Search for single production of vector-like quarks decaying to a b quark and a Higgs boson

A search is presented for single production of heavy vector-like quarks (B) that decay to a Higgs boson and a b quark, with the Higgs boson decaying to a highly boosted $\mathrm{b\overline{b}}$ pair reconstructed as a single collimated jet. The analysis is based on data collected by the CMS experiment in proton-proton collisions at $\sqrt{s} =$ 13 TeV, corresponding to an integrated luminosity of 35.9 fb$^{-1}$. The data are consistent with background expectations, and upper limits at 95% confidence level on the product of the B quark cross section and the branching fraction are obtained in the range 1.28-0.07 pb, for a narrow B quark with a mass between 700 and 1800 GeV. The production of B quarks with widths of 10, 20 and 30% of the resonance mass is also considered, and the sensitivities obtained are similar to those achieved in the narrow width case. This is the first search at the CERN LHC for the single production of a B quark through its fully hadronic decay channel, and the first study considering finite resonance widths of the B quark.


Introduction
With the discovery of the Higgs boson (H) by the ATLAS [1] and CMS [2,3] experiments at the CERN LHC, the standard model (SM) of particle physics has now been completely confirmed. However, the SM does not address, for example, problems related to the nature of the electroweak symmetry breaking and the hierarchy between the electroweak and the Planck mass scales. Several extensions of the SM address such issues through the introduction of new particles that allow the cancellation of loop corrections to the mass of the Higgs boson [4]. Supersymmetric theories propose bosonic partners of the top quark to address the hierarchy problem; other models such as Little Higgs or Composite Higgs boson models [5][6][7][8] overcome the hierarchy problem by introducing heavy fermionic resonances called vector-like quarks (VLQs) [4,[9][10][11]. The vector-like nature of these quarks does not exclude their having a fundamental mass, in contrast to chiral fermions, which acquire mass via electroweak symmetry breaking in the SM. The VLQs are therefore not excluded by present searches, unlike a fourth generation of SM quarks that is ruled out by electroweak precision measurements [12,13], and by the measured properties of the SM Higgs boson [14][15][16]. Previous searches for VLQs have been performed by the ATLAS [17][18][19][20][21][22] and CMS [23][24][25][26][27][28][29] experiments in proton-proton collisions recorded at centre-of-mass energies of 7, 8, and 13 TeV.
We present a search for electroweak production of single vector-like B quarks with electrical charge −1/3 e, with e the proton charge, that decay to a bottom (b) quark and a Higgs boson. The search uses pp events collected by the CMS experiment at a centre-of-mass energy of 13 TeV, corresponding to an integrated luminosity of 35.9 fb −1 . We study the fully hadronic final state with the Higgs boson decaying to a pair of b quarks. Figure 1 illustrates the electroweak production of a B quark in association with a b and a light-flavour quark, typically emitted into the forward region of the detector.
The B decay channel considered in this analysis is B → Hb. However, the B quark can also decay into Zb, Wt, and possibly into lighter states predicted in models beyond the SM that have model-dependent branching fractions. Our results are interpreted assuming that the B quark belongs to a singlet or doublet representation and that it decays exclusively to SM particles. The singlet branching fractions of the B quark into Hb, Zb, and Wt are B ≈ 25, 25, and 50%, and the doublet branching fractions are 50, 50, and 0%, and all depend on the vector-like quark mass m B .
Previous CMS searches for vector-like B quarks relied on the assumption of a decay width that is narrow compared to the experimental resolution. The present analysis, in addition to searching for B quarks with narrow decay widths, also explores the possibility that B quarks have a non-negligible width, with values up to 30% of the resonance mass. In comparison, the experimental resolution in the reconstructed B mass, defined as the ratio between the rootmean-square width of the peak and its mean position, ranges between 8 and 15%, depending on the mass hypothesis. In addition to broadening the width of the observed signal, the intrinsic width of the resonance would modify the kinematic distributions of the final state, thus changing the selection efficiency. These effects are taken into account in this analysis.
The cross section for single production of a B quark depends on m B and its electroweak couplings to SM particles. The kinematic distributions depend only on the total width of the B quark. The benchmark model in this analysis assumes a weak coupling of the B quark to the Z boson and b quark. Because of the mixing between B and the SM bottom quark in models where B is a singlet or part of a doublet, the BbZ electroweak coupling has a predominant chirality, respectively, right-or left-handed. The coupling chirality can potentially affect the kinematic distributions. We explicitly checked and found that these effects are negligible for the channel discussed in this work, and our results can therefore be interpreted in both singlet and doublet models.  Figure 1: The leading-order Feynman diagram for the production of a single vector-like B quark in association with a b quark and light-flavour quark, and its decay to a Higgs boson and a b quark.

The CMS detector and particle reconstruction
The central feature of the CMS apparatus is a superconducting solenoid of 6 m internal diameter, providing a magnetic field of 3.8 T. A silicon pixel and strip tracker, a lead tungstate crystal electromagnetic calorimeter (ECAL), and a brass and scintillator hadron calorimeter (HCAL), each composed of a barrel and two end sections, reside within the solenoid. Forward calorimeters extend the pseudorapidity (η) coverage provided by the barrel and end detectors. Muons are measured in gas-ionization detectors embedded in the steel flux-return yoke outside the solenoid. A more detailed description of the CMS detector, together with a definition of the coordinate system and kinematic variables, can be found in Ref. [30].
Events of interest are selected using a two-tiered trigger system [31]. The first level, composed of specialized hardware processors, uses information from the calorimeters and muon detectors to select events at a rate of ≈100 kHz within a time interval of less than 4 µs. The second level, known as the high-level trigger (HLT), consists of a farm of processors running a version of the full event-reconstruction software optimized for fast processing that reduces the event rate to ≈1 kHz before data storage.
Event reconstruction is based on the CMS particle-flow (PF) algorithm [32], which reconstructs and identifies each individual particle through an optimized combination of information from the various elements of the CMS detector. The energy of electrons is defined through the combination of the electron momentum at the primary interaction vertex determined in the tracker, the energy of the corresponding ECAL cluster, and the energy sum of all bremsstrahlung photons spatially compatible with originating from the electron track from the primary pp collision vertex. The energy of muons is obtained from the curvature of the corresponding track. The reconstructed energy of charged hadrons is extracted from the reconstructed tracks in the tracker and their matching energy depositions in ECAL and HCAL. Energy depositions are corrected for ignoring calorimeter readouts that are close to threshold (zero suppression) and for the Simulated events are used throughout the analysis to define selection strategy and to determine the expected sensitivity to vector-like quarks. The background from multijet events is estimated using data in control regions. Simulation is also used to cross-check the multijet background prediction and to evaluate its validity. The contributions from other backgrounds, such as tt events and W or Z boson production in association with jets, are estimated through MC simulation.
Multijet events, as well as electroweak backgrounds from virtual or on-mass shell Z or γ+jets and W+jets production, are simulated at leading order (LO) using the MADGRAPH5 aMC@NLO 2.2.2 generator [42], interfaced to PYTHIA 8.2 [43] with the CUETP8M1 [44,45] underlying-event tune for parton-shower simulation and evolution. The background tt events are generated using POWHEG v2 at next-to-leading order (NLO) [46][47][48][49], also interfaced to PYTHIA. The mass of the top quark is set to 172.5 GeV, and the cross section is calculated at next-to-next-to-leading order (NNLO) in perturbative QCD using a next-to-next-to-leading-logarithmic (NNLL) softgluon approximation (NNLO+NNLL) in the TOP++ 2.0 program [50]. The cross sections for Z or γ+jets and W+jets processes are calculated at NNLO using the FEWZ MC program [51].
The B → Hb → bbb events are simulated at LO, modelled using the universal FEYNRULES output [52,53] and the MC generator MADGRAPH5 aMC@NLO, interfaced to PYTHIA 8 for parton-shower simulation. Several mass hypotheses are considered for signals in the range 700 < m B < 1800 GeV, in steps of 100 GeV for total decay widths of 1 GeV, representing the narrow-width categories. Signal events for B quarks with large widths (10, 20, or 30% of the mass hypothesis) are also generated in the same mass range. All B quarks are generated with left-handed chirality, but the effect on the kinematic distributions of only considering one chirality is found to be negligible. Interference between the signal and the SM background is negligible.
Simulations using LO and NLO calculations, respectively, use the LO and NLO NNPDF3.0 [54] sets of parton distribution functions (PDFs). All signal and background events are processed using GEANT 4 [55] to provide a full simulation of the CMS detector. The generated events are also reweighted to account for the dependence of the reconstruction efficiency on the number of pileup interactions in the collisions.

Interpretation framework
The total cross section for the single production and decay of a B quark with final state X can be written as: where C 1 and C 2 are the production and decay couplings corresponding to the interactions through which a B quark is produced and decays, andσ AW is the reduced cross section for a resonance of arbitrary width (AW). This width can be written as Γ B = Γ(C i , m B , m decays ), as it depends on the B quark mass, on the masses of all its decay products, and on its couplings to all decay channels, C i .
Equation (1) is valid in all width regimes. However, when Γ B /m B approaches zero, it is possible to factorize production and decay and to write the cross section as: where C 1 is the B production coupling, and information for the parameters C 2 and Γ B are included in the branching fraction for the specific B quark decay, in this case B B→X , whilê σ NWA (m B ) is the reduced cross section in the narrow-width approximation (NWA).
Our assumptions have the B quark decaying into Hb, Zb, and Wt with branching fractions that are specified in the model. The couplings of the B quark to SM bosons and quarks can be parametrized as: c Z = e/(2c w s w κ Z ), c W = e/( √ 2s w κ W ), and c H = (m B κ H )/v, where e is the electric charge of the proton, v = 246 GeV is the vacuum-expectation value for the field of the Higgs boson, c w and s w are the cosine and sine of the weak mixing angle θ W , and κ is a coupling strength that can be fixed to obtain the desired width. Numerically, e/(2c w s w ) = 0.370, and c W = e/( √ 2s w ) = 0.458. For the process under consideration, we can set C 1 ≡ c Z and C 2 ≡ c H .
The κ values can be related to the mixing angle between the vector-like B quark and the b quark [56], and correspond to left-and right-handed couplings, which are the dominant chiralities for a singlet or part of a doublet B quark, respectively. For small values of κ, corresponding to the NWA regime, the following relations hold to excellent approximation: for a B singlet κ Z ≈ κ H ≈ κ W ≈ κ, while for a (T,B) doublet (where T is a vector-like quark with electrical charge 2/3) with no vector-like top quark Yukawa coupling, κ Z ≈ κ H ≈ κ, and κ W = 0. By imposing these relations among the κ values, and fixing the Γ B /m B ratio to 1%, κ is ≈0.1 in the whole range of explored masses. Table 1 provides the values forσ NWA and the physical cross sections in the NWA for the pp → Bbq process. The CTEQ6L PDF set [57] is used in this calculation.
To interpret the results in a model-independent way, the mechanism through which the B quarks achieve large widths is not specified, and Γ B is considered as a free parameter. The relations among the κ X (with X = W, Z, H), corresponding to the NWA limit (κ Z = κ H = κ W = κ), are imposed for the large-width regime. With this assumption, the total width Γ B is always proportional to κ 2 , and therefore κ can be chosen to obtain a specific Γ B /m B ratio. However, with the assumption relaxed, in a simplified model, new physics can be invoked to generate the required couplings.  Table 2 reports the cross sections integrated over the phase space of q and b, the particles produced in association with the B quark (see Fig. 1), for fixed values of Γ B /m B , with configurations of κ corresponding to singlet (σ S ) and doublet (σ D ) representations. Given the yields for a doublet in the Zb and Hb decay modes, these couplings at fixed width are larger than for singlets, and as a consequence σ D > σ S .
For each Γ B /m B , we provide the values ofσ AW and of the physical cross sections for both the singlet and doublet models, σ S and σ D respectively. The uncertainties in the production cross sections correspond to the halving and doubling of the QCD renormalization and factorization scales. The values of κ are listed in the parentheses.

Event selection
This analysis searches for a Higgs boson and a bottom quark arising from the decay of a B quark, and the decay of the Higgs boson into a pair of b quarks. An additional light-flavour quark, resulting from the production mechanism and produced in the forward direction (see Fig. 1), is also present. For values of m B much larger than the Higgs boson mass, the decay products of the B quark are expected to have large p T . The two b quarks originating from the Higgs boson tend therefore to emerge very close to each other in η-φ space, resulting in a single large jet.
The data are collected through an online selection (trigger) based on jet activity H T , defined as the scalar p T sum of all AK4 jets with p T > 30 GeV and |η| < 3. The jet activity threshold for this trigger is 900 GeV. Collisions containing at least one jet reconstructed through the HLT system with p T > 450 GeV are also selected, to increase the HLT efficiency. At the analysis level, H T is recalculated using AK4 jets with p T > 50 GeV and |η| < 2.4, and H T > 950 GeV is required. This offline selection corresponds to a trigger efficiency in excess of 87%.
Events are preselected if they contain three or more AK4 jets with p T > 30 GeV and |η| < 4, among which there must be at least one b-tagged jet with |η| < 2.4. A veto is applied to events with one or more leptons to ensure that the selection criteria do not overlap with those used for searches for the B quark in leptonic final states. Selected events are further required to have at least one large Higgs-tagged AK8 jet, fulfilling the Higgs boson tagging requirements as described in Section 2. The Higgs boson tagging efficiency is 10-20%, depending on the value of m B . Figure 2 compares to data the b-tagged subjet multiplicity expected for simulated background and for signal processes.
The B quark is reconstructed from the Higgs jet candidate along with a nonoverlapping btagged jet. The b quark from B quark decay is usually highly energetic (p T > 200 GeV), thus the b jet with the highest p T is chosen, and this reduces significantly the combinatorial background.
Furthermore, to reduce overlaps with the decay products of the Higgs boson, a condition is applied on the distance between the two objects in (η, φ), requiring ∆R(b, H) > 1.2. Background events are normalized to data. Only the statistical uncertainties are taken into consideration here, and they are too small to be visible.
To further reduce the multijet background and the contamination from gluon-like jets, H T is required to be in excess of 950 GeV for smaller mass values of 700 < m B < 1500 GeV, while for 1500 < m B < 1800 GeV, a trigger with a threshold of H T > 1250 GeV is chosen. In what follows, we refer to the former as the "low-mass analysis" and to the latter as the "high-mass analysis".
The signal to background discrimination is enhanced by exploiting the distinctive presence of a forward jet. Events are therefore separated into categories based on the forward-jet multiplicity. A high-purity category is obtained by requiring at least one forward jet. A second category that contains a large fraction of events from both signal and background, is defined requiring no forward jets. The forward-jet multiplicity expected for background and signal events after preselection is compared to data in Fig. 3. After all the selections are implemented, we reach signal efficiencies ranging from 2% or less at low masses, to larger values at larger m B , as a result of the optimization of the analysis for highly-boosted topologies. The disagreement between data and simulation at large forward-jet multiplicities does not affect the analysis, as the background contribution in the signal region is estimated from data. Moreover, the effect on the measurement is negligible since the majority of vector-like B quark events contain less than 2 forward jets, for which the simulated and observed yields are consistent after preselection.

Signal extraction
A potential signal would manifest itself as a localized excess over the expected background in the spectrum of the reconstructed mass m bH . A binned maximum likelihood fit is performed to the m bH distribution to extract a signal, exploiting the characteristic structure of the reconstructed B quark mass spectrum.
Multijet events constitute the dominant source of background in this search. An additional contribution of 5-7% arises from tt events. To reduce the dependence of the maximum-likelihood fit on the modelling of the multijet background in simulation, the contribution from this background is obtained from data. The procedure we use to estimate the yield of such events in the signal region is referred to as the "ABCD method" (discussed below), but its dependence on m bH is taken from a background-enriched control region in data. A minor contribution (≈1%) to the background arises from other SM sources, such as Z+jets and W+jets events. Both tt events and these minor backgrounds are estimated from simulation. The normalization of multijet events in the signal region is estimated using three data control regions, enriched in background events. These regions, in addition to the one enriched in signal events, are sampled in a two-dimensional phase space defined by two variables: the b-tagged subjet multiplicity of the Higgs jet and its reconstructed mass, m J . From a check on the simulation, the number of b-tagged subjets is not correlated with m J . The four regions used to define the ABCD method are: (i) region A, with two b-tagged subjets, and 105 < m J < 135 GeV, (ii) region B, with two b-tagged subjets, and 75 < m J < 105 GeV or m J > 135 GeV, (iii) region C, with one btagged subjet, and 105 < m J < 135 GeV, and (iv) region D, with one b-tagged subjet, and 75 < m J < 105 GeV or >135 GeV.
Region A is the signal region, defined by the selection criteria described in the previous section. The multijet background yield in the signal region is obtained from regions B, C, and D, which are background enriched. Assuming that the b-tagged subjet multiplicity and the Higgs boson mass are uncorrelated, the number of background events in the four regions follows the relationship: where N A , N B , N C , and N D are the yields in regions A, B, C, and D, respectively. Thus, the number of background events in the signal region A is given by: after subtracting the tt contribution predicted in the MC simulation. The contributions from Z+jets and W+jets backgrounds are not subtracted as they are negligible.
The m B distribution of the multijet background in the signal region is estimated from the m bH distribution in region C, since the reconstructed m bH spectrum is not expected to be correlated with the b jet multiplicity. The compatibility of the distributions in regions A and C is verified using simulated multijet events, and cross-checked in data.
In addition, the method is validated using a signal-depleted region from sidebands at large mass. Here, two regions (A' and C') are defined, similar to A and C in the mass region 135 < m J < 165 GeV. Two control regions (B' and D') are defined requiring 75 < m J < 105 GeV or m J > 165 GeV, respectively, with 2 or 1 b-tagged subjets. The background distribution estimated in region A', using the method described above, agrees with the observed data in region A'. The systematic uncertainty in the normalization of the multijet background is taken to be equal to the observed difference between the predicted and the measured yields in region A'. It amounts to 10% in the high purity category, and 5% in the category with no forward jets.

Systematic uncertainties
The systematic effect of each source of uncertainty is evaluated by propagating the uncertainty in the input parameters to the reconstructed B quark mass distribution and to the event yield. Then, the uncertainties in the event yield and in the m bH distribution for signal and background processes are taken into account as "nuisance" parameters that are integrated over in the statistical process of inferring the resultant parameters.
The statistical uncertainties in the background estimate of multijet production from control samples in data are propagated to m bH in the signal region by changing the observed event yields in regions B and D, up and down by one standard deviation, and recalculating the expected distribution in the signal region. As the expected multijet distribution in m bH is estimated from region C, its statistical uncertainty in this region is considered in the signal extraction. In addition to the normalization, this uncertainty affects the distribution of the background m bH in the signal region. Therefore, a systematic uncertainty in the estimated shape of this distribution, arising from the limited number of events in the observed m bH spectrum in region C, was derived by allowing the content of each bin to fluctuate independently according to Poisson statistics.
An additional systematic uncertainty in the estimated multijet background is obtained from the difference between the observed and predicted yields in the check, in the validation step that uses large-mass sideband regions, described in Section 6, and corresponds to ≈5-10%.
The systematic uncertainties from the limited number of simulated events and background estimates from simulation are also included by fluctuating each bin of the m bH distribution independently, according to Poisson statistics.
Additional systematic uncertainties in simulated signal and background distributions originate from the corrections applied to rescale simulated distributions to data. Other such uncertainties are listed below. An uncertainty of 2.5% [58] in the measured integrated luminosity is used just to account for the total event yields.
The corrections to account for the difference between the b tagging efficiency measured in data and in simulation are changed up and down by their uncertainties in both AK4 jets and subjets. The reconstructed four-momenta of the AK4 and AK8 jets are also shifted by ±1 standard deviation in the jet energy scale and resolution, and propagated to m bH . In addition, the pruned mass scale and resolution of the Higgs-tagged jet are changed within their uncertainties, affecting the m bH spectrum by 0.5-5.5%.
All simulated events are weighted to match the distribution of pileup interactions. The corresponding uncertainty is obtained by changing the total inelastic cross section by ±4.6%, which is used to calculate the pileup distribution in data. Scale factors are applied to account for differences between the trigger efficiency measured in data and in simulated events, with the uncertainties in the scale factors applied as a function of H T and propagated to the m bH distribution.
An additional uncertainty is applied to the simulated signal and backgrounds to account for discrepancies in the modelling of the forward jet multiplicity. The magnitude of this effect is obtained by considering the difference between the event yield in data and in MC, and results in an uncertainty of 0.5% for the category with no forward jets, and 2.0% for the category with at least one jet in the forward region.
The uncertainties from the choice of factorization and renormalization scales, µ F and µ R , are taken into account by halving and doubling the nominal values and using the combination of µ F and µ R leading to the maximal change. The resulting uncertainty in signal acceptance is as small as 1.3%, depending on the mass hypothesis. Larger effects (15-25%) are observed in the overall normalization and acceptance in simulated backgrounds. In addition, the uncertainty from the choice of PDF is estimated by reweighting the simulated signal and background events using the NNPDF3.0 [59][60][61] set of eigenvectors.
A summary of the systematic uncertainties considered in this analysis, along with their effect when propagated to the reconstructed B mass, is presented in Table 3.

Results
A binned maximum likelihood fit is performed to the m bH distribution in Fig. 4, where the dominant multijet background is estimated from data, as discussed in Section 6. The fitted m bH distributions are presented in Fig. 5, while the expected yields are listed in Table 4 for the backgrounds, and for two signal hypotheses (m bH = 1000 and 1800 GeV), together with their observed yields. The observed distributions are consistent with the background-only hypothesis in all the categories. Upper limits are set therefore on the product of the cross section and branching fraction of a B quark decaying to Hb, produced in association with another b quark and a light-flavoured quark, as a function of m bH . Exclusion limits at 95% confidence level (CL) are calculated using a modified frequentist approach and a profile likelihood ratio as test statistic, in an asymptotic approximation [62][63][64]. The combination of the two forward-jet multiplicity-based categories increases the sensitivity of the analysis by up to 20% relative to that obtained when only requiring at least one jet in the |η| > 2.4 region of the detector.
Systematic uncertainties described in Section 7 are treated as nuisance parameters affecting the rate of the expected m bH distribution. Both the uncertainties affecting the normalization, modelled using log-normal priors, and uncertainties in distributions are included in the fit [65].
The observed and expected combined upper limits from the two categories are given in Fig. 6. Assuming a narrow width, values of σ B(Hb) between 0.07-1.28 pb are excluded at the 95% Table 3: Summary of systematic uncertainties in background events. The quantification of the effects quoted in the table reflects the uncertainties in the event yields. All uncertainties are considered in the simulated background events, except the one on background estimation that affects only the data-based estimate of the multijet process. All the systematic uncertainties apply to both categories of forward-jet multiplicity, except for the case of the modelling of the forward jets, where the first entry corresponds to the category with no forward jets, and the second entry to the category with at least one jet in the forward region.    Table 4: Observed and expected fitted number of events in the signal ranges of 700 < m B < 1500 and 1500 < m B < 1800 GeV, and expected signal at m B = 1000 and 1800 GeV. The multijet background is obtained from data, while the yields for the other sources of background are obtained from MC simulation. The combined statistical and systematic uncertainties correspond to the quadrature of the statistical and systematic uncertainties.

Summary
A search has been presented for electroweak production of vector-like B quarks with charge −1/3 e, decaying to a bottom quark and a Higgs boson (H). The analysis uses a data sample corresponding to an integrated luminosity of 35.9 fb −1 , collected in pp collisions at √ s = 13 TeV.
No significant deviations are observed relative to the standard model prediction, and upper limits are placed on the product of the cross section and the branching fraction of the B quark.
Expected and observed limits at 95% confidence level vary from 1.20 to 0.07 pb and from 1.28 to 0.07 pb, respectively, for B quark masses in the range considered, which extends from 700 to 1800 GeV. The search is performed under the hypothesis of a singlet or doublet B quark of narrow width decaying to Hb with a branching fraction of approximately 25%. The possibility of having non-negligible resonant widths is also studied. Limits obtained on the production of B quarks with widths of 10, 20, and 30% of the resonance mass are comparable to those found for the narrow-width approximation. This search extends existing knowledge on vectorlike quarks, by interpreting the results in a new theoretical framework with non-negligible resonance widths, and investigating the final state with a bottom quark and a Higgs boson for the first time.

Acknowledgments
We congratulate our colleagues in the CERN accelerator departments for the excellent performance of the LHC and thank the technical and administrative staffs at CERN and at other CMS institutes for their contributions to the success of the CMS effort. In addition, we gratefully acknowledge the computing centers and personnel of the Worldwide LHC Computing Grid for delivering so effectively the computing infrastructure essential to our analyses.  [14] ATLAS and CMS Collaborations, "Measurements of the Higgs boson production and decay rates and constraints on its couplings from a combined ATLAS and CMS analysis of the LHC pp collision data at √ s = 7 and 8 TeV", JHEP 08 (2016) 045, doi:10.1007/JHEP08(2016)045, arXiv:1606.02266.