Gavin P. Salam* Rudolf Peierls Centre for Theoretical Physics & All Souls

Home , Rudolf Peierls

Physics motivations for sub-percent absolute cross sections

Gavin P. Salam* Rudolf Peierls Centre for Theoretical Physics & All Souls College, Oxford

* on leave from CERN and CNRS LHCb UK annual meeting,

University of Warwick, 4 January 2019 1 UNDERLYING EXPERIMENTAL THEORY DATA

can we achieve sub- percent precision?

2 is it experimentally feasible?

3 EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN)

CERN-PH-EP-2015-301 LHCb-PAPER-2015-049 February 3, 2016

Measurement of forward W and Z boson production in pp collisions at ps =8TeV

1000 The LHCb collaboration† LHCb, s = 8 TeV

[pb] + µ 800 Datastat (W ) CT14 η Data (W +) MMHT14 Abstract

/d tot

ν − µ Data (W ) NNPDF30 600 Measurementsstat are presented of electroweak boson production using data from pp → Data (W −) CT10 W collisionstot at a centre-of-mass energy of ps =8TeV. The analysis is based on an σ 400 ABM12 1

d integrated luminosity of 2.0 fb recorded with the LHCb detector. The bosons are identiﬁed in the WHERA15µ⌫ and Z µ+µ decay channels. The cross-sections are ! ! 200 measured for muons in the pseudorapidity range 2.0 < ⌘ < 4.5, with transverse p µ > 20 GeV/c T momenta pT > 20 GeV/c and, in the case of the Z boson, a dimuon mass within 2 1.1 60

Theory/Data = 818.4 1.9 5.0 7.0 9.5pb, 2 2.5 3 3.5 4 W µ4.5⌫ µ ! µη ± ± ± ± Z µ+µη = 95.0 0.3 0.7 1.1 1.1pb, ! ± ± ± ± 100100 where the ﬁrst uncertainties are statistical,stat. syst. the secondbeam E arelumi systematic, the third 1511.08039 areLHCb, due to s the= 8 TeV knowledge2.0 fb-1 of the LHC beam energy and the fourth are due to the [pb]

Z 80 y luminosityData determination.(Z) CT14 The evolution of the W and Z boson cross-sections [pb] stat

/d 80 Z with centre-of-massData (Z) MMHT14 energy is studied using previously reported measurements with µ tot

y LHCb achieves 0.8% µ 1 µ 60 arXiv:1511.08039v2 [hep-ex] 2 Feb 2016 1.0 fb of data atNNPDF30 7 TeV.Diηηµ ↵erential distributions are also presented. Results are → /d Z uncertainty, aside from µ 60 in good agreementCT10 with theoretical predictions at next-to-next-to-leading order in σ d µ 40 perturbative quantumABM12 chromodynamics.luminosity uncertainty → HERA15 Z 4020 (beam energy systematic σ

d Published in JHEPgreatly 01 improved (2016) 155 since 200 1.6 this paper) 1.4 2 2.5 3 3.5 c CERN4 on behalf4.5 of the LHCb collaboration, licence CC-BY-4.0. 1.2 y 1 Z 0.80 †Authors are listed at the end of this paper. 0.6 Theory/Data 2 2.5 3 3.5 4 4.5 y Z 4 + Figure 2: (top) Di↵erential W and W boson production cross-section in bins of muon pseudorapidity. (bottom) Di↵erential Z boson production cross-section in bins of boson rapidity. Measurements, represented as bands, are compared to (markers, displaced horizontally for presentation) NNLO predictions with di↵erent parameterisations of the PDFs.

10 ] ]

-1 -1 -1 -1 −1 ATLAS s=8 TeV, 20.3 fb −1 ATLAS s=8 TeV, 20.3 fb 10 12 GeV ≤ m < 20 GeV, |y | < 2.4 10 20 GeV ≤ m < 30 GeV, |y | < 2.4 ll ll ll ll [GeV [GeV ll 10T −2 ll 10T −2 /dp /dp σ −3 ee-channel σ −3 ee-channel d 10 d 10 σ µµ-channel σ µµ-channel 1/ 1/ 4 10− Combined 10−4 Combined Statistical uncertainty Statistical uncertainty −5 10 Total uncertainty 10−5 Total uncertainty

1.2 2 2 10 1.1 10 1.1 1 1

Channel 0.9 Channel Combined χ2/NDF= 8/8 Combined 0.9 χ2/NDF= 6/8 0.8 ] 2 ] 2 σ 2 10 σ 2 10 0 0

Pull [ −2 Pull [ −2 50 100 500 50 100 500 pll [GeV] pll [GeV] T T ] ] -1 s=8 TeV, 20.3 fb-1 -1 s=8 TeV, 20.3 fb-1 −1 ATLAS 1 ATLAS 10 30 GeV ≤ m < 46 GeV, |y | < 2.4 46 GeV ≤ m < 66 GeV, |y | < 2.4 ll ll ll ll −1

[GeV [GeV 10 ll 10T −2 ll T −2 /dp /dp 10 σ σ −3 ee-channel ee-channel d 10 d −3 σ µµ-channel σ 10 µµ-channel 1/ 1/ 10−4 Combined 10−4 Combined Statistical uncertainty Statistical uncertainty 10−5 10−5 Total uncertainty Total uncertainty 10−6 102 1 10 102 1.1 1.05 1 1 Channel Channel Combined 0.9 χ2/NDF= 7/8 Combined 0.95 χ2/NDF=13/20

] 2 ] 2 σ 2 10 σ 2 1 10 10 0 0

Pull [ −2 Pull [ −2 50 100 500 1 10 102 pll [GeV] pll [GeV] T T ] ] -1 1 ATLAS s=8 TeV, 20.3 fb-1 -1 1 ATLAS s=8 TeV, 20.3 fb-1 −1 66 GeV ≤ m < 116 GeV, |y | < 2.4 Other116 experiments? GeV ≤ m < 150 GeV, |y | < 2.4 10 ll ll ll ll −1

[GeV −2 [GeV 10 ll 10T ll T −3 One−2 of the most precise results are /dp 10 /dp 10 σ ee-channel σ ee-channel d 10−4 d perhaps for the Z transverse σ µµ-channel σ 10−3 µµ-channel 1/ −5 1/ momentum (ATLAS) 10 Combined Combined −6 10−4 10 Statistical uncertainty ➤ normalisedStatistical to uncertainty Z ﬁducal σ −7 10 Total uncertainty 10−5 Total uncertainty 10−8 ➤ achieves <1%, from 1 10 102 1 10 102 1.01 1.05 pT = 1 to 200 GeV, 1 ±1% 1and <0.5% in some regions Channel Channel

Combined 0.99 Combined 0.95 χ2/NDF=43/43 χ2/NDF=27/20 ] 2 ] 2 σ 2 1 10 10 σ 2 1 10 10 0 0

Pull [ −2 Pull [ −2 Ratio to total cross section cancels 1 10 102 1 10 102 pll [GeV] lumi & some lepton-eﬃciency pll [GeV] T T systematics. 5 `` Figure 6: The Born-level distributions of (1/)d/dpT for the combination of the electron-pair and muon-pair channels, shown in six m regions for y < 2.4. The central panel of each plot shows the ratios of the values from `` | ``| the individual channels to the combined values, where the error bars on the individual-channel measurements represent the total uncertainty uncorrelated between bins. The light-blue band represents the data statistical uncertainty on the combined value and the dark-blue band represents the total uncertainty (statistical and systematic). The 2 per degree of freedom is given. The lower panel of each plot shows the pull, deﬁned as the di↵erence between the electron-pair and muon-pair values divided by the uncertainty on that di↵erence.

18 Total L systematics: vdM or BGI - & more

16 One limiting factor: luminosity uncertainty

 Appdx

From LHCP 2016 talk by W. Kozanecki

LHCb Adaptedcan do from±1.1%. ref. Can [17], that Table be 14 improved upon? Other experiments have ~2% lumi uncertainty. Can new hardware help them match LHCb? W. Kozanecki 13 June 2016 If LHCb remains the best, can its lumi uncertainty be transferred to other experiments (e.g. J/ψ & Z → μμ measurements)

luminosity is potential keystone measurement for LHC precision programme 6 is QCD theory up to it?

7 Hard processes: to 3rd order (NNLO) in perturbation theory strong coupling constant (αs)

�� 2 � �� + c.c. � �� × �� + c.c. �� × �� 2 �� explosion of calculations �� in past 3 years ��

��

�� as of April 2017 �� 8 NNLO precision d.) d.) d.) d.) d.) d.) d.) d.) d.) d.) d.) (f (f (f (f (f (f (f (f (f (f (f (tot.) (tot.) (tot.) (tot.) (tot.) (tot.) γ γ γ H Zj - Hj Wj ZZ γ Z- WZ WH W- WW W/Z WW VBF VBF ttbar @50

10% tZ p

-2%

-5%

-10%

-15%

NNLO NLO -20% 9 NNLO precision (magnified) d.) d.) d.) d.) d.) d.) d.) d.) d.) d.) d.) (f (f (f (f (f (f (f (f (f (f (f (tot.) (tot.) (tot.) (tot.) (tot.) (tot.) γ γ γ H Zj - Hj Wj ZZ γ Z- WZ WH W- WW W/Z WW VBF VBF ttbar 5% @50 tZ p NNLO just barely gets us to 1% accuracy in some 2% cases 1% (insofar as you believe in -1% uncertainties from -2% scale variation)

NNLO NLO -5% 10 5 /dY 1.1 /dY 1.1 LO 3 best N best NNLO d d 1 1 /dY /dY LO 3 approx approx N NNLO 0.9 0.9 3 d d pp H + X 0 1 pp H + X 0 1 ! z¯ z¯ ! z¯ z¯ 2 3 2 3 z¯ z¯ z¯ z¯ µ = µ = m /2 4 4 0.8 F R h z¯ 0.8 µF = µR = mh/2 z¯

4 3 2 1 0 1 2 3 4 4 3 2 1 0 1 2 3 4 Y Y

FIG. 1: Approximate Higgs boson rapidity distribution with threshold expansion truncated at di↵erent orders. The left panel shows the ratio of the approximate NNLO to the exact result, the right panel shows the approximate N3LO result to the best prediction obtained in this work. N3LO Higgs rap. dist.

ISSN 0370-2693, URL http://www.sciencedirect.com/ 12 pp H + X ! NNLO → N3LO science/article/pii/S037026931200857X. [4] S. Chatrchyan, V. Khachatryan, A. Sirunyan, µ = µ = m /2 10 F R h 3 ➤ A.Technology Tumasyan, is W. there Adam, for simple E. Aguilo, processes T. Bergauer, M. Dragicevic, J. Er, C. Fabjan, et al., Physics 8 Letters B 716,30(2012),ISSN0370-2693,URL ] ➤ Demonstrated in Higgs case (poorly

[ http://www.sciencedirect.com/science/article/ 6 pii/S0370269312008581convergent perturbative. expansion → + 0.9%

/dY −3.4% n [5] S. Dawson, Nucl. Phys. B359,283(1991).

d uncertainty). 4 [6] D. Graudenz, M. Spira, and P. M. Zerwas, Phys. Rev. Lett. 70,1372(1993). ➤ Expect DY/Z production @N3LO 2 [7] M. Spira, A. Djouadi, D. Graudenz, and P. M. Zerwas, Nucl. Phys. B453,17(1995),hep-ph/9504378. arXiv:1810.09462 soonish, with hope of genuine sub- [8] A.percent Djouadi, uncertainties. M. Spira, and P. M. Zerwas, Phys. Lett. 0 Dulat, Mistlberger, Pelloni B264,440(1991). [9] C. Anastasiou and K. Melnikov, Nuclear Physics B http:

/dY 646,220(2002),ISSN05503213,0207004,URL LO

3 1 //arxiv.org/abs/hep-ph/0207004. N

[10] R. V. Harlander and W. B. Kilgore, Phys. Rev. Lett. 88, d . see201801 also Cieri, (2002), Chen, hep-ph/0201206. Gehrmann, Glover, Huss 0.9 /dY [11]arXiv:1807.11501 V. Ravindran, J. Smith, and W. L. van Neerven, Nucl. Phys. B665,325(2003),hep-ph/0302135. NNLO

[12] C. Anastasiou, C. Duhr, F. Dulat, and B. Mistlberger, d 0.8 p. 78 (2013), 1302.4379, URL http://arxiv.org/abs/ 4 3 2 1 0 1 2 3 4 1302.4379. Y [13] C. Anastasiou, C. Duhr, F. Dulat, F. Herzog,11 and B. Mistlberger, JHEP 12,088(2013),1311.1425. [14] C. Anastasiou, C. Duhr, F. Dulat, E. Furlan, FIG. 2: The Higgs boson rapidity distribution at di↵erent or- T. Gehrmann, F. Herzog, and B. Mistlberger, Phys. Lett. ders in perturbation theory. The lower panel shows the N3LO B737,325(2014),1403.4616. and NNLO predictions normalised to the N3LO prediction for [15] C. Anastasiou, C. Duhr, F. Dulat, E. Furlan, µ = mh/2. T. Gehrmann, F. Herzog, and B. Mistlberger, JHEP 03, 091 (2015), 1411.3584. [16] F. Dulat and B. Mistlberger (2014), 1411.3586. [17] C. Anastasiou, C. Duhr, F. Dulat, E. Furlan, F. Herzog, and B. Mistlberger, JHEP 08,051(2015),1505.04110. [18] C. Anastasiou, C. Duhr, F. Dulat, F. Herzog, and [1] M. Aaboud et al. (ATLAS), Phys. Lett. B786,114 B. Mistlberger, Phys. Rev. Lett. 114,212001(2015), (2018), 1805.10197. 1503.06056. [2] A. M. Sirunyan et al. (CMS) (2018), 1809.10733. [19] C. Anastasiou, C. Duhr, F. Dulat, E. Furlan, [3] G. Aad, T. Abajyan, B. Abbott, J. Abdallah, S. A. T. Gehrmann, F. Herzog, A. Lazopoulos, and B. Mistl- Khalek, A. Abdelalim, O. Abdinov, R. Aben, B. Abi, berger, JHEP 05,058(2016),1602.00695. M. Abolins, et al., Physics Letters B 716,1 (2012), [20] F. Dulat, A. Lazopoulos, and B. Mistlberger, Comput. + -0.22 d ln q / d ln 2 NLO NS µf -0.12 N2LO -0.23 3 N LOA,B -0.24 -0.13

-0.25 x = 0.8 x = 0.5 -0.14 -0.025

+ -0.22 d ln q / d ln 2 0.05 NLO NS µf -0.12 N2LO -0.03-0.23 3 N LOA,B -0.24 0.045-0.13

-0.035-0.25 x = 0.01 xx == 0.150.8 x = 0.5 -0.140.04 -0.025 N3LO splitting functions 0.08 + -0.22 2 0.09 NLO d ln q / d ln µf -0.12 NS 0.05 ➤ non-singlet splitting functions (e.g. for N2LO -0.23 DGLAP evolution of u-ubar) are now 3 -0.03 essentiallyN LOA,B known 0.07 -0.130.08 -0.24 ➤ uncertainties on evolution are ~ ±0.5% 0.045 in the ﬁgure (~ half those at NNLO) -0.25 ±1% ➤ x = 0.8 singlet (e.g. Pggx) =still 0.5 in progress -0.035 −3 -0.140.07 −4 0.06 xx = 100.15 xx == 0.0110 -0.025 -1 0.04 -1 10 1 10 10 1 10

2 2 0.05 2 2 + 2 0.08 µr / µf µr / µf -0.22 d ln q / d ln NLO 0.09 NS µf -0.12 Figure-0.03 9: The dependence2 of the NLO, NNLO and N3LO predictions forq ˙+ d lnq+ d lnµ2 on the N LO ns ≡ ns/ f -0.23 renormalization scale µr at six typical values of x for the initial conditions1707.08315 (5.6) Moch, and Ruijl, (5.7). Ueda, The Vermaseren, effect of the Vogt 3 0.045 (3)+ remaining uncertaintyN LO of the four-loop splitting function Pns ,indicatedbythedifferenceofthesolidand A,B 0.08 N4LO progress: 1812.11818 12 dotted0.07 curves, is practically invisible except for the last two panel. -0.24 -0.13 -0.035 x = 0.15 x = 0.01 0.04 -0.25 −3 0.07 −4 x = 0.8 0.06 x = x0.5 = 10 29 x = 10 0.08 -0.14 0.09 -1 -1 10 1 10 10 1 10 -0.025 2 2 2 2 µr / µf µr / µf

0.07 0.08 3 + + 2 0.05Figure 9: The dependence of the NLO, NNLO and N LO predictions forq ˙ns ≡ d lnqns/d lnµf on the renormalization scale µr at six typical values of x for the initial conditions (5.6) and (5.7). The effect of the (3)+ -0.03 remaining uncertainty of the four-loop splitting function Pns ,indicatedbythedifferenceofthesolidand dotted curves, is practically invisible except −3 for the last0.07two panel. −4 0.06 x = 10 x = 10

0.045 -1 -1 10 1 10 10 1 10 2 2 2 2 µr / µf µr / µf -0.035 29 x = 0.15 x = 0.01 3 + + 2 Figure 9: The dependence of the NLO, NNLO and N LO predictions forq ˙ns ≡ d lnqns/d lnµf on the renormalization scale µr at six typical values of x for the initial conditions (5.6) and (5.7). The effect of the 0.04 (3)+ remaining uncertainty of the four-loop splitting function Pns ,indicatedbythedifferenceofthesolidand dotted curves, is practically invisible except for the last two panel. 0.08 0.09

0.07 0.08

−3 0.07 −4 0.06 x = 10 x = 10

-1 -1 10 1 10 10 1 10 2 2 2 2 µr / µf µr / µf

3 + + 2 Figure 9: The dependence of the NLO, NNLO and N LO predictions forq ˙ns ≡ d lnqns/d lnµf on the renormalization scale µr at six typical values of x for the initial conditions (5.6) and (5.7). The effect of the (3)+ remaining uncertainty of the four-loop splitting function Pns ,indicatedbythedifferenceofthesolidand dotted curves, is practically invisible except for the last two panel.

29 Non-perturbative effects in Z (& H?) pT

➤ Inclusive Z/W, H,VV cross sections impact�� of ��0.5�� GeV�� shift�� of�� Z pT 2 2 �� should have ~ΛQCD /M corrections �� -4 � (~10 ?) ��

➤ Z (&H) pT not inclusive so corrections �� can be ~Λ/M.

➤ Size of eﬀect can’t be probed by turning �� MC hadronisation on/oﬀ � � [maybe by modifying underlying MC ��

parameters?] �� 0.5 GeV is perhaps conservative(?) ➤ Shifting Z pT by a ﬁnite amount Suggests up to 2% effects could be �� present. illustrates what could happen �� ?� �� Techniques to answer this: Ferrario-Ravasio, Jezo, Nason & Oleari, 1810.10931 13 what might we do with it? 0. many processes, much phase space within 1% statistical reach

14 To calculate the extrapolated cross section, the combined fiducial cross section is divided by AZZ and by the leptonic branching fraction 4 (3.3658%)2 [58], where the factor of four accounts for the di↵erent ⇥ flavor combinations of the decays. The result is obtained using the same maximum-likelihood method as A huge range of phasespace willfor thereach combined < fiducial 1% cross statistical section, but now precision including the uncertainties @ HL-LHC of AZZ as additional nuisance parameters. The used leptonic branching fraction value excludes virtual-photon contributions. Based on + + a calculation with Pythia, including these would increase the branching fraction ZZ ` ``0 `0 by ! For example, processes with two orabout more 1–2%. leptons The extrapolated cross section is found to be 17.3 0.9[ 0.6 (stat.) 0.5 (syst.) 0.6 (lumi.)] pb. The ± ± ± ±ZZ ➤ NNLO prediction from Matrix, with the gg-initiated process multiplied by a global NLO correction Drell-Yan: mμμ up to 1 TeV 36.1fb+0-1.6 @ 13 TeV: factor of 1.67 [3] is 16.9 0.5 pb, where the uncertainty is estimated by performing QCD scale variations. A comparison of the17.3 extrapolated ± 0.9 [±0.6 cross section (stat.) to ±0.5 the NNLO (syst.) prediction ±0.6 (lumi.)] as well as topb. previous measurements ➤ high pT Z, up to 1 TeV is shown in Figure→6. expect stat errs of 0.3% for 3000 fb-1

24 ➤ LHC Data 2016+2015 ( s=13 TeV) top-antitop (di-lepton channels) ATLAS ZZ→ llll (m 66-116 GeV) 36.1 fb-1 22 ll

[pb] LHC Data 2015 ( s=13 TeV) ATLAS 20 CMS ZZ llll (m 60-120 GeV) 2.6 fb-1 tot ZZ → 1709.07703 ➤ ll Mtt up to 2 TeV, σ 18 LHC Data 2012 ( s=8 TeV) ATLAS ZZ→ ll(ll/νν) (m 66-116 GeV) 20.3 fb-1 ll 16 CMS ZZ→ llll (m 60-120 GeV) 19.6 fb-1 ll ➤ 14 LHC Data 2011 ( s=7 TeV) pT, t o p up to 1 TeV ATLAS ZZ→ ll(ll/νν) (m 66-116 GeV) 4.6 fb-1 ll 12 CMS ZZ→ llll (m 60-120 GeV) 5.0 fb-1 ll Tevatron Data ( s=1.96 TeV) ➤ 10 -1 VV production CDF ZZ→ ll(ll/νν) (on-shell) 9.7 fb D0 ZZ→ ll(ll/νν) (m 60-120 GeV) 8.6 fb-1 8 ll

6 MATRIX CT14 NNLO ZZ (pp) 4 ZZ (pp) 2

0 0 2 4 6 8 10 12 14 s [TeV]

15 Figure 6: Extrapolated cross section compared to other measurements at various center-of-mass energies by AT- LAS, CMS, CDF, and D0 [13, 14, 16, 84–86], and to pure NNLO predictions from Matrix (with no additional higher-order corrections applied). The total uncertainties of the measurements are shown as bars. Some data points are shifted horizontally to improve readability.

9 Di↵erential cross sections

Di↵erential cross sections are obtained by counting candidate events in each bin of the studied observable, subtracting the expected background, and unfolding to correct for detector e↵ects. The unfolding takes into account events that pass the selection but are not in the fiducial phase space (which may occur due to detector resolution or misidentification), bin migrations due to limited detector resolution, as well as detector ineciencies. To minimize the model dependence of the measurement, the unfolding corrects and extrapolates the measured distributions to the fiducial phase space, rather than extrapolating to non- fiducial regions. For each given observable distribution, all of the above detector e↵ects are described by a response matrix R whose elements Rij are defined as the probability of an event in true bin j being

18 e-μ events (a mix of ttbar and VV) at LHCb

number of e-μ events with 300 fb-1@ 14 TeV (LHCb upgrade II) 20000 ttbar WW this is a quick study to gauge orders of magnitude 15000 (Pythia8, LO, no showering)

LO, Pythia8

bin N(N)LO K-factors will increase /

10000 2 < ηe/μ < 5, pT,μ/e > 20 GeV rates significantly

events there will also be other VV channels (probably smaller) 5000

0 2 2.5 3 3.5 4 4.5 5

rapeμ 16 what might we do with it? 1. As an input to everything else: PDFs

17 Drell-Yan as input to PDF fits

NNPDF3.1 MMHT14 ATLAS 13 TeV NNPDF3.0 ABMP16 CT14 data experimental uncertainty 1.10 ±

1.05 PDFs in the most basic Heavy: NNLO QCD + NLO EW Light: NNLO QCD process, Drell–Yan/Z,

meas. would be tightly 1.00 / constrained (~0.9%) were pred.

it not for the luminosity 0.95 uncertainty (~2%)

+ 0.90 W W Z

0.85

NNPDF3.1, 1706.00428 (data from 1603.09222) 18 Drell-Yan as input to PDF fits

NNPDF3.1 MMHT14 ATLAS 13 TeV NNPDF3.0 ABMP16 CT14 data experimental uncertainty 1.10 ±

1.05 PDFs in the most basic Heavy: NNLO QCD + NLO EW Light: NNLO QCD ⊕ lumi process, Drell–Yan/Z,

meas. would be tightly 1.00 / constrained (~0.9%) were pred.

it not for the luminosity 0.95 uncertainty (~2%)

+ 0.90 W W Z

0.85

NNPDF3.1, 1706.00428 (data from 1603.09222) 18 UNCERTAINTIES ON PARTONIC LUMINOSITIES — V. RAPIDITY(Y) AND MASS

�� gluon‒gluon �� quark‒gluon ��

��

� ��

�� quark‒quark �� quark-antiquark ��

��

� ��

�� 19 what might we do with it? 2. Higgs

20 Higgs precision (H → γγ) : semi-optimistic estimate v. luminosity & time

extrapolation of µγγ precision from 7+8 TeV results 50 7+8 TeV one expt. two-expt. combination [%] 20 ATLAS CMS 10 µ precision 5

extrapolated ana- 2013 lysed 1 10 30 100 300 1000 3000 -1 1 fb-1 = 1014 collisions lumi at 13/14 TeV [fb ] 21 Higgs precision (H → γγ) : semi-optimistic estimate v. luminosity & time

extrapolation of µγγ precision from 7+8 TeV results 50 7+8 TeV one expt. The LHC has the two-expt. combination [%] 20 ATLAS statistical potential to CMS take Higgs physics from 10 µ “observation” to precision 1–2% precision 5 But only if we learn how to connect experimental 2 observations with theory extrapolated ana- on 2013 2024 2038 lysed tape at that precision 1 10 30 100 300 1000 3000 -1 1 fb-1 = 1014 collisions lumi at 13/14 TeV [fb ] 21 s = 14 TeV, 3000 fb-1 per experiment 1 1 CERN-LPCC-2018-04CERN-LPCC-2018-04 December 21,December 2018 21, 2018 3000 fb-1 Total ATLAS and CMS Statistical HL-LHC Projection ATLAS and CMS Stat. + Exp. Experimental 2 2 HiggsHiggs Physics Physics HL-LHC Projection + Theory Theory Uncertainty [%] 3 3 at the HL-LHCat the HL-LHC and HE-LHC and HE-LHC ATLAS CMS 2% 4% Tot Stat Exp Th 4 Report from4 Report Working from Group Working 2 on Group the Physics 2 on of the the Physics HL-LHC, of the and HL-LHC, Perspectives and Perspectives at the HE-LHC at the HE-LHC σggH 1.6 0.7 6 0.8 1.2 6 5 5 Official Higgs projections for HL-LHC indicate σggH 7 Convenors:7 Convenors: 1,2 31,2 4 3 5,46,7 58,6,7 8 8 M. Cepeda8 M., S. Cepeda Gori1%, P., Ilten S.experimental Gori, M., P. Kado Ilten , M., F. Kadouncertainties Riva , , F. Riva , from ATLAS/CMS σ 3.1 1.8 1.3 2.1 VBF 9 Contributors:9 Contributors: 9,10 combination9,10 11 for main1112 production1312 1413 cross section,14 σggH σVBF 10 R. Abdul10 KhalekR. Abdul, A. Khalek Aboubrahim, A. Aboubrahim, J. Alimena , J., S. Alimena Alioli ,, A. S. Alves Alioli ,, A. Alves , 15 1516,17 1618,17 1918 19 20 20 21 21 11 C. Asawatangtrakuldee11 C. Asawatangtrakuldee, A. Azatov , A., Azatov P. Azzi , S., P. Bailey Azzi , S. BanerjeeBailey , S., Banerjee E. L. Barberio, E. L., Barberio , 17 1722 2220 20 23 2324 2425 25 26 26 12 D. Barducci12 D., Barducci G. Barone ,, G. M. Barone Bauer ,, M. C. Bauer Bautista, C., P. Bautista Bechtle ,, P. K. Bechtle Becker ,, K. A. Becker Benaglia, A., Benaglia , σ 5.7 3.3 2.4 4.0 27 2728 2829 2930 2430 2415 3115 31 WH 13 M. Bengala13 M., N. Bengala Berger ,, N. C. Berger Bertella, C., A. Bertella Bethani ,, A. A. Bethani Betti ,, F. A. Bishara Betti ,, F. D. Bishara Bloch , D. Bloch , σ 32 3233 335 5 34,35 34,35 36 3637 37 WH 14 P. Bokan14 , O.P. Bokan Bondu ,, O. M. Bondu Bonvini,, M. L. Bonvini Borgonovi, L. Borgonovi, M. Borsato, M., S. Borsato Boselli ,, S. Boselli , Credibility34,35 34,35 38 will be38 greatly39 39enhanced40 40 41 if,42 we can41,42 15 S. Braibant-Giacomelli15 S. Braibant-Giacomelli, G. Buchalla, G., BuchallaL. Cadamuro, L. Cadamuro, C. Caillol ,, C. A. Caillol Calandri, A. Calandri, , 43 43 44 2044 2045 4544,46 44,46 16 A. Calderon16 A. Tazon Calderon, J. M. Tazon Campbell, J. M.,demonstrate Campbell F. Caola ,, M. F. Caola Capozi <1%,, M. M. Capozi Carena control, M., Carena C. M. Carloni, C. M. Carloni σ 4.2 2.6 1.3 3.1 47 47 48 4849 49 50 50 ZH 17 Calame 17, A.Calame Carmona, A., Carmona E. Carquin, E., A. Carquin Carvalho, A. Antunes Carvalho De Oliveira Antunes De, A. Oliveira Castaneda, A. Castaneda σZH 51 5251 5352 5453 5455,56 55,56 20,57 20,57 58 58 18 Hernandez18 Hernandez, O. Cata ,, A. O. Celis Cata ,, A. A.across Cerri Celis ,, F. A. a Cerutti Cerri range, F., Cerutti G. of S. Chahalprocesses, G. S., Chahal G. Chaudhary, G. Chaudhary, , 59 59 1,4 1,460 60 61,62 61,6362 6364 6465 65 19 X. Chen 19, A.X. S. Chen Chisholm, A. S., Chisholm R. Contino, R., R. Contino Covarelli, R. Covarelli, N. Craig ,, D. N. CurtinCraig ,, D. N. Curtin P. Dang, N., P. Dang , 66 6622 22 1 181 ,67 18,67 68 68 20 P. Das ,20 S.P. Dawson Das , S., O. Dawson A. De Aguiar, O. A. Francisco De Aguiar, FranciscoJ. de Blas , J., de S. Blas De Curtis, S., De N. Curtis De , N. De σ 4.3 1.3 1.8 3.7 69,70 69,7033 33 71 71 33 3372 72 28 28 σ ttH 21 Filippis 21 ,Filippis C. Delaere, C., H. Delaere De la Torre, H. De, M. la Delcourt Torre , M., L.Delcourt de Lima, L., M. de Delmastro Lima , M., Delmastro , ttH 73 73 74 7574 7515 1576 76 77 7778 78 22 S. Demers22 ,S. L. Demers D’Eramo, L., D’Eramo N. Dev , A., N. De Dev Wit ,, A. S. De Dildick Wit ,, S. R. Dildick Di Nardo, R., Di S. Di Nardo Vita ,, S. Di Vita , 79 79 23,80 8123,80 81 1 1 15,82 15,82 83 83 23 M. Donadelli23 M., Donadelli L. A. F. do, Prado L. A. F. do, D. Prado Du , M., D. Dührssen Du , M., G. Dührssen Durieux, G. Durieux, O. Eberhardt, O., Eberhardt , 0 0.1 0.2 0.3 0.4 0.5 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 1 1,501,84 1,50,84 85 8586 2886 2887 8788 88 24 J. Elias-Miro24 J., Elias-Miro J. Ellis , J., K. Ellis El Morabit, K. El, C. Morabit Englert ,, C. S. Englert Falke ,, M. S. FarinaFalke ,, M. A. Farina Ferrari ,, A. Ferrari , 89 89 90 90 34,35 34,35 74,91 74,91 92,93 92,93 Expected relative uncertainty Expected uncertainty25 M. Flechl25 ,M. S. FolguerasFlechl , S., Folgueras E. Fontanesi, E. Fontanesi, P. Francavilla, P. Francavilla, R. Franceschini, R. Franceschini, , 94 9495 95 55,56 55,56 1 1 27 27 96 96 26 R. Frederix26 R., S. Frederix Frixione, S., A. Frixione Gabrielli, A. Gabrielli, S. Gadatsch, S., M. Gadatsch Gallinaro, M., Gallinaro A. Gandrakota, A. Gandrakota, , 97 97 98 98 59 59 96 9699 991 1001 100 27 J. Gao ,27 F.J. M. Gao Garay, F. Walls M. Garay, T. Gehrmann Walls , T. Gehrmann, Y. Gershtein, Y., Gershtein T. Ghosh , T.A. Ghosh Gilbert ,, A. R. Gilbert Glein ,, R. Glein , Fig. 28: (left) Summary plot showing the total expected 1 uncertainties in S2 (with YR18 systematic20 20 20 20 101 101 102 102 103 103 28 E. W. N.28 GloverE. W., N. R. Glover Gomez-Ambrosio, R. Gomez-Ambrosio, G. Gómez-Ceballos, G. Gómez-Ceballos, D. Gonçalves, D. Gonçalves, M. Gorbahn, M., Gorbahn , ± 27 27 104 10413,26 13,26 59 59 96 96105 105 29 E. Gouveia29 E., M. Gouveia Gouzevitch, M. Gouzevitch, P. Govoni , P., Govoni M. Grazzini, M., B. Grazzini Greenberg, B. Greenberg, K. Grimm , K., Grimm , uncertainties) on the per-production-mode cross sections normalized to the SM predictions for ATLAS106 106 15 15 15 107 15 10725 25 20 20 30 A. V. Gritsan30 A. V., A. Gritsan Grohsjean, A., Grohsjean C. Grojean, C., J. Grojean Gu , R., J. Gugel Gu , R. S.Gugel Gupta, R., S. Gupta , 108 10898 9857 5745 45 109 102109 102 (blue) and CMS (red). The ﬁlled coloured box corresponds to the statistical and experimental31 systematicC. B. Gwilliam31 C. B., Gwilliam M. Haacke, M., Y. Haacke Haddad ,, Y. U. Haddad Haisch ,, U. G. Haisch N. Hamity, G. N., T. Hamity Han , , T. Han , 19 1944 44 43,110,111 43,110,111 45 4542 42 32 L. A. Harland-Lang32 L. A. Harland-Lang, R. Harnik ,, R. S. Harnik Heinemeyer, S. Heinemeyer, G. Heinrich, G., V.Heinrich Hirschi ,, V. S. Hirschi Höche , S. Höche 112 112 113 113 114,115 114,115 22,116 22,116117 1117 661 66 uncertainties, while the hatched grey area represent the additional contribution to the total uncertainty33 , K. Hoepfner33 due, K., Hoepfner J. M. Hogan, J. M. Hogan, S. Homiller, S. Homiller, Y. Huang ,, Y. A. Huang Huss , Sa., A. Jain Huss,, Sa. Jain , 28 281 1 118 118 119,120 119,120101 101 59 59 34 S. Jézéquel34 S., S. Jézéquel P. Jones ,, S. J. P. Kalinowski Jones , J. Kalinowski, J. F. Kamenik, J. F. Kamenik, M. Kaplan, M., A. Kaplan Karlberg, A., Karlberg , 1 58 5885 8559 59121 1211 1221 123122 123 to theoretical systematic uncertainties. (right) Summary plot showing the total expected 35 uncertain-M. Kaur 35 , P.M. Keicher KaurDRAFT,, P. M. Keicher KernerDRAFT,, M. A. Kerner Khanov , A., J. Khanov Kieseler ,, J. J. Kieseler H. Kim ,, J. M. H. Kim Kim ,, M. Kim , 42 12442 124101 101 125 12525 12225 122 ± 36 T. Klijnsma36 T., Klijnsma F. Kling , F.M. Kling Klute ,, M. J. R. Klute Komaragiri, J. R. Komaragiri, K. Köneke ,, K. K. Köneke Kong ,, K. Kong , ties in S2 (with YR18 systematic uncertainties) on the per-production-mode cross sections normalized126 to 126118 11844 32 44 32 57 21 57 2189 89 37 J. Kozaczuk37 J., Kozaczuk P. Kozow ,, P. C. Kozow Krause ,, C. S. Krause Lai , J., Langford S. Lai , J., Langford B. Le , L., Lechner B. Le , L., Lechner , 127 127 128 24128 12924 81 129 130 81 130131 13120 20 38 W. A. Leight38 W., A. K. Leight J. C. Leney, K. J., C. T. Leney Lenz , C-Q., T. Lenz Li ,, H. C-Q. Li Li, Q., Li H. Li, S., Liebler Q. Li , S., J. Liebler Lindert ,, J. Lindert , the SM predictions for the combination of ATLAS and CMS extrapolations. For each measurement,132 the133132 134133 44,134135 44,136135 13640 13240,137 132,13745 45 39 D. Liu 39, J.D. Liu Liu , Y., J.Liu Liu , Z., Y. Liu Liu ,, Z. A. Liu Long , K.A. Long Long ,, I. K. Low Long , I., G. Low Luisoni ,, G. Luisoni , 81 81 57 57 122 122 19 19138 138 1 1 total uncertainty is indicated by a grey box while the statistical, experimental and theory uncertainties40 L. L. Ma40 , A.-M.L. L. Ma Magnan, A.-M., D. Magnan Majumder, D. Majumder, A. Malinauskas, A. Malinauskas, F. Maltoni , F., M.Maltoni L. Mangano, M. L., Mangano , 74 74 101 10175 7595,139 95,139 1 1 39,140 39,140 41 G. Marchiori41 G., Marchiori A. C. Marini, A. C., A. Marini Martin ,, A. S. Martin Marzani, S. Marzani, A. Massironi, A., Massironi K. T. Matchev, K. T. Matchev, , 23 23 66 66 21 21116 116 5 5 15 15 are indicated by a blue, green and red line respectively. 42 R. D. Matheus42 R. D., K. Matheus Mazumdar, K. Mazumdar, A. E. Mcdougall, A. E. Mcdougall, P. Meade ,, P. P. Meade Meridiani, P., Meridiani A. B. Meyer, A., B. Meyer , 18 18 1,141 1,141 57 57138 138 142 142 43 E. Michielin43 E., Michielin P. Milenovic, P. Milenovic, V. Milosevic, V., Milosevic K. Mimasu, K., Mimasu B. Mistlberger, B. Mistlberger, , 143 143 1 1 47,144 47,14413,26 13,26 1 1451 145 44 M. Mlynarikova44 M. Mlynarikova, P. F. Monni,, P. G. F. Montagna Monni , G. Montagna, F. Monti , F., M. Monti Moreno, Llacer M. Moreno, A. Mueck Llacer ,, A. Mueck ,

SM j 2 j 1561 with the coupling modifiers as / = B  /(1 B ), where B is the SM branching H H j SM j BSM SM 1562 fraction for the H jj channel and B is the Higgs boson branching fraction to BSM final states. In ! BSM P 1563 the results for the j parameters presented here BBSM is fixed to zero and only decays to SM particles 1564 are allowed. Projections are also given for the upper limit on BBSM when this restriction is relaxed, in 1565 which an additional constraint that  < 1 is imposed. | V| 1566 The expected uncertainties for the coupling modifier parametrization for ATLAS, CMS ?? and 1567 their combination for scenario S2 are summarized in Figure 30 while the numerical values are provided 1568 in Table 36. For the combined measurement in S2, the uncertainty components contribute at a similar

1569 level for g , W, Z and t . The signal theory remains the main component for t and g, while µ and 1570 Zg are limited by statistics. SM 1571 The expectedDRAFT uncertainties on BBSM and H/H for the parametrisation with BBSM 0 and 1572  1. The 1 uncertainty on B is 0.035 (0.048) in S1 and 0.027 (0.032) in S2 for CMS | V|  BSM 1573 (ATLAS), where in the latter case the statistical uncertainty is the largest component. The expected 1574 uncertainty for the ATLAS-CMS combination on BBSM is XXXX in S2.

1575 The correlation coefficients between the coupling modifiers are in general larger compared to the 1576 one related to the signal strength (up to +XX ). One reason for this is that the normalisation of any 1577 signal process depends on the total width of the Higgs boson, which in turn depends on the values of the 1578 other coupling modifiers. The largest correlations involve b, as this gives the largest contribution to the 1579 total width in the SM. Therefore improving the measurement of the H bb process will improve the ! 1580 sensitivity of many of the other coupling modifiers at the HL-LHC.

63 DI-HIGGS PRODUCTION AT HL-LHC (HH → 4b, 3ab-1)

Figure 22. Same as Fig. 19 for the PU80+SK+Trim case. Behr, Bortoletto, Frost, Hartland, Issever & Rojo, 1512.08928

Category signal background S/pBtot S/pB4b S/Btot S/B4b tot 4b Nev Nev Nev no PU 290 1.2 104 8.0 103 2.7 3.2 0.03 0.04 Boosted · · PU80+SK+Trim 290 3.7 104 1.2 104 1.5 2.7 0.01 0.02 · · no PU 130 3.1 103 1.5 103 2.3 3.3 0.04 0.08 Intermediate · · PU80+SK+Trim 140 5.6 103 2.4 103 1.9 2.9 0.03 0.06 · · no PU 630 1.1 105 5.8 104 1.9 2.7 0.01 0.01 Resolved · · PU80+SK 640 1.0 105 7.0 104 2.0 2.6 0.01 0.01 · · no PU 4.0 5.3 Combined PU80+SK+Trim 3.1 4.7

1 Table 9. Post-MVA number of signal and background events with = 3 ab . For the back- tot 4bL grounds, both the total number, Nev , and the 4b component only, Nev , are shown. Also provided are the values of the signal signiﬁcance and the signalKey over signal background channels ratio, bothwill need separated ~1% in categories and for their combination. We quote the results withoutcontrol PU of and complex for PU80+SK+Trim. bkgds 23

mhh,thepT of the AKT03 subjets and the substructure variables, with a similar weighting among them. In Table 9 we provide the post-MVA number of signal and background events expected 1 tot for = 3 ab . For the backgrounds, we quote both the total number, Nev , and the L 4b QCD 4b component only, Nev . We quote results for the no PU and PU80+SK+Trim cases. We also quote in each case the corresponding values for the signal signiﬁcance and the signal over background ratio. Note that the MVA is always trained to the inclusive background sample, though di↵erences in the kinematic distributions of the 4b and 2b2j processes are moderate, see Fig. 14. From Table 9 one observes that all categories exhibit a marked improvement from eliminating the contamination from light and charm jet mis- identiﬁcation. For instance, in the intermediate category, S/pB increases from 2.3 to 3.3 (1.9 to 2.9) in the no PU (PU80) case, with similar improvements in the resolved and boosted categories. In Table 9 we also provide the results for S/pB obtained by combining the three categories. Taking into account all background components, we obtain for the case of

– 36 – 2 4 the SM, the light-quark e↵ects are small. Specifically, in 130 [9], respectively. It is however not competitive | c| . comparison to the Higgs e↵ective field theory (HEFT)with the bound c . 6.2 that derives from a global prediction, in gg hj the bottom contribution hasanalysis of Higgs| data| [9], which unlike (2) depends on ! an e↵ect of around 5% on the di↵erential distribu-fit assumptions and hence is more model dependent. tions for pT . mh/2 while the impact of the charm Turning our attention to the allowed modifications of quark is at the level of 1%. Likewise, the combinedthe bottom Yukawa coupling, one observes that our pro- gQ hQ, QQ¯ hg channels (with Q = b, c) lead to aposal leads to  [ 3.2, 8.3]. This limit is thus signifi- ! ! b shift of roughly 2%. Precision measurements of the Higgscantly weaker than2 the constraints from the LHC Run I distributions for moderate pT values combined with pre-measurements of pp W/Zh (h b¯b), pp tth¯ (h cision calculations of these observables are thus neededb¯b) and h b¯b in vector! boson fusion! that! already re-! ! to probe (1) deviations in yb and yc. Achieving suchstrict the relative shifts in y to around 50% [1, 2]. O b ± an accuracy is both a theoretical and experimental chal- Future prospects. In order to investigate the future lenge, but it seems possible in view of foreseen advancesprospects of our method in constraining the bottom and in higher-order calculations and the large statistics ex-charm Yukawa couplings, we study two benchmark cases. 1 indirect constraints on Hcc pected at future LHC upgrades. Our LHC Run II scenario employs 0.3 ab of integrated Theoretical framework. The goal of our work isluminosity and assumes a systematic error of 3% on impact of modified Hcc± joint limits on κc & κb to explore the sensitivity of the Higgs-boson (p ) andthe experimental side and a total theoretical uncertainty T,h coupling on Higgs+jet p dist @ HL-LHC leading-jet (p ) transverse momentum distributions inof 5%. This means that we envision that theT non- T,j ± inclusive Higgs production to simultaneous modificationsstatistical uncertainties1.4 present at LHC Run I can be 3 HL-LHC of the light Yukawa couplings. We consider final stateshalved in the coming years, which seems plausible.κc =-10 Our

SM 1

) 1.3 where the Higgs boson decays into a pair of electroweakHL-LHC scenarioj instead uses 3 ab of data and foresees , 2

T κ =-5 bosons. In order to be insensitive to the variations ofa reduction of both systematic and theoreticalc errors by

dp 1.2 the corresponding branching ratios due to light Yukawaanother factor/ of two, leading to uncertaintiesκc of= 0 1.5% σ ± modiﬁcations, we normalise the distributions to the in-and 2.5%,d respectively. Reaching such precisions will 1 × κ = 5 ± σ 1.1 c clusive cross section in the considered channels. The ef-clearly require/ a dedicated experimental and theoretical b 1 e↵ort. In both benchmarks, we employ ps = 13 TeV and κ

fect on branching ratios can be included in the context of )/( j the PDF4LHC15, 1.0nnlo mc set [58–61], consider the range 0 a global analysis, jointly with the method proposed here. T 3 pT [0, 70] GeVdp in bins of 5 GeV, and take into account

The gg hj channel has been analysed in depth in / 2 0.9 ! h , h σ ZZ⇤ 4` and h WW⇤ 2`2⌫`. × SM the HEFT framework where one integrates out the domi- ! d ! ! ! ! -1 We assume that the future measurements will be cen- 2 nant top-quark loops and neglects the contributions from σ Δχ = 2.3 / 0.8 tred around1 the SM predictions. These channels sum to lighter quarks. While in this approximation the two spec- ( Δχ2 = 5.99 tra and the total cross section have been studied exten-a branching ratio of 1.2%, but given the large amount -2 0.7 sively, the e↵ect of lighter quarks is not yet known withof data the statistical10 errors20 per30 bin will40 be at50 the 602% -10 -5 0 5 10 ( 1%) level in our LHC Run II (HL-LHC) scenario.± We the same precision for pT mh/2. Within the SM, the . ± pT, j [GeV] κ LO distribution for this process has been derived longmodel the correlation matrix as in the 8 TeV case. c The results of our 2 fits are presented in Figure 3 24 ago [14, 15], and the next-to-leading-order (NLO) cor- Fady Bishara, Ulrich Haisch, Pier FrancescoFigure 3: Monni Projected and Emanuele future constraints Re, arXiv:1606.09253 in the c –b plane. Figure 1: The p normalised spectrum of inclusive Higgs rections to the total cross section have been calculatedwith the upper (lower)T,j panel showing the constraints in The SM point is indicated by the black crosses. The upper production divided by the SM prediction for di↵erent values 1 1 in [16–20]. In the context of analytic resummations ofthe c –b plane for the LHC Run II (HL-LHC) scenario. (lower) panel shows our projection for 0.3ab (3 ab )of of b (upper panel) and c (lower panel). In each panel onlyintegrated luminosity at ps = 13TeV. The remaining as- the Sudakov logarithms ln (p /m ), the inclusion of massBy profiling over b, we find in the LHC Run II scenario T h the indicated Q is modified, while the remaining Yukawasumptions entering our future predictions are detailed in the the following 95% CL bound on the yc modifications corrections to the HEFT has been studied both for the couplings are kept at their SM values. main text. p and p distributions [21–23]. More recently, the T,h T,j  [ 4.7, 5.5] (LHC Run II) , (3) first resummations of some of the leading logarithms (1) c 2 have been accomplished both in the abelian [24] and while the corresponding HL-LHC bound reads pp W/Zh (h cc¯) and pp hc channels, which in the high-energy [25] limit. The reactions gQ to NNLL order both for pT,h [38–40] and pT,j [41–43], ! ! ! hQ, QQ¯ hg have been computed at NLO [26, 27]in! treating mass corrections following [23]. The latter ef-are either limited by small signal-to-background ratios c [ 2.9, 4.2] (HL-LHC) . (4) the five-flavour! scheme that we employ here, and the re- fects will be2 significant, once the spectra have been pre-or by the charm-bottom discrimination of heavy-flavour tagging. We notice that at future LHC runs our method summation of the logarithms ln (pT,h/mh)inQQ¯ hThesecisely limits measured compare down well not to p onlyT values with theof projected(5 GeV). The ! O will allow one to set relevant bounds on the modifications has also been performed up to next-to-next-to-leading-reachgQ of otherhQ, proposed QQ¯ strategieshg contributions but also to have the the distributions nice ! ! of y . For instance, in the HL-HLC scenario we obtain logarithmic (NNLL) order [28]. featureare that calculated they are at NLO controlled with byMG5aMC@NLO the accuracy[44] and that cross- b  [0.3, 1.4] at 95% CL. In the case of gg hj, we generate the LO spectratheoreticalchecked predictions against MCFM can. reach The obtained in the future. events are This showered is b 2 with MG5aMC@NLO [29!]. We also include NLO correctionsnot thewith casePYTHIA for extractions 8 [45] and of jetsy using are reconstructed the h J/ with, the Finally, we have also explored the possibility of con- c ! straining modifications  = y /ySM of the strange to the spectrum in the HEFT [30–32]usingMCFM [33]. anti-kt algorithm [46] as implemented in FastJet [47] s s s Yukawa coupling by means of our proposal. Under the The total cross sections for inclusive Higgs production are using R =0.4 as a radius parameter. assumption that the bottom Yukawa coupling is SM-like obtained from HIGLU [34], taking into account the next- Our default choice for the renormalisation (µR), fac- 3 Enlarging the bin size leads to a minor reduction of the sensitivity but profiling over c, we find that at the HL-LHC one to-next-to-leading order corrections in the HEFT [35– torisation (µF ) and the resummation (QR,presentinthe to the Yukawa modifications, because shape information is lost. should have a sensitivity to y values of around 30 times 37]. Sudakov logarithms ln (p /m ) are resummed up gg hj case) scales is m /2. Perturbative uncertainties s T h ! h what might we do with it? 3. New physics searches

25 See also e.g. Biekötter, Knochel, Krämer, Liu, Riva, arXiv:1406.7320 VH production at large m(VH) [specific illustration of a generic point]

➤ Higher-dimension operators cause deviations that grow as, e.g. 2 dim-6 pT ⇠ ⇤2

➤ In some relevant range of pT, Λ value to which you’re sensitive grows as ⇤ (Lumi)1/4 ⇠ ➤ that’s faster than most direct searches (x100 in lumi → x1.5 in reach for Z’)

Mimasu, Sanz, Williams, arXiv:1512.02572v

26 WH at large Q2 with dim-6 BSM effect

-1 3000 fb ��

�� new physics isn’t just a

�� single number that’s � � �� wrong (think g-2) � � � �

�� but rather a distinct � �� scaling pattern of ��

�� 2 deviation (~ pT ) � ��

�� moderate and high pT’s ��

�� have similar statistical � ��

� signiﬁcance — so it’s �� schematic useful to understand GPS 2016-10 �� whole pT range ��

��

27 WH at large Q2 with dim-6 BSM effect

-1 3000 fb ��

�� new physics isn’t just a

�� single number that’s � � �� wrong (think g-2) � � � �

�� but rather a distinct � �� scaling pattern of ��

�� 2 deviation (~ pT ) � ��

�� moderate and high pT’s ��

�� have similar statistical improved systematic precision gives� enhanced ��

access to lower pT end of such scaling� patternssigniﬁcance — so it’s �� schematic (where pT2 pattern is most robust) useful to understand GPS 2016-10 �� ̶> crucially important in deciding if a signalwhole of pT range �� this kind is real ��

27 conclusions

28 Conclusions

➤ Experimentally: relative sub-percent precision looks feasible for many leptonic processes. Open question is absolute precision, especially luminosity. ➤ Theoretically: expect substantial further progress on higher-order perturbative calculations. For a subset of processes (non-inclusive), open questions about nonperturbative eﬀects. [Technology probably exists to answer this questions] ➤ Varied applications: ➤ many processes, over wide phasespace, will have statistical precision < 1% ➤ could bring PDFs into new era of precision (& complementary to low-Q2 DIS?) ➤ will build conﬁdence in precision needed for precision Higgs physics

➤ can provide crucial low-pT end of lever-arm in searches for higher-dim. BSM operators 29 backup

30 Input parameters? Especially, αs

(almost) all theory predictions for LHC are based on perturbation theory, e.g.

2 σ = αsσ1 + αs σ2 + …

not so critical for DY, VV (LO doesn’t depend on αs), 2 but is crucial for ttbar and other processes that start at O(αs ) 31 32 Bethke, Bethke, Dissertori & GPS in PDG ‘16 ) = 0.1181 ± 0.0013 (1.1%) 0.0013 ± 0.1181 ) = Z (M s α Average: World PDG ts structure f τ-decays lattice e+e- annihilation collider functions hadron electroweak precision (Wilson loops) (3j) (vac. pol. fctns.) pol. (vac. (static energy) (static (Wilson loops) (c-c correlators)

(T) (T) (jets&shapes) (Adler functions) (Adler

(3j) (j&s) (j&s) (ghost-gluon vertex) (T)

(C)

Hoang

JADE

OPAL

ALEPH ALEPH

BBGPSV

ETM

PACS-CS PACS-CS

Abbate Abbate

JADE

Dissertori

MMHT

NNPDF

BBG

ABM

JLQCD JLQCD

Maltmann

HPQCD

SM review

Boito

Pich

Davier

Baikov

(tt cross section) (tt cross

CMS

GFitter Gehrm. 36 1. Quantum chromodynamics τ

Baikov -decays Davier Pich Boito SM review

HPQCD (Wilson loops) HPQCD (c-c correlators) lattice Maltmann (Wilson loops) 36 JLQCD1. Quantum (Adler functions) chromodynamics PACS-CS (vac. pol. fctns.) ETM (ghost-gluon vertex) τ BBGPSV (staticBaikov energy) -decays Davier Pich functions structure τ

Baikov -decays ABM Boito Davier BBG SM review Pich PDG World Average: αs(MZ) = 0.1181 ± 0.0013 (1.1%) JR HPQCD (Wilson loops) Boito HPQCD (c-c correlators) SM review NNPDF lattice ➤ Maltmann (Wilson loops) Most consistent set of independentMMHT determinationsJLQCD (Adler functions) is from lattice HPQCD (Wilson loops) PACS-CS (vac. pol. fctns.) (c-c correlators) e+e- annihilation e+e- HPQCD ETM (ghost-gluon vertex) τ lattice ➤ Maltmann (Wilson loops) Two best determinations are fromBaikovALEPH same (jets&shapes) groupBBGPSV (static (HPQCD, energy) -decays JLQCD (Adler functions) OPAL(j&s) functions structure PACS-CS (vac. pol. fctns.) 1004.4285, 1408.4169) Davier ABM BBG ETM (ghost-gluon vertex) PichJADE(j&s) JR BBGPSV (static energy) αs(MZ) = 0.1183 ± 0.0007 (0.6%) [heavy-quark correlators] BoitoDissertori (3j)NNPDF ABM functions structure αs(MZ) = 0.1183 ± 0.0007 (0.6%)SMJADE [Wilson review (3j) MMHT loops] BBG annihilation e+e- ➤ DW (T) ALEPH (jets&shapes) JR Error criticised by FLAG, who OPAL(j&s) HPQCDAbbate (Wilson(T) loops) NNPDF JADE(j&s) MMHT suggest HPQCDGehrm. (c-c(T) Dissertoricorrelators) (3j) lattice JADE (3j) e+e- annihilation e+e- MaltmannHoang (Wilson loops) ALEPH (jets&shapes) DW (T) (C) OPAL(j&s) αs(MZ) = 0.1184 ± 0.0012(1%) JLQCD (AdlerAbbate functions) (T) electroweak JADE(j&s) Gehrm. (T) PACS-CSGFitter (vac.Hoang pol. fctns.) Dissertori (3j) ➤ (C) precision fts JADE (3j) Worries include missing ETM (ghost-gluon vertex) electroweak CMS GFitter hadron DW (T) BBGPSV (static energy) precision fts perturbative contributions, non- (tt cross section) Abbate (T) CMS hadroncollider (tt cross section) (T) collider Gehrm. perturbative effects in 3–4 functions structure Hoang ABM (C) electroweak flavour transition at charm mass BBG GFitter precision fts JR CMS hadron [addressed in some work], etc. (tt cross section) collider NNPDF 2 ➤ FigureMMHT 1.2: Summary of determinations of αs(MZ )fromthesixsub-fields New ALPHA extraction (1706.03821)discussed inis the cleaner text. The yellowin many (light shaded) respects bands and dashed lines indicate the

pre-average values of each sub-field. The dotted line and gre2 annihilation e+e- y(darkshaded)band αs(MZ) = 0.1185 ± 0.00084(0.7%)Figure 1.2: Summary of determinations of αs(M )fromthesixsub-fields representALEPH the(jets&shapes) final world average value of α (M 2 ). 33 Z discussed in the text. The yellow (light shaded)s Z bands and dashed lines indicate the OPAL(j&s) pre-average valuesJADE(j&s) of each sub-field. The dotted line and grey(darkshaded)band represent the finalSo far, world only one average analysis is available value of whichα involves(M 2 ). the determination of αs from hadronDissertori collider data(3j) in NNLO of QCD: from as measurementZ of the tt cross section at JADE (3j) DW (T) February 22, 2016 10:28 Abbate (T) So far, only oneGehrm. analysis (T) is available which involves the determination of αs from hadron collider dataHoang in NNLO of QCD: from a measurement of the tt cross section at (C) electroweak GFitter February 22, 2016 10:28precision fts CMS hadron (tt cross section) collider 36 1. Quantum chromodynamics τ 36 1. QuantumBaikov chromodynamics -decays Davier Pich Boito τ

SMBaikov review -decays Davier HPQCD (Wilson loops) Pich HPQCD (c-c correlators) Boito lattice MaltmannSM review (Wilson loops) 36 JLQCD1. Quantum (Adler functions) chromodynamics PACS-CSHPQCD (Wilson (vac. pol. loops) fctns.) ETM (ghost-gluon(c-c correlators) vertex)

HPQCD τ lattice (staticBaikov energy) -decays BBGPSV (Wilson loops) Maltmann Davier JLQCD (AdlerPich functions) functions structure ABM Boito PACS-CS (vac.SM pol. review fctns.) BBGETM (ghost-gluon vertex) HPQCD (Wilson loops) JRBBGPSV (static energy) HPQCD (c-c correlators) lattice NNPDF Maltmann (Wilson loops) MMHTABM JLQCD (Adler functions) functions structure PACS-CS (vac. pol. fctns.)

2 Figure 1.2: Summary of determinations of αs(MZ )fromthesixsub-fields discussed in the text. The yellow (light shaded) bands and dashed lines indicate the 2 Figure 1.2: pre-averageSummary values of of determinations each sub-field. The dotted of lineαs and(M grey(darkshaded)band)fromthesixsub-fields represent the final world average value of α (M 2 ). Z discussed in the text. The yellow (light shaded)s Z bands and dashed lines indicate the pre-average values of each sub-field. The dotted line and grey(darkshaded)band So far, only one analysis is available which involves2 the determination of αs from represent the final world average value of αs(MZ). 2 Figure 1.2:hadronSummary collider data of in determinations NNLO of QCD: from a of measurementαs(MZ of)fromthesixsub-fields the tt cross section at

discussed in the text. The yellow (lightFebruary shaded) 22, 2016 10:28 bands and dashed lines indicate the pre-average values of each sub-ﬁeld. The dotted line and grey(darkshaded)band 2 Sorepresent far, only the one ﬁnal analysis world average is available value which of αs(M involvesZ). the determination of αs from hadron collider data in NNLO of QCD: from a measurement of the tt cross section at

34 February 22, 2016 10:28 So far, only one analysis is available which involves the determination of αs from hadron collider data in NNLO of QCD: from a measurement of the tt cross section at

February 22, 2016 10:28 36 1. Quantum chromodynamics τ 36 1. QuantumBaikov chromodynamics -decays Davier Pich Boito τ

BBG ETM (ghost-gluon vertex) annihilation e+e- ALEPHJR (jets&shapes)BBGPSV (static energy) OPALNNPDF(j&s) ABM functions structure JADE(j&s) BBG MMHT JR Dissertori (3j)NNPDF e+e- annihilation e+e- MMHT JADEALEPH (3j) (jets&shapes) e+e- annihilation e+e- DWOPAL (T)(j&s) ALEPH (jets&shapes) E+E- EVENT SHAPES AND JET RATES OPAL(j&s) Abbate(j&s) (T) JADE JADE(j&s) ➤ Gehrm.Dissertori (T) (3j)Dissertori (3j) Two “best” determinations are from same group Hoang JADE (3j) JADE (3j) DW (T) (C) (Hoang et al, 1006.3080,1501.04111) DW (T) Abbate (T) Gehrm. (T) electroweak GFitterAbbate (T) Hoang αs(MZ) = 0.1135 ± 0.0010 (0.9%) [thrust] (C) precision fts Gehrm. (T) electroweak GFitter hadron CMSHoang precision fts αs(MZ) = 0.1123 ± 0.0015 (1.3%) [C-parameter] (tt cross section) (C) CMS30 hadroncollider (tt cross section) colliderelectroweak dependence on fit range GFitter 5 ± 2 precision fts 0 0" 38 ')& :ELQV αs(mZ ) (pert. error) χ /(dof) Q %<& ττPLQ0 PD[ 3 ′ Rgap 0 09 0 33 ± CMS hadron !"*! " 0 " N LLthrustwith Ω1 & “best”0.1135 lattice0.0009 are 0.91 4-σ apart 3 ′ MS (tt cross section) collider N LL with Ω¯ 1 0.1146 ± 0.0021 1.00 " ! 5 '#( 0 0" 33 3 ′ mod ± 2 Q N LL without Sτ 0.1241 0.0034Figure 1.2: 1.26Summary of determinations of αs(M )fromthesixsub-fields +*H9/ 6 '&' Z 0 0" 38 O 3 Q 8 (αs)fixed-order discussed in the text. The yellow (light shaded) bands and dashed lines indicate the !")! 0 0" 33 &&$ Comments: 0.1295 ± 0.0046 1.12 Q without Smod pre-average values of each sub-field. The dotted line and gre2y(darkshaded)band 6 %<$ τ Figure 1.2: Summary of determinations of2 αs(M )fromthesixsub-fields 0 0" 25 &%% represent the final world average value of αs(M ). Z Q VWULFW vary0 0" 33 Z TABLE➤ VII: Comparisondiscussed of global fit in results the for text. our full The anal- yellow (light shaded) bands and dashed lines indicate the 6 &*) thrust & C-parameter are highly correlated observables 0 0" 33 ¯ Q VWULFW ysis to a fit where the renormalon is not canceled with Ω1,a mod pre-average values of each sub-field. The dotted line and grey(darkshaded)band !"(! fit without Sτ (meaning without powerSo corrections far, only one with analysis is available which involves2 the determination of α from mod➤ Analysis validrepresent far from the final3-jet world region, average but valuenot too of αdeeps(M ). 2 s Sτ (k)=δ(k)), and aFigure fit at fixed 1.2: orderhadron withoutSummary collider power data cor- of in determinations NNLO of QCD: from a of measurementαZs(M of)fromthesixsub-fields the tt cross section at rections and log resummation. All results include bottom Z into 2-jet regiondiscussed — inat theLEP, text. not The clear yellow how (light much shaded) of bands and dashed lines indicate the 8 mass and QED corrections. February 22, 2016 10:28 0 0" 33 &$) !"'! Q VWULFW distributionpre-average satisfies this values requirement of each sub-field. The dotted line and grey(darkshaded)band 8 %%' 39% CL 0 0" 25 2 Q VWULFW Sorepresent far, only the one final analysis world average is available value which of αs(M involvesZ). the determination of αs from 68% CL eff➤ect of power corrections is a shift of the distribution in τ, wethrust have estimated fithadron shows in Sec. collider noticeable I that data a 300 MeV insensitivity NNLO power of to QCD: fit region from a (C- measurement of the tt cross section at !"&! correction will lead to an extraction of α from Q = m !"### !"##$ !"##% !"##& !"##' !"##( parameter doesn't) s Z data that is δα /α ( 9 3)% lower than an anal- 34 s s ≃ − ± February 22, 2016 10:28 ysis without power corrections.So far, only In our one theory analysis code we is available which involves the determination of αs from can easily eliminate all nonperturbative effects by set- FIG. 17: The smaller elongated ellipses show the experimental hadron collider data in NNLO of QCD: from a measurement of the tt cross section at ting Smod(k)=δ(k)and∆¯ = δ =0.AtN3LL′ or- 39% CL error (1-sigma for αs)andbestfitpointsfordifferent τ global data sets at N3LL′ order in the R-gap scheme and der and using our scan method to determine the per- including bottom quark mass and QED effects. The default turbative uncertainty a global fit to our default data set February 22, 2016 10:28 theory parameters given in Tab. III are employed. The larger yields αs(mZ )=0.1241 (0.0034)pert which is indeed ± ellipses show the combined theoretical plus experimental error 9% larger than our main result in Eq. (68) which ac- for our default data set with 39% CL (solid, 1-sigma for one counts for nonperturbative effects. It is also interesting dimension) and 68% CL (dashed). to do the same fit with a purely fixed-order code, which we can do by setting µS = µJ = µH to eliminate the summation of logarithms. The corresponding fit yields experimental error ellipses, hence to larger uncertainties. αs(mZ )=0.1295 (0.0046)pert, where the displayed error ± It is an interesting but expected outcome of the fits has again been determined from the theory scan which in this case accounts for variations of µ and the numerical that the pure experimental error for αs (the uncertainty H uncertainties associated with ϵ2 and ϵ3. (A comparison of αs for fixed central Ω1) depends fairly weakly on the τ range and the size of the global data sets shown in with Ref. [22] is given below in Sec. IX.) Fig. 17. If we had a perfect theory description then we These results have been collected in Tab. VII together would expect that the centers and the sizes of the error with the αs results of our analyses with power corrections ellipses would be statistically compatible. Here this is in the R-gap and the MS schemes. For completeness we 2 not the case, and one should interpret the spread of the have also displayed the respective χ /dof values which ellipses shown in Fig. 17 as being related to the theo- were determined by the average of the maximal and the retical uncertainty contained in our N3LL′ order predic- minimum values obtained in the scan. tions. In Fig. 17 we have also displayed the combined (experimental and theoretical) 39% CL standard error ellipse from our default global data set which was al- VIII. FAR-TAIL AND PEAK PREDICTIONS ready shown in Fig. 11a (and is 1-sigma, 68% CL, for either one dimensional projection). We also show the The factorization formula (4) can be simultaneously used 68% CL error ellipse by a dashed red line, which corre- in the peak, tail, and far-tail regions. To conclude the sponds to 1-sigma knowledge for both parameters. As discussion of the numerical results of our global analysis we have shown above, the error in both the dashed and in the tail region, we use the results obtained from this solid larger ellipses is dominated by the theory scan un- tail fit to make predictions in the peak and the far-tail certainties, see Eqs. (68). The spread of the error ellipses regions. from the different global data sets is compatible with the In Fig. 18 we compare predictions from our full N3LL′ 1-sigma interpretation of our theoretical error estimate, code in the R-gap scheme (solid red line) to the accurate and hence is already represented in our final results. ALEPH data at Q = mZ in the far-tail region. As input for α (m )andΩ we use our main result of Eq. (68) Analysis without Power Corrections s Z 1 and all other theory parameters are set to their default Using the simple assumption that the thrust distribution values (see Tab. III). We find excellent agreement within in the tail region is proportional to αs and that the main the theoretical uncertainties (pink band). Key features WHAT WAY FORWARDS FOR αs?

➤ We need to settle question of whether “small” (0.113) αs is possible. LHC data already weighing in on this (top data), ATLAS-CONF-2015-049

further info in near future (Z pT, cf. later slides) αs(MZ)=0.113

➤ To go beyond 1%, best hope is probably lattice QCD — on a 10-year timescale, there will likely be enough progress that multiple groups will have high-precision determinations NB: top-quark mass choice affects this plot

35 DATA-DRIVEN BKGD ESTIMATES: NON-SMOOTHNESS AT 1% LEVEL

Predictions at high invariant masses. As we all know, bump hunts in the diphoton system assume a smooth function which can be ﬁtted to the data. Begging the question, Standard t How smooth is smooth? :-) experimental C. Williams techniques, like Moriond QCD ‘16 t data-driven bkgd Figure 2.RepresentativeFeynmandiagramsforthecalculationof=( +gg) at LO (top left) and ! estimates, can be NLO (the remainder). The virtual two-loop corrections are shown in the top right, while the bottom =( )+ row corresponds to real radiation contributions.=( + )+(( )) skewed by O(1%)

) Figure 2.RepresentativeFeynmandiagramsforthecalculationof( soft [51, 52] and beam [53] functions,gg togetherat LO (top with left) the and process-dependent hard function. theoretical ! NLO (the remainder). The virtual two-loopVarious corrections component are pieces shown of in this the calculation, top right, while including the bottom explicit results for the hard function,

row corresponds to real radiation contributions. 1% / are given in Appendix A subtleties. 2.2 gg initiated loops at LO and NLO soft [51, 52] and beam [53] functions, together with the process-dependent hard function. Various component pieces of this calculation,The NNLO including calculation explicit of resultsproduction for the represents hard function, the first order in perturbation theory that is sensitive to gg initial states. One class of gg configurations corresponds to real-real are given in Appendix A corrections, i.e. thegg qqmatrix element that is related to the contribution shown in ! 20 2.2 gg initiated loops at LOfigure and NLO1 (right) by crossing. These [ pieces] are combined with contributions from the DGLAP 36 evolution of the parton distribution functions in the real-virtual and double-virtual terms The NNLO calculation of production represents the first order in perturbation theory Figure 10.Theratioofvariousdito ensure an IR-finitefferent result. theoretical The second predictions type of to contribution the NNLO isnF due=5 to dinFfferential“box” loops, for that is sensitivecross to section.gg initial The states.which diff Oneerent a class representative predictions of gg configurations correspond Feynman diagram corresponds to: the is inclusion shown to real-real in of the the top top left quark cornergg of Figure 2. This corrections, i.e. the gg qq matrix elementN3LO that is related to the contributionN3LO shown in ! box diagrams! (green),contribution the gg,n hasF nocorrection tree-level (red) analogue and andthe is thusgg,nF separatelyand the finite. top boxes with the figure 1 (right) byN3LO crossing.correction These re-scaled piecesThe are box by combined thediagrams ratio with result(m contributions) indescribed a sizeable from in cross the the section text DGLAP (blue). ( ), primarily due to the large gg,nF K t ⇡ LO evolution of the parton distributiongluon functions flux at in LHC the energies real-virtual and and the double-virtual fact that this contribution terms sums over different quark to ensure an IR-finite result. The second type of contribution is due to nF “box” loops, for flavors in the loop. In this section, we focus on nF =5light quark loops. Since this which a representative Feynmancontribution diagram is shown is clearly in the important top left corner for phenomenology of Figure 2. This it is interesting to try to isolate and analyses the Standard Model background is accounted for by using a data-driven approach contribution has no tree-level analoguecompute and higher is thus order separately corrections finite. to it. We illustrate typical component pieces of these The boxthat diagrams fits result a smooth in aNLO sizeable polynomial corrections cross section function in the ( remaining toLO), the primarily diagramsdata across due in to Figure the the large entire2. Theym comprise spectrum. two-loop Agg ⇡ ! gluon flux atresonance LHC energies might and then theamplitudes, fact be that observed and this one-loop contribution as a localggg sums excessand overgq inq di thisffamplitudes.erent spectrum, quark A deviatingNLO calculation from of thegg ! flavors in thefitted loop. form. In this Although section,including we well-motivated, focus the two-loop on nF =5 andonelight one-loop might quark beggg concerned loops. amplitudes Since that this was the presented spectrum in may refs. [ not20, 21]. An contributionbe is clearly correctly important modeledinfrared-finite for at phenomenology high energies, calculation it where is interesting can bethere obtained tois little try from to data, isolate the andgg and that two small loop fluctuations amplitudes and the compute higher order corrections to it. We illustrate typical component pieces of these! could unduly influenceggg theone-loop form amplitudes, of the fit and provided result that in a misinterpretation suitable modification of to the the data. quark Such PDFs is used NLO corrections in the remaining diagrams in Figure 2. They comprise two-loop gg worries could be lessened(essentially by using using a a LO first-principles evolution for the theoretical quark PDFs prediction! and a NLO for evolution the spectrum for the gluon amplitudes, and one-loop ggg PDFs).and gqq On amplitudes. the other hand A NLO if the calculationqqg amplitudes of gg are included then the corresponding and it is this issue that we aim to address in this section. ! including the two-loop and one-loopcollinearggg singularityamplitudes can was be presented absorbed in into refs. the [20 quark, 21]. An PDFs as normal at NLO, allowing infrared-finite calculationAs a concrete can be obtained example, from we will the gg produce two NNLO loop amplitudespredictions and for the the invariant mass spec- for a fully consistent treatment.! In the original calculation [20, 21] (and the corresponding ggg one-looptrum amplitudes, at high provided energiesimplementation that using a suitable cuts in MCFM modification that are [46]) inspired the to the first quark approach by the PDFs recent was is used taken. ATLAS Here we analysis will follow [16 the]. second (essentially usingSpecifically, a LO evolution these are: for the quark PDFs and a NLO evolution for the gluon PDFs). On the other hand if the qqg amplitudes are included then the corresponding collinear singularity can be absorbed,hard into the quark PDFs,soft as normal at NLO, allowing pT > 0.4 m pT > 0.3 m for a fully consistent treatment. In the original calculation [20, 21] (and the–4– corresponding ⌘ < 2.37, excluding the region, 1.37 < ⌘ < 1.52 (4.6) implementation in MCFM [46]) the| first| approach was taken. Here we will follow the| second|

We will only be interested in the region m > 150 GeV, so these represent hard cuts on the photon momenta. The small region of rapidity that is removed corresponds to the transition from barrel to end-cap calorimeters.–4– We maintain the same isolation requirements as the previous section, which again differs slightly from the treatment in the ATLAS paper. Our first concern is to address the impact of the gg pieces at NLO, represented by N3LO the contribution gg,nF defined previously, and the contribution of the top quark loop. We summarize our results in Figure 10, in which we present several different theoretical predictions, each normalized to the the default NNLO prediction with 5 light flavors. The

–15– e-μ events (a mix of ttbar and VV) at HL-LHC ATLAS/CMS

number of e-μ events with 3000 fb-1@ 14 TeV (ATLAS/CMS HL-LHC) 3x106 ttbar WW this is a quick study to gauge orders of magnitude (Pythia8, LO, no showering) 2x106 LO, Pythia8

bin N(N)LO K-factors will increase /

|ηe/μ| < 4, pT,μ/e > 20 GeV rates significantly

events there will also be other VV channels 1x106 (probably smaller)

0 -4 -2 0 2 4

rapeμ 37 Photon PDF (NNPDF31luxQED, using Manohar, Nason, GPS & Zanderighi, 1607.04266)

- p p → l+ l @ s = 13 TeV, 0 < |y | < 2.5 ll 0.06 PI 3.1luxQED

0.05 PI 3.0QED PDF errors 0.04 barely any uncertainty on

0.03 γγ-induced contribution

0.02

0.01 DY measurements can genuinely constraint Ratio to central value of NNPDF3.1luxQED 0 15 20 25 30 35 40 45 50 55 60 other PDF contributions

Mll (GeV) 38 Figure 4.3. Similar representation as the right panel of Fig. 4.2 for the low (left) and high (right plot) invariant mass regions, deﬁned as 15 GeV Mll 60 GeV and Mll 400 GeV respectively. Please note that the right ﬁgure is plotted in a larger y-axis range in comparison to previous plots.

Mll region the contribution of the PI channel exceeds the level of PDF uncertainty, highlighting the sensitivity of this distribution to the photon PDF. We find that NNPDF3.1luxQED and LUXqed17 lead to a larger PI contribution as compared to NNPDF3.0QED at low Mll. As the PI contribution is only significant away from the Z-peak, where the bulk of the cross-section lies, these e↵ects may be reasonably neglected in the integrated cross-sections. Fig. 4.2 demonstrates that the PI contributions in NNPDF3.1luxQED and LUXqed17 lead to very similar results for Drell-Yan production around the Z peak. We have verified that this similarity holds also for the low and high mass kinematic regions, as well as for the rest of processes studied in this section. In the following discussion we will therefore restrict ourselves to comparisons between NNPDF3.0 and NNPDF3.1luxQED. We now move to study the low- and high-mass regions, defined as 15 M 60 GeV and  ll  M 400 GeV respectively. Drell-Yan low-mass measurements have been presented by ATLAS, ll CMS, and LHCb [84,127,128], with the two-fold motivation of providing input for PDF fits and to study QCD in complementary kinematic regimes. The high-mass region is relevant for BSM searches that exploit lepton-pair final states, such as those expected in the presences of new heavy gauge bosons W 0 or Z0 [129, 130]. In Fig. 4.3 we show the same comparison as in the right panel of Fig. 4.2 for the low- and high-mass regions. In the low-mass case, the PI e↵ects are more significant than in the Z-peak region, being between 3% and 4% for most of the Mll range, consistently larger that the PDF uncertainty. We find that PI e↵ects in NNPDF3.1luxQED can be up to a factor three larger than in the NNPDF3.0QED case due to the corresponding di↵erences in the photon PDF at small x. In the case of the high-mass region, we observe that the e↵ect of the PI contribution computed with NNPDF3.1luxQED is comparable to the PDF uncertainties for Mll > 3 TeV, eventually becoming as large as 10% of the QCD cross-section. These e↵ects are markedly⇠ smaller than ' in NNPDF3.0QED, where shifts in the cross-section up to 80% due to PI contributions were ' allowed within uncertainties. To conclude this discussion on Drell-Yan at the LHC, we have evaluated the ratio of the LO PI contributions to the NLO QCD cross-sections for the kinematics of the ATLAS high-mass Drell-Yan measurements at 8 TeV [40]. Both the Bayesian reweighting study of the ATLAS paper [40] and the analysis of Ref. [22] indicate that this dataset has a considerable sensitivity to PI contributions if NNPDF3.0QED is used as a prior. Here we revisit this process to assess how the picture changes when using NNPDF3.1luxQED. In Fig. 4.4 we show the ratio of PI over QCD contributions for the lepton-pair rapidity distributions y in Drell-Yan at 8 TeV for two invariant mass bins, 250 GeV M 300 | ll|  | ll|  GeV and 300 GeV M 1500 GeV. As can be seen, with NNPDF3.0QED the e↵ects of  | ll|  the PI contribution at large invariant masses can be as large as 25% of the QCD cross-section. This shift is larger than the corresponding experimental uncertainties, which are typically at the percent level, explaining the sensitivity of NNPDF3.0QED to this dataset. From Fig. 4.4 we observe that the PI contribution becomes smaller when using NNPDF3.1luxQED, though its