measurements of inclusive w and z production in atlas

Measurements of Inclusive W and Z Production in ATLAS

Jack Goddard on behalf of the ATLAS Collaboration

LHCP 2013, Barcelona

17th May 2013

1

Outline

• Present a number of current W and Z inclusive measurements made at ATLAS

• Inclusive W and Z Cross Sections

• W Charge Asymmetry: Comparing to LHCb and CMS

• Determination of the Strange Quark Density of the Proton

• High Mass Drell-Yan

• Other talks at LHCP on ATLAS standard model measurements are:

• Lea Caminada: “Measurements of vector bosons plus jets production with the ATLAS detector” - QCD Parallel 1

• Suen Hou: “Measurement of diboson production with the ATLAS detector” - Higgs 3 + EW

2

Measurement of the Inclusive W and Z/γ* Cross-Sections in e/μ Decay Channels

• Cross section measurements have been made of the W and Z bosons

• electron and muon channels

• using 35pb-1 of 2010 data.

• integrated cross sections

• dσ/d|ηl| for W decays

• dσ/d|yZ| for Z decays

• both fiducially and extrapolated to the full phase space

• compared to NNLO theory using various PDF sets

• W± charge asymmetry is also measured 3

Phys. Rev. D85 (2012) 072004arXiv:1109.5141

http://prd.aps.org/abstract/PRD/v85/i7/e072004


http://arxiv.org/abs/1109.5141


• Combined e/μ cross sections with kinematic requirements:

• Z: 66 < mll < 116, pTl > 20 GeV

• W: pTl > 20 GeV, pTν > 25 GeV, mTlν > 40 GeV

• Data is broadly described by the theory


4

Phys. Rev. D85 (2012) 072004arXiv:1109.5141

• Comparisons to NNLO theory with FEWZ and DYNNLO using the Gμ scheme with different PDFs• FEWZ and DYNNLO agree to 1% for fiducial cross sections• FEWZ: yZ distribution, ηl normalisation• DYNNLO: ηl shape (as DYNNLO has higher statistical

precision)• Uncertainties on the theory:

• 68% C.L PDF uncertainties• <0.5% numerical uncertainty





• Integrated cross sections are all measured to ~1% systematic uncertainty (excluding 3.4% luminosity)

• Theory comparisons using NNLO FEWZ

• Exhibits sensitivity to differences in the predicted cross sections


5

Phys. Rev. D85 (2012) 072004arXiv:1109.5141

• Combination also performed with the total integrated cross sections.

• Extrapolation uncertainty ~2% making the total integrated cross sections less useful for discriminating between PDFs

• Ratios of cross sections also examined

• For example, ratio of the two decay channels for each process, serving as a test for electron-muon universality





Extrapolation of W charge Asymmetry • Measurement of the W charge asymmetry made in the 35

pb-1 2010 W/Z paper for electron and muon channels

• pTl > 20 GeV, pTν > 25 GeV, mTlν > 40 GeV

• Selection gives excellent purity.

• But cannot be compared to measurements made by CMS or LHCb due to mTlν cut

• Fiducial cross-section have been extrapolated to a new fiducial region corresponding to pTl > 20 GeV

• Extrapolation factors determined entirely from simulation.

• Using MC@NLO with CTEQ6.6

• Cross sections increase by 20% with an uncertainty of 1% of the extrapolation factors 6

ATLAS-CONF-2011-129

17

are compared in Figs. 12 and 13 with the calculatedNNLO predictions using the JR09, ABKM09, HERA-PDF1.5 and MSTW08 NNLO PDF sets. The uncertain-ties of the bin-wise predictions are a convolution of thePDF uncertainties, considered by the authors of the vari-ous PDF sets 2 to correspond to 68% C.L., and a residualnumerical uncertainty of below 0.5%. One observes thatthe measured y

Z

and ⌘

`

dependencies are broadly de-scribed by the predictions of the PDF sets considered.Some deviations, however, are visible, for example thelower Z cross section at central rapidities in the case ofthe JR09 PDF set, or the tendency of the ABKM09 pre-diction to overshoot the Z and the W cross sections atlarger y

Z

and ⌘

`

, respectively. It thus can be expectedthat the di↵erential cross sections presented here will re-duce the uncertainties of PDF determinations and alsoinfluence the central values.

The combined electron and muon data allow for anupdate of the measurement of the W charge asymmetry

A

`

(⌘`

) =d�

W

+

/d⌘`

� d�W

�/d⌘

`

d�W

+

/d⌘`

+ d�W

�/d⌘

`

, (5)

which was previously published [26] by ATLAS based

|Z

|y0 0.5 1 1.5 2 2.5 3 3.5

Theo

ry/D

ata

0.91

1.1|

Z|y

0 0.5 1 1.5 2 2.5 3 3.5

| [pb

]Z

/d|y

σd

20

40

60

80

100

120

140

160

= 7 TeV)sData 2010 (

MSTW08

HERAPDF1.5

ABKM09

JR09

-1 L dt = 33-36 pb∫-l+ l→Z

Uncorr. uncertainty

Total uncertainty

ATLAS

FIG. 12. Di↵erential d�/d|yZ

| cross section measurement forZ ! `` compared to NNLO theory predictions using vari-ous PDF sets. The kinematic requirements are 66 < m

``

<116 GeV and p

T,`

> 20 GeV. The ratio of theoretical predic-tions to data is also shown. Theoretical points are displacedfor clarity within each bin.

2 The HERAPDF analysis considers explicitly uncertainties due toparameterisation and fit parameter choices. This leads to some-what enlarged and asymmetric errors as compared to the genuineexperimental uncertainties, which in the HERAPDF analysis cor-respond to a change of �2 by one unit.

on initial muon measurements alone. The asymmetryvalues, obtained in the W fiducial region of this analy-sis, and their uncertainties are listed in Tab. XXVI. Themeasurement accuracy ranges between 4 and 8%. Theprevious and the new measurements are consistent. Sincethe present measurement is more precise and relies on thesame data taking period, it supersedes the previous re-sult.

Figure 14 shows the measured W charge asymmetrytogether with the NNLO predictions obtained from the

|lη|

0 0.5 1 1.5 2 2.5

Theo

ry/D

ata

0.91

1.1|lη|

0 0.5 1 1.5 2 2.5

| [pb

]lη

/d|

σd

300

400

500

600

700

800

= 7 TeV)sData 2010 (MSTW08HERAPDF1.5ABKM09JR09

-1 L dt = 33-36 pb∫ lν+ l→+W

Uncorr. uncertainty

Total uncertainty

ATLAS

|lη|

0 0.5 1 1.5 2 2.5

Theo

ry/D

ata

0.91

1.1|lη|

0 0.5 1 1.5 2 2.5

| [pb

]lη

/d|

σd

100

200

300

400

500

600

= 7 TeV)sData 2010 (MSTW08HERAPDF1.5ABKM09JR09

-1 L dt = 33-36 pb∫ lν- l→

-W

Uncorr. uncertainty

Total uncertainty

ATLAS

FIG. 13. Di↵erential d�/d|⌘`

+

| (top) and d�/d|⌘`

� | (bot-tom) cross section measurements for W ! `⌫ compared tothe NNLO theory predictions using various PDF sets. Thekinematic requirements are p

T,`

> 20 GeV, pT,⌫

> 25 GeVand m

T

> 40 GeV. The ratio of theoretical predictions todata is also shown. Theoretical points are displaced for clar-ity within each bin.

http://cdsweb.cern.ch/record/1383832


Extrapolation of W charge Asymmetry

• Comparison to LHCb and CMS is made

• See good agreement with the other experiments and theory curves

• Maximum in asymmetry around |ηl|~2.0 clearly seen

• Turn over in asymmetry at |ηl|~3.0 also clearly seen

• Both features related to the kinematics of the V-A decay of the W boson, its polarisation, and different fractions of W+ and W- produced by higher-x sea quarks

• For detailed comparisons to theory the un-extrapolated measurement is more appropriate due to smaller uncertainties.

7

ATLAS-CONF-2011-129



Determination of the Strange Quark Density of the Proton from W→lν and Z→ll Cross-Sections• Interprets the 35 pb-1 2010 W/Z analysis

• Flavour SU(3) symmetry suggests that the light sea quark distributions are equal

• But strange quarks might be suppressed due to their larger mass

• Analysis of neutrino scattering measurements at NuTeV and CCFR give results both in favour and against suppression

• HERAFitter framework used to fit the ATLAS data and the HERA e±p measurements.

• Quark and gluon distributions in the initial scale (Q02 = 1.9 GeV) are expressed as:

• Two types of NNLO fit are performed

• Fixed s: strange quark distribution fully coupled to the down sea quarks by setting s/d = 0.5 at the initial scale.

• Free s: xs is parameterised with Ps = 1 and Bs = Bd , leaving 2 free free strangeness parameters.

8Phys.Rev.Lett. 109 (2012) 012001arXiv:1203.4051

2

in Ref. [14], and the results are cross-checked betweenthe FEWZ [19] and the DYNNLO [18] programs. TheHERAFitter package uses the APPLGRID code [25] in-terfaced to the MCFM program [26] for fast calculationof the di↵erential W and Z boson cross sections at NLOand a K-factor technique to correct from NLO to NNLOpredictions. The data are compared to the theory usingthe �

2 function defined in Refs. [27–29].The evolution equations yield the PDFs at any value

of Q2 given that they are parametrized as functions of xat an initial scale Q

2

0

. In the present analysis, this scaleis chosen to be Q

2

0

= 1.9 GeV2 such that it is below thecharm mass threshold m

2

c

. The heavy quark masses arechosen to be m

c

= 1.4 GeV and m

b

= 4.75 GeV. Thestrong coupling constant is fixed to ↵

S

(MZ

) = 0.1176, asin Ref. [5]. A minimum Q

2 cut of Q2

min

� 7.5 GeV2 isimposed on the HERA data.

The quark distributions at the initial scale are repre-sented by the generic form

xq

i

(x) = A

i

x

Bi(1� x)CiP

i

(x), (1)

where P

i

(x) denotes polynomials in powers of x. Theparametrized quark distributions, xq

i

, are chosen to bethe valence quark distributions (xu

v

, xd

v

) and the lightanti-quark distributions (xu, xd, xs). The gluon dis-tribution is parametrized with the more flexible formxg(x) = A

g

x

Bg (1� x)CgP

g

(x)� A

0g

x

B

0g (1� x)C

0g , where

C

0g

is set to 25 to suppress negative contributions at highx. The parameters A

uv and A

dv are fixed using the quarkcounting rule and A

g

using the momentum sum rule. Thenormalization and slope parameters, A and B, of u andd are set equal such that xu = xd at x ! 0. Terms areadded in the polynomial expansion P

i

(x) only if requiredby the data, following the procedure described in Ref. [5].This leads to one additional term, P

uv (x) = 1 + E

uvx2.

Two types of NNLO fit, termed epWZ, are performedwith di↵erent treatments of strangeness. First, thestrange quark distribution is fully coupled to the downsea quark distribution and suppressed by fixing s/d = 0.5at the initial scale Q

2

0

(“fixed s fit”) as suggested byRefs. [5, 8–11]. In a second fit, xs is parametrized asin Eq. 1, with P

s

= 1 and B

s

= B

¯

d

, leaving two freestrangeness parameters, A

s

and C

s

(“free s fit”). Bydefault it is assumed that xs = xs.

Both fits result in good overall �

2

/N

DF

values of546.1/567 with 13 free parameters, for fixed s, and of538.4/565 with 15 free parameters, for free s. For thefixed s fit, the partial �

2 of the ATLAS data is 44.5for 30 data points. This improves significantly to 33.9for the fit with free s. This fit determines the value ofr

s

= 0.5(s+ s)/d to be

r

s

= 1.00±0.20exp±0.07mod

+0.10

�0.15

par

+0.06

�0.07

↵S±0.08th, (2)

at Q2

0

and x = 0.023, the x value, which corresponds tox = 0.013 at Q

2 = M

2

Z

as a result of QCD evolution.The combined result is r

s

= 1.00+0.25

�0.28

.

The uncertainty of rs

(Eq. 2) is dominated by the ex-perimental (exp) uncertainty, which is mostly driven bythe statistical and systematic uncertainties of the W andZ cross section measurements. The model (mod) uncer-tainty includes e↵ects due to variations (1.25 < m

c

<

1.55GeV and 4.5 < m

b

< 5.0GeV) of the charm andbeauty quark masses following Ref. [30], of the minimumQ

2 cut value (5 < Q

2

min

< 10GeV2), and the value of thestarting scale (Q2

0

lowered to 1.5GeV2). The largest con-tribution to the model uncertainty of ±0.05 comes fromthe variation of the charm quark mass. The parametriza-tion (par) uncertainty corresponds to the envelope of theresults obtained with the polynomials P

i

, in Eq. 1, ex-tended by one or two terms, resulting in somewhat di↵er-ent parton distributions with similar �2 as for the nom-inal fit. The parametrization uncertainty also includesa fit with B

s

free. The ↵

s

uncertainty corresponds toa variation of ↵

s

(MZ

) from 0.114 to 0.121. Finally, atheoretical (th) uncertainty is assessed by comparing theDYNNLO and FEWZ predictions on the Z, W+ andW

�

fiducial cross sections, which agree at the level of 0.2, 0.5and 1.0%, respectively. In addition, remaining missingpure electroweak corrections may alter the QCD predic-tions at the per mille level. Both e↵ects are well coveredby an uncertainty of 1% on the W/Z cross section ratioand this results in a theoretical uncertainty on r

s

of 0.08.The fits impose small shifts, typically much smaller

than one standard deviation, on the correlated system-atic uncertainties of the data. The global normalizationis observed to be shifted upwards for both fits by aboutthe size of the luminosity measurement uncertainty. TheW

± and Z cross section measurements are compared inFig. 1 to the NNLO fit results, after these shifts are ap-plied to the predictions. Also shown are the ratios of thefits with free s and with fixed s. It is apparent that theenhanced strange quark fraction in the free s fit has nosignificant e↵ect on the prediction of the ⌘

l

distributionsfor both the W

+ and W

� decay leptons, while it leadsto an improvement in the prediction of the y

Z

distribu-tion. An improvement is also observed in the descriptionof the ratio of the (W+ +W

�) to the Z boson cross sec-tions in the fiducial phase space. This is predicted to be11.10 in the fit with fixed s, while the measured value of10.70± 0.15 is almost exactly reproduced in the fit withfree s, which gives a value of 10.74.

In order to check the robustness of the present resultfor r

s

, a series of cross checks is performed. A fit withoutallowing an adjustment of the correlated errors yields avalue of r

s

= 0.97±0.26exp, in good agreement with Eq. 2.A fit with identical input parameters is repeated at NLOand also yields a consistent result: r

s

= 1.03 ± 0.19exp.If this NLO fit is performed with a massless heavy quarktreatment then r

s

= 1.05 ± 0.19exp is obtained. In aseparate NLO study, the constraint (xu � xd) ! 0 forx ! 0 is relaxed. The xd(x) distribution is found tobe consistent with xu(x), albeit with large uncertainties

2




2

0


2

0


2

c


c

= 1.4 GeV and m

b


S

(MZ


2 cut of Q2

min



xq

i

(x) = A

i

x

Bi(1� x)CiP

i

(x), (1)

where P

i


i


v

, xd

v


g

x

Bg (1� x)CgP

g

(x)� A

0g

x

B

0g (1� x)C

0g , where

C

0g


uv and A


g


i


uv (x) = 1 + E

uvx2.


2

0


s

= 1 and B

s

= B

¯

d


s

and C

s



2

/N

DF



s

= 0.5(s+ s)/d to be

r

s

= 1.00±0.20exp±0.07mod

+0.10

�0.15

par

+0.06

�0.07

↵S±0.08th, (2)

at Q2

0


2 = M

2

Z


s

= 1.00+0.25

�0.28

.



c

<

1.55GeV and 4.5 < m

b


2 cut value (5 < Q

2

min


0


i


s

free. The ↵

s


s

(MZ


�


s




l


+ and W


Z




s


s


s


s


http://prl.aps.org/abstract/PRL/v109/i1/e012001




Determination of the Strange Quark Density of the Proton from W→lν and Z→ll Cross-Sections

• Free sbar fit determines value of

• Found to be:

• at Q02 and x = 0.023 ( x value that corresponds to x = 0.013 at Q2 = MZ2 due to PDF evolution)

• Compared to results from a number of different PDF sets.

• Enlarged fraction of s quark sea leads to decrease in up and down quark densities (by ~10%)

• Total sea xΣ is correspondingly enhanced by ~8%

9

2




2

0


2

0


2

c


c

= 1.4 GeV and m

b


S

(MZ


2 cut of Q2

min



xq

i

(x) = A

i

x

Bi(1� x)CiP

i

(x), (1)

where P

i


i


v

, xd

v


g

x

Bg (1� x)CgP

g

(x)� A

0g

x

B

0g (1� x)C

0g , where

C

0g


uv and A


g


i


uv (x) = 1 + E

uvx2.


2

0


s

= 1 and B

s

= B

¯

d


s

and C

s



2

/N

DF



s

= 0.5(s+ s)/d to be

r

s

= 1.00±0.20exp±0.07mod

+0.10

�0.15

par

+0.06

�0.07

↵S±0.08th, (2)

at Q2

0


2 = M

2

Z


s

= 1.00+0.25

�0.28

.



c

<

1.55GeV and 4.5 < m

b


2 cut value (5 < Q

2

min


0


i


s

free. The ↵

s


s

(MZ


�


s




l


+ and W


Z




s


s


s


s


Phys.Rev.Lett. 109 (2012) 012001arXiv:1203.4051

2




2

0


2

0


2

c


c

= 1.4 GeV and m

b


S

(MZ


2 cut of Q2

min



xq

i

(x) = A

i

x

Bi(1� x)CiP

i

(x), (1)

where P

i


i


v

, xd

v


g

x

Bg (1� x)CgP

g

(x)� A

0g

x

B

0g (1� x)C

0g , where

C

0g


uv and A


g


i


uv (x) = 1 + E

uvx2.


2

0


s

= 1 and B

s

= B

¯

d


s

and C

s



2

/N

DF



s

= 0.5(s+ s)/d to be

r

s

= 1.00±0.20exp±0.07mod

+0.10

�0.15

par

+0.06

�0.07

↵S±0.08th, (2)

at Q2

0


2 = M

2

Z


s

= 1.00+0.25

�0.28

.



c

<

1.55GeV and 4.5 < m

b


2 cut value (5 < Q

2

min


0


i


s

free. The ↵

s


s

(MZ


�


s




l


+ and W


Z




s


s


s


s






Measurement of Angular Correlations in Drell-Yan Lepton Pairs

• Observable Φη* defined as:

• Dependant only on the direction of the two leptons, which is better measured than their momenta.

• Has been used previously at the DØ experiment

• Correlated to pTZ/mll so probes the same physics as pTZ

• Differential cross-section measured as a function of Φη* in the electron and muon channel

• Compared to RESBOS which resums the NNNL and matches to LO result. k-factors correct to NNLO. 10

Phys. Lett. B 720 (2013) 32-51arXiv:1211.6899

Measurement of angular correlations in Drell–Yan lepton pairs to probe Z/γ∗ boson

transverse momentum at√s = 7 TeV with the ATLAS detector

The ATLAS Collaboration

Abstract

A measurement of angular correlations in Drell–Yan lepton pairs via the φ∗

η observable is presented. This variable probesthe same physics as the Z/γ∗ boson transverse momentum with a better experimental resolution. The Z/γ∗ → e+e−

and Z/γ∗ → µ+µ− decays produced in proton–proton collisions at a centre-of-mass energy of√s = 7 TeV are used.

The data were collected with the ATLAS detector at the LHC and correspond to an integrated luminosity of 4.6 fb−1.Normalised differential cross sections as a function of φ∗

η are measured separately for electron and muon decay channels.These channels are then combined for improved accuracy. The cross section is also measured double differentially as afunction of φ∗

η for three independent bins of the Z boson rapidity. The results are compared to QCD calculations andto predictions from different Monte Carlo event generators. The data are reasonably well described, in all measuredZ boson rapidity regions, by resummed QCD predictions combined with fixed-order perturbative QCD calculations orby some Monte Carlo event generators. The measurement precision is typically better by one order of magnitude thanpresent theoretical uncertainties.

Keywords:

Z Boson, Differential Cross Section, Perturbative QCD, Event Generators, Monte Carlo Models

1. Introduction

In hadron collisions at TeV energies the vector bosonsW and Z/γ∗ are copiously produced with non-zero mo-mentum transverse to the beam direction (pT) becauseof radiation of quarks and gluons from the initial-statepartons. In this context the signatures Z/γ∗ → e+e−

and Z/γ∗ → µ+µ− provide an ideal testing ground forQCD due to the absence of colour flow between the initialand final state [1–3]. The study of the low pZT spectrum(pZT < mZ), which dominates the cross section, has im-portant implications on the understanding of Higgs bosonproduction since the transverse-momentum resummationformalism required to describe the Z/γ∗ boson cross sec-tion is valid also for the Higgs boson [4–7]. A precise un-derstanding of the pZT spectrum is also necessary to furtherimprove the modelling ofW boson production in QCD cal-culations and Monte Carlo (MC) event generators, sincethe measurement of the W mass is directly affected byuncertainties in the pWT shape [8, 9].The transverse momentum spectra of W and Z/γ∗

bosons produced via the Drell–Yan mechanism have beenextensively studied by the Tevatron collaborations [10–14]and, recently, also by the LHC experiments [15–17]. How-ever, the precision of direct measurements of the Z/γ∗

spectrum at low pZT at the LHC and the Tevatron is lim-ited by the experimental resolution and systematic uncer-tainties rather than by the size of the available data sam-ples. This limitation affects the choice of bin widths andthe ultimate precision of the pZT spectrum. In recent years,

additional observables with better experimental resolutionand smaller sensitivity to experimental systematic uncer-tainties have been investigated [18–21]. The optimal ex-perimental observable to probe the low-pZT domain of Z/γ∗

production was found to be φ∗

η which is defined [20] as:

φ∗

η ≡ tan(φacop/2) · sin(θ∗η) , (1)

where φacop ≡ π − ∆φ, ∆φ being the azimuthal open-ing angle between the two leptons, and the angle θ∗η isa measure of the scattering angle of the leptons with re-spect to the proton beam direction in the rest frame ofthe dilepton system. The angle θ∗η is defined [20] bycos(θ∗η) ≡ tanh[(η− − η+)/2] where η− and η+ are thepseudorapidities1 of the negatively and positively chargedlepton, respectively. Therefore, φ∗

η depends exclusively onthe directions of the two lepton tracks, which are bettermeasured than their momenta. The φ∗

η variable is posi-tive by definition. It is correlated to the quantity pZT/m"",where m"" is the invariant mass of the lepton pair, andtherefore probes the same physics as the transverse mo-mentum pZT [22]. Values of φ∗

η ranging from 0 to 1 probe

1ATLAS uses a right-handed coordinate system with its origin atthe nominal pp interaction point (IP) in the centre of the detectorand the z-axis along the beam pipe. The x-axis points from theIP to the centre of the LHC ring, and the y-axis points upward.Cylindrical coordinates (r,φ) are used in the transverse plane, φbeing the azimuthal angle around the beam pipe. The pseudorapidityis defined in terms of the polar angle θ as η = − ln tan(θ/2) and therapidity is defined as y = ln[(E + pz)/(E − pz)]/2.

Preprint submitted to Physics Letters B November 29, 2012




Abstract







Keywords:


1. Introduction








φ∗











Abstract







Keywords:


1. Introduction








φ∗








http://www.sciencedirect.com/science/article/pii/S0370269313000956




Measurement of Angular Correlations in Drell-Yan Lepton Pairs

• Electron and muon channels are combined with a χ2 minimisation technique

• Compared to RESBOS, FEWZ and a QCD calculation by A. Banfi et al.

• Difference between RESBOS and data is smaller than the PDF uncertainty on the RESBOS predictions (4% for Φη* <0.1 and 6% above)

• Double differential cross-section also measured in three |yZ| bins, with comparisons to different Monte Carlo generators (see backup slides)

11

bare reference points at particle level regarding QED FSR.The QED FSR corrections for the three levels are calcu-lated using Photos. The measured cross sections definedat the Z/γ∗ Born level are shown in Fig. 1 for the e+e−

and µ+µ− channels and are compared to predictions fromResBos.The normalised differential cross sections measured in

the fiducial acceptance for the two channels are combinedusing a χ2 minimisation method which takes into accountthe point-to-point correlated and uncorrelated systematicuncertainties [63–65] and correlations between electronand muon channels. The procedure allows a model inde-pendent check of the electron and muon data consistencyand leads to a significant reduction of the correlated un-certainties.

Table 2: The combined normalised differential cross section 1/σfid ·dσfid/dφ∗

η in bins of φ∗

η at Born level. The statistical (δstat) and totalsystematic (δsys) uncertainties are given in percent. The normaliseddifferential cross section extrapolated to the full lepton acceptance1/σtot ·dσtot/dφ∗

η is obtained at Born level by multiplication with the

inverse acceptance correction factor A−1c . The uncertainty δ(A−1

c )on this acceptance correction factor is also given in percent. Theoverall point-to-point uncorrelated additional uncertainty in QEDFSR of 0.3% is not included.

φ∗

η 1/σfid · dσfid/dφ∗

η δstat δsys A−1c δ(A−1

c )bin range [%] [%] [%]

0.000 – 0.004 9.77 0.30 0.21 1.06 3.80.004 – 0.008 9.73 0.30 0.20 1.06 3.00.008 – 0.012 9.41 0.31 0.18 1.06 3.70.012 – 0.016 9.21 0.31 0.22 1.06 2.40.016 – 0.020 8.82 0.31 0.16 1.05 2.50.020 – 0.024 8.49 0.32 0.18 1.05 2.20.024 – 0.029 8.01 0.29 0.18 1.05 1.80.029 – 0.034 7.56 0.30 0.14 1.04 2.40.034 – 0.039 7.07 0.31 0.15 1.04 2.20.039 – 0.045 6.52 0.30 0.14 1.03 2.20.045 – 0.051 5.97 0.31 0.13 1.02 2.80.051 – 0.057 5.52 0.32 0.16 1.01 2.10.057 – 0.064 5.02 0.31 0.13 1.01 1.90.064 – 0.072 4.54 0.31 0.18 1.00 2.00.072 – 0.081 4.03 0.31 0.13 0.99 1.80.081 – 0.091 3.56 0.31 0.15 0.99 1.00.091 – 0.102 3.15 0.32 0.16 0.98 1.10.102 – 0.114 2.731 0.32 0.17 0.97 1.30.114 – 0.128 2.347 0.32 0.19 0.97 1.30.128 – 0.145 1.996 0.32 0.16 0.96 1.70.145 – 0.165 1.677 0.32 0.19 0.95 2.00.165 – 0.189 1.355 0.32 0.16 0.95 2.70.189 – 0.219 1.084 0.32 0.15 0.94 2.30.219 – 0.258 8.24 · 10−1 0.33 0.15 0.94 2.90.258 – 0.312 5.95 · 10−1 0.33 0.14 0.93 2.90.312 – 0.391 3.96 · 10−1 0.33 0.14 0.92 3.40.391 – 0.524 2.282 · 10−1 0.34 0.15 0.92 3.50.524 – 0.695 1.169 · 10−1 0.42 0.18 0.92 4.40.695 – 0.918 5.78 · 10−2 0.52 0.23 0.93 4.00.918 – 1.153 2.92 · 10−2 0.71 0.29 0.94 5.31.153 – 1.496 1.52 · 10−2 0.81 0.33 0.98 10.51.496 – 1.947 7.13 · 10−3 1.04 0.40 1.04 10.31.947 – 2.522 3.54 · 10−3 1.30 0.49 1.11 17.52.522 – 3.277 1.77 · 10−3 1.61 0.58 1.19 16.2

The uncertainties due to the unfolding procedure, thepile-up, and QED FSR are considered to be completelycorrelated between the e+e− and µ+µ− channels. Theminimisation yields a total χ2 per degree of freedom

(ndof) of χ2/ndof = 33.2/34, indicating a good consis-tency between the electron and muon data. Measured val-ues of the combined normalised differential cross section1/σfid ·dσfid/dφ∗

η within the fiducial lepton acceptance arepresented in Table 2. At lower φ∗

η values the statisticaland systematic uncertainties are of the same order, whilstfor large φ∗

η values statistical uncertainties are dominating.The acceptance correction factors Ac needed to extrapo-late the measurement to the full lepton acceptance aredetermined using the Powheg simulation with the CT10PDF set and reweighted as a function of pZT to ResBospredictions. The uncertainty in Ac is estimated from theextreme differences among predictions obtained with Res-Bos, Mc@nlo, Sherpa, Alpgen, Herwig and Powheginterfaced to Pythia8. Uncertainties in Ac resulting fromPDF uncertainties are below 1%.

η*φ

-310 -210 -110 1

Data

(Pre

dict

ion)

/ RE

SBO

S

0.8

0.9

1

1.1

1.2

η*φ

-310 -210 -110 1

Data

(Pre

dict

ion)

/ RE

SBO

S

0.8

0.9

1

1.1

1.2 Data 2011-µ+µ + -e+eRESBOSA. Banfi et al.

= 7 TeVs| < 2.4η| > 20 GeVTp

< 116 GeV66 GeV < m

ATLAS-1 L dt = 4.6 fb∫

!

!

!!

η*φ

-310 -210 -110 1

Data

(Pre

dict

ion)

/ RE

SBO

S

0.8

0.9

1

1.1

1.2

η*φ

-310 -210 -110 1

Data

(Pre

dict

ion)

/ RE

SBO

S

0.8

0.9

1

1.1

1.2 Data 2011-µ+µ + -e+eRESBOS

)s αFEWZ 2.1 O( 2

= 7 TeVs| < 2.4η| > 20 GeVTp

< 116 GeV66 GeV < m

ATLAS-1 L dt = 4.6 fb∫

!

!

!!

Figure 2: The ratio of the combined normalised differential crosssection 1/σfid ·dσfid/dφ∗

η to ResBos predictions as a function of φ∗

η .The inner and outer error bars on the data points represent the sta-tistical and total uncertainties, respectively. The uncertainty due toQED FSR is included in the total uncertainties. The measurementsare also compared to predictions, which are represented by a dashedline, from Ref. [22] and from Fewz in the top and bottom panels,respectively. Uncertainties associated with these two calculationsare represented by shaded bands. The prediction from Fewz is onlypresented for φ∗

η > 0.1.

6

Phys. Lett. B 720 (2013) 32-51arXiv:1211.6899





Measurement of the High-Mass Drell-Yan Differential Cross-Section

• Measuring dσ/dmee (di-electron only)

• Invariant mass range: 116 < mee < 1500 GeV

• Fiducial region: |ηe| < 2.5, pTe > 25 GeV

• Full 2011 data set (4.9 fb-1) at √s = 7 TeV

• Complementary to the Z’ searches

• Extending Standard Model Z measurement (66–116 GeV)

• Measurement improving background for searches (DY: irreducible background for di-electron final states) 12

ATLAS-CONF-2012-159

• Kinematic Range:• Higher scale: x large → sensitive to

behaviour of parton distribution function (PDFs) in these regions

• Sensitive to higher order electroweak corrections




• Measured cross-section compared to FEWZ 3.1 at NNLO using the Gμ electroweak scheme.

• NLO electroweak corrections are applied to the theory comparisons

• LO Photon induced corrections have also been applied to the predictions (using MSTW2005qed PDF set)

• Comparison to a number of different PDFs is made

• Data is consistent with the nominal calculations for all PDF sets considered

13

ATLAS-CONF-2012-159



Summary

14

• Covered a number of W Z inclusive measurements at ATLAS.

• Much more could have been talked about

Measurement of the Z → ττ Cross-Section

Measurement of the Polarisation of the W Bosons Produced with Large pT

Measurement of τ Polarisation in the W → τν Decays

ATLAS-CONF-2012-006

Eur.Phys.J. C72 (2012) 2062arXiv:1204.6720

Eur. Phys. J. C72 (2012) 2001arXiv:1203.2165

• Much more to come: use of the 2011 and 2012 data sets will allow greater improvements on both the statistics and systematics of measurements



http://www.springerlink.com/content/3h24r0j573181876/

http://www.springerlink.com/content/3h24r0j573181876/



http://www.springerlink.com/content/v358r2163k706543/

http://www.springerlink.com/content/v358r2163k706543/



Back-up Slides

15


16


17


18

Determination of the Strange Quark Density of the Proton from W→lν and Z→ll Cross-Sections: Uncertainties

19


• Result for rs evolves to:

• At Q2 = MZ and x = 0.013

• More than twice as precise than at Q02.

• Smaller uncertainties due to gluon splitting probability into qq pairs is flavour independent - thus reducing any initial asymmetries

• For the s fixed fit, rs increase from 0.5 to 0.8 at Q2 = MZ

20Phys.Rev.Lett. 109 (2012) 012001

3

|lη|

0 0.5 1 1.5 2 2.5

sfre

e/fix

ed

0.981

1.02|lη|

0 0.5 1 1.5 2 2.5

| [pb

]lη

/d|

σd

500

550

600

650

700

-1 L dt = 33-36 pb∫

lν+ l→+W

= 7 TeV)sData 2010 ( stat. uncertainty)⊕(uncorr. sys.

sepWZ fixed sepWZ free

ATLAS

|lη|

0 0.5 1 1.5 2 2.5

sfre

e/fix

ed

0.981

1.02|lη|

0 0.5 1 1.5 2 2.5

| [pb

]lη

/d|

σd

350

400

450

500

-1 L dt = 33-36 pb∫

lν- l→-W



ATLAS

|Z

|y0 0.5 1 1.5 2 2.5 3 3.5

sfre

e/fix

ed

0.981

1.02|

Z|y

0 0.5 1 1.5 2 2.5 3 3.5

| [pb

]Z

/d|y

σd

60

80

100

120

140

-1 L dt = 33-36 pb∫

-l+ l→Z



ATLAS

FIG. 1. Di↵erential d�/d|⌘`+ | (left) and d�/d|⌘`� | (middle) cross section measurements for W ! `⌫ and d�/d|yZ | crosssection measurement for Z ! `` (right). The error bars represent the statistical and uncorrelated systematic uncertaintiesadded in quadrature while the theoretical curves are adjusted to the correlated error shifts (see text). The NNLO fit resultswith free and fixed strangeness are also indicated, and their ratios are shown below the cross section plots.

(⇠ 15% at x ⇠ 0.01 and Q

2

0

). The fraction of strangenessis again consistent with unity, r

s

= 0.96±0.25exp. Finallythe data are fitted, to NNLO, with separate strange andanti-strange normalizations. The resulting value of r

s

isconsistent with unity and the ratio s/s is 0.93 ± 0.15expat x = 0.023 and Q

2 = Q

2

0

.

W,Z cross section measurements performed at theTevatron may potentially have sensitivity to r

s

similarto that of the ATLAS data. A NLO fit to the HERAwith the CDF W asymmetry [31] and Z rapidity [32]data gives r

s

= 0.66 ± 0.29exp at a mean x of about0.081. This is consistent within uncertainties with bothsuppressed strangeness and with the present result. ANLO fit to the combined HERA, ATLAS and CDF datayields r

s

= 0.95± 0.17exp.

The provision of the full di↵erential cross sections forboth W

+

, W

� and Z boson production, besides the ep

cross sections, is essential for the determination of xs: ifthe ATLAS Z cross section data are fitted together withthe ATLAS W charge asymmetry data, rather than withthe separate W+ and W

� cross section measurements, aless precise result is obtained with r

s

= 0.92± 0.31exp.

In Fig. 2 the present result for rs

is compared with pre-dictions obtained from four global PDF determinations.The CT10 (NLO) [12] determination gives a large frac-tion consistent with the present result. On the otherhand, the MSTW08 [8] and ABKM09 [9] determina-tions give a much lower value of r

s

' 0.5, and theNNPDF2.1 [10, 11] result of r

s

' 0.25 is even lower.

The enlarged fraction of the strange quark sea leads toa decrease of the down and up quark sea densities at theinitial scale Q

2

0

, because xs, xd and xu are tied togetherat low x by the precise F

2

data. In compensation for theincrease of xs, the xd and xu distributions are dimin-ished by ' 10%. The total sea, x⌃, is correspondinglyenhanced by ' 8%, as illustrated in Fig. 3.

sr-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4

ABKM09NNPDF2.1MSTW08CT10 (NLO)total uncertaintyexperimental uncertainty

ATLAS, x=0.0232 = 1.9 GeV2Q sepWZ free

FIG. 2. Predictions for the ratio rs = 0.5(s + s)/d, atQ

2 = 1.9GeV2, x = 0.023. Points: global fit results us-ing the PDF uncertainties as quoted; bands: this analysis;inner band, experimental uncertainty; outer band, total un-certainty.

The result on r

s

, Eq. 2, evolves to

r

s

= 1.00±0.07exp±0.03mod

+0.04

�0.06

par±0.02↵S±0.03th (3)

at Q

2 = M

2

Z

and x = 0.013, corresponding to a valueof r

s

(0.013,M2

Z

) = 1.00+0.09

�0.10

, which is more than twiceas precise as at the initial scale Q

2

0

. Uncertainties aresmaller at Q

2 = M

2

Z

because the gluon splitting proba-bility into qq pairs is flavor independent, thus reducingany initial flavor asymmetries. This also causes r

s

to in-crease from 0.5 at Q2

0

to a value of about 0.8 at Q2 = M

2

Z

in the fixed s fit.In summary, a NNLO pQCD analysis is performed of

the first di↵erential ATLAS W

±, Z pp cross sections with

HERA e

±p data. The W, Z measurements introduce a

novel sensitivity to the strange quark density at x ⇠ 0.01,which is exploited here for the first time. The ratio ofthe strange to the down sea quark density is found to ber

s

= 1.00+0.25

�0.28

, at Bjorken x = 0.023 and the initial scaleof the QCD fit Q

2

0

= 1.9GeV2. This is consistent withthe prediction that the light quark sea at low x is flavor

3

|lη|

0 0.5 1 1.5 2 2.5

sfre

e/fix

ed

0.981

1.02|lη|

0 0.5 1 1.5 2 2.5

| [pb

]lη

/d|

σd

500

550

600

650

700

-1 L dt = 33-36 pb∫

lν+ l→+W



ATLAS

|lη|

0 0.5 1 1.5 2 2.5

sfre

e/fix

ed

0.981

1.02|lη|

0 0.5 1 1.5 2 2.5

| [pb

]lη

/d|

σd

350

400

450

500

-1 L dt = 33-36 pb∫

lν- l→-W



ATLAS

|Z

|y0 0.5 1 1.5 2 2.5 3 3.5

sfre

e/fix

ed

0.981

1.02|

Z|y

0 0.5 1 1.5 2 2.5 3 3.5

| [pb

]Z

/d|y

σd

60

80

100

120

140

-1 L dt = 33-36 pb∫

-l+ l→Z



ATLAS

FIG. 1. Di↵erential d�/d|⌘`+ | (left) and d�/d|⌘`� | (middle) cross section measurements for W ! `⌫ and d�/d|yZ | crosssection measurement for Z ! `` (right). The error bars represent the statistical and uncorrelated systematic uncertaintiesadded in quadrature while the theoretical curves are adjusted to the correlated error shifts (see text). The NNLO fit resultswith free and fixed strangeness are also indicated, and their ratios are shown below the cross section plots.

(⇠ 15% at x ⇠ 0.01 and Q

2

0

). The fraction of strangenessis again consistent with unity, r

s

= 0.96±0.25exp. Finallythe data are fitted, to NNLO, with separate strange andanti-strange normalizations. The resulting value of r

s

isconsistent with unity and the ratio s/s is 0.93 ± 0.15expat x = 0.023 and Q

2 = Q

2

0

.

W,Z cross section measurements performed at theTevatron may potentially have sensitivity to r

s

similarto that of the ATLAS data. A NLO fit to the HERAwith the CDF W asymmetry [31] and Z rapidity [32]data gives r

s

= 0.66 ± 0.29exp at a mean x of about0.081. This is consistent within uncertainties with bothsuppressed strangeness and with the present result. ANLO fit to the combined HERA, ATLAS and CDF datayields r

s

= 0.95± 0.17exp.

The provision of the full di↵erential cross sections forboth W

+

, W

� and Z boson production, besides the ep

cross sections, is essential for the determination of xs: ifthe ATLAS Z cross section data are fitted together withthe ATLAS W charge asymmetry data, rather than withthe separate W+ and W

� cross section measurements, aless precise result is obtained with r

s

= 0.92± 0.31exp.

In Fig. 2 the present result for rs

is compared with pre-dictions obtained from four global PDF determinations.The CT10 (NLO) [12] determination gives a large frac-tion consistent with the present result. On the otherhand, the MSTW08 [8] and ABKM09 [9] determina-tions give a much lower value of r

s

' 0.5, and theNNPDF2.1 [10, 11] result of r

s

' 0.25 is even lower.

The enlarged fraction of the strange quark sea leads toa decrease of the down and up quark sea densities at theinitial scale Q

2

0

, because xs, xd and xu are tied togetherat low x by the precise F

2

data. In compensation for theincrease of xs, the xd and xu distributions are dimin-ished by ' 10%. The total sea, x⌃, is correspondinglyenhanced by ' 8%, as illustrated in Fig. 3.

sr-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4

ABKM09NNPDF2.1MSTW08CT10 (NLO)total uncertaintyexperimental uncertainty

ATLAS, x=0.0232 = 1.9 GeV2Q sepWZ free

FIG. 2. Predictions for the ratio rs = 0.5(s + s)/d, atQ

2 = 1.9GeV2, x = 0.023. Points: global fit results us-ing the PDF uncertainties as quoted; bands: this analysis;inner band, experimental uncertainty; outer band, total un-certainty.

The result on r

s

, Eq. 2, evolves to

r

s

= 1.00±0.07exp±0.03mod

+0.04

�0.06

par±0.02↵S±0.03th (3)

at Q

2 = M

2

Z

and x = 0.013, corresponding to a valueof r

s

(0.013,M2

Z

) = 1.00+0.09

�0.10

, which is more than twiceas precise as at the initial scale Q

2

0

. Uncertainties aresmaller at Q

2 = M

2

Z

because the gluon splitting proba-bility into qq pairs is flavor independent, thus reducingany initial flavor asymmetries. This also causes r

s

to in-crease from 0.5 at Q2

0

to a value of about 0.8 at Q2 = M

2

Z

in the fixed s fit.In summary, a NNLO pQCD analysis is performed of

the first di↵erential ATLAS W

±, Z pp cross sections with

HERA e

±p data. The W, Z measurements introduce a

novel sensitivity to the strange quark density at x ⇠ 0.01,which is exploited here for the first time. The ratio ofthe strange to the down sea quark density is found to ber

s

= 1.00+0.25

�0.28

, at Bjorken x = 0.023 and the initial scaleof the QCD fit Q

2

0

= 1.9GeV2. This is consistent withthe prediction that the light quark sea at low x is flavor




21Phys.Rev.Lett. 109 (2012) 012001



Measurement of Angular Correlations in Drell-Yan Lepton Pairs.

22


• Best prediction of the ϕη* distribution is given by SHERPA and POWHEG+PYTHIA8

•

23

• The inner and outer error bars on the data points represent the statistical and total uncertainties, respectively.


24


25

27

Source of systematic uncertainty Bin: 116-130 GeV [%] Bin: 1000-1500 GeV [%]Total background estimate 1.3 8.2Electron reconstruction and identification 2.8 3.0Electron energy scale and resolution 2.1 3.3Unfolding method 1.5 1.5Trigger efficiency 0.8 0.8MC modelling 0.2 0.3MC statistics 0.7 0.4Total experimental uncertainty 4.2 9.8Luminosity uncertainty 3.9 3.9Theoretical CDY 0.1 0.3Theoretical CDY/EDY 0.3 0.4

Table 1: Summary of systematic uncertainties on the cross-section measurement, shown for the lowestand highest bin in mee.

performed by correcting the MC@NLO signal sample using the PYTHIA-derived CDY factor. Due to theslightly different shapes of the mee distribution between the two generators, considered to represent thepossible shape difference between data and PYTHIA, a non-closure of around 1.5% is found. This isadded as a systematic uncertainty on the cross-section in all mee bins.

Trigger efficiency Scale factors to account for the difference in data and MC of the EM signal triggerefficiency were obtained by comparing the efficiency in MC to that measured on data using a tag-and-probe method. Z → ee events were tagged by selecting events passing an alternative single-electrontrigger, thus providing one electron probe free of trigger bias to test against the signal trigger require-ments. The effect on the cross-section of varying the trigger efficiency scale factors up and down withintheir systematic uncertainties is approximately ±1%.

MC modelling and MC statistics Systematic uncertainties are associated to the reweighting of thePYTHIA signal MC events in order to better match the data in terms the transverse momentum distributionof the Z bosons, the mean number of interactions per bunch crossing, and also to the use of K-factors.These uncertainties enter into the calculation of CDY and result in an overall uncertainty on the cross-section of less than 1%. Consistent results are found using the MC@NLO and SHERPA samples so no furtheruncertainty is added for the choice of generator. The statistical uncertainties from the PYTHIA signal MCsamples, that also apply to CDY, are between 0.4 and 2.4%.

Luminosity The uncertainty on the luminosity is 3.9% [31].

The contributions from the above sources of systematic uncertainty to the uncertainty on the mea-sured cross-section are summarised in Table 1 for the lowest and highest bin in the mee range considered.The overall experimental uncertainty, except for luminosity, rises from 4.2% at low mee to 9.8% in thehighest mee bin.

6

measurements of inclusive w and z production in atlas

Documents