ATL
-PH
YS-
PRO
C-2
017-
041
09M
ay20
17
MEASUREMENTS OF INTEGRATED AND DIFFERENTIAL CROSS1
SECTIONS FOR ISOLATED PHOTON PAIR PRODUCTION IN 8 TEV2
PROTON-PROTON COLLISIONS AT THE ATLAS EXPERIMENT3
M. SAIMPERT, ON BEHALF OF THE ATLAS COLLABORATION.4
DESY, Hamburg and Zeuthen, Germany.5
6
A measurement of the production cross section for two isolated photons in proton–protoncollisions at a center-of-mass energy of
√s = 8 TeV is presented. The results are based on
an integrated luminosity of 20.2 fb−1 recorded by the ATLAS detector at the Large HadronCollider. The measurement considers photons with pseudorapidities satisfying |ηγ | < 1.37 or1.56 < |ηγ | < 2.37 and transverse energies of respectively EγT,1 > 40 GeV and EγT,2 > 30 GeVfor the two leading photons produced in the interaction. The fiducial cross sections are cor-rected for detector effects and measured differentially as a function of six kinematic observ-ables. The measured cross section integrated within the fiducial volume is 16.8± 0.8 pb. Thedata are compared to four calculations with relative uncertainties varying from 5% to 20%.
7
1 General description of the measurement8
More than 99% of the high-energy photon pairs produced at the Large Hadron Collider originate9
from processes predicted by perturbative quantum chromodynamics (pQCD), offering a natural10
testing ground for strong interactions. We present in this document ATLAS 1 cross-section11
measurements for isolated photon pair production in√s = 8 TeV proton–proton collisions 2.12
Significant improvements in the analysis were achieved so that the uncertainties were reduced13
by up to a factor of two compared to the previous ATLAS publication 3 based on√s = 7 TeV14
data.15
Integrated fiducial and differential cross section measurements are performed. Both require16
the determination of the number of background events in the selected sample (Nbkg), the cor-17
rections related to detector inefficiencies and resolution, and the integrated luminosity of the18 √s = 8 TeV ATLAS data sample 4, L = 20.2 ± 0.4 fb−1. The critical part of the measurement19
consists in getting accurate estimates of Nbkg. Six observables of interest are studied: the dipho-20
ton invariant mass mγγ , the absolute value of the cosine of the scattering angle with respect to21
Residual [GeV]
1− 0.5− 0 0.5 1
Fra
ction o
f events
/ 0
.04 G
eV
0
0.05
0.1
γγT,p
)2 (x T
a
)γγm2* (x η
φ
ATLAS Simulation
= 8 TeVs
(a)
4− 2− 0 2 4 6
Eve
nts
/ 0
.4 G
eV
0
5
10
15
20
25
30
35
310×
Data
Full model
Syst. unc.
γγ
jγ
γj
jj
ee
ATLAS
1 = 8 TeV, 20.2 fbs
[GeV]isoT,1E
4− 2− 0 2 4 6
Data
/ M
odel
0.8
1
1.2
(b)
Figure 1: (a) Distributions of pT,γγ , aT and φ∗η residuals using simulated diphoton events7 in theselected sample. Since aT is a component of pT,γγ and φ∗η is approximately aT/mγγ when both
photons are back-to-back, the residuals of aT and φ∗η are scaled, respectively, by√
2 and√
2mγγ
for comparison. (b) Distribution of EisoT,1 in the selected sample. Also shown are the projections
of the signal and various background components used in the two-dimensional template fit andthe full model, corresponding to their sum, after fit. The ratio of the data to the model afterthe fit is also shown, including the uncertainties on the full model, shown as a gray band. Thedata is shown only with statistical uncertainties.
the direction of the proton beamsa | cos θ∗η |, the diphoton transverse momentum pT,γγ , the open-22
ing angle between the photons in the azimuthal plane ∆φγγ , and two observables which were23
not included in the previous analysis, aT and φ∗η. The aT and φ∗η observables5,6 are less sensitive24
to the energy resolution of the individual photons and therefore are more precisely determined25
than pT,γγ , as shown in Figure 1 (a). Hence, they are ideally suited to probe the region of low26
pT,γγ , in which QCD resummation effects are most significant. Measurements of pT,γγ , aT and27
φ∗η for diphoton production (which originates from both the quark–antiquark and gluon–gluon28
initial states) are important benchmarks to test the description of the low transverse-momentum29
region by pQCD and complementary to similar measurements performed for Drell-Yan events 830
(in which quark–antiquark initial states dominate).31
2 Event selection32
The data used in this analysis were recorded using a diphoton trigger with transverse energy33
thresholds of 35 GeV and 25 GeV for the ET-ordered leading and subleading photon candi-34
dates, respectively. The shapes of the energy depositions in the electromagnetic calorimeter are35
required to match those expected for electromagnetic showers initiated by photons 9. Photons36
reconstructed within |η| < 2.37 are retained, while those near the region between the barrel and37
end-caps (1.37 < |η| < 1.56) are excluded from the analysis. After the final energy calibration 1038
has been applied, only events with ET,1 and ET,2 greater than 40 GeV and 30 GeV, respectively,39
and angular separation between the two photons ∆Rγγ > 0.4 are selected. In addition, the two40
photon candidates must be isolated from additional activity in the detector. For this purpose,41
the track isolation energy pisoT is defined as the scalar sum of the pT of tracks with pT > 1 GeV42
and within a cone of size ∆R = 0.2 around the photon candidate, and the calorimeter isolation43
energy EisoT is defined as the scalar sum of the ET of positive energy topological clusters11 within44
aThe ATLAS experiment uses a right-handed coordinate system with its origin at the nominal interactionpoint (IP) in the center of the detector and the z-axis along the beam pipe. The x-axis points from the IP to thecenter of the Large Hadron Collider ring, and the y-axis points upward. Cylindrical coordinates (r, φ) are used inthe transverse plane, φ being the azimuthal angle around the beam pipe. The pseudorapidity is defined in termsof the polar angle θ as η = − ln tan(θ/2). Angular distance is measured in units of ∆R =
√(∆η)2 + (∆φ)2. The
transverse energy is defined as ET = E/ cosh(η).
[GeV]γγm0 500 1000 1500
Eve
nts
/ G
eV
1−10
1
10
210
310
410Data
γγ
+ jjγj + jγ
ee
Data
γγ
+ jjγj + jγ
ee
Data
γγ
+ jjγj + jγ
ee
Data
γγ
+ jjγj + jγ
ee
Data
γγ
+ jjγj + jγ
ee
1 = 8 TeV, 20.2 fbs
ATLAS
(a)
Integrated fiducial cross section [pb]
0 5 10 15 20
Nam
e a
nd type o
f com
puta
tion
Sherpa 2.2.1 (ME+PS merged at NLO)
pb2.2
+3.116.4
DIPHOX (qq/qg at NLO, gg at LO)
pb1.1
+1.510.7
RESBOS (full NLO + NNLL)
12.1 pb
NNLO (NNLO)γ2
pb0.9
+1.314.2
0.8 pb)±Data (16.8
bandσ 1 ±
bandσ 2 ±
ATLAS 1 = 8 TeV, 20.2 fbs
(b)
Figure 2: (a) Distribution of the reconstructed invariant mass in the selected sample, togetherwith the yields associated with the various components estimated using the two-dimensionaltemplate fit method. (b) Measured fiducial cross section compared to the predictions from thefour event generators described in the text.
a cone of size ∆R = 0.4. Photon candidates satisfying pisoT < 2.6 GeV and Eiso
T < 6 GeV are45
selected.46
3 Sample composition47
The main background in the selected sample originates from high pT neutral mesons such as48
π0 → γγ carrying most of the energy of the associated jet. About 22% of the events in the49
selected sample include such objects among the selected photon candidates. Another source of50
background originates from misreconstructed electrons and represents typically 3% of the events51
in the selected sample. It is dominated by Z → ee decays and thus located in terms of invariant52
mass near the Z boson mass.53
Two data-driven methods giving compatible results, and validated using pseudo-data gener-54
ated with known signal and background composition, are used to substract the jet and electron55
background. The method used to derive the final results is an extended maximum-likelihood fit56
to the two-dimensional distribution of the calorimeter isolation variables (EisoT,1, Eiso
T,2) of events57
passing the signal selection. The yields associated with five components are extracted simulta-58
neously: diphoton (γγ), γ + jet (γj), jet + γ (jγ), jet + jet (jj) and dielectron (ee) events. The59
fit is performed in the integrated signal region and in each bin of the observables studied. The60
main uncertainty in the signal yield arises from the modeling of EisoT for photons. Figure 1 (b),61
shows the distributions of EisoT,1 in the integrated signal region and the projections of the five62
components after the maximization of the likelihood. The composition in the different bins for63
which the mγγ differential cross section is measured is shown in Figure 2 (a).64
4 Final results and comparison to theory65
The cross sections are measured in a fiducial region defined at particle level to closely follow the66
criteria used in the event selection. The same requirements on the photon kinematics are applied.67
The photons must not come from hadron or τ decays and the transverse isolation energy of each68
photon at particle level must be below 11 GeV. The estimated number of diphoton events in69
each bin is corrected for detector resolution, reconstruction and selection efficiencies using an70
iterative Bayesian unfolding method 12,13. The measured fiducial cross section is:71
σfid.tot = 16.8± 0.1 (stat)± 0.7 (syst)± 0.3 (lumi) pb = 16.8± 0.8 pb. (1)
The main uncertainties originate from uncertainties in the photon identification efficiency72
(±2.5%), the modeling of the calorimeter isolation (±2.0%) and the integrated luminosity73
210 310
[fb /
GeV
]γγ
/dm
σd
2−10
1−10
1
10
210
310
410Data + stat. unc.Total exp. uncertainty
NNLO (NNLO)γ2SHERPA 2.2.1 (ME+PS at NLO)
-1= 8 TeV, 20.2 fbsATLAS
0 100 2000
100
200
Theo
ry /
Dat
a
0.5
1
1.5
[GeV]γγm210 310
0.5
1
1.5
Theo
ry /
Dat
a DIPHOX (qq/qg at NLO, gg at LO)RESBOS (NLO + NNLL)
(a)
2−10 1−10 1 10 210 310 410
* [fb
]ηφ
/dσd
4−10
2−10
1
210
410
610Data + stat. unc.Total exp. uncertainty
NNLO (NNLO)γ2SHERPA 2.2.1 (ME+PS at NLO)
-1= 8 TeV, 20.2 fbsATLAS
0 0.1
50
100
150
310×
0.5
1
1.5
*η
φ2−10 1−10 1 10 210 310 410
1
2
Theo
ry /
Dat
aTh
eory
/ D
ata
DIPHOX (qq/qg at NLO, gg at LO)RESBOS (NLO + NNLL)
(b)
Figure 3: (a) mγγ and (b) φ∗η differential cross section measurements, compared with the fourQCD predictions described in the text.
(±1.9%). The measurements are compared to fixed-order predictions at NLO (Diphox) 14 and74
NNLO (2γNNLO) 15 precision in pQCD, and to computations combining NLO matrix elements75
and resummation of initial-state gluon radiation to NNLL (Resbos) 16 or matched to a parton76
shower (Sherpa 2.2.1) 17, see Figure 2 (b) and Figure 3. The theoretical uncertainties are domi-77
nated by missing higher-order corrections. Fixed-order computations are unable to describe the78
regions sensitive to soft gluon emissions (e.g. low φ∗η), whereas the inclusion of soft-gluon re-79
summation or a parton shower provides a good description of the latter. Sherpa 2.2.1 provides80
an improved description of the data for all observables compared to the other computations and81
gives predictions in good agreement with all the measurements.82
References83
1. ATLAS Collaboration, JINST 3, S08003 (2008).84
2. ATLAS Collaboration, arXiv:1704.03839 [hep-ex] (2017).85
3. ATLAS Collaboration, JHEP 01, 086 (2013).86
4. ATLAS Collaboration, EPJC 76, 653 (2016).87
5. M. Vesterinen and T.R. Wyatt, Nucl. Instrum. Methods A 602, 432 (2009).88
6. A. Banfi et al, EPJC 71, 1600 (2011).89
7. T. Gleisberg et al, JHEP 02, 007 (2009).90
8. ATLAS Collaboration, EPJC 76, 291 (2016).91
9. ATLAS Collaboration, EPJC 76, 666 (2016).92
10. ATLAS Collaboration, EPJC 74, 3071 (2014).93
11. ATLAS Collaboration, arXiv:1603.02934 [hep-ex] (2016).94
12. G. D’Agostini, Nucl. Instrum. Methods A 362, 487 (1995).95
13. T. Adye, arXiv:1105.1160 [physics.data-an] (2011).96
14. T. Binoth et al, EPJC 16, 311 (2000).97
15. S. Catani et al, Phys. Rev. Lett. 108, 072001 (2012).98
16. P. Nadolsky et al, Phys. Rev. D 76, 013008 (2007).99
17. F. Siegert, arXiv:1611.07226 [hep-ph] (2016).100