematicuncertainties. (c) The normalized residual or pull: (data � fit)/(fit uncertainty).

validate the fit procedure by performing pseudo-experiments. In each experiment, we generatethe mjj pseudo-data for the SM processes, taking into account the correlations between theyields, and then perform a fit to each pseudo-data mjj distribution. The results for both themuon and electron channels indicate that there is a small bias (-8.6% and -6.6%) in the WW+WZyield, corresponding to less than 0.4 standard deviations, and that the fit slightly overestimatesthe uncertainty on the yield. These effects are corrected for in the final result. The validationprocedure shows that biases in all background yields and errors are small. The fit results for thebackground components are statistically consistent with the expectations, with the exceptionof W+jets, where 11% fewer events for muons and 15% fewer events for electrons, compared tothe expectation, are observed. Overall, the approach produces a high quality model of the data(Fig. 1(a)), where the pull distribution is consistent with 0 (Fig. 1(c)), and allows us to extractthe diboson peak (Fig. 1(b)).

EP

JC 73 (2013) 2283

CMS 7 TeV

ATLAS

-CO

NF-2012-157

0 20 40 60 80 100 120 140 160 180 200

Entri

es /

5 G

eV

0

5000

10000

15000

20000

25000DataMultijet

+single-t ttW/Z + HFW/Z + jetsWW/WZ

-1L dt = 4.7 fb∫ = 7 TeVs

ATLAS Preliminary

+ 2 jetsν) µ l (e,→W

[GeV]T

Leading jet p0 20 40 60 80 100 120 140 160 180 200(D

ata-

MC

)/MC

-0.10.00.1

0 20 40 60 80 100

Entri

es /

5 G

eV

0

10000

20000

30000

40000

50000

60000

70000

80000 DataMultijet


-1L dt = 4.7 fb∫ = 7 TeVs

ATLAS Preliminary


[GeV]T

SubLeading jet p0 20 40 60 80 100(D

ata-

MC

)/MC

-0.10.00.1

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Entri

es /

0.15

0

5000

10000

15000

20000

25000

30000

35000 DataMultijet


-1L dt = 4.7 fb∫ = 7 TeVs

ATLAS Preliminary


R ΔDijet 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5(D

ata-

MC

)/MC

-0.10.00.1

50 100 150 200 250

Entri

es /

5 G

eV

0

2000

4000

6000

8000

10000

12000

14000

16000

18000 DataMultijet


-1L dt = 4.7 fb∫ = 7 TeVs

ATLAS Preliminary


Dijet Mass [GeV]50 100 150 200 250(D

ata-

MC

)/MC

-0.10.00.1

Figure 2: Distributions of the leading (top left) and sub-leading (top right) jet transverse momenta, of their angulardistance ∆R (bottom left) and of the di-jet invariant mass distribution of reconstructed W/Z → j j candidates(bottom right). The distributions are for the electron and muon channels summed together. Filled circles show theexperimental data and the stacked histograms are SM predictions. The rightmost bins include overflow. In eachplot, the lower panel displays the fractional difference between the data and the MC expectation. The yellow bandsshow the JES systematic uncertainty, which dominates.

6

Dijet Mass [GeV]

ATLAS 7 TeV

10

(l) (GeV)T

p0 20 40 60 80 100 120 140

Even

ts /

(5 G

eV)

0

2000

4000

6000

8000

10000

12000

14000

16000 DataWVW/Z+LPW/Z+HFTopMultijetsUncert.

-1DØ 4.3 fb

) (GeV)1

(jetT

p0 20 40 60 80 100 120 140 160 180 200

Even

ts /

(5 G

eV)

0

2000

4000

6000

8000

10000

12000DataWVW/Z+LPW/Z+HFTopMultijetsUncert.

-1DØ 4.3 fb

) (GeV)2

(jetT

p0 20 40 60 80 100 120 140

Even

ts /

(5 G

eV)

05000

10000

1500020000250003000035000

4000045000 Data

WVW/Z+LPW/Z+HFTopMultijetsUncert.

-1DØ 4.3 fb

(GeV)TE0 20 40 60 80 100 120 140

Even

ts /

(5 G

eV)

02000400060008000

1000012000140001600018000 Data

WVW/Z+LPW/Z+HFTopMultijetsUncert.

-1DØ 4.3 fb

(GeV)jjM0 50 100 150 200 250 300

Even

ts /

(5 G

eV)

0

1000

2000

3000

4000

5000

6000DataWVW/Z+LPW/Z+HFTopMultijetsUncert.

-1DØ 4.3 fb

(GeV)νlTM

0 20 40 60 80 100 120 140 160 180 200

Even

ts /

(5 G

eV)

0

2000

4000

6000

8000

10000


-1DØ 4.3 fb

(GeV)TH0 50 100 150 200 250 300

Even

ts /

(5 G

eV)

0100020003000400050006000700080009000

DataWVW/Z+LPW/Z+HFTopMultijetsUncert.

-1DØ 4.3 fb

(GeV)W)1

(dijet,jetrelT

p0 10 20 30 40 50 60 70 80 90 100

Even

ts /

(2 G

eV)

0

2000

4000

6000

8000

10000


-1DØ 4.3 fb

FIG. 6: (color online) Distributions of the variables (next eight of fifteen) used as inputs to the RF classifier for electron andmuon channels combined, and before b-tagging. The signal and background predictions and the systematic uncertainty bandare evaluated after the fit of the total WV cross section in the RF output distribution. Definitions for each variable are providedin the text (LP denotes light partons such as u, d, s or gluon, and HF denotes heavy-flavor such as cc or bb).

PR

L 108 (2012) 181803

DØ 4.3 fb-1


Cross sections at the LHC

13

W Z WW Wt

[pb]

tota

lσ

1

10

210

310

410

510

-120 fb

-113 fb

-15.8 fb

-15.8 fb

-14.6 fb

-12.1 fb-14.6 fb

-14.6 fb-11.0 fb

-11.0 fb

-135 pb

-135 pb

tt t WZ ZZ

= 7 TeVsLHC pp Theory

)-1Data (L = 0.035 - 4.6 fb


)-1Data (L = 5.8 - 20 fb

ATLAS PreliminaryATLAS PreliminaryATLAS Preliminary

[p

b]to

tσ

Prod

uctio

n C

ross

Sec

tion,

1

10

210

310

410

510

CMSNov 2012

W

1j≥

2j≥

3j≥

4j≥

Z

1j≥

2j≥

3j≥

4j≥

> 30 GeV jetTE

| < 2.4 jetη|

γW

> 15 GeVγ TE

,l) > 0.7γR(Δ

γZ WW+WZ WW

WZZZ

-136, 19 pb -15.0 fb -11.1 fb-15.0 fb -14.9 fb

-13.5 fb

-14.9 fb-15.3 fb

JHEP10(2011)132JHEP01(2012)010

CMS-PAS-SMP-12-011 (W/Z 8 TeV)

CMS EWK-11-009 CMS-PAS-EWK-11-010 (WZ)CMS-PAS-SMP-12-005 (WW7),

007(ZZ7), 013(WW8), 014(ZZ8), 015(WV)

syst) ⊕7 TeV CMS measurement (stat


7 TeV Theory prediction


W Z WW Wt

[pb]

tota

lσ

1

10

210

310

410

510

-120 fb

-113 fb

-15.8 fb

-15.8 fb

-14.6 fb

-12.1 fb-14.6 fb

-14.6 fb-11.0 fb

-11.0 fb

-135 pb

-135 pb

tt t WZ ZZ


)-1Data (L = 0.035 - 4.6 fb


)-1Data (L = 5.8 - 20 fb


[p

b]to

tσ

Prod

uctio

n C

ross

Sec

tion,

1

10

210

310

410

510

CMSNov 2012

W

1j≥

2j≥

3j≥

4j≥

Z

1j≥

2j≥

3j≥

4j≥

> 30 GeV jetTE

| < 2.4 jetη|

γW

> 15 GeVγ TE

,l) > 0.7γR(Δ

γZ WW+WZ WW

WZZZ

-136, 19 pb -15.0 fb -11.1 fb-15.0 fb -14.9 fb

-13.5 fb

-14.9 fb-15.3 fb

JHEP10(2011)132JHEP01(2012)010

CMS-PAS-SMP-12-011 (W/Z 8 TeV)

CMS EWK-11-009 CMS-PAS-EWK-11-010 (WZ)CMS-PAS-SMP-12-005 (WW7),

007(ZZ7), 013(WW8), 014(ZZ8), 015(WV)





WW

WZ

ZZ

W Z WW Wt

[pb]

tota

lσ

1

10

210

310

410

510

-120 fb

-113 fb

-15.8 fb

-15.8 fb

-14.6 fb

-12.1 fb-14.6 fb

-14.6 fb-11.0 fb

-11.0 fb

-135 pb

-135 pb

tt t WZ ZZ


)-1Data (L = 0.035 - 4.6 fb


)-1Data (L = 5.8 - 20 fb


ZZ


Triple Gauge CouplingsWWV (V = Z/γ) couplings ⬌ WW and WZ (also Wγ)

n 5 parameters:n Additional constraints may be imposed

ZZV (V = Z/γ) couplings ⬌ ZZ (also Zγ)

n 4 parameters:

Parameters in red (anomalous TGCs) are zero in the SM

14

W

q′

q

Z

W

TGC

Z /γ

W

W

LWWV

gWWV

= ig1V (Wµν

+W µV ν −Wµ+VνW

µν ) + iκVWµ+WνV

µν +iλV

mW2 Wλµ

+WνµV νλ

LZZV = −

eMZ

2 f4V (∂4

VV µβ )Zα (∂αZβ ) + f5V (∂σVσµ ) Z µβZβ

⎡⎣ ⎤⎦

Δg1

Z (≡ g1Z −1), Δκ Z (≡κ Z −1), Δκγ (≡κγ −1), λZ , λγ

f4Z , f4

γ , f5Z , f5

γ

W

q′

q

Z

W

TGC

W

W

Z

W

q′

q

Z

W

TGC

Z /γ

Z

Z

Equal coupling Δg1Z = 0, Δκ Z = Δκγ , and λZ = λγ

LEP scenario Δg1Z − Δκ Z = Δκγ tan2θW and λZ = λγ

HISZ scenario Δκ Z = Δg1Z (cos2θW − sin2θW ), Δκγ = 2Δg1

Z cos2θW and λZ = λγ


Triple Gauge CouplingsEffects of anomalous TGCs increase with n Increase sensitivity by binning in, or selecting the upper tail of n Observables: mZZ (for ZZ), pT of Z (for WZ or ZZ), pT of leading lepton (for WW)

n Also used: pT of di-‐jet in WW/WZ → ℓνjj

Extraction of TGC relies on NLO calculations: Powheg, MC@NLO, MCFMn LO-‐to-‐NLO correction is substantial at large

15

s s

7

include final states with hadronically decaying t leptons.

(GeV)2l2lm500 1000 1500

Even

ts/b

in

-110

1

10

DATAZZWZ/Z + jets

-1 = 7 TeV, L = 5.0 fbsCMS

= 0Z4f = 0.015Z

4f

Figure 2: Distribution of the four-lepton reconstructed mass for the sum of the 4e, 4µ, and the2e2µ channels. Points represent the data, and the shaded histograms represent the expected ZZsignal and the reducible background. The dashed and dotted histograms represent the resultsof the SHERPA simulation for the SM ( f Z

4 = 0) and in the presence of an ATGC ( f Z4 = 0.015),

while all the other anomalous couplings are set to zero.

The limits on ATGCs are calculated with the modified frequentist construction CLs [37–39]based on the shape of the four-lepton invariant mass distributions, including the 4e, 4µ, and2e2µ channels in the likelihood combination. Figure 2 presents the distribution of the four-lepton reconstructed mass for the sum of the 4e, 4µ, and 2e2µ channels. The dashed and dottedhistograms represent the results of the SHERPA simulation for the SM ( f Z

4 = 0) and in thepresence of an ATGC ( f Z

4 = 0.015), while all the other anomalous couplings are set to zero. Thepresence of ATGCs would be manifested in an increased yield of events at high four-leptonmasses. The invariant mass distributions are interpolated from SHERPA simulation for differentvalues of the anomalous couplings. For each distribution only one or two couplings are varied,while all others are set to zero. The fit is performed to find the maximum likelihood value andlimits are calculated. To avoid unitarity violation at energies above the scale L of new physics,the ATGCs are often modified with a form-factor parametrization of the type 1/(1 + s/L2)2,where

ps ⇡ m2`2` is the effective center-of-mass energy of the collision. However, no unitarity

violations occur in the sensitive region m2`2` . 1.5 TeV for bare anomalous couplings of order0.05 or smaller [40], so we calculate the limits without form-factor scaling. This choice has theadvantage of avoiding any bias from energy-dependence assumptions and is exact in the limitin which the scale of new physics is much larger than

ps.

Figure 3 presents the expected and observed two-dimensional exclusion limits at 95% confi-dence level (CL) on the anomalous neutral trilinear ZZZ and ZZg couplings. The green andyellow bands represent the one and two standard-deviation variations from the expected limit.The present limits are dominated by statistical uncertainties. Systematic uncertainties arisingfrom the uncertainty on the theoretical cross section, PDFs, detector efficiencies, and luminosityare introduced in the form of nuisance parameters with log-normal probability density func-tions. One-dimensional 95% CL limits for the f Z,g

4 and f Z,g5 anomalous coupling parameters

are measured to be

CMS ZZ 7 TeV

JHE

P 1301 (2013) 063

9

[GeV]ZT

p0 30 60 90 120 150 180 2000

Even

ts /

bin

0

20

40

60

80

100

120

140

160

DataSM WZBkg

stat + systσ = 0.05Zλ

= 0.57ZκΔ = 0.10Z

1gΔ

ATLAS

∫ -1Ldt = 4.6 fb = 7 TeVs

Fig. 4 Transverse momentum pZT of the Z boson inW±Z can-didate events. Data are shown together with expected back-ground and signal events, assuming the Standard Model. Ex-pected events in the case of anomalous TGC without formfactor are also shown for illustration. The last bin is short-ened for display purposes.

squared of the W±Z system, approaches infinity. Toachieve this, an arbitrary form factor may be intro-duced [32]. Here the dipole form factor adopted is

α(s) =α0

(1 + s/Λ2)2(8)

where α stands for ∆gZ1 , ∆κZ , or λZ , α0 is the valueof the anomalous coupling at low energy, and Λ is thecut-off scale, the scale at which new physics enters. Theresults are reported both with and without this formfactor.

Since an enhancement in the cross-section due to ananomalous coupling would grow with s, measurementsensitivity to anomalous TGCs is enhanced by binningthe data in a kinematic variable related to s. The trans-verse momentum pZT of the Z boson provides a natu-ral choice for such binning as it is strongly correlatedwith s and can be directly reconstructed from the mea-sured lepton momenta with good precision. The dataare therefore divided into six bins in pZT of width 30GeVfollowed by a wide bin that includes 180–2000GeV.

MC@NLO [11] is used to generateW±Z events withnon-SM TGC. The generator computes, for each event,a set of weights that can be used to reweight the fullsample to any chosen set of anomalous couplings. Thisfunctionality is used to express the predicted signalyields in each bin of pZT as a function of the anoma-lous couplings. Figure 4 shows the pZT distribution ofthe selected events together with the SM prediction.Also shown for illustration are predictions with non-zero anomalous couplings without form factor: each cou-pling is increased to the expected 99% confidence-levelupper limit while keeping the other two couplings atthe SM value. For this plot the 99%, rather than 95%,

Table 6 Expected and observed 95% confidence intervals onthe anomalous couplings ∆gZ1 , ∆κZ , and λZ . The expectedintervals assume the Standard Model values for the couplings.

Observed Observed ExpectedΛ = 2 TeV no form factor no form factor

∆gZ1 [−0.074, 0.133] [−0.057, 0.093] [−0.046, 0.080]∆κZ [−0.42, 0.69] [−0.37, 0.57] [−0.33, 0.47]λZ [−0.064, 0.066] [−0.046, 0.047] [−0.041, 0.040]

confidence-level upper limits are used to accentuate dif-ferences in shape. As expected, the largest deviationsfrom the SM are in the last bin of pZT, while the devia-tions in the lower-pZT bins depend on which coupling isvaried.

Frequentist confidence intervals are obtained on theanomalous couplings by forming a profile likelihood testincorporating the observed number of candidate eventsin each pZT bin, the expected signal as a function ofthe anomalous couplings and the estimated number ofbackground events [33]. The systematic uncertaintiesare included in the likelihood function as nuisance pa-rameters with correlated Gaussian constraints. A pointin the anomalous TGC space is accepted (rejected) atthe 95% confidence level if less (more) than 95% of ran-domly generated pseudo-experiments exhibit a value ofthe profile likelihood ratio larger than that observed indata.

Table 6 summarizes the observed 95% confidenceintervals on the anomalous couplings ∆gZ1 , ∆κZ , andλZ , with the cut-off scale Λ = 2 TeV and without theform factor. The limits on each anomalous TGC param-eter are obtained with the other two anomalous TGCparameters set to zero. The expected intervals in Ta-ble 6 are medians of the 95% confidence-level upper andlower limits obtained in pseudo-experiments that as-sume the SM coupling. The widths of the expected andobserved confidence intervals are dominated by statisti-cal uncertainty. Figure 5 compares the observed limitswith the Tevatron results [34, 35].

The 95% confidence regions are shown as contourson the (∆gZ1 ,∆κZ), (∆gZ1 ,λZ), and (∆κZ ,λZ) planesin Figure 6. In each plot the remaining parameter is setto the SM value. The limits were derived with no formfactor.

6.3 Normalized Fiducial Cross-Sections

The effective Lagrangian adopted in the TGC analysisin Section 6.2 allows us to probe non-SM physics withlittle model dependence. An alternative approach is tomeasure kinematic distributions, such as the pZT spec-trum, that could be compared with model-dependent

ATLAS WZ 7 TeVE

PJC

72 (2012) 2173

7

0 50 100 150 200 250 300 350 400 450 500-210

-110

1

10

210

-W+SM W

Background

Data

= 0.16ZλΔ

= 0.34Z1

gΔ

= 0.72γκΔ

]-1 [GeV cT

Lepton p

-1Ev

ents

/ 20

GeV

c

FIG. 2: Leading-lepton pT distribution for data compared tothe SM expectation. Also shown is how the expectation wouldbe modified by anomalous couplings near the observed limits.

TABLE II: Expected and observed limits on anomalousTGCs. For each coupling limit set, the two other couplingsare fixed at their SM values. Values of the couplings outside ofthe given observed range are excluded at the 95% confidencelevel.

Λ (TeV) λZ ∆gZ1 ∆κγ

Expected 1.5 (-0.05,0.07) (-0.09,0.17) (-0.23,0.31)Observed 1.5 (-0.16,0.16) (-0.24,0.34) (-0.63,0.72)Expected 2.0 (-0.05,0.06) (-0.08,0.15) (-0.20,0.27)Observed 2.0 (-0.14,0.15) (-0.22,0.30) (-0.57,0.65)

observed 95% confidence limits, shown in Table II, areweaker than expected. The probability of observing theselimits in the presence of only standard model W+W−

production ranges from 7.1% to 7.6% depending on thecoupling constants (λZ , gZ

1 ,κγ).In summary, the W+W− production cross section has

been measured in pp collisions at√

s = 1.96 TeV fromreconstructed events in the dilepton final state using alikelihood ratio formed from matrix-element-based eventprobabilities. This result is the most precise measure-ment at this energy with an overall uncertainty of lessthan 15%. The same event sample is also used to per-form the most sensitive probe to date at this energy ofanomalous WWZ and WWγ couplings. The leading-lepton pT distribution of the sample is found to be inmoderate agreement with the SM expectation and usedto place limits on anomalous triple gauge couplings.

We thank the Fermilab staff and the technical staffsof the participating institutions for their vital contribu-tions. This work was supported by the U.S. Departmentof Energy and National Science Foundation; the ItalianIstituto Nazionale di Fisica Nucleare; the Ministry ofEducation, Culture, Sports, Science and Technology ofJapan; the Natural Sciences and Engineering Research

Council of Canada; the National Science Council of theRepublic of China; the Swiss National Science Founda-tion; the A.P. Sloan Foundation; the Bundesministeriumfur Bildung und Forschung, Germany; the World ClassUniversity Program, the National Research Foundationof Korea; the Science and Technology Facilities Coun-cil and the Royal Society, UK; the Institut National dePhysique Nucleaire et Physique des Particules/CNRS;the Russian Foundation for Basic Research; the Minis-terio de Ciencia e Innovacion, and Programa Consolider-Ingenio 2010, Spain; the Slovak R&D Agency; and theAcademy of Finland.

[1] J. M. Campbell and R. K. Ellis, Phys. Rev. D 60, 113006(1999).

[2] S. Frixione and B. R. Webber, J. High Energy Phys. 06(2002) 029.

[3] J. Ellison and J. Wudka, Ann. Rev. Nucl. Part. Sci. 48,33 (1998).

[4] F. Abe et al. (CDF Collaboration), Phys. Rev. Lett. 78,4536 (1997).

[5] D. E. Acosta et al. (CDF Collaboration), Phys. Rev. Lett.94, 211801 (2005).

[6] V. M. Abazov et al. (D0 Collaboration), Phys. Rev. Lett.94, 151801 (2005) [Erratum-ibid. 100, 139901 (2008)].

[7] V. M. Abazov et al. (D0 Collaboration), Phys. Rev. Lett.(to be published) [arXiv:0904.0673].

[8] J. Abdallah et al. (DELPHI Collaboration), Eur. Phys.J. C 54, 345 (2008); S. Schael et al. (ALEPH Collabora-tion), Phys. Lett. B 614, 7 (2005); P. Achard et al. (L3Collaboration), Phys. Lett. B 586, 151 (2004); G. Abbi-endi et al. (OPAL Collaboration), Eur. Phys. J. C 33, 463(2004); V. M. Abazov et al. (D0 Collaboration), Phys.Rev. D 74, 057101 (2006); T. Aaltonen et al. (CDF Col-laboration), Phys. Rev. D 76, 111103 (2007).

[9] K. Hagiwara, S. Ishihara, R. Szalapski, and D. Zeppen-feld, Phys. Rev. D 48, 2182 (1993).

[10] A. Abulencia et al. (CDF Collaboration), J. Phys. G 34,2457 (2007).

[11] T. Aaltonen et al. (CDF Collaboration), Phys. Rev. Lett.102, 021802 (2009).

[12] T. Sjostrand, S. Mrenna, and P. Skands, J. High EnergyPhys. 05 (2006) 026.

[13] U. Baur, T. Han, and J. Ohnemus, Phys. Rev. D 57,2823 (1998).

[14] R. Brun, R. Hagelberg, M. Hansroul, and J. C. Lassalle,version 3.15, CERN-DD-78-2-REV.

[15] S. Moch and P. Uwer, Nucl. Phys. Proc. Suppl. 183, 75(2008).

[16] C. Anastasiou, L. J. Dixon, K. Melnikov, and F. Petriello,Phys. Rev. D 69, 094008 (2004).

[17] D. Acosta et al., Nucl. Instrum. Methods Phys. Res.,Sect. A 494, 57 (2002).

[18] F. James and M. Roos, Comput. Phys. Commun. 10, 343(1975).

CDF WW 3.6 fb-1

PR

L 104 (2010) 201801 7

(GeV)jjT

p0 50 100 150 200 250 300

Even

ts / B

in210

310

410DataWVW/Z+l.f.W/Z+h.f.TopMultijetsUncert.

= 0.1,λ = 0.2γκΔ

-1DØ, L = 4.3 fb(a)

(GeV)llT

p0 20 40 60 80 100 120140 160 180200

Even

ts / B

in

05

101520253035 Data

WZDYOther bgd.Uncert.

= -0.05,λ = -0.06Z

1 gΔ

-1DØ, L = 8.6 fb(b)

FIG. 1: (color online) (a) The pjjT distribution summed over electron and muon channels from WW +WZ → !νjj (l = µ, e)production for data and SM MC predictions (“l.f.” denotes light partons such as u, d, s or gluon, and “h.f.” denotes heavy-flavor such as c or b). Also shown are expected distributions for an ATGC model with ∆κγ = 0.2, and λ = 0.1. (b) The pllTdistribution summed over eee, eµµ, µee and µµµ channels from WZ → !ν!! production for data, SM MC predictions and forATGC model with λ = −0.05 and ∆gZ1 = −0.06.

γκ Δ-0.4 -0.2 0 0.2 0.4

λ

-0.15-0.1

-0.050

0.050.1

0.150.2

0.250.3 -1DØ, L = 4.3 fb

(a)

LEP parameterization

Standard ModelMinimum95% Contour68% Contour

γκ Δ-0.4 -0.2 0 0.2 0.4

Z 1 gΔ

-0.15-0.1

-0.050

0.050.1

0.150.2

0.250.3

0.35-1DØ, L = 4.3 fb

(b)



λ-0.1 -0.05 0 0.05 0.1

Z 1 gΔ

-0.15-0.1

-0.050

0.050.1

0.150.2

0.250.3

-1DØ, L = 4.3 fb(c)



γκ Δ-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3

λ

-0.1

-0.05

0

0.05

0.1

0.15

0.2-1DØ, L = 4.3 fb

(d)

Equal couplings parameterization


FIG. 2: WW +WZ → !νjj (l = µ, e). The 68% and 95% C.L. two-parameter limits on the γWW/ZWW coupling parametersassuming the LEP (a, b, c) and equal couplings parameterization (d) with Λ = 2 TeV. Black circles indicate the most probablevalues of an ATGCs from the two-parameter fit.

of systematic uncertainties on separate samples and sub-channels due to the same uncertainty are assumed to be100% correlated but different uncertainties are assumedto be uncorrelated.

The 68% and 95% C.L. limits on ATGCs from the4.3 fb−1 analysis of WW + WZ → !νjj final states inthe two-parameter space are shown in Fig. 2. The limitsfrom the 8.6 fb−1 analysis of WZ → !ν!! final states

are presented only in the λ − ∆gZ1 space as shown inFig. 3, because WZ production is weakly sensitive to∆κγ via the relation given by Eq. (3). The 95% C.L.one-parameter limits, obtained from single parameter fitswith all other parameters fixed to their SM values arepresented in Table I.

The resulting 68% and 95% C.L. one-parameter lim-its from the combined fit of !νγ, !ν!ν, !νjj, and !ν!!

DØ ℓνjj 4.3 fb-1

PLB

718 (2012) 451

s


TGC results

TGCs consistent with the SMn Four of the WWZ and WWγ couplings are constrained to O(0.05)▶ Caveat: LEP scenario is used▶ Δκγ remains less precise

n ZZZ and ZZγ couplings are constrained by the LHC results to O(0.01)

8 TeV data not included yet

16

WWZ couplings WWγ couplings

ZZZ and ZZγ couplings


Does it make sense?Shouldn’t the WW “excess” show up as anomalous TGCs?

n TGC sensitivity is concentrated in the highest bins of leading lepton pT

n Excesses are mostly at low pT where anomalous TGCs don’t contribute

If the excess is real, it’s not a kind of physics described with aTGCsn i.e. not a heavy new particle in s-‐channel loop diagrams

17

arXiv:1210.2979

ATLAS 7 TeV

5

(GeV)Tmax

p0 20 40 60 80 100 120 140 160 180 200

even

ts/b

in

0

20

40

60

80

100

120

140

160

180 Data WW WZ + ZZ Top Fakes Z+jets

syst.) ⊕ (stats.σ

CMS preliminary 2012-1 = 7 TeV, L = 4.92 fbs

(GeV)Tmax

p0 20 40 60 80 100 120 140 160 180 200

Rat

io

0

1

2 (GeV)

Tminp

0 20 40 60 80 100 120 140 160

even

ts/b

in

0

50

100

150

200

250

300

350




(GeV)Tmin

p0 20 40 60 80 100 120 140 160

Rat

io

0

1

2

(GeV)llT

p40 60 80 100 120 140 160 180 200

even

ts/b

in

0

50

100

150

200




(GeV)llT

p40 60 80 100 120 140 160 180 200

Rat

io

0

1

2

(GeV)invM0 20 40 60 80 100 120 140 160 180 200

even

ts/b

in

0

20

40

60

80

100

120




(GeV)invM0 20 40 60 80 100 120 140 160 180 200

Rat

io

0

1

2

Figure 1: Distributions of the leading lepton transverse momentum (pTmax), the trailing lep-ton transverse momentum (pTmin), and the dilepton transverse momentum (p``T ) and invariantmass (M``) at the final selection level, reweighted to the data-driven estimates. All four chan-nels (ee, µµ, eµ and µe) are combined, and the uncertainty band corresponds to the statisticaland systematic uncertainties on the predicted yield.

5 Efficiencies and Systematic Uncertainties

The signal efficiency is estimated using simulation, considering both the qq ! W+W� andgg ! W+W� prcosses. Residual discrepancies in the lepton reconstruction and identificationefficiency between data and simulation are corrected by determining data-to-simulation scalefactors measured using Z/g⇤! `+`� events in the Z peak region [25], recorded with dedicatedunbiased triggers. These factors depend on the lepton pT and |h| and are within four percent

CM

S-PA

S-S

MP

-12-005

CMS 7 TeV


SummaryWW, WZ, and ZZ measurements continue to improven CDF/DØ results with full Run-‐II datan ATLAS/CMS results with (full or partial) 8 TeV data

Data are largely consistent with the SM predictionn WW cross sections at the LHC slightly higher than expected

▶ This has spurred theoretical investigations▶ Jet veto is the leading source of experimental/theoretical uncertainties▶ Analyze full 8 TeV data!

TGCs show no deviation from the SMn Four (out of 5) WWZ and WWγ couplings constrained at ~0.05 level

▶ Δκγ remains less precise, but improving

n All ZZZ and ZZγ couplings constrained at ~0.01 leveln 8 TeV data have yet to be used ➔ Expect improvements soon

18

Backup Slides


Two-‐photon WW productionNew measurement from CMS of γγ → WWn Final state is e±μ∓ with no other tracks from the event vertex

n pT(eμ) > 30 GeV suppresses ττ background

n 2 events observed. 2.2 ± 0.5 signal and 0.84 ± 0.13 (stat.) background expected

Limits are set on anomalous quartic gauge couplings (aQGCs)

20

14 8 Results

Correcting for efficiency, acceptance, and backgrounds, the result interpreted as a cross sectiontimes branching fraction is:

s(pp ! p(⇤)W+W�p(⇤) ! p(⇤)µ±e⌥p(⇤)) = 2.1+3.1�1.9 fb,

with a significance of 1.1s. With statistical uncertainties only, the resulting value of the crosssection times branching fraction is 2.1+3.0

�1.9(stat.) fb. The SM prediction is 3.8 ± 0.9 fb, includingthe uncertainty on the contribution of proton dissociation.

The pT(µ±e⌥) distribution for events with zero extra tracks is shown in Figure 12. In the AQGCsearch region pT(µ±e⌥) > 100 GeV, zero events are observed in data, consistent with the Stan-dard Model expectation of 0.14, dominated by pp ! p(⇤)W+W�p(⇤).

) [GeV]µ(eT

p0 50 100 150 200 250 300

Even

ts/3

0 G

eV

0

5

10

15

20

25

30

=500GeV)Λ (a0W=2E-4, aCW=0, -W+ W→ γγ

=500GeV)Λ (a0W=-2E-4, aCW=-8E-4, -W+ W→ γγ

Data -τ+τDrell-Yan -W+Inclusive W -W+Diffractive W

-τ+τ → γγElastic -τ+τ → γγInelastic

tt W+jets

(SM)-W+ W→ γγ

-1=7 TeV, L=5.05 fbsCMS Preliminary 2011,

vertex)µN(extra tracks, e0 2 4 6 8 10 12 14

Even

ts

0

5

10

15

20

25

=500GeV)Λ (a0W=2E-4, aCW=0, -W+ W→ γγ

=500GeV)Λ (a0W=-2E-4, aCW=-8E-4, -W+ W→ γγ

Data -τ+τDrell-Yan -W+Inclusive W -W+Diffractive W

-τ+τ → γγElastic -τ+τ → γγInelastic

tt W+jets

(SM)-W+ W→ γγ


Figure 12: Full pT(µ±e⌥) distribution for events with 0 extra tracks (left) and multiplicity of extratracks for events with pT(µ±e⌥) > 100 GeV (right). The backgrounds (solid histograms) are stackedwith statistical uncertainties indicated by the shaded region, the signal histogram (open histogram) isstacked on top of the backgrounds. The expected signal is shown for the SM gg ! W+W� signal (solidlines), and for two representative values of the anomalous couplings aW

0 /L2 and aWC /L2 (dotted and

dashed lines).

We find the selection efficiency does not vary strongly between the SM and AQGC samplestested within the acceptance (Table 5), and therefore set an upper limit on the partial cross sec-tion times branching fraction for gg ! W+W� ! µ±e⌥ with pT(µ, e) > 20 GeV, |h(µ, e)| < 2.4,and pT(µ±e⌥) > 100 GeV. We treat the residual SM pp ! p(⇤)W+W�p(⇤) signal as a back-ground, resulting in a total of 0.14 ± 0.02 expected events, and include an additional systematicuncertainty of 10% based on the maximum relative variation of the efficiency between the SMand samples generated with two values of the anomalous couplings.

We find an upper limit of 3 events, corresponding to a 95% CL upper limit on the partial crosssection times branching fraction with the selection cuts pT(µ, e) > 20 GeV, |h(µ, e)| < 2.4, andpT(µ±e⌥) > 100 GeV:

s(pp ! p(⇤)W+W�p(⇤) ! p(⇤)µ±e⌥p(⇤)) < 1.9 fb.

In Figure 13, the excluded value is compared to the predicted cross section for non-zero valuesof aW

0 /L2 and aWC /L2 within the defined acceptance, scaled to include the contribution from

proton dissociation. The theoretical prediction is determined from the fully simulated samples,

9

Selection step Signal e ⇥ A Visible cross section (fb) Events in dataTrigger and preselection 28.5% 1.4 9086m(µ±e⌥) > 20 GeV 28.0% 1.4 8200Muon ID and Electron ID 22.6% 1.1 1222µ±e⌥ vertex with 0 extra tracks 13.7% 0.7 6pT(µ±e⌥) > 30 GeV 10.6% 0.5 2

Table 2: Signal efficiency ⇥ acceptance and number of events selected in data at each stage of theselection. The preselection corresponds to requiring a reconstructed muon and electron of oppositecharge, each having pT > 15 GeV and |h| < 2.4, matched to a common primary vertex with less than 15additional tracks.

) [GeV]µ(eT

p0 50 100 150 200 250 300

Even

ts/6

GeV

-210

-110

1

10

210

310Data -τ+τDrell-Yan

-W+Inclusive W -W+Diffractive W-τ+τ → γγElastic -τ+τ → γγInelastic

tt W+jets (SM)-W+ W→ γγ


Figure 7: µ±e⌥ pair transverse momentum. The events plotted are required to pass the trigger andpreselection requirements, and the lepton identification. The shaded bands indicate the statistical un-certainty on the background estimation.

to the other backgrounds, without accounting for any survival probabilities or overlap with theinclusive W+W� sample. To study the W+jets backgrounds, where the contribution is mainlyfrom fake leptons or non-prompt leptons in jets, we select a control sample of events withpT(µ±e⌥) > 30 GeV and at least one of the two leptons failing the nominal offline identification.This sample is then normalized to the simulation in the high-multiplicity (more than 6 extratracks) region, and used to estimate the W+jets background in the signal and inclusive W+W�

control regions. Figure 8 shows the distribution of the number of extra tracks for the W+W�

region with pT(µ±e⌥) > 30 GeV. In Figure 9, the invariant mass and acoplanarity of the eventswith 1-6 extra tracks are plotted. In general the data is consistent with the sum of simulatedbackgrounds.

The corresponding extra track multiplicity distribution for the Drell-Yan t+t�-dominated re-gion with pT(µ±e⌥) < 30 GeV is shown in Figure 8. We find a deficit in data compared tosimulation in the 1-6 tracks region. In the t+t� sample with zero extra tracks we find 4events in data, compared to an MC background expectation of 2.5 events, plus 0.9 events ofgg ! W+W� signal. The expected contribution to the background from gg ! t+t� is ap-proximately 0.7 events. The invariant mass and acoplanarity distributions are plotted in Fig-ure 10.

Table 3 summarizes the observed and expected background event yields for the three orthogo-nal control regions. Due to tracks from pile-up vertices being wrongly associated to the µ±e⌥

CMS 7 TeV

CM

S-PA

S-FS

Q-12-010CMS 7 TeV

No othertracks

15

with one anomalous quartic coupling parameter fixed to zero, and a second order polynomialinterpolation of the other. With a dipole form factor of L = 500 GeV, the limits obtained on theAQGC parameters are:

�0.00017 < aW0 /L2 < 0.00017 GeV�2 (aW

C /L2 = 0, L = 500 GeV),�0.0006 < aW

C /L2 < 0.0006 GeV�2 (aW0 /L2 = 0, L = 500 GeV),

which are approximately two orders of magnitude more stringent than the limits obtained atLEP [7, 10, 12].

We perform a similar procedure to scan the two dimensional space of aW0 /L2 and aW

C /L2, usinga large number of samples generated with a fast simulation of the CMS detector [44]. For eachpoint the cross section and the number of events passing generator selection requirements isused to calculate the observed cross section times branching fraction. This procedure onlydepends on the generator level prediction, with the fast simulation used to confirm that thesignal efficiency of all trigger, reconstruction, and analysis selections, relative to the acceptanceis flat across AQGC sample space. The result is shown in Figure 14, where the area outside theellipse corresponds to values of the anomalous couplings that would result in a partial crosssection times branching fraction above 1.9 fb, including the form factor with L = 500 GeV.The result of Ref. [7], obtained from a maximum likelihood fit to a combination of WWg andWW ! gg channels in e+e� collisions, is shown in the inset.

We also obtain the corresponding limits without form factors. In this case the cross section isdominated by the region of high energy gg interactions, above the unitarity bound. This leadsto limits on the anomalous couplings much smaller than in the scenario with form factors:

�2.80 ⇥ 10�6 < aW0 /L2 < 2.80 ⇥ 10�6 GeV�2 (aW

C /L2 = 0, no form factor),�1.02 ⇥ 10�5 < aW

C /L2 < 1.02 ⇥ 10�5 GeV�2 (aW0 /L2 = 0, no form factor),

]-2 [GeV2Λa0W/-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25

-310×

BF

[fb]

× σ

0

0.5

1

1.5

2

2.5


= 500 GeVΛ

) > 20 GeVµ(e,T

p

)| < 2.4µ(e,η|

e) > 100 GeVµ(T

p

]-2 [GeV2ΛaCW/-0.0015 -0.001 -0.0005 0 0.0005 0.001 0.0015

BF

[fb]

× σ

0

0.5

1

1.5

2

2.5


= 500 GeVΛ

) > 20 GeVµ(e,T

p

)| < 2.4µ(e,η|

e) > 100 GeVµ(T

p

Figure 13: Expected partial cross section times branching fraction with L = 500 GeV as a functionof aW

0 /L2 with aWC /L2 fixed to 0 (left), and aW

C /L2 with aW0 /L2 fixed to 0 (right). The prediction

(solid line) and excluded value (dashed line) are both shown for pT(µ, e) > 20 GeV, |h(µ, e)| < 2.4,pT(µ±e⌥) > 100 GeV. The prediction is rescaled to include the contribution from proton dissociation.

Documents

WZ,and ZZProduction - Harvard University · 2016. 7. 29. · ATLAS ATLAS-CONF-2013-021 WZ → ℓνℓℓ [13 fb-1] 8 TeV ATLAS-CONF-2013-020 ZZ → ... 00 50 100 150 200 250 300