Towards ultimate parton distributions at the highluminosity LHC
Abstract
Since its start of data taking, the LHC has provided an impressive wealth of information on the quark and gluon structure of the proton. Indeed, modern global analyses of parton distribution functions (PDFs) include a wide range of LHC measurements of processes such as the production of jets, electroweak gauge bosons, and top quark pairs. In this work, we assess the ultimate constraining power of LHC data on the PDFs that can be expected from the complete dataset, in particular after the HighLuminosity (HL) phase, starting in around 2025. The huge statistics of the HLLHC, delivering \({\mathcal {L}}=3\hbox { ab}^{1}\) to ATLAS and CMS and \({\mathcal {L}}=0.3\hbox { ab}^{1}\) to LHCb, will lead to an extension of the kinematic coverage of PDFsensitive measurements as well as to an improvement in their statistical and systematic uncertainties. Here we generate HLLHC pseudodata for different projections of the experimental uncertainties, and then quantify the resulting constraints on the PDF4LHC15 set by means of the Hessian profiling method. We find that HLLHC measurements can reduce PDF uncertainties by up to a factor of 2 to 4 in comparison to stateoftheart fits, leading to fewpercent uncertainties for important observables such as the Higgs boson transverse momentum distribution via gluonfusion. Our results illustrate the significant improvement in the precision of PDF fits achievable from hadron collider data alone, and motivate the continuation of the ongoing successful program of PDFsensitive measurements by the LHC collaborations.
1 Introduction
A detailed understanding of the quark and gluon structure of the proton [1, 2, 3] is an essential ingredient of theoretical predictions for hadron colliders such as the LHC. This is quantified by the parton distribution functions (PDFs), which determine how the proton’s momentum is shared among its constituents in a hard–scattering collision. PDF uncertainties represent one of the dominant theoretical systematic errors in many important LHC processes, including the profiling of the Higgs boson sector [4]; direct searches for new heavy beyond the Standard Model (BSM) particles [5]; indirect BSM searches by means of the SM Effective Field Theory (SMEFT) [6]; as well as in the measurement of fundamental SM parameters such as the W boson mass [7], the Weinberg mixing angle [8], and the strong coupling constant [9] and its running [10].
Since the start of data taking in 2009, the LHC has provided an impressive wealth of information on the proton’s PDFs. Indeed, modern global PDF fits [11, 12, 13, 14] include a wide range of LHC measurements in processes such as the production of jets, weak gauge bosons, and top quark pairs. Crucially, the recent breakthroughs in the calculation of NNLO QCD and NLO QED and electroweak corrections (including photon–induced ones) to most PDF–sensitive processes have been instrumental in allowing for the full exploitation of the information provided by the LHC measurements. The impact of high precision LHC data combined with state–of–the art perturbative calculations has been quantified for many of the processes of interest, such as top quark pair production [15, 16], the transverse momentum spectrum of Z bosons [17], direct photon production [18, 19], D meson production in the forward region [20, 21], W production in association with charm quarks [22, 23, 24], and inclusive jet production [25, 26]. See the reviews [1, 2] for a more extensive list of references.
With experimentalists warming up to analyse the complete Run II dataset, the high energy physics community is already busy looking ahead to the future. Following Run III, around 2023, a major upgrade of the LHC accelerator and detector systems will make the start of its High Luminosity (HL) operation phase possible. The ten–fold increase in its instantaneous luminosity will lead to the collection of huge datasets, with the HL–LHC expected to deliver around \({\mathcal {L}}=3\hbox { ab}^{1}\) to ATLAS and CMS and around \({\mathcal {L}}=0.3\hbox { ab}^{1}\) to LHCb. This unprecedented dataset will open new exciting physics opportunities, such as the measurement of the Higgs boson couplings to second generation fermions as well as of its self–interactions. These opportunities will be summarised in a CERN Yellow Report [27] to be presented before the end of 2018, in order to contribute to the update of the European Strategy for Particle Physics (ESPP).^{1}
From the point of view of PDF determinations, the availability of these immense data samples will permit a significant extension of the kinematic coverage of PDF–sensitive measurements as well as a marked improvement in their statistical and systematic uncertainties. With this motivation, the goal of this work is to quantify the impact of the future HL–LHC measurements on the proton PDFs. In other words, we aim to assess the ultimate constraining power of hadron collider data on the PDFs. In turn, the resulting projections for the expected PDF uncertainties will feed into other related projections for HL–LHC processes, which will benefit from the associated reduction of theoretical errors.
It is important to emphasise here that while this type of study has previously been carried out in the context of lepton–hadron colliders such as the Large Hadron electron Collider (LHeC) and the Electron Ion Collider (EIC) [29, 30, 31, 32, 33, 34, 35, 36, 37], this is the first time that such a systematic effort has been devoted to determine the PDF–constraining potential of a future hadron collider. Clearly, being able to compare the information on PDFs that will be provided by the HL–LHC with that from proposed electron–proton colliders such as the LHeC represents an important input to inform the upcoming ESPP update.
Our analysis has been carried out as follows. First, we have generated HL–LHC pseudo–data for a number of PDF–sensitive processes: Drell–Yan production (both at high dilepton invariant mass and in the forward rapidity regions); W production in association with charm quarks (central and forward regions); inclusive jet and prompt photon production; the transverse momentum of Z bosons; and differential distributions in top quark pair production. We have selected those processes that should benefit more directly from the increased statistics available at the HL–LHC. We consider measurements such as inclusive W, Z production in the central region, which are already completely limited by systematic uncertainties [38, 39], with no significant improvement anticipated from increased statistics alone. For each process, the binning and kinematic cuts applied to the pseudo–data is constructed from a suitable extension of reference measurements at \(\sqrt{s}=8\) and 13 TeV. We consider different scenarios for the expected systematic uncertainties, from a conservative one with approximately the same systematics as the corresponding baseline measurements from Run I and a factor of 2 reduction for those from Run II, to an optimistic one with a reduction by a factor 2.5 (5) as compared to Run I (II).
By performing this analysis, we find that the legacy HL–LHC measurements can reduce the uncertainties in the PDF luminosities by a factor between 2 and 5 in comparison to state–of–the–art fits, depending on the specific flavour combination of the initial state and the invariant mass of the produced final state. We also show that our projections for the PDF error reduction, which are predominantly driven by the increased statistics of the HL–LHC data sample, depend only moderately on the specific scenario adopted for the reduction of the experimental systematic errors.
We then explore the implications of the profiled PDFs for representative LHC cross sections at \(\sqrt{s}=14\) TeV, both within the Standard Model (SM) and beyond it. Our analysis highlights how \({\mathcal {O}}\left( 1\%\right) \) PDF uncertainties are within the reach of the HL–LHC for key observables such as the transverse momentum distribution in Higgs production from gluon fusion. Therefore, our study illustrates the significant improvement in the precision of PDF determinations achievable from hadron collider data alone, and motivates the continuation of the ongoing successful program of PDF–sensitive measurements at the LHC.
The outline of the paper is the following. First, in Sect. 2 we describe the features of the PDF–sensitive processes used to generate the HL–LHC pseudo–data. Then in Sect. 3 we quantify the constraints on the PDFs of individual processes using the Hessian profiling method. The full set of HL–LHC pseudo–data is combined in Sect. 4 to construct the ultimate HL–LHC parton distributions, which is then used to assess their phenomenological implications for different processes both in the SM and beyond it. Finally, in Sect. 5 we summarise our results and indicate how they are made publicly available.
2 Pseudo–data generation
In this section we present the PDF–sensitive processes for which HL–LHC pseudo–data have been generated, provide details about the binning and kinematic cuts, and also describe the baseline Run I and II measurements that are used to model the experimental systematic uncertainties expected in the HL–LHC era.
2.1 PDF–sensitive processes
We start by describing the PDF–sensitive processes that will be considered in this study to generate HL–LHC pseudo–data. Our analysis is based on six different types of processes: the production of top quark pairs, jets, direct photons, and W bosons in association with charm quarks, the transverse momentum of Z bosons, and the neutral and charged current Drell–Yan processes. In Fig. 1 we show representative Feynman diagrams at the Born level for all of these processes, in order to illustrate their sensitivity to the different partonic initial states. For instance, we see that jets, photon, and top quark pair production are directly dependent on the gluon content of the proton, while W+charm is sensitive to strangeness, and the Drell–Yan process to the quark–antiquark luminosity.
This choice of input processes is driven by the fact that some types of hard–scattering reactions should benefit more directly from the increased statistics offered by the HL–LHC than others. Indeed, some of the existing LHC measurements, such as inclusive W, Z production in the central region [38, 39], are already limited by systematic uncertainties, and therefore are unlikely to improve significantly at higher luminosities. On the other hand, our selection of processes will greatly benefit from the huge HL–LHC dataset either because they are relatively rare, such as W+charm, or because their kinematic coverage can be extended to regions of large invariant masses and transverse momentum or forward rapidities where event rates exhibit a steep fall–off. While these pseudo–data sets do include some regions which are currently systematics dominated, i.e. towards central rapidity and lower mass/transverse momentum, as we will see the dominant PDF impact comes from the regions which are not, where the existing data are less constraining and the contributing PDFs are currently less well determined.

High–mass Drell–Yan, specifically the dilepton invariant mass differential distribution \(d\sigma (pp\rightarrow ll)/dm_{ll}\) for \(m_{ll}\gtrsim 110\) GeV for a central rapidity acceptance, \(\eta _{l}\le 2.4\). This process is particularly useful for quark flavour separation, specifically to constrain the poorly known large–x sea quarks. Here the ATLAS 8 TeV measurement of differential Drell–Yan cross sections [47] is taken as reference, with additional bins in the high \(m_{ll}\) region included to benefit from the enhanced kinematic coverage.

The differential distributions for on–peak W and Z boson production in the forward region, \(2.0 \le \eta _{l} \le 4.5\), covered by the LHCb experiment. These measurements constrain quark flavour separation, including the strange and charm content of the proton, in the large and small x region [48], complementary to the data from the central region. The reference analysis is the LHCb measurement of the rapidity distributions of W and Z bosons in the muon final state at 8 TeV [49]. In comparison to the reference measurement, a finer binning by a factor of 2 to 5 has been adopted as allowed by the increased event rates. Events are selected if \(p_T^l \ge 20\) GeV, the lepton rapidities fall in the LHCb acceptance, and, in the case of Z production, there is the additional requirement that \(60~\mathrm{GeV}\le m_{ll} \le 120~\mathrm{GeV}\).

Differential distributions in top quark pair production, providing direct information on the large x gluon [15]. Specifically, we consider the top quark transverse momentum \(p_T^t\) and rapidity \(y_t\), and the top quark pair rapidity \(y_{t\bar{t}}\) and invariant mass \(m_{t\bar{t}}\). The reference measurements here are the ATLAS 8 TeV differential distributions in the lepton+jets final state [50]. We assume that the statistical correlations between different distributions will be available, as is the case for the 8 TeV data [51], and therefore include the four distributions simultaneously in the fit. To account for the increased statistics of the HL–LHC, the number of bins in the rapidity distributions is doubled, while the \(p_T^t\) and \(m_{t\bar{t}}\) distributions are extended to higher values in the TeV region.

The transverse momentum distribution of the Z bosons in the dilepton final state, \(20 \,\mathrm{GeV}<p_T^{ll}<3.5\,\mathrm{TeV}\) region for central rapidities \(\eta _{Z}\le 2.4\) and different bins of the dilepton invariant mass \(m_{ll}\). This process is relevant to constrain the gluon and the antiquarks at intermediate values of x [17]. For the reference analysis, we take the ATLAS measurements of the transverse momentum of lepton pairs at 8 TeV [52]. The pseudo–data is generated for six different bins of the dilepton invariant mass \(m_{ll}\), with boundaries 12, 20, 30, 40, 66, 116, and 150 GeV respectively. In each of the invariant mass \(m_{ll}\) bins, additional bins are added to the \(p_T^{ll}\) distribution to exploit the improved coverage of the large transverse momentum region.

The production of W bosons in association with charm quarks. This process provides a sensitive handle on the strangeness content of the proton [23, 53], which is the least well known of the light quark PDFs. The pseudo–data for this process has been generated as a function of the lepton psuedorapidity \(\eta _l\) from the W boson decay, and is inclusive over the kinematics of the charm quark provided it satisfies the selection cuts. For this process, pseudo–data have been generated both for the central rapidity region relevant for ATLAS and CMS as well as for the forward region covered by LHCb. In the central rapidity region, \(\eta ^l\le 2.4\), the reference measurement is the CMS analysis at 13 TeV [24], where events are selected provided that \(p_T^c \ge 5 \) GeV and \(p_T^l \ge 26\) GeV with l indicating the result of the \(W\rightarrow l\nu \) decay. At forward rapidities, \(2\le \eta ^l \le 4.5\), we use a dedicated selection strategy with \(2.2 \le \eta ^c \le 4.2\), \(p_T^\mu \ge 20\) GeV, \(p_T^c \ge 20\) GeV, and \(p_T^{\mu +c}\ge 20\) GeV [54]. We take the acceptance to be 30% due to c–jet tagging and an overall normalisation error of 5%.

Prompt isolated photon production represents a complementary probe of the gluon PDF at intermediate values of x [19]. Here the pseudo–data has been generated as differential distributions in the photon transverse momentum \(p_T^\gamma \) for different bins in the photon rapidity \(\eta ^\gamma \). The reference measurements here is the ATLAS 13 TeV analysis of [55], where additional bins have been added to the \(p_T^\gamma \) distribution in each rapidity bins to benefit from the improved coverage of the large \(p_T^\gamma \) region.

Finally, we consider the inclusive production of hadronic jets in different bins of their rapidity up to \(y_\mathrm{jet}\le 3\) as a function of their \(p_T^{\mathrm{jet}}\). This process provides direct information on the gluon and the valence quarks at large–x. Here the jets are reconstructed using the anti–\(k_T\) algorithm with \(R=0.4\) as radius parameter. The reference measurement here is the 13 TeV ATLAS analysis of inclusive jet and dijet production based on a luminosity \({\mathcal {L}}=3.2\) \(\hbox {fb}^{1}\) from the 2015 data–taking period. The coverage of the high–\(p_T\) region has been extended to the \(p_T^{\mathrm{jet}}\simeq 23\) TeV in comparison to these.
In addition, one should take into account that progress from both the experimental and theoretical sides could lead to novel processes being added to the PDF fitting toolbox, for instance more exclusive processes or processes for which the standard DGLAP description breaks down. With these caveats, the set of processes adopted in this work is representative enough to provide a reasonable snapshot of the PDF–constraining potential of the HL–LHC.
It is also important to mention that the HL–LHC projections presented in this work are based on pseudo–data generated specifically for this study, and that they are not endorsed by the LHC collaborations. However, we have taken into account all the feedback and suggestions received from the ATLAS, CMS, and LHCb contacts involved in the Yellow Report studies.
2.2 Theory calculations and pseudo–data generation
Summary of the features of the HL–LHC pseudo–data generated for the present study For each process we indicate the kinematic coverage, the number of pseudo–data points used across all detectors \(N_\mathrm{dat}\), the values of the correction factors \(f_\mathrm{corr}\) and \(f_\mathrm{red}\); and finally the reference from the 8 TeV or 13 TeV measurement used as baseline to define the binning and the systematic uncertainties of the HL–LHC pseudo–data, as discussed in the text
Process  Kinematics  \(N_\mathrm{dat}\)  \(f_\mathrm{corr}\)  \(f_\mathrm{red}\)  Baseline 

\(Z\,p_T\)  \(20\,\mathrm{GeV}\le p_T^{ll} \le 3.5\) TeV  338  0.5  \(\left( 0.4, 1\right) \)  [52] (8 TeV) 
\(12\,\mathrm{GeV}\le m_{ll} \le 150\) GeV  
\(y_{ll}\le 2.4\)  
Highmass DrellYan  \(p_T^{l1(2)}\ge 40(30)\,\mathrm{GeV}\)  32  0.5  \(\left( 0.4, 1\right) \)  [47] (8 TeV) 
\(\eta ^l\le 2.5\), \(m_{ll}\ge 116\,\mathrm{GeV}\)  
Top quark pair  \(m_{t\bar{t}}\simeq 5\) TeV, \(y_t\le 2.5\)  110  0.5  \(\left( 0.4, 1\right) \)  [50] (8 TeV) 
W+charm (central)  \(p_T^\mu \ge 26\,\mathrm{GeV}\), \(p_T^c \ge 5\,\mathrm{GeV}\)  12  0.5  \(\left( 0.2, 0.5\right) \)  [24] (13 TeV) 
\(\eta ^\mu \le 2.4\)  
W+charm (forward)  \(p_T^\mu \ge 20\,\mathrm{GeV}\), \(p_T^c \ge 20\,\mathrm{GeV}\)  10  0.5  \(\left( 0.4, 1\right) \)  LHCb projection 
\(p_T^{\mu +c} \ge 20\,\mathrm{GeV}\)  
\(2\le \eta ^\mu \le 4.5\), \(2.2\le \eta ^c \le 4.2\)  
Direct photon  \(E_T^\gamma \lesssim 3\) TeV, \(\eta _{\gamma }\le 2.5\)  118  0.5  \(\left( 0.2, 0.5\right) \)  [55] (13 TeV) 
Forward W, Z  \(p_T^{l}\ge 20\,\mathrm{GeV}\), \(2.0\le \eta ^l\le 4.5\)  90  0.5  \(\left( 0.4, 1\right) \)  [49] (8 TeV) 
\(60\,\mathrm{GeV}\le m_{ll}\le 120\,\mathrm{GeV}\)  
Inclusive jets  \(y \le 3\), \(R = 0.4\)  58  0.5  \(\left( 0.2, 0.5\right) \)  [61] (13 TeV) 
Total  768 
Theoretical predictions are computed at next–to–leading order (NLO) in the QCD expansion using MCFM [58] interfaced to APPLgrid [59] to produce the corresponding fast grids. The only exception is inclusive jet production, for which the NLO calculation is obtained from the NLOJET++ program [60]. The central value of the pseudo–data initially coincides with the corresponding prediction obtained using this NLO calculation with the PDF4LHC15 NNLO set as input. Subsequently, this central value is fluctuated according to the corresponding experimental uncertainties. This implies that, by construction, one should find \(\chi ^2/N_\mathrm{dat}\simeq 1\) from the fit to the pseudo–data.
In Eq. (2.2), \(\delta ^{\mathrm{exp}}_{\mathrm{sys},i}\) indicates the total systematic error of bin i taken from the reference LHC measurement at either 8 TeV or 13 TeV, while \(f_\mathrm{red}\le 1\) is a correction factor that accounts for the fact that on average systematic uncertainties will decrease at the HL–LHC in comparison to Run II due to both detector improvements and the enlarged dataset for calibration. Finally, \(f_\mathrm{corr}\) represents an effective correction factor that accounts for the fact that data with correlated systematics may be more constraining than the same data where each source of error is simply added in quadrature, as we do in this analysis. We discuss below in Sect. 2.3 how the value of \(f_\mathrm{corr}\) can be determined by means of available LHC measurements for which the full information on correlated systematics is available.
Concerning the theoretical calculations adopted here, since the present study relies on pseudo–data, it is not necessary to account for higher–order QCD effects or electroweak corrections. Indeed, by far the dominant contribution to the PDF sensitivity of hadron collider processes is contained within the NLO calculation. As in the case of PDF closure tests [62], here we are only interested in the relative reduction of the PDF uncertainties once the HL–LHC data is included in the fit, while the central value itself will be essentially unaffected. Note that this also holds for the contribution of photon–initiated (PI) processes, since the photon PDF is very well know [63, 64, 65]. Therefore, PI processes effectively induce an overall rescaling of the cross section which becomes irrelevant when generating pseudo–data.
In Table 1 we present the summary of the main features of the HL–LHC pseudo–data generated for the present study. For each process, we indicate the kinematic coverage, the number of pseudo–data points used \(N_\mathrm{dat}\), the values of the correction factors \(f_\mathrm{acc}\), \(f_\mathrm{corr}\), and \(f_\mathrm{red}\); and finally the reference for the 8 TeV or 13 TeV measurement used as baseline to define the binning and the systematic uncertainties of the HL–LHC pseudo–data. A total of around \(N_\mathrm{dat}= 768\) pseudo–data points are then used in the PDF profiling. The values of the reduction factor for the systematic errors \(f_\mathrm{red}\) is varied between 1 (0.5) and 0.4 (0.2) in the conservative and optimistic scenarios for a 8 TeV (13 TeV) baseline measurement. This different treatment is motivated by the fact that available 13 TeV measurements are based on a smaller dataset and therefore tend to have larger systematic errors in comparison to the 8 TeV case. Thus we can expect some improvement here at the HL–LHC even in the most conservative scenario; Run II measurements based on the complete integrated luminosity will certainly benefit from reduced systematics.
2.3 Impact of correlating uncertainties
As we will also discuss in Sect. 3, when constructing the \(\chi ^2\) estimator for the HL–LHC pseudo–data we will not explicitly include the correlations between the systematic errors. Instead, we add statistical and systematic uncertainties in quadrature as indicated in Eq. (2.2). This choice is motivated by the fact that it is already challenging to estimate how specific systematic uncertainties will be reduced at the HL–LHC, let alone how their mutual correlations will be modified. Note that even restricting ourselves to Run I measurements, the determination of the experimental correlation model is a delicate problem, and can in some cases complicate the PDF interpretation of measurements such as inclusive jet production [66].
In Fig. 3 we compare the baseline PDF4LHC15 set and the sets profiled with these two LHC datasets, with or without the correlations between the experimental systematic uncertainties accounted for. In the latter case, the \(f_\mathrm{corr}\) factor is chosen to reproduce the results of the profiling when the correlations are included. We can see that for the two considered datasets, rather different values of \(f_\mathrm{corr}\) are preferred; for the top data, we require \(f_\mathrm{corr}\sim 0.25\) while for the \(W+\)charm data we require instead \(f_\mathrm{corr}\sim 1\). Clearly the precise value of this correction therefore appears to depend quite sensitively on the considered datasets, in terms of the corresponding breakdown of systematic uncertainties and overall PDF impact.
The results of Fig. 3 might suggest that, for projections which are dominantly driven by the potential improvement in systematic uncertainties, our approach could be questionable and require a more complete treatment of experimental correlations. However, here we have explicitly chosen our input dataset to be composed of those processes for which the PDF impact will be driven instead by the improvement in the statistics and extension to unconstrained kinematic regions. Indeed, we will see later on that the specific value of this parameter does not have a large impact on the final results, and we will simply take \(f_\mathrm{corr}=0.5\) in what follows as an average, somewhat weighted towards the value required by the top quark differential data, as this shows a larger PDF impact and would therefore be more important to account for accurately.
3 HL–LHC constraints from individual processes
In this section, we study the constraints on the PDFs that are expected from individual HL–LHC measurements listed in Table 1. First of all, we review the formulation of the Hessian profiling used in this work to quantify the PDF constraints. Then we present the results for the various HL–LHC processes and study how the description of the pseudo–data is affected. The complete set of processes is combined together into a single profiled PDF set in the next section.
3.1 The Hessian profiling method
The minimisation of Eq. (3.1) produces approximately equivalent results to carrying out the corresponding Hessian fit from scratch, provided settings such as the input PDF parameterisations, the tolerance factor T, and the theoretical calculations are unchanged. An advantage of the Hessian profiling method in comparison to related techniques such as the Bayesian reweighting method [69, 70], relevant for Monte Carlo PDF sets, is that there is no information loss even when the added measurements provide significant new information. This property is crucial in the present analysis, since the HL–LHC pseudo–data induces significant constraints on the PDFs.
At this point, the minimisation of Eq. (3.2) with respect to the Hessian PDF nuisance parameters \(\beta _{k,\mathrm th}\) can be interpreted as leading to PDFs that have been optimized to describe this new specific measurement. The resulting Hessian matrix in the \(\beta _{k,\mathrm th}\) parameter space at the minimum can be diagonalized to construct the new eigenvector directions, and PDF uncertainties are determined from the \(\Delta \chi ^2=T^2\) criteria. In the studies presented here, we use \(T=3\), which roughly corresponds to the average tolerance determined dynamically in the CT14 and MMHT14 analyses. The resulting profiled PDF set^{2} can be straightforwardly used for phenomenology using the uncertainty prescription of symmetric Hessian sets, and the default output format is compliant with the LHAPDF interface.
3.2 Inclusive gauge boson production
In this section, we will use the same structure to discuss the impact on the PDFs of the individual HL–LHC processes that are being considered. First, we will display representative examples of the correlations between the PDFs and the pseudo–data, to illustrate the sensitivity of the latter. Second, we will show how the description of the HL–LHC pseudo–data is modified once it is included in the PDF4LHC15 set by means of profiling. Finally, we will assess its impact on the PDFs in a specific scenario for the projections of the experimental systematic errors. In particular, we adopt the ‘optimistic’ choice of Table 2, i.e. \(F\equiv f_\mathrm{corr}\cdot f_\mathrm{red}=0.2\), which corresponds to a value \(f_\mathrm{red}=0.4\) for the reduction of the systematic uncertainties compared to the 8 TeV baseline measurements. As discussed above, for 13 TeV baselines, in this scenario we take a lower value of \(f_\mathrm{red}=0.2\), to account for the smaller 13 TeV datasets these are based on.
We start by discussing the correlations. In Fig. 4 we show the correlation coefficients \(\rho \) between the PDFs and the HL–LHC pseudo–data on the Drell–Yan process. The left (right) plot displays the correlation between the anti–up (anti–down) quark as a function of x for \(Q=100\) GeV for the high–mass (forward) Drell–Yan pseudo–data. A value of \(\rho \) close to 1 (\(1\)) in a given region of x indicates that this process is strongly (anti–) correlated with the input PDFs in this same region, and thus that could potentionally be used to reduce PDF uncertainties there.
As we can see from Fig. 4, in the case of high–mass Drell–Yan we have \(\rho \ge 0.9\) for \(0.05 \lesssim x \lesssim 0.5\), indicating that this process can provide information on the large–x antiquarks. In the case of the forward W, Z production measurements the correlation coefficient for the \(\bar{d}\) PDF peaks at \(x\simeq 10^{4}\), highlighting that the forward kinematic coverage of LHCb allows the quark flavour separation to be pinned down to small values of x.
From these comparisons, we see that the impact of the high–mass Drell–Yan pseudo–data on the PDFs is rather moderate, presumably because even at the HL–LHC the expected precision of the measurements is comparable or larger than current PDF uncertainties, in particular in the high \(m_{ll}\) range. On the other hand, for the W, Z measurements that will be carried out by LHCb we can observe a marked error reduction of up to a factor two, highlighting the usefulness of the forward kinematic coverage. Note that in both cases the central values of the theoretical predictions are relatively unaffected, with the dominant impact being on the uncertainties. This is expected, as by construction we assume the datasets are consistent with the underlying theory and PDFs.
Concerning the corresponding impact of the HL–LHC pseudo–data on the PDFs, in Fig. 6 we show the reduction of the PDF uncertainties found upon the inclusion of the high–mass Drell–Yan (left) and the forward W, Z (right) pseudo–data on the PDF4LHC15 set. We display the same PDF flavours as those used in the calculation of the correlation coefficients in Fig. 4, namely the up and down antiquarks respectively. What we find is consistent with Fig. 5: a rather moderate effects on the up antiquark from the high–mass Drell–Yan process, while a more marked effect on the down antiquark from the forward W, Z process specially in the small–x region.
3.3 Top quark pair production
Next, in Fig. 8 (left) we show the same comparison as in Fig. 5 now for the \(m_{t\bar{t}}\) distribution. We can observe a very marked PDF uncertainty reduction at large values of the invariant mass. As expected, we find in Fig. 8 (right) that the addition of the HL–LHC \(t\bar{t}\) pseudo–data leads to a significant reduction in the PDF uncertainties in the gluon PDF at large–x, highlighting the good constraining power of this type of measurements.
3.4 Jet and photon production
3.5 W production in association with charm quarks
The comparison between the HL–LHC pseudo–data and the corresponding theoretical predictions for W+charm production both in the central and forward regions are collected in Fig. 13. In the central region, we see a clear reduction of the PDF uncertainties after including the pseudo–data into the fit, by around a factor two. This reduction of uncertainty is approximately constant as a function of the lepton rapidity. At forward rapidities instead, we find that before adding the pseudo–data the PDF uncertainties grow very fast with rapidity, reaching up to 30% for \(\eta _l \simeq 4.5\), while after including it they are markedly reduced and become more or less constant with rapidity as in the central region. Taking into account the correlation coefficients shown in Fig. 12, these results indicates that W+charm production in the forward region provides valuable constraints on the large–x strangeness, which is currently affected by large uncertainties.
3.6 The transverse momentum of Z bosons
The three scenarios for the systematic uncertainties of the HL–LHC pseudo–data that we assume in the present study. These scenarios, ranging from conservative to optimistic, differ among them in the reduction factor \(f_\mathrm{red}\), Eq. (2.2), applied to the systematic errors of the reference 8 TeV or 13 TeV measurements. We also indicate in each case the name of the corresponding LHAPDF grid
Scenario  \(f_\mathrm{red}\) (8 TeV)  \(f_\mathrm{red}\) (13 TeV)  LHAPDF set  Comments 

A  1  0.5  PDF4LHC_nnlo_hllhc_scen1  Conservative 
B  0.7  0.36  PDF4LHC_nnlo_hllhc_scen2  Intermediate 
C  0.4  0.2  PDF4LHC_nnlo_hllhc_scen3  Optimistic 
The comparison between HL–LHC pseudo–data and theoretical predictions in the on–peak bin defined by \(66~\mathrm{GeV}\le m_{ll} \le 116~\mathrm{GeV}\) is shown in Fig. 16, where we can see that coverage up to \(p_T^{ll}\simeq 3\) TeV is expected, similar as in the case of direct photon production. We find a moderate reduction in the PDF uncertainties once the HL–LHC pseudo–data is added to the fit by means of Hessian profiling. Concerning its effects on the gluon, we see that the Z \(p_T\) measurements provide valuable information in the intermediate x region between \(10^{3}\) and \(10^{2}\) with a clear reduction of PDF uncertainties even if in this region these were quite small to begin with.
4 Ultimate PDFs with HL–LHC pseudo–data
In this section we combine the complete set of HL–LHC pseudo–data listed in Table 1 to produce the final profiled PDF sets, which quantify the impact of future HL–LHC measurements on our knowledge of the quark and gluon structure of the proton.
In Table 2 we list the three scenarios for the systematic uncertainties of the HL–LHC pseudo–data that we assume in the present analysis. These scenarios, ranging from more conservative to more optimistic, differ among them in the reduction factor \(f_\mathrm{red}\), Eq. (2.2), applied to the systematic errors of the reference 8 TeV or 13 TeV measurements. In particular, in the optimistic scenario we assume a reduction of the systematic errors by a factor 2.5 (5) as compared to the reference 8 TeV (13 TeV) measurements, while for the conservative scenario we assume no reduction in systematic errors with respect to 8 TeV reference. We also indicate in each case the name of the corresponding LHAPDF grid. Reassuringly, as we show below, the qualitative results of our study depend only mildly in the specific assumption for the values of \(f_\mathrm{red}\).
In this section, we study how the HL–LHC pseudo–data constraints the parton distributions and the PDF luminosities for proton–proton collisions at \(\sqrt{s}=14\) TeV. Then we present an initial study with some representative implications of the ultimate PDFs for LHC phenomenology.
4.1 Parton distributions
In Fig. 17 we present a comparison of the baseline PDF4LHC15 set with the profiled sets based on HL–LHC pseudo–data from scenarios A (conservative) and C (optimistic) as defined in Table 2. Specifically, we show the gluon, down quark, up anti–quark, and total strangeness at \(Q=10\) GeV, normalized to the central value of the PDF4LHC15 baseline. In this comparison, the bands correspond to the one–sigma PDF uncertainties.
First of all, we observe that the impact of the HL–LHC pseudo–data is reasonably similar in the conservative and optimistic scenarios. This is not so surprising, as we have explicitly chosen those datasets which will benefit from a significant improvement in statistics, and these tend to lie in kinematic regions where the PDFs themselves are generally less well determined, see the discussion in Sect. 2. Therefore, the dominant reason for the observed reduction of PDF uncertainties is the increased statistics and the corresponding extended kinematic reach that becomes available at the HL–LHC, rather than the specific assumptions about the systematic uncertainties. This demonstrates that our results are robust against the details of the projections of how the experimental systematic uncertainties will be reduced in the HL–LHC era.
From Fig. 17 we observe a marked reduction of the PDF uncertainties in all cases. This is particularly significant for the gluon and the sea quarks, for the reason that these are currently affected by larger uncertainties than in the case of the valence quarks. In the case of the gluon PDF, there is an improvement of uncertainties across a very broad range of x. This is a direct consequence of the fact that we have included several HL–LHC processes that have direct sensitivity to the gluon content of the proton, namely jet, direct photon, and top quark pair production, as well as the transverse momentum of Z bosons.
Another striking feature of Fig. 17 concerns the strange PDF. In this case, the PDF uncertainties are reduced by almost a factor 4, from around 15% to a few percent, in a wide region of x. This result highlights the importance of the W+charm measurements at the HL–LHC, specially those in the forward region by LHCb, see Fig. 12, which represent a unique handle on the poorly known strange content of the proton. In turn, such an improved understanding of the strange PDF will feed into a reduction of theory uncertainties in crucial HL–LHC measurements such as those of \(M_W\) or \(\sin ^2\theta _W\).
4.2 Partonic luminosities
Next we take a look at the partonic luminosities, to quantify the improvement in the PDF uncertainties in different initial–state partonic combinations from the HL–LHC pseudo–data. In Fig. 18 we show the reduction of PDF uncertainties in the gg, qg, \(q\bar{q}\), and qq, \(s\bar{s}\), and \(s\bar{u}\) luminosities at \(\sqrt{s}=14\) TeV that can be expected as a consequence of adding the HL–LHC pseudo–data on top of the PDF4LHC15 baseline. Note that a value of 1 in these plots corresponds to no uncertainty reduction. As in the case of the PDF comparisons, results are shown both for the conservative (A) and optimistic (C) scenarios for our projections of the experimental systematic uncertainties.
The reduction of the PDF uncertainties compared to the PDF4LHC15 baseline for different initial partonic combinations (that is, a value of 1 corresponds to no reduction at all). Results are presented for three different bins of the invariant mass \(M_X\) of the produced system, averaging over 10 points logarithmically spaced within each bin. The values shown outside (inside) the brackets correspond to the optimistic (conservative) scenario. The corresponding results differential in \(M_X\) are presented in Fig. 18
Ratio to baseline  \(10~\mathrm{GeV}\le M_X\le 40~\mathrm{GeV}\)  \(40~\mathrm{GeV}\le M_X\le 1~\mathrm{TeV}\)  \(1~\mathrm{TeV}\le M_X\le 6~\mathrm{TeV}\) 

gluon–gluon  0.50 (0.60)  0.28 (0.40)  0.22 (0.34) 
gluon–quark  0.66 (0.72)  0.42 (0.45)  0.28 (0.37) 
quark–quark  0.74 (0.79)  0.37 (0.46)  0.43 (0.59) 
quark–antiquark  0.71 (0.76)  0.31 (0.40)  0.50 (0.60) 
strange–antistrange  0.34 (0.44)  0.19 (0.30)  0.23 (0.27) 
strange–antiup  0.67 (0.73)  0.27 (0.38)  0.38 (0.43) 
From the comparisons in Fig. 18 and Table 3, we observe again that the reduction in the uncertainties of the PDF luminosities is rather robust with respect to the assumed projections for the experimental systematic uncertainties. For instance, for intermediate values of the final–state invariant mass, \(40~\mathrm{GeV}\le M_X\le 1~\mathrm{TeV}\), we find that for all the partonic initial states the reduction factor varies between 0.28 and 0.40 (0.42 and 0.45, 0.31 and 0.40) in the optimistic and conservative scenario for the gluon–gluon (gluon–quark, quark–antiquark) luminosities. These results again reinforce our conclusion that the results of this study are only mildly sensitive to the details of the projected pseudo–data.
We find that in the intermediate \(M_X\) bin the reduction of PDF uncertainties ranges approximately between a factor 2 and a factor 5, depending on the specific partonic channel and the scenario for the systematic errors. For example, for the gluon–gluon luminosity in the range relevant for Higgs production in gluon fusion, one finds a reduction by almost a factor 4 in the optimistic scenario. The improvement in the strange–initiated processes is also remarkable, for example the PDF uncertainties in the \(s\bar{s}\) luminosity are expected to be reduced by a factor 5 (3) in the optimistic (conservative) scenario. Recall that strange–initiated processes are important for a variety of LHC analysis, from measurements of \(M_W\) and \(\sin ^2\theta _W\) to searches for BSM \(W'\) bosons. We also find that the uncertainties in quark–antiquark luminosities, relevant for example for precision electroweak measurements, are expected to be reduced by up to a factor 3 in this invariant mass range.
Similar improvements in the PDF luminosities are found in the high mass region, \(M_X\ge 1\) TeV, directly relevant for BSM searches. For instance, in the optimistic scenario, the PDF error reduction at higher masses is expected to be as large as a factor 5 for the gluon–gluon luminosity. Again this is a consequence of the inclusion in the profiling of gluon–dominated processes such as \(t\bar{t}\) and inclusive jets that at the HL–LHC, which cover the region up to 6 TeV, see Fig. 2. The impact of the HL–LHC pseudo–data is less marked for the quark–quark and quark–antiquark luminosities in this high–mass region, due to the fact that of the data points included in the profiling only a fraction of them are both quark–initiated and cover the large–x region.
It is worth emphasizing again here that the list of processes studied in this work and summarised in Table 3 are just a subset of those HL–LHC measurements with PDF–constraining potential. Therefore, it is conceivable that the actual reduction of PDF errors presented in Table 3 would actually be more significant than our estimates here.
4.3 Implications for HL–LHC phenomenology
We now turn to present some representative results of the phenomenological implications that these “ultimate” PDFs will have at the HL–LHC, both for processes within the SM and beyond it. It is beyond the scope of this work to carry out a comprehensive phenomenological study, and we refer the reader to the upcoming Yellow Report [27] describing the physics case of the HL–LHC, where more detailed projections and analyses will be presented.
Let us begin by assessing the PDF impact of HL–LHC measurements on representative Standard Model processes. In particular, we consider diphoton production, dijet production, and Higgs production in gluon fusion, both inclusive and in association with a hard jet. In the following all cross sections have been computed at \(\sqrt{s}=14\) TeV using leading order (LO) matrix elements with MCFMv8.2 [58] and applying the standard ATLAS/CMS central acceptance cuts. Since the comparison is restricted to ratios of cross sections, the LO calculation is sufficient to illustrate the impact of the improvement in the PDF uncertainties in each of these processes. Indeed, we are only interested here in illustrating the relative impact of the PDF error reduction, rather than providing state–of–the–art predictions for the rates, which will be presented elsewhere in the Yellow Report [27].
First of all, we show the production cross sections of pairs of photons (left) and of jets (right) in the upper panels of Fig. 19. We compare the PDF4LHC15 baseline with the HL–LHC profiled PDF sets in the conservative (A) and optimistic (C) scenarios of Table 2, normalised to the central value of PDF4LHC15. In the considered kinematic regions, these two processes are mostly sensitive to the quark–antiquark initial state, and to the quark–gluon and quark–(anti)quark initial states, respectively. The cross sections are presented as a function of the minimum invariant mass of the final state, \(M_{\gamma \gamma }^{\mathrm{min}}\) and \(M_{jj}^{\mathrm{min}}\) respectively, in order to facilitate their comparison with the corresponding PDF luminosities shown in Fig. 18.
In the two lower plots of Fig. 19, we present the corresponding comparisons for the case of Higgs boson production via gluon fusion, using heavy top quark effective theory. In the case of inclusive production with decay into bottom quarks (left plot), we find that the constraints from HL–LHC measurements are expected to reduce PDF uncertainties down to the \(1\%\) level. Needless to say, this will directly benefit the characterisation of the Higgs sector at the HL–LHC, where a few percent is the typical uncertainty target for the determination of its couplings. In the case of Higgs boson production in association with a hard jet (right plot), also there we find a marked error reduction, indicating that PDF uncertainties in the Higgs transverse momentum distribution could be reduced down to the \(\simeq \)2% level in the entire kinematical range relevant at the HL–LHC. We recall that the large Higgs transverse momentum region is sensitive to new heavy particles running in the loops as well as to BSM effects such as partial Higgs compositeness [75].
As we have discussed above in Sects. 4.1 and 4.2, the impact of the HL–LHC pseudo–data is also significant in the large–x region, which in turn corresponds to large invariant masses for the PDF luminosities. This is of course an important region for the searches of BSM heavy particles, where PDF uncertainties often represent the dominant source of theoretical uncertainty. With this motivation, to illustrate the benefits that HL–LHC measurements will provide for BSM searches we consider here high–mass supersymmetric (SUSY) particle production at \(\sqrt{s}=14\) TeV, where the HL–LHC reach extends to sparticles masses up to around \(M\simeq 3\) TeV. While we use SUSY production as a benchmark process, our results also apply to the production of other heavy particles predicted in different BSM scenarios.
In Fig. 20 we show the comparison between the PDF4LHC15 predictions with the corresponding results from the profiled PDF sets with HL–LHC pseudo–data, normalised to the central value of the PDF4LHC15 baseline. As in Fig. 19, we provide results for scenarios A and C, the conservative and optimistic ones respectively. Specifically, we show the cross sections for gluino–gluino and squark–gluino production at \(\sqrt{s}=14\) TeV – similar conclusions are derived from squark–squark and squark–antisquark production. The theoretical calculations have been obtained using leading order (LO) matrix elements with Pythia8.235 [76] and assuming the SLHA2 benchmark point [77], for a range of sparticle masses within the HL–LHC reach. For simplicity, underlying event and multiple interactions have been switched off in the calculation. Again, we are not interested here in providing state–of–the–art predictions for the event rates, which can be found elsewhere [78].
From the comparisons in Fig. 20, we can see that the constraints on the PDFs expected from the HL–LHC data permit a significant reduction of the uncertainties in the high–mass SUSY cross sections. The size of this reduction is consistent with the corresponding results at the level of luminosities, reported in Fig. 18 and Table 3, recalling that gluino–gluino and gluino–squark production are driven by the gluon–gluon and gluon–quark initial states respectively [5]. For instance, for gluino pair–production with \(M_{\widetilde{g}}=3\) TeV, the PDF uncertainties are reduced from \(\simeq 60\%\) to \(\simeq 20\%\) in the optimistic scenario. A somewhat milder reduction is found for the squark–gluino cross sections. For squark–squark and squark–antisquark production, driven by the quark–quark and quark–antiquark initial states respectively, a PDF uncertainty reduction by around a factor two at high masses is found, consistently with Table 3.
To summarise, the initial phenomenological study presented in this section nicely illustrates the internal coherence of the HL–LHC physics program: high precision SM measurements will lead to a much improved understanding of the quark and gluon structure of protons, which in turn will benefit many other important analyses, from the characterisation of the Higgs sector to the searches of new heavy particles.
5 Summary
In this study, we have quantified the expected constraints that precision HL–LHC measurements will impose on the quark and gluon structure of the proton. To achieve this goal, we have assessed the impact of a range of relevant PDF–sensitive processes, from weak gauge boson and jet production to top quark and photon production. Moreover, we have studied the robustness of our results with respect to different projections for the experimental systematic uncertainties, from a more conservative one, where systematics are assumed to have the same size as in current measurements, to a more optimistic one, where they are markedly reduced.
Our main finding is that HL–LHC data has the potential to significantly reduce the PDF uncertainties in a wide kinematic range and for all relevant partonic final states. This is true both for the region of intermediate invariant masses, relevant for precision Higgs, electroweak, and top quark measurements, as well as in the TeV region relevant for searches of new heavy particles. Even in the most conservative scenario, in the region \(M_X\gtrsim 40\) GeV we find that HL–LHC measurements can reduce PDF uncertainties by at least a factor between 2 and 3 as compared to the current PDF4LHC15 baseline. The PDF constraining information from the HL–LHC is expected to be specially significant for gluon– and for strange–initiated processes. We also find that the quark–antiquark luminosity at the electroweak scale, a central input for legacy LHC measurements such as \(M_W\) and \(\sin ^2\theta _W\), could be improved by more than a factor 3 in the optimistic scenario.
This improved knowledge of the quark and gluon structure of the proton which will become possible at the HL–LHC will directly benefit a number of phenomenologically important process, due to the reduction of the associated theoretical errors. For instance, the PDF uncertainties in Higgs production in gluon fusion can be reduced down to \(\lesssim 2\%\) for the entire range of Higgs transverse momenta accessible at the HL–LHC. Likewise, PDF uncertainties in high–mass supersymmetric particle production can be decreased by up to a factor 3, with a similar impact expected for other BSM scenarios. This improvement should strengthen the bounds derived in the case of null searches, or facilitate their characterisation in the case of an eventual discovery. Similar improvements are found for Standard Model process, for example dijet production, which provides a unique opportunity to measurement the running of the strong coupling constant at the TeV scale. More detailed studies of the phenomenological implications of our study will be presented in the upcoming HL–LHC Yellow Report. Two caveats are relevant at this point. First, it should be emphasised again that in this study we have only considered a subset of all possible measurements of relevance for PDF fits. There are certainly processes for which data is and will be available, such as multijet production and single top production, that we have not considered here. Moreover, we can also reasonably expect that various new processes may be added to the PDF toolbox on the rather long timescales we consider here. Thus, we may certainly expect further constraints to become available for PDF studies by the end of HL–LHC running.
Second, in this study we have ignored any possible issues such as data incompatibilities, limitations of the theoretical calculations, or issues affecting the data correlation models. These are common in PDF fits, and indeed have already been found when comparing theory calculations against existing LHC data from Runs I and II. Such potential problems may eventually limit the PDF constraining power, in comparison to the estimates presented in this work, when the actual global fit with real HL–LHC data is performed. Clearly, such questions can only be tackled once the HL–LHC measurements are carried out, and indeed doing so will present an important programme of experimental and theoretical PDF–related work on its own. We cannot anticipate such work in our present study, which instead represents our best quantitative projections using our current knowledge. The results of this study are made publicly available in the LHAPDF6 format [46], with the grid names listed in Table 2 for the three scenarios that have been considered. These three grid files can be downloaded from:
https://data.nnpdf.science/HLLHC_YR/PDF4LHC15_nnlo_hllhc_scen1.tgz
https://data.nnpdf.science/HLLHC_YR/PDF4LHC15_nnlo_hllhc_scen2.tgz
https://data.nnpdf.science/HLLHC_YR/PDF4LHC15_nnlo_hllhc_scen3.tgz
The “ultimate” PDFs produced in this exercise can then be straightforwardly applied to other physics projections of HL–LHC processes, taking into account our improved knowledge of the partonic structure of the proton which is expected by then. We believe that the results of this work represent an important ingredient towards sharpening as much as possible the physics reach of the LHC in its upcoming high–luminosity era.
Footnotes
Notes
Acknowledgements
We are grateful to W. Barter, M. Campanelli, C. Gwenlan, S. Farry, and K. Lipka for discussion about the projections of future HL–LHC measurements at ATLAS, CMS, and LHCb. We thank P. Starovoitov for providing the APPLgrids for the inclusive jet measurements at the HL–LHC. R. A. K. and J. R. are supported by the European Research Council (ERC) Starting Grant “PDF4BSM” and by the Dutch Organization for Scientific Research (NWO). The work of J. G. is sponsored by Shanghai Pujiang Program and by the National Natural Science Fundation of China under the Grant No. 11875189. S. B. is supported by the Science and Technology Facilities Council (STFC). L. H. L thanks the Science and Technology Facilities Council (STFC) for support via grant award ST/L000377/1.
References
 1.J. Gao, L. HarlandLang, J. Rojo, The structure of the proton in the LHC precision era. Phys. Rept. 742, 1–121 (2018). arXiv:1709.04922 ADSMathSciNetCrossRefGoogle Scholar
 2.J. Rojo et al., The PDF4LHC report on PDFs and LHC data: Results from Run I and preparation for Run II. J. Phys. G 42, 103103 (2015). arXiv:1507.00556 ADSCrossRefGoogle Scholar
 3.S. Forte, G. Watt, Progress in the determination of the partonic structure of the proton. Ann. Rev. Nucl. Part. Sci. 63, 291 (2013). arXiv:1301.6754 ADSCrossRefGoogle Scholar
 4.LHC Higgs Cross Section Working Group Collaboration, D. de Florian et al., Handbook of LHC Higgs Cross Sections: 4. Deciphering the Nature of the Higgs Sector. arXiv:1610.07922
 5.W. Beenakker, C. Borschensky, M. Kramer, A. Kulesza, E. Laenen, S. Marzani, J. Rojo, NLO+NLL, squark and gluino production crosssections with thresholdimproved parton distributions. Eur. Phys. J. C 76(2), 53 (2016). arXiv:1510.00375 ADSCrossRefGoogle Scholar
 6.S. Alioli, M. Farina, D. Pappadopulo, J.T. Ruderman, Precision probes of QCD at high energies. JHEP 07, 097 (2017). arXiv:1706.03068 ADSCrossRefGoogle Scholar
 7.ATLAS Collaboration, M., Aaboud et al., Measurement of the \(W\)boson mass in pp collisions at \(\sqrt{s}=7\) TeV with the ATLAS detector. Eur. Phys. J. C 78(2), 110 (2018). arXiv:1701.07240
 8.CDF, D0 Collaboration, T. A. Aaltonen et al., Tevatron Run II combination of the effective leptonic electroweak mixing angle. Phys. Rev. D 97(11), 112007 (2018). arXiv:1801.06283
 9.NNPDF Collaboration, R.D. Ball, S. Carrazza, L. Del Debbio, S. Forte, Z. Kassabov, J. Rojo, E. Slade, M. Ubiali, Precision determination of the strong coupling constant within a global PDF analysis. Eur. Phys. J. C 78(5), 408 (2018). arXiv:1802.03398
 10.D. Becciolini, M. Gillioz, M. Nardecchia, F. Sannino, M. Spannowsky, Constraining new colored matter from the ratio of 3 to 2 jets cross sections at the LHC. Phys. Rev. D 91 (1), 015010 (2015). arXiv:1403.7411 [Addendum: Phys. Rev.D92,no.7,079905(2015)
 11.NNPDF Collaboration, R.D. Ball et al., Parton distributions from highprecision collider data. Eur. Phys. J. C 77(10), 663 (2017). arXiv:1706.00428
 12.S. Dulat, T.J. Hou, J. Gao, M. Guzzi, J. Huston, P. Nadolsky, J. Pumplin, C. Schmidt, D. Stump, C .P. Yuan, New parton distribution functions from a global analysis of quantum chromodynamics. Phys. Rev. D 93(3), 033006 (2016). arXiv:1506.07443 ADSCrossRefGoogle Scholar
 13.L.A. HarlandLang, A.D. Martin, P. Motylinski, R.S. Thorne, Parton distributions in the LHC era: MMHT 2014 PDFs. Eur. Phys. J. C 75, 204 (2015). arXiv:1412.3989 ADSCrossRefGoogle Scholar
 14.S. Alekhin, J. Blümlein, S. Moch, R. Placakyte, Parton distribution functions, \(alpha _s\), and heavyquark masses for LHC Run II., Phys. Rev. D 96(1), 014011 (2017). arXiv:1701.05838
 15.M. Czakon, N.P. Hartland, A. Mitov, E.R. Nocera, J. Rojo, Pinning down the largex gluon with NNLO topquark pair differential distributions. JHEP 04, 044 (2017). arXiv:1611.08609 ADSCrossRefGoogle Scholar
 16.M. Guzzi, K. Lipka, S.O. Moch, Topquark pair production at hadron colliders: differential cross section and phenomenological applications with DiffTop. JHEP 01, 082 (2015). [arXiv:1406.0386]ADSCrossRefGoogle Scholar
 17.R. Boughezal, A. Guffanti, F. Petriello, M. Ubiali, The impact of the LHC Zboson transverse momentum data on PDF determinations. JHEP 07, 130 (2017). arXiv:1705.00343 ADSCrossRefGoogle Scholar
 18.D. d’Enterria, J. Rojo, Quantitative constraints on the gluon distribution function in the proton from collider isolatedphoton data. Nucl. Phys. B 860, 311–338 (2012). arXiv:1202.1762 ADSCrossRefGoogle Scholar
 19.J.M. Campbell, J. Rojo, E. Slade, C. Williams, Direct photon production and PDF fits reloaded. Eur. Phys. J. C 78(6), 470 (2018). arXiv:1802.03021 ADSCrossRefGoogle Scholar
 20.PROSA Collaboration, O. Zenaiev et al., Impact of heavyflavour production cross sections measured by the LHCb experiment on parton distribution functions at low x. Eur. Phys. J. C 75(8), 396 (2015). arXiv:1503.04581
 21.R. Gauld, J. Rojo, Precision determination of the small\(x\) gluon from charm production at LHCb, Phys. Rev. Lett. 118(7), 072001 (2017). arXiv:1610.09373
 22.ATLAS Collaboration, G. Aad, et al., Measurement of the production of a \(W\) boson in association with a charm quark in \(pp\) collisions at \(\sqrt{s}\) = 7 TeV with the ATLAS detector, JHEP 1405, 068 (2014). arXiv:1402.6263
 23.CMS Collaboration, S. Chatrchyan et al., Measurement of the muon charge asymmetry in inclusive pp to WX production at \(\sqrt{s}\) = 7 TeV and an improved determination of light parton distribution functions. Phys. Rev. D 90, 032004 (2014). arXiv:1312.6283
 24.CMS Collaboration, Measurement of associated production of W bosons with charm quarks in protonproton collisions at \(\sqrt{s}=13~\rm TeV\) with the CMS experiment at the LHC, Tech. Rep. CMSPASSMP17014, CERN, Geneva (2018)Google Scholar
 25.J. Currie, E.W.N. Glover, J. Pires, Nexttonextto leading order QCD predictions for single jet inclusive production at the LHC. Phys. Rev. Lett. 118(7), 072002 (2017). arXiv:1611.01460 ADSCrossRefGoogle Scholar
 26.J. Rojo, Constraints on parton distributions and the strong coupling from LHC jet data. Int. J. Mod. Phys. A 30, 1546005 (2015). arXiv:1410.7728 ADSMathSciNetCrossRefGoogle Scholar
 27.P. Azzi et al., The Physics at the HL/HELHC Yellow ReportGoogle Scholar
 28.LHCb Collaboration, I. Bediaga et al., Physics case for an LHCb Upgrade II—opportunities in flavour physics, and beyond, in the HLLHC era. arXiv:1808.08865
 29.LHeC Study Group Collaboration, J. Abelleira Fernandez et al., A large hadron electron collider at cern: report on the physics and design concepts for machine and detector. J. Phys. G 39 (2012) 075001. arXiv:1206.2913
 30.LHeC Study Group Collaboration, J. L. Abelleira Fernandez et al., On the Relation of the LHeC and the LHC. arXiv:1211.5102
 31.LHeC study Group Collaboration, H. Paukkunen, An update on nuclear PDFs at the LHeC. PoS DIS 2017, 109 (2018). arXiv:1709.08342
 32.NNPDF Collaboration, R.D. Ball, S. Forte, A. Guffanti, E.R. Nocera, G. Ridolfi, J. Rojo, Polarized Parton Distributions at an ElectronIon. Collider. Phys. Lett. B 728, 524–531 (2014). arXiv:1310.0461
 33.C. Marquet, M.R. Moldes, P. Zurita, Unveiling saturation effects from nuclear structure function measurements at the EIC. Phys. Lett. B 772, 607–614 (2017). arXiv:1702.00839 ADSCrossRefGoogle Scholar
 34.E.C. Aschenauer, S. Fazio, M.A.C. Lamont, H. Paukkunen, P. Zurita, Nuclear structure functions at a future electronion collider. Phys. Rev. D 96(11), 114005 (2017). arXiv:1708.05654 ADSCrossRefGoogle Scholar
 35.E.C. Aschenauer, R. Sassot, M. Stratmann, Unveiling the proton spin decomposition at a future electronion collider. Phys. Rev. D 92(9), 094030 (2015). arXiv:1509.06489 ADSCrossRefGoogle Scholar
 36.D. Boer, M. Diehl, R. Milner, R. Venugopalan, W. Vogelsang, et al., Gluons and the quark sea at high energies: distributions, polarization, tomography. arXiv:1108.1713
 37.LHeC study Group Collaboration, A. M. CooperSarkar, Improved measurement of parton distribution functions and \(alpha _s(M_Z)\) with the LHeC. PoS DIS2016 274 (2016). arXiv:1605.08579
 38.ATLAS Collaboration, M. Aaboud et al., Precision measurement and interpretation of inclusive \(W^+\), \(W^\) and \(Z/gamma ^*\) production cross sections with the ATLAS detector. Eur. Phys. J. C 77(6), 367 (2017). arXiv:1612.03016
 39.CMS Collaboration, V. Khachatryan et al., Measurement of the differential cross section and charge asymmetry for inclusive \(\rm p\rm p\rm \rightarrow \rm W^{\pm }+X\) production at \({\sqrt{s}} = 8\) TeV. Eur. Phys. J. C 76(8), 469 (2016). arXiv:1603.01803
 40.J. Butterworth et al., PDF4LHC recommendations for LHC Run II. J. Phys. G 43, 023001 (2016). arXiv:1510.03865 ADSCrossRefGoogle Scholar
 41.J. Gao, P. Nadolsky, A metaanalysis of parton distribution functions. JHEP 1407, 035 (2014). arXiv:1401.0013 ADSCrossRefGoogle Scholar
 42.S. Carrazza, J.I. Latorre, J. Rojo, G. Watt, A compression algorithm for the combination of PDF sets. Eur. Phys. J. C 75, 474 (2015). arXiv:1504.06469 ADSCrossRefGoogle Scholar
 43.S. Carrazza, S. Forte, Z. Kassabov, J.I. Latorre, J. Rojo, An unbiased hessian representation for Monte Carlo PDFs. Eur. Phys. J. C 75(8), 369 (2015). arXiv:1505.06736 ADSCrossRefGoogle Scholar
 44.H. Paukkunen, P. Zurita, PDF reweighting in the Hessian matrix approach. JHEP 12, 100 (2014). arXiv:1402.6623 ADSCrossRefGoogle Scholar
 45.C. Schmidt, J. Pumplin, C. P. Yuan, P. Yuan, Updating and Optimizing Error PDFs in the Hessian Approach. arXiv:1806.07950
 46.A. Buckley, J. Ferrando, S. Lloyd, K. Nordstrm, B. Page et al., LHAPDF6: parton density access in the LHC precision era. Eur. Phys. J. C 75, 132 (2015). arXiv:1412.7420 ADSCrossRefGoogle Scholar
 47.ATLAS Collaboration, G. Aad et al., Measurement of the doubledifferential highmass DrellYan cross section in pp collisions at \( sqrt{s}=8 \) TeV with the ATLAS detector. JHEP 08, 009 (2016). arXiv:1606.01736
 48.NNPDF Collaboration, J. Rojo, Improving quark flavor separation with forward W and Z production at LHCb. PoS DIS 2017, 198 (2018). arXiv:1705.04468
 49.LHCb Collaboration, R. Aaij, et al., Measurement of forward W and Z boson production in \(pp\) collisions at \( sqrt{s}=8 \). TeV, JHEP 01, 155 (2016). arXiv:1511.08039
 50.ATLAS Collaboration, G., Aad et al., Measurements of topquark pair differential crosssections in the lepton+jets channel in \(pp\) collisions at \(sqrt{s}=8\) TeV using the ATLAS detector. Eur. Phys. J. C 76(10), 538 (2016). arXiv:1511.04716
 51.ATLAS Collaboration Collaboration, Determination of the parton distribution functions of the proton from ATLAS measurements of differential \(W\) and \(Z/\gamma ^{*}\) and \(t{\bar{t}}\) cross sections, Tech. Rep. ATLPHYSPUB2018017, CERN, Geneva (2018)Google Scholar
 52.ATLAS Collaboration, G., Aad et al., Measurement of the transverse momentum and \(phi ^*_{eta }\) distributions of DrellYan lepton pairs in protonproton collisions at \(sqrt{s}=8\) TeV with the ATLAS detector. Eur. Phys. J. C 76(5), 291 (2016). arXiv:1512.02192
 53.W. Stirling, E. Vryonidou, Charm production in association with an electroweak gauge boson at the LHC. Phys. Rev. Lett. 109, 082002 (2012). arXiv:1203.6781 ADSCrossRefGoogle Scholar
 54.Will Barter, Stephen Farry, private communicationGoogle Scholar
 55.ATLAS Collaboration, M., Aaboud, et al., Measurement of the cross section for inclusive isolatedphoton production in \(pp\) collisions at \(sqrt{s}=13\) TeV using the ATLAS detector. Phys. Lett. B 770, 473–493 (2017). arXiv:1701.06882
 56.J. Currie, A. GehrmannDe Ridder, T. Gehrmann, E.W.N. Glover, A. Huss, J. Pires, Precise predictions for dijet production at the LHC. Phys. Rev. Lett. 119(15), 152001 (2017). arXiv:1705.10271 ADSCrossRefGoogle Scholar
 57.E.L. Berger, J. Gao, H.X. Zhu, Differential distributions for tchannel single topquark production and decay at nexttonexttoleading order in QCD. JHEP 11, 158 (2017). arXiv:1708.09405 ADSCrossRefGoogle Scholar
 58.R. Boughezal, J.M. Campbell, R.K. Ellis, C. Focke, W. Giele, X. Liu, F. Petriello, C. Williams, Color singlet production at NNLO in MCFM. Eur. Phys. J. C 77(1), 7 (2017). arXiv:1605.08011 ADSCrossRefGoogle Scholar
 59.T. Carli et al., A posteriori inclusion of parton density functions in NLO QCD finalstate calculations at hadron colliders: the APPLGRID Project. Eur. Phys. J. C 66, 503 (2010). arXiv:0911.2985 ADSCrossRefGoogle Scholar
 60.Z. Nagy, Nexttoleading order calculation of threejet observables in hadron hadron collision. Phys. Rev. D 68, 094002 (2003). hepph/0307268ADSCrossRefGoogle Scholar
 61.CMS Collaboration, V. Khachatryan et al., Measurement of the doubledifferential inclusive jet cross section in proton–proton collisions at \(\sqrt{s}\) = 13 TeV. Eur. Phys. J. C 76(8), 451 (2016). https://doi.org/10.1140/epjc/s1005201642863
 62.NNPDF Collaboration, R.D. Ball et al., Parton distributions for the LHC Run II. JHEP 04 040, (2015). arXiv:1410.8849
 63.A.V. Manohar, P. Nason, G.P. Salam, G. Zanderighi, The photon content of the proton. JHEP 12, 046 (2017). arXiv:1708.01256 ADSCrossRefGoogle Scholar
 64.NNPDF Collaboration, V. Bertone, S. Carrazza, N.P. Hartland, J. Rojo, Illuminating the photon content of the proton within a global PDF analysis. SciPost Phys. 5, 008 (2018). arXiv:1712.07053
 65.R. Nathvani, L. HarlandLang, R. Thorne, A. Martin, Ad Lucem: The Photon in the MMHT PDFs, in 26th International Workshop on Deep Inelastic Scattering and Related Subjects (DIS 2018) Port Island, Kobe, Japan, April 16–20, 2018, (2018). arXiv:1807.07846
 66.L.A. HarlandLang, A.D. Martin, R.S. Thorne, The impact of LHC jet data on the MMHT PDF Fit at NNLO. Eur. Phys. J. C 78(3), 248 (2018). arXiv:1711.05757 ADSCrossRefGoogle Scholar
 67.CMS Collaboration, V. Khachatryan et al., Measurement of the differential cross section for top quark pair production in pp collisions at \(\sqrt{s} = 8\,\text{TeV} \). Eur. Phys. J. C 75(11), 542 (2015). arXiv:1505.04480
 68.CMS Collaboration, S. Chatrchyan et al., Measurement of associated W + charm production in pp collisions at \(\sqrt{s}\) = 7 TeV. JHEP 02, 013 (2014). arXiv:1310.1138
 69.R.D. Ball, V. Bertone, F. Cerutti, L. Del Debbio, S. Forte et al., Reweighting and unweighting of parton distributions and the LHC W lepton asymmetry data. Nucl. Phys. B 855, 608–638 (2012). arXiv:1108.1758 ADSCrossRefGoogle Scholar
 70.The NNPDF Collaboration, R.D. Ball et al., Reweighting NNPDFs: the W lepton asymmetry. Nucl. Phys. B 849, 112–143 (2011). arXiv:1012.0836
 71.The NNPDF Collaboration, R.D. Ball et al., Fitting parton distribution data with multiplicative normalization uncertainties. JHEP 05, 075 (2010). arXiv:0912.2276
 72.R.D. Ball, S. Carrazza, L. Del Debbio, S. Forte, J. Gao et al., Parton distribution benchmarking with LHC data. JHEP 1304, 125 (2013). arXiv:1211.5142 ADSCrossRefGoogle Scholar
 73.J. Gao, C.S. Li, C.P. Yuan, NLO QCD corrections to dijet production via quark contact interactions. JHEP 07, 037 (2012). arXiv:1204.4773 ADSCrossRefGoogle Scholar
 74.CMS Collaboration, V. Khachatryan et al., Search for quark contact interactions and extra spatial dimensions using dijet angular distributions in protonproton collisions at \(\sqrt{s} =\) 8 TeV. Phys. Lett. B 746, 79–99 (2015). arXiv:1411.2646
 75.C. Grojean, E. Salvioni, M. Schlaffer, A. Weiler, Very boosted Higgs in gluon fusion. JHEP 05, 022 (2014). arXiv:1312.3317 ADSCrossRefGoogle Scholar
 76.T. Sjostrand, S. Mrenna, P.Z. Skands, A brief introduction to PYTHIA 8.1. Comput. Phys. Commun. 178, 852–867 (2008). arXiv:0710.3820 ADSCrossRefGoogle Scholar
 77.B.C. Allanach, SUSY Les Houches Accord 2. Comput. Phys. Commun. 180, 8–25 (2009). arXiv:0801.0045 ADSCrossRefGoogle Scholar
 78.W. Beenakker, C. Borschensky, M. Krmer, A. Kulesza, E. Laenen, NNLLfast: predictions for coloured supersymmetric particle production at the LHC with threshold and Coulomb resummation. JHEP 12, 133 (2016). arXiv:1607.07741 ADSCrossRefGoogle Scholar
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Funded by SCOAP^{3}