Measurement of differential cross sections for the production of top quark pairs and of additional jets in lepton + jets events from pp collisions at ffiffi s p = 13 TeV A.M. Sirunyan et al. * (CMS Collaboration) (Received 23 March 2018; published 15 June 2018) Differential and double-differential cross sections for the production of top quark pairs in proton-proton collisions at ffiffiffi s p ¼ 13 TeV are measured as a function of kinematic variables of the top quarks and the top quark-antiquark (tt̄) system. In addition, kinematic variables and multiplicities of jets associated with the tt̄ production are measured. This analysis is based on data collected by the CMS experiment at the LHC in 2016 corresponding to an integrated luminosity of 35.8 fb−1. The measurements are performed in the leptonþ jets decay channels with a single muon or electron and jets in the final state. The differential cross sections are presented at the particle level, within a phase space close to the experimental acceptance, and at the parton level in the full phase space. The results are compared to several standard model predictions that use different methods and approximations. The kinematic variables of the top quarks and the tt̄ system are reasonably described in general, though none predict all the measured distributions. In particular, the transverse momentum distribution of the top quarks is more steeply falling than predicted. The kinematic distributions and multiplicities of jets are adequately modeled by certain combinations of next-to-leading- order calculations and parton shower models. DOI: 10.1103/PhysRevD.97.112003 I. INTRODUCTION Measurements of differential production cross sections of top quark pairs (tt̄) provide important information for testing the standard model and searching for phenomena beyond the standard model. Precise theoretical predictions of these measurements are challenging since higher-order effects of quantum chromodynamics (QCD) and electro- weak (EW) corrections [1] are important. Moreover, the generation of tt̄ events requires a realistic modeling of the parton shower (PS). The measured kinematic properties and multiplicities of jets allow for a detailed comparison of different PS models to the data and provide insight into their tuning. In this paper, differential and double-differential pro- duction cross sections as a function of kinematic variables of the top quarks and the tt̄ system are reported. In addition, measurements of multiplicities and kinematic properties of jets in tt̄ events are presented. The measurements are based on proton-proton (pp) collision data at a center-of-mass energy of 13 TeV corresponding to an integrated luminosity of 35.8 fb−1 [2]. The data were recorded by the CMS experiment at the CERN LHC in 2016. Only tt̄ decays into the lþ jets (l ¼ e, μ) final state are considered, where, after the decay of each top quark into a bottom quark and a W boson, one of theW bosons decays hadronically and the other one leptonically. Hence, the experimental signature consists of two jets coming from the hadronization of b quarks (b jets), two jets from a hadronically decaying W boson, a transverse momentum imbalance associated with the neutrino from the leptonically decayingW boson, and a single isolated muon or electron. This measurement continues a series of differential tt̄ production cross section measurements in pp collisions at the LHC. Previous measurements of differential cross sections at ffiffiffi s p ¼ 7 TeV [3,4] and 8 TeV [5–11] have been performed in various tt̄ decay channels. First measurements at 13 TeV are available [12–14]. Previous studies of multiplicities and kinematic properties of jets in tt̄ events can be found in Refs. [15–17]. With about 15 times more data and an improved understanding of systematic uncer- tainties, we provide an update and extension to the previous CMS analysis in the lþ jets channel at 13 TeV [18]. We measure differential cross sections defined in two ways: at the particle level and the parton level. For the particle-level measurement a proxy of the top quark is defined based on experimentally accessible quantities, such as properties of jets, which are made up of quasistable *Full author list given at the end of the article. Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. Funded by SCOAP3. PHYSICAL REVIEW D 97, 112003 (2018) 2470-0010=2018=97(11)=112003(91) 112003-1 © 2018 CERN, for the CMS Collaboration https://crossmark.crossref.org/dialog/?doi=10.1103/PhysRevD.97.112003&domain=pdf&date_stamp=2018-06-15 https://doi.org/10.1103/PhysRevD.97.112003 https://doi.org/10.1103/PhysRevD.97.112003 https://doi.org/10.1103/PhysRevD.97.112003 https://doi.org/10.1103/PhysRevD.97.112003 https://creativecommons.org/licenses/by/4.0/ https://creativecommons.org/licenses/by/4.0/ particles with a mean lifetime greater than 30 ps. These quantities are described by theoretical predictions that require modeling of the PS and hadronization, in addition to the matrix-element calculations. The kinematic require- ments on these objects are chosen to closely reproduce the experimental acceptance. Muons and electrons stemming from τ lepton decays are not treated separately and can contribute to the particle-level signal. A detailed definition of particle-level objects is given in Sec. III. The particle- level approach has the advantage that it reduces theoretical uncertainties in the experimental results by avoiding theory-based extrapolations from the experimentally acces- sible portion of the phase space to the full range, and from jets to partons. For the parton-level measurement top quarks in the lþ jets decay channel are defined as signal directly before their decays into a bottom quark and a W boson. The τ þ jets decay channel is not considered here as signal even in cases where the τ lepton decays into a muon or electron. No restriction on the phase space is applied for parton-level top quarks. The corrections and extrapolations used in this measurement are based on a next-to-leading-order (NLO) calculation of tt̄ production, combined with a simulation of the PS. For both particle- and parton-level measurements the tt̄ system is reconstructed at the detector level with a likelihood-based approach using the top quark andW boson mass constraints to identify the corresponding top quark decay products. The differential cross sections are measured at the particle and parton levels as a function of the transverse momentum pT and the absolute rapidity jyj of the top quarks, separately for the hadronically (labeled th) and leptonically (labeled tl) decaying W bosons, and the pT, jyj, and invariant mass M of the tt̄ system. In addition, the differential cross sections at the parton level are determined as a function of the lower- and higher-pT values of the top quarks in an event. Double-differential cross sections for the following combinations of variables are determined at both levels: jyðthÞj vs. pTðthÞ, Mðtt̄Þ vs. jyðtt̄Þj, and pTðthÞ vs. Mðtt̄Þ. At particle level, the differential cross sections as a function of pTðthÞ, pTðtt̄Þ, and Mðtt̄Þ are measured in bins of jet multiplicity. Using the four jets identified as the tt̄ decay products and the four highest-pT additional jets, the cross sections are determined as a function of the jet pT and absolute pseudorapidity jηj, the minimal separation ΔRjt of jets from another jet in the tt̄ system, and the separation ΔRt of jets from the closest top quark. Here ΔR ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðΔϕÞ2 þ ðΔηÞ2 p , where Δϕ and Δη are the differences in azimuthal angle (in radians) and pseudora- pidity between the directions of the two objects. Finally, we determine the gap fraction, defined as the fraction of events that do not contain jets above a given pT threshold, and the jet multiplicities for various thresholds of the jet pT. This paper is organized as follows: In Sec. II, we provide a description of the signal and background simulations, followed by the definition of the particle-level top quarks in Sec. III. After a short overview of the CMS detector and the particle reconstruction in Sec. IV, we describe the object and event selections in Sec. V. Section VI contains a detailed description of the reconstruction of the tt̄ system. Details on the background estimation and the unfolding are presented in Secs. VII and VIII. After a discussion of systematic uncertainties in Sec. IX, the differential cross sections as a function of observables of the top quark and the tt̄ system are presented in Sec. X. Finally, observables involving jets are discussed in Sec. XI. The results are summarized in Sec. XII. II. SIGNAL AND BACKGROUND MODELING The Monte Carlo generator POWHEG [19–22] (v2,hvq) is used to calculate the production of tt̄ events at NLO accuracy in QCD. The renormalization μr and factorization μf scales are set to the transverse massmT ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi m2 t þ p2 T p of the top quark, where a top quark mass mt ¼ 172.5 GeV is used in all simulations. The result is combined with the PS simulations of PYTHIA8 [23,24] (v8.219) using the under- lying event tune CUETP8M2T4 [25,26], and of HERWIG++ [27] (v2.7.1) using the tune EE5C [28]. In addition, MADGRAPH 5_aMC@NLO [29] (v2.2.2) (MG5_aMC@NLO) is used to produce a simulation of tt̄ events with additional partons. All processes with up to two additional partons are calculated at NLO and combined with the PYTHIA8 PS simulation using the FxFx [30] algorithm. The scales are selected as μr ¼ μf ¼ 1 2 ðmTðtÞ þmTðt̄ÞÞ. The default para- metrization of the parton distribution functions (PDFs) used in all simulations is NNPDF30_nlo_as_0118 [31]. The simulations are normalized to an inclusive tt̄ production cross section of 832þ40 −46 pb [32]. This value is calculated with next-to-NLO (NNLO) accuracy, including the resum- mation of next-to-next-to-leading-logarithmic soft-gluon terms. Its uncertainty is evaluated by varying the choice of μr and μf and by propagating uncertainties in the PDFs. Distributions that correspond to variations in the PDFs or the scales μr and μf are obtained by applying different event weights. These distributions are used for the corresponding uncertainty estimates. For additional uncertainty estima- tions we use POWHEG+PYTHIA8 simulations with top quark masses of 171.5 and 173.5 GeV, with initial and final PS scales varied up and down by a factor of two, with variations of the underlying event tune, and a simulation with an alternative color-reconnection model. The main backgrounds are simulated using the same techniques. The MG5_aMC@NLO generator is used for the simulation of W boson production in association with jets, t-channel single top quark production, and Drell–Yan (DY) production in association with jets. The generator POWHEG [33] is used for the simulation of single top quark associated production with aW boson (tW), and PYTHIA8 is used for multijet production. In all cases, the PS and the A.M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-2 hadronization are described by PYTHIA8. The W boson and DY backgrounds are normalized to their NNLO cross sections calculated with FEWZ [34] (v3.1). The t-channel single top quark production is normalized to the NLO calculation obtained from HATHOR [35] (v2.1). The pro- duction of tW is normalized to the NLO calculation [36], and the multijet simulation is normalized to the LO calculation obtained with PYTHIA8 [24]. The detector response is simulated using GEANT4 [37]. The simulations include multiple pp interactions per bunch crossing (pileup). The simulated events are weighted, depending on their number of pileup interactions, to reproduce the measured pileup distribution. Finally, the same reconstruction algorithms that are applied to the data are used for the simulated events. III. PARTICLE-LEVEL TOP QUARK DEFINITION The definitions of particle-level objects constructed from quasi-stable simulated particles, obtained from the predic- tions of tt̄ event generators before any detector simulation, are summarized below. These particle-level objects are further used to define the particle-level top quarks. Detailed studies on particle-level definitions can be found in Ref. [38]. (i) All simulated muons and electrons are corrected for effects of bremsstrahlung by adding the photon momenta to the momentum of the closest lepton if their separation is ΔR < 0.1. All photons are considered for the momentum correction. A cor- rected lepton is selected if it fulfills the isolation requirement that the pT sum of all quasi-stable particles, excluding corrected leptons and neutrinos, within ΔR ¼ 0.4 is less than 35% of the corrected lepton pT. In addition, we require the lepton to have pT > 15 GeV and jηj < 2.4. (ii) Simulated photons with pT > 15 GeV and jηj < 2.4 that are not used in the momentum correction of a lepton are considered if their isolation, defined analogously to the lepton isolation, is below 25%. (iii) All neutrinos are selected including those stemming from decays of hadrons. (iv) Jets are clustered by the anti-kT jet algorithm [39,40] with a distance parameter of 0.4. All quasistable particles with the exception of neutrinos are clus- tered. Jets with pT > 25 GeV and jηj < 2.4 are selected if there is no isolated lepton or photon, as defined above, within ΔR ¼ 0.4. (v) b jets at the particle level are defined as those jets that contain a b hadron. As a result of the short lifetime of b hadrons, these are not quasistable particles and only their decay products should be considered for the jet clustering. However, to allow their association with a jet, the b hadrons are also included with their momenta scaled down to a negligible value. This preserves the information of their directions, but removes their impact on the jet clustering. Based on the invariant masses of these objects, we construct a pair of particle-level top quarks in the lþ jets final state. Events with exactly one muon or electron with pT > 30 GeV and jηj < 2.4 are selected. Simulated events with an additional muon or electron with pT > 15 GeV and jηj < 2.4 are rejected. We take the sum of the four-momenta of all neutrinos as the neutrino candidate momentum pν from the leptonically decaying top quark and find the permutation of jets that minimizes the quantity ½Mðpν þ pl þ pblÞ −mt�2 þ ½MðpjW1 þ pjW2 Þ −mW �2 þ ½MðpjW1 þ pjW2 þ pbhÞ −mt�2; ð1Þ where pjW1;2 are the four-momenta of two light-flavor jet candidates, considered as the decay products of the hadronically decayingW boson; pbl;h are the four-momenta of two b jet candidates; pl is the four-momentum of the lepton; and mW ¼ 80.4 GeV [41] is the mass of the W boson. All jets with pT > 25 GeV and jηj < 2.4 are considered. At least four jets are required, of which at least two must be b jets. The remaining jets with pT > 30 GeV and jηj < 2.4 are defined as additional jets. Events with a hadronically and a leptonically decaying particle-level top quark are not required to be lþ jets events at the parton level, e.g., tt̄ dilepton events with additional jets can be identified as lþ jets event at the particle level if one lepton fails to pass the selection. As an example, the comparison between the pTðthÞ distributions at the particle and parton levels are shown in Fig. 1 and demonstrates the direct relation between particle-level and parton-level top quarks. To obtain an unambiguous nomenclature for the jets, we define jW1 to be the jet in theW boson decay with the higher pT. The additional jets ji are sorted by their transverse momenta where j1 has the highest pT. IV. THE CMS DETECTOR The central feature of the CMS detector is a super- conducting solenoid of 6 m internal diameter, providing a magnetic field of 3.8 T. Within the solenoid volume are 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 endcap sections. Forward calorimeters extend the η coverage provided by the barrel and endcap 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 relevant kinematic variables, can be found in Ref. [42]. The particle-flow (PF) event algorithm [43] reconstructs and identifies each individual particle with an optimized MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-3 combination of information from the various elements of the CMS detector. The energy of muons is obtained from the curvature of the corresponding track. The energy of electrons is determined from a combination of the electron momentum at the primary interaction vertex as determined by 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. The energy of photons is directly obtained from the ECAL measurement, corrected for zero-suppression effects. The energy of charged hadrons is determined from a combination of their momentum measured in the tracker and the matching ECAL and HCAL energy deposits, corrected for zero-suppression effects and for the response function of the calorimeters to hadronic showers. Finally, the energy of neutral hadrons is obtained from the corre- sponding corrected ECAL and HCAL energy. V. PHYSICS OBJECT RECONSTRUCTION AND EVENT SELECTION The measurements presented in this paper depend on the reconstruction and identification of muons, electrons, jets, and missing transverse momentum associated with a neutrino. Muons and electrons are selected if they are compatible with originating from the primary vertex, which, among the reconstructed primary vertices, is the one with the largest value of summed physics-object p2 T. The physics objects are jets, clustered using the jet finding algorithm [39,40] with the tracks assigned to the primary vertex as inputs, and the associated missing transverse momentum, taken as the negative vector sum of the p⃗T of those jets. Since leptons from tt̄ decays are typically isolated, a requirement on the lepton isolation is used to reject leptons produced in decays of hadrons. The lepton isolation variables are defined as the sum of the pT of neutral hadrons, charged hadrons, and photon PF candidates within a cone of ΔR ¼ 0.4 for muons and ΔR ¼ 0.3 for electrons. It is required to be less than 15% (6%) of the muon (electron) pT. Event-by-event corrections are applied to maintain a pileup-independent isolation efficiency. The muon and electron reconstruction and selection efficiencies are measured in the data using tag-and-probe techniques [44–46]. Depending on the pT and η, their product is 75–85% for muons and 50–80% for electrons. Jets are clustered from PF objects using the anti-kT jet algorithm with a distance parameter of 0.4 implemented in the FASTJET package [40]. Charged particles originating from a pileup interaction vertex are excluded. The total energy of the jets is corrected for energy depositions from pileup. In addition, pT- and η-dependent corrections are applied to correct for the detector response effects [47]. If an isolated lepton with pT > 15 GeV within ΔR ¼ 0.4 around a jet exists, the jet is assumed to represent the isolated lepton and is removed from further consideration. For the identification of b jets, the combined secondary vertex algorithm [48] is used. It provides a discriminant between b and non-b jets based on the combined informa- tion of secondary vertices and the impact parameter of tracks at the primary vertex. A jet is identified as a b jet if the associated value of the discriminant exceeds a threshold criterion with an efficiency of about 63% and a combined charm and light-flavor jet rejection probability of 97%. The missing transverse momentum p⃗miss T is calculated as the negative of the vectorial sum of transverse momenta of all PF candidates in the event. Jet energy corrections are also propagated to improve the measurement of p⃗miss T . 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T Particle level p 0 0.2 0.4 0.6 0.8 1 1.2 F ra ct io n Purity Stability Bin efficiency Non-parton-level signal CMS Simulation 0 0.05 0.1 0.15 0.2 0.25 N or m al iz ed a t p ar to n le ve l p er b in 0.09 0.03 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.11 0.05 0.03 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.04 0.12 0.06 0.03 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.03 0.13 0.07 0.03 0.02 0.01 0.01 0.01 0.01 0.01 0.00 0.00 0.01 0.02 0.14 0.08 0.03 0.02 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.01 0.02 0.16 0.08 0.03 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.01 0.02 0.17 0.08 0.02 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.01 0.03 0.21 0.09 0.03 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.22 0.09 0.02 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.22 0.07 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.03 0.23 0.05 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.03 0.23 1 2 3 4 5 6 7 8 9 10 11 12 ) h (t T Bin at parton level: p 1 2 3 4 5 6 7 8 9 10 11 12 ) h (t T B in a t p ar tic le le ve l: p CMSSimulation FIG. 1. Comparison between the pTðthÞ distributions at the particle and parton level, extracted from the POWHEG+PYTHIA8 simulation. Left: fraction of parton-level top quarks in the same pT bin at the particle level (purity), fraction of particle-level top quarks in the same pT bin at the parton level (stability), ratio of the number of particle- to parton-level top quarks (bin efficiency), and fraction of events with a particle-level top quark pair that are not considered as signal events at the parton level (non-parton- level signal). Right: pT -bin migrations between particle and parton level. The pT range of the bins can be taken from the left panel. Each column is normalized such that the sum of its entries corresponds to the fraction of particle-level events in this bin at the parton level in the full phase space. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-4 Events considered for this analysis are selected by single-lepton triggers. These require pT > 24 GeV for muons and pT > 27 GeV for electrons, as well as various quality and isolation criteria. To reduce the background contributions and optimize the tt̄ reconstruction, additional requirements are imposed on the recorded events. Events with exactly one muon or electron with pT > 30 GeV and jηj < 2.4 are selected. No additional muons or electrons with pT > 15 GeV and jηj < 2.4 are allowed. In addition to the lepton, at least four jets with pT > 30 GeV and jηj < 2.4 are required. At least two of these jets must be identified as b jets. We compare several kinematic distributions in the data to the simulation separately for the muon and electron channels to verify that there are no unexpected differences. The ratios of the measured to the expected event yields in the two channels agree within the uncertainty in the lepton reconstruction and selection efficiencies. In the remaining steps of the analysis, the two channels are combined by adding their distributions. VI. RECONSTRUCTION OF THE TOP QUARK-ANTIQUARK SYSTEM The reconstruction of the tt̄ system follows closely the methods used in Ref. [18]. The goal is the correct identification of detector-level objects as parton- or particle-level top quark decay products. In the simulation, a jet or lepton at the particle level can be spatially matched to the corresponding detector-level object. If no one-to-one assignment to a corresponding detector-level object is possible for any of the objects in the particle-level tt̄ system, the event is considered as “nonreconstructable” in the particle-level measurement. For the parton-level meas- urement a quark from the tt̄ decay is assigned to the detector-level jet with the highest pT within ΔR ¼ 0.4 around the parton. If no one-to-one correspondence at detector level is found for any of these quarks or the leptons, the event is “nonreconstructable” in the parton- level measurement. In particular, this includes events with merged topologies where at least two quarks are matched to the same jet. Based on these relations between detector level and parton or particle level, the efficiencies of the tt̄ reconstruction are studied. A detailed discussion on the relationship between quantities at the parton or particle level and detector level is presented in Sec. VIII. For the reconstruction all possible permutations of assigning detector-level jets to the corresponding tt̄ decay products are tested and a likelihood that a certain permu- tation is correct is evaluated. Permutations are considered only if the two jets with the highest b identification probabilities are the two b jet candidates. In each event, the permutation with the highest likelihood is selected. The likelihoods are evaluated separately for the particle- and the parton-level measurements. For each tested permutation the neutrino four-momentum pν is reconstructed using the algorithm of Ref. [49]. The idea is to find all possible solutions for the three components of the neutrino momentum vector using the two mass constraints ðpνþplÞ2¼m2 W and ðpν þ pl þ pblÞ2 ¼ m2 t . Each equation describes an ellipsoid in the three- dimensional momentum space of the neutrino. The inter- section of these two ellipsoids is usually an ellipse. We select pν as the point on the ellipse for which the distance Dν;min between the ellipse projection onto the transverse plane and p⃗miss T is minimal. This algorithm leads to a unique solution for the longitudinal neutrino momentum and an improved resolution of its transverse component. For the cases where the invariant mass of the lepton and bl candidate is above mt no solution can be found and the corresponding permutation is discarded. The minimum distance Dν;min is also used to identify the correct bl, as described below. The value of Dν;min from the neutrino reconstruction and the mass constraints on the hadronically decaying top quark are combined in a likelihood function λ, given by − log½λ� ¼ − log½Pmðm2; m3Þ� − log½PνðDν;minÞ�; ð2Þ where Pm is the two-dimensional probability density of the invariant masses of W bosons and top quarks that are correctly reconstructed, based on the matching criteria described above. The value of λ is maximized to select the permutation of jets. The probability density Pm is calculated as a function of the invariant mass of the two jets, m2, tested as the W boson decay products, and the invariant mass of the three jets, m3, tested as the decay products of the hadronically decaying top quark. The distributions for the correct jet assignments, taken from the POWHEG+PYTHIA8 simulation and normalized to unit area, are shown in Fig. 2 (upper) for the particle- and parton-level measurements. This part of the likelihood function is sensitive to the correct reconstruction of the hadronically decaying top quark. Permutations with prob- abilities less than 0.1% of the maximum value of the probability density Pm are rejected. This selection criterion discards less than 1% of the correctly reconstructed events. Especially in the parton-level measurement, it removes events that are incompatible with the hypothesis of a hadronically decaying top quark and reduces the back- ground contribution. This is caused by the stringent mass constraints for a parton-level top quark, where, in contrast to the particle-level top quark, close compatibility with the top quark and W boson masses are required. The probability density Pν describes the distribution of Dν;min for a correctly selected bl. In Fig. 2 (lower), the normalized distributions of Dν;min for bl and for other jets are shown. On average, the distance Dν;min for a correctly selected bl is smaller and has a smaller tail compared to the distance obtained for other jets. Permutations with values of MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-5 Dν;min > 150 GeV are rejected since they are very unlikely to originate from a correct bl association. This part of the likelihood function is sensitive to the correct reconstruction of the leptonically decaying top quark. Since the likelihood function λ combines the probabil- ities from the reconstruction of the leptonically and hadronically decaying top quarks, it provides information on the reconstruction of the whole tt̄ system. The 6−10 5−10 4−10 3−10 0 50 100 150 200 250 300 350 400 450 500 ) [GeV] h M(t 0 50 100 150 200 250 300 350 400 450 500 M (W ) [G eV ] CMS Simulation parton level +jetsμe/ 0 50 100 150 200 250 300 350 400 450 500 ) [GeV] h M(t 0 50 100 150 200 250 300 350 400 450 500 M (W ) [G eV ] 7−10 6−10 5−10 4−10 CMS Simulation particle level +jetsμe/ 0 20 40 60 80 100 120 140 [GeV],minνD 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 lCorrect b lIncorrect b CMS Simulation parton level +jetsμe/ 0 20 40 60 80 100 120 140 [GeV],minνD 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 lCorrect b lIncorrect b CMS Simulation particle level +jetsμe/ FIG. 2. Upper: normalized two-dimensional mass distribution of the correctly reconstructed hadronically decaying W bosons MðWÞ and the correctly reconstructed top quarks MðthÞ for the (left) parton- and the (right) particle-level measurements. Lower: normalized distributions of the distance Dν;min for correctly and incorrectly selected b jets from the leptonically decaying top quarks. The distributions are taken from the POWHEG+PYTHIA8 tt̄ simulation. 0 1 2 3 4 5≥ Additional jets 0 0.2 0.4 0.6 0.8 1 r ec on st ru ct io n ef f. tt P8OWHEGP P8 FSR downOWHEGP P8 FSR upOWHEGP H++OWHEGP MG5_aMC@NLO P8 CMS Simulation parton level +jetsμe/ 0 1 2 3 4 5≥ Additional jets 0 0.2 0.4 0.6 0.8 1 r ec on st ru ct io n ef f. tt P8OWHEGP P8 FSR downOWHEGP P8 FSR upOWHEGP H++OWHEGP MG5_aMC@NLO P8 CMS Simulation particle level +jetsμe/ FIG. 3. Reconstruction efficiency of the tt̄ system as a function of the number of additional jets for the (left) parton- and (right) particle-level measurements. The efficiencies are calculated based on the simulations with POWHEG+PYTHIA8 (P8) with scale variations up and down of the final-state PS, POWHEG+HERWIG++ (H++), and MG5_aMC@NLO+PYTHIA8. The vertical bars represent the statistical uncertainties in each simulation. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-6 performance of the reconstruction as a function of jet multiplicity is shown for several tt̄ simulations in Fig. 3, where we use the input distributions Pm and Pν from the POWHEG+PYTHIA8 simulation. The reconstruction effi- ciency of the algorithm is defined as the probability that the most-likely permutation, as identified through the maximization of the likelihood λ, is the correct one, given that all tt̄ decay products are reconstructed and selected. The performance deteriorates with the increase in the number of jets, since the number of permutations increases drastically and the probability of selecting a wrong per- mutation increases. The differences observed between the various simulations are taken into account in the estimation of the systematic uncertainties. We observe a lower reconstruction efficiency for the particle-level measure- ment. This is caused by the less powerful mass constraints for a particle-level top quark. This can be seen in the mass distributions of Fig. 2 and the likelihood distributions in Fig. 4, where the simulations are normalized to the measured integrated luminosity of the data sample, and the tt̄ simulation is divided into the following categories: correctly reconstructed tt̄ systems (tt̄ right reco); events where all decay products are available, but the algorithm failed to identify the correct permutation (tt̄ wrong reco); the nonreconstructable events (tt̄ nonreconstructable); and events that are according to the parton- or particle-level definitions not tt̄ signal events (tt̄ nonsignal). Only the last category is treated as tt̄ background, while the other categories are considered as signal. The lower reconstruction efficiency of the particle-level top quark is compensated by the higher number of reconstructable events. In Fig. 5, the pT of the jets from the tt̄ system, as identified by the reconstruction algorithm, and of the additional jets are presented and compared to the simu- lation. In Fig. 6, the distributions of pT and jyj of the reconstructed top quarks, and in Fig. 7, the distributions of pTðtt̄Þ, jyðtt̄Þj, and Mðtt̄Þ for the parton- and particle-level measurements are shown. The simulations are normalized according to the measured integrated luminosity of the data. In general, good agreement is observed between the data and the simulation, although all measured pT spectra are softer than predicted by the simulation. VII. BACKGROUND SUBTRACTION After the event selection and tt̄ reconstruction, we observe about 450 000 and 570 000 events at the parton and particle levels, respectively, where total background contributions of 4.5 and 6.0% from single top quark, DY,W boson, and multijet events are predicted. These back- grounds have to be estimated and subtracted from the selected data. In addition, a residual contamination from nonsignal tt̄ events is expected and estimated from the simulation, as detailed below. The predictions of the single top quark background are taken from simulations. Its overall contribution corre- sponds to about 2.7 and 3.3% of the selected data in the parton- and particle-level measurements, respectively. Single top quark production cross sections are calculated with a precision of a few percent [35,36]. However, these calculations do not consider the production of additional jets as required by the tt̄ selection. Therefore, we use an overall uncertainty of 50%, which represents a conservative estimate of the PS modeling, scale, and PDF uncertainties. Even with such a conservative estimate, its impact on the precision of the final results is negligible. After the full tt̄ selection, the numbers of events in the simulations of multijet, DY, and W boson production are not sufficient to obtain smooth background distributions. Therefore, we extract a common distribution for these 0 10000 20000 30000 40000 50000 60000 E ve nt s / 0 .4 Data right recott wrong recott nonreconstructablett nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb parton level +jetsμe/CMS 10 12 14 16 18 20 22 )λ-log( 0.6 0.8 1 1.2 1.4 P re d. D at a 0 10000 20000 30000 40000 50000 60000 70000 E ve nt s / 0 .4 Data right recott wrong recott nonreconstructablett nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 10 12 14 16 18 20 22 )λ-log( 0.6 0.8 1 1.2 1.4 P re d. D at a FIG. 4. Distribution of the negative log-likelihood for the selected best permutation in the (left) parton- and the (right) particle-level measurements in data and simulations. Events generated with POWHEG+PYTHIA8 are used to describe the tt̄ production. The contribution of multijet, DY, andW boson plus jets background events is extracted from the data (cf. Sec. VII). Combined experimental (cf. Sec. IX) and statistical uncertainties (hatched area) are shown for the total predicted yields. The data points are shown with statistical uncertainties. The ratios of data to the sum of the predicted yields are provided at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-7 210 310 410 510 610 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 100 200 300 400 500 ) [GeV] l (b T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 100 200 300 400 500 ) [GeV] h (b T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 100 200 300 400 500 ) [GeV] W1 (j T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 710 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 50 100 150 200 250 ) [GeV] W2 (j T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 100 200 300 400 500 600 ) [GeV] 1 (j T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 50 100 150 200 250 300 350 400 450 ) [GeV] 2 (j T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 50 100 150 200 250 300 ) [GeV] 3 (j T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 E ve nt s / 1 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 40 60 80 100 120 140 160 180 ) [GeV](jp 0.6 0.8 1 1.2 1.4 P re d. D at a 4T FIG. 5. Comparisons between data and simulation at the particle level of the reconstructed distributions of the pT of jets as identified by the tt̄ reconstruction algorithm. The simulation of POWHEG+PYTHIA8 is used to describe the tt̄ production. The contribution of multijet, DY, andW boson plus jets background events is extracted from the data (cf. Sec. VII). Combined experimental (cf. Sec. IX) and statistical uncertainties (hatched area) are shown for the total predicted yields. The data points are shown with statistical uncertainties. The ratios of data to the predicted yields are given at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-8 backgrounds from a control region in the data. Its selection differs from the signal selection by the requirement of having no b-tagged jet in the event. In this control region, the contribution of tt̄ events is estimated to be about 15%, while the remaining fraction consists of multijet, DY, andW boson events. The background distributions are obtained after applying exactly the same tt̄ reconstruction algorithm as in the signal region. The two b jet candidates still have the highest value of the b identification discriminant to maintain a similar number of allowed permutations of jets in the control and signal regions. To estimate the shape dependency on the selection of the control region, we vary the selection threshold of the b identification discriminant. This changes the tt̄ signal contribution and the flavor composition. However, we find the observed shape variations to be small. In addition, we verify in the simulation that the shapes of the distributions obtained from the control region are compatible with the background distributions in the signal region. For the background subtraction the distributions extracted from the control region are normalized individually in each bin of jet multiplicity to the yield of multijet, DY, and W boson events predicted by the simulation in the signal region. In the control region, the expected and measured event yields agree within their statistical uncertainties. Taking into account the statistical uncertainty in the normalization factor and the shape differences between the signal and control regions in the simulation, we derive an overall uncertainty of 20% in this background estimation. Special care has to be taken with the contribution of nonsignal tt̄ events. For the parton-level measurement these are dilepton, all-jets, and τ þ jets events. For the particle- level measurement all tt̄ events for which no pair of particle- level top quarks exists are considered as nonsignal tt̄ events. The corresponding contributions are about 11.5% for both the parton- and the particle-level measurements. The behav- ior of these backgrounds depends on the tt̄ cross section, and a subtraction according to the expected value can result in a bias of the measurement, especially if large differences between the simulation and the data are observed. However, the shapes of the distributions from data and simulation are consistent within their uncertainties, and we subtract the predicted relative fractions from the remaining event yields. 210 310 410 510 610 E ve nt s / 4 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb parton level +jetsμe/CMS 0 100 200 300 400 500 600 700 800 900 ) [GeV] l (t T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 4 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 0 100 200 300 400 500 600 700 800 900 ) [GeV] h (t T p 0.6 0.8 1 1.2 1.4 P re d. D at a 0 20 40 60 80 100 120 310× E ve nt s / 0 .2 5 Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb parton level +jetsμe/CMS 0 0.5 1 1.5 2 2.5 l )y(t 0.6 0.8 1 1.2 1.4 P re d. D at a 0 20 40 60 80 100 120 140 160 310× E ve nt s / 0 .2 5 Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 0 0.5 1 1.5 2 2.5 h )y(t 0.6 0.8 1 1.2 1.4 P re d. D at a FIG. 6. Comparisons of the reconstructed pT (upper) and jyj (lower) in data and simulations for the tl (left) at the parton level and the th (right) at the particle level. The simulation of POWHEG+PYTHIA8 is used to describe the tt̄ production. The contribution of multijet, DY, and W boson plus jets background events is extracted from the data (cf. Sec. VII). Combined experimental (cf. Sec. IX) and statistical uncertainties (hatched area) are shown for the total predicted yields. The data points are shown with statistical uncertainties. The ratios of data to the predicted yields are given at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-9 VIII. CORRECTIONS TO PARTICLE AND PARTON LEVELS After the subtraction of the backgrounds, an unfolding procedure is used to correct the reconstructed distributions for detector-specific effects, e.g., efficiency and resolutions, and to extrapolate either to the parton or particle level. We do not subtract the fractions of wrongly reconstructed or nonreconstructable events, since in many of these events a rather soft jet is misidentified, which has little impact on the resolution of the measured quantities. The iterative D’Agostini method [50] is used to unfold the data. The migration matrices, which relate the quantities at the parton or particle level and at detector level, and the acceptances are needed as the input. However, not only the detector simulation, but also the theoretical description of tt̄ events 210 310 410 510 610 E ve nt s / 2 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb parton level +jetsμe/CMS 0 100 200 300 400 500 600 700 800 900 ) [GeV]t(t T p 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 2 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 0 100 200 300 400 500 600 700 800 900 ) [GeV]t(t T p 0.6 0.8 1 1.2 1.4 P re d. D at a 0 20 40 60 80 100 120 140 160 310× E ve nt s / 0 .2 5 Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb parton level +jetsμe/CMS 0 0.5 1 1.5 2 2.5 t)y(t 0.6 0.8 1 1.2 1.4 P re d. D at a 0 20 40 60 80 100 120 140 160 180 200 310× E ve nt s / 0 .2 5 Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 0 0.5 1 1.5 2 2.5 t)y(t 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 8 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb parton level +jetsμe/CMS 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 ) [GeV]tM(t 0.6 0.8 1 1.2 1.4 P re d. D at a 210 310 410 510 610 E ve nt s / 8 0 G eV Data signaltt nonsignaltt Single t Multijet, DY/W+jets Exp. uncertainty (13 TeV)-135.8 fb particle level +jetsμe/CMS 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 ) [GeV]tM(t 0.6 0.8 1 1.2 1.4 P re d. D at a FIG. 7. Comparisons of the reconstructed distributions of pTðtt̄Þ (upper), jyðtt̄Þj (middle), and Mðtt̄Þ (lower) for the (left) parton- and the (right) particle-level measurements in data and simulation. The simulation of POWHEG+PYTHIA8 is used to describe the tt̄ production. The contribution of multijet, DY, and W boson plus jets background events is extracted from the data (cf. Sec. VII). Combined experimental (cf. Sec. IX) and statistical uncertainties (hatched area) are shown for the total predicted yields. The data points are shown with statistical uncertainties. The ratios of data to the predicted yields are given at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-10 affects the migration matrix. This dependence is reduced in the particle-level measurement, where no extrapolation to parton-level top quarks is needed. For the central results the migration matrices and the acceptances are taken from the POWHEG+PYTHIA8 simulation, and other simulations are used to estimate the uncertainties. The binning of the distributions is optimized based on the resolution in the simulation. The minimal bin widths are selected such that, according to the resolution, at least 50% of the events are reconstructed in the correct bin. As an example, the migration matrices for the parton- and particle-level mea- surements of pTðthÞ are shown in the right-hand plots of Fig. 8. For the measured parton-level distributions of any quantity we define the purity as the fraction of parton-level top quarks in the same bin at the detector level, the stability as the fraction of detector-level top quarks in the same bin at the parton level, and the bin efficiency as the ratio of the number of detector- to parton-level top quarks in the same bin. Similar parameters are defined for the particle-level distributions. The purity, stability, and bin efficiency are shown for the pTðthÞmeasurements in the left-hand plots of Fig. 8. These illustrate the improved agreement between the reconstructed and the unfolded quantities, as well as the reduced extrapolation in the particle-level measurement. To control the level of regularization, the iterative D’Agostini method takes the number of iterations as an input parameter. The initial distributions for the D’Agostini unfolding are taken from the POWHEG +PYTHIA8 simulation. The number of iterations is chosen such that the compatibility between a model and the unfolded data at either the parton or particle level is the same as the compatibility between the folded model and the data at detector level. The compatibilities are deter- mined by χ2 tests at each level that are based on all the available simulations and on several modified spectra obtained by reweighting the POWHEG+PYTHIA8 distribu- tions of pTðtÞ, jyðtÞj, pTðtt̄Þ, or pTðj1Þ before the detector simulation. The modified spectra are chosen such that the effect of the reweighting corresponds roughly to the observed differences between the data and the unmodified simulation at detector level. We have found that the number of iterations needed to fulfill the above criterion is such that a second χ2 test between the detector-level spectrum with its statistical 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T Detector level p 0 0.2 0.4 0.6 0.8 1 F ra ct io n Purity Stability Bin efficiency parton level +jetsμe/CMS Simulation 0 0.5 1 1.5 2 2.5 3 3.5 N or m al iz ed p er u nf ol de d bi n [% ] 1.32 0.49 0.23 0.15 0.09 0.07 0.05 0.04 0.03 0.04 0.05 0.05 1.06 1.74 0.90 0.45 0.28 0.20 0.16 0.13 0.11 0.12 0.12 0.14 0.47 0.76 1.86 1.12 0.50 0.30 0.23 0.18 0.15 0.13 0.14 0.15 0.18 0.26 0.63 2.10 1.35 0.49 0.27 0.19 0.16 0.14 0.13 0.14 0.06 0.08 0.16 0.62 2.43 1.55 0.47 0.22 0.14 0.11 0.10 0.10 0.02 0.03 0.05 0.13 0.71 2.79 1.67 0.43 0.15 0.10 0.10 0.07 0.01 0.01 0.02 0.04 0.12 0.85 3.10 1.61 0.32 0.12 0.05 0.04 0.01 0.01 0.01 0.01 0.04 0.13 1.10 3.76 1.74 0.32 0.12 0.04 0.00 0.00 0.00 0.00 0.01 0.02 0.09 1.10 3.80 1.62 0.28 0.04 0.00 0.00 0.00 0.00 0.00 0.01 0.02 0.10 1.28 3.57 1.22 0.09 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.02 0.13 1.47 3.62 0.57 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.02 0.07 0.94 2.55 1 2 3 4 5 6 7 8 9 10 11 12 ) at parton level h (t T Bins of p 1 2 3 4 5 6 7 8 9 10 11 12 ) at d et ec to r le ve l h (t T B in s of p +jets, parton levelμe/CMSSimulation 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T Detector level p 0 0.2 0.4 0.6 0.8 1 F ra ct io n Purity Stability Bin efficiency particle level +jetsμe/CMS Simulation 0 2 4 6 8 10 12 14 N or m al iz ed p er u nf ol de d bi n [% ] 8.10 1.88 0.89 0.51 0.30 0.18 0.12 0.11 0.08 0.09 0.07 0.09 4.91 9.79 3.18 1.49 0.85 0.53 0.38 0.29 0.22 0.21 0.21 0.24 2.35 3.82 10.01 3.36 1.33 0.79 0.54 0.40 0.33 0.32 0.26 0.29 1.08 1.31 3.21 10.53 3.44 1.13 0.65 0.46 0.33 0.30 0.25 0.29 0.42 0.51 0.76 3.08 11.11 3.62 0.93 0.51 0.34 0.26 0.22 0.25 0.17 0.21 0.29 0.50 3.32 11.40 3.74 0.77 0.36 0.27 0.21 0.17 0.08 0.09 0.13 0.19 0.38 3.62 11.25 3.43 0.56 0.22 0.16 0.14 0.04 0.05 0.07 0.10 0.16 0.35 4.06 12.31 3.68 0.58 0.25 0.15 0.02 0.02 0.03 0.04 0.06 0.10 0.22 3.46 11.57 3.67 0.46 0.15 0.01 0.01 0.01 0.02 0.03 0.04 0.07 0.19 3.65 10.68 3.02 0.22 0.01 0.01 0.01 0.01 0.02 0.02 0.04 0.06 0.23 3.95 11.49 1.81 0.00 0.01 0.01 0.01 0.01 0.02 0.02 0.04 0.07 0.14 2.56 10.83 1 2 3 4 5 6 7 8 9 10 11 12 ) at particle level h (t T Bins of p 1 2 3 4 5 6 7 8 9 10 11 12 ) at d et ec to r le ve l h (t T B in s of p +jets, particle levelμe/CMSSimulation FIG. 8. Migration studies of the (upper) parton- and (lower) particle-level measurements of pTðthÞ, extracted from the POWHEG +PYTHIA8 simulation. Left: purity, stability, and bin efficiency. Right: bin migrations between detector and parton (particle) level. The pT range of the bins can be taken from the left panels. Each column is normalized such that the sum of its entries corresponds to the percentage of reconstructed events in this bin at the parton (particle) level. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-11 uncertainty and the refolded spectrum with zero uncertainty exceeds a probability of 99.9%. The refolded spectrum is obtained by inverting the unfolding step. This consists of a multiplication with the response matrix and does not need any regularization. The algorithm needs between 4 and 56 iterations depending on the distribution. The numbers of iterations are higher for measurements with lower purities and stabilities of the migration matrices. This is the case for the measurements of pTðtlÞ and jyðtlÞj, whose resolutions are significant worse than those of pTðthÞ and jyðthÞj due to the missing neutrino information. For the two-dimensional measurements with n bins in one quantity and mi, i ¼ 1…n bins in the other the D’Agostini unfolding can be generalized using a vector of B ¼ P n i mi entries of the form: b1;1; b2;1…bn;1;… b1;m1 ; b2;m2 …bn;mn , with a corresponding B × B migration matrix. The number of iterations is optimized in the same way. In the measurements of jet kinematic properties, we do not unfold the measured spectra of each jet separately, but do correct for the effect of misidentified jets. The response matrix showing the migration among the identified jets is given in Fig. 9 for the measurements of the jet pT spectra. IX. SYSTEMATIC UNCERTAINTIES Several sources of experimental and theoretical system- atic uncertainty are considered. Uncertainties in the jet and p⃗miss T calibrations, pileup modeling, b identification and lepton selection efficiencies, and integrated luminosity fall into the first category. The total uncertainty in the jet energy calibration is the combination of 19 different sources of uncertainty and the jet-flavor-specific uncertainties [47], where the uncertainty for b jets is evaluated separately. For each uncertainty source the energies of jets in the simulation are shifted up and down. At the same time, p⃗miss T is recalculated accord- ingly to the rescaled jet energies. The recomputed back- grounds, response matrices, and acceptances are used to unfold the data. The observed differences between these and the original results are taken as an uncertainty in the unfolded event yields. The same technique is used to calculate the impact of the uncertainties in the jet energy resolution, the uncertainty in p⃗miss T not related to the jet energy calibration, the b identification, the pileup model- ing, and the lepton reconstruction and selection. The b identification efficiency in the simulation is corrected using scale factors determined from data [51]. These have an uncertainty of about 1–3% depending on the pT of the b jet. The effect on the measurement due to the uncertainty in the modeling of pileup in simulation is estimated by varying the average number of pileup events per bunch crossing by 4.6% [52] and reweighting the simulated events accordingly. The trigger, reconstruction, and identification efficien- cies of leptons are evaluated with tag-and-probe techniques using Z boson dilepton decays [45,46]. The uncertainties in the scale factors, which are used to correct the simulation to match the data, take into account the different lepton selection efficiencies in events with high jet multiplicities as in tt̄ events. The uncertainty in the lepton reconstruction and selection efficiencies depends on pT and η and is below 2% in the relevant phase-space region. The relative uncertainty in the integrated luminosity measurement is 2.5% [2]. Uncertainties in the choice of μr and μf , the combination of the matrix-element calculation with the PS, the modeling of the PS and hadronization, the top quark mass, and the PDFs fall into the second category of uncertainties. The effects of these theoretical uncertainties are estimated either by using the event weights introduced in Sec. II, or by using a Bins of jet type at detector level 0 0.2 0.4 0.6 0.8 1 F ra ct io n ) l (b T p ) h (b T p ) W1 (j T p ) W2 (j T p ) 1 (j T p ) 2 (j T p ) 3 (j T p ) 4 (j T p Purity Stability Bin efficiency +jets, particle levelμe/CMSSimulation Bins of jet type at particle level B in s of je t t yp e at d et ec to r le ve l 0 2 4 6 8 10 12 14 16 18 N or m al iz ed p er u nf ol de d bi n [% ] +jets, particle levelμe/CMSSimulation ) l (b T p ) l (b T p ) h (b T p ) h (b T p ) W1 (j T p ) W1 (j T p ) W2 (j T p ) W2 (j T p ) 1 (j T p ) 1 (j T p ) 2 (j T p ) 2 (j T p ) 3 (j T p ) 3 (j T p ) 4 (j T p ) 4 (j T p FIG. 9. Migration studies of the particle-level measurement of the jet pT spectra, extracted from the POWHEG+PYTHIA8 simulation. Left: purity, stability, and bin efficiency. Right: bin migrations between detector and particle level. On the axes the pT bins for each jet are shown. Each column is normalized in the way that the sum of its entries corresponds to the percentage of reconstructed events in this bin at the particle level. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-12 tt̄ signal simulation with varied settings. Again, the uncer- tainties are assessed using the recomputed backgrounds, response matrices, and acceptances to unfold the data. The scales μr and μf are varied up and down by a factor of two individually and simultaneously in the same directions. Afterwards, the envelope of the observed variations is quoted as the uncertainty. The uncertainty in the combination of the matrix- element calculation with the PS is estimated from an ≈40% variation of the hdamp parameter in POWHEG, normally set to hdamp ¼ 1.58mt. This variation has been found to be compatible with the modeling of jet multi- plicities in previous measurements at ffiffiffi s p ¼ 8 TeV [16]. To estimate the uncertainty in the PS, several effects have been studied and are assessed individually. The scale of the initial- (ISR) and final-state (FSR) radiation is varied up and down by a factor of 2 and ffiffiffi 2 p , respectively. These variations are motivated by the uncertainties in the PS tuning [25]. The effect of multiple parton interactions and the parametrization of color reconnection have been studied in Ref. [26] and are varied accordingly. In addition, we use a simulation with activated color reconnection of resonant decays. This enables the color reconnection of top quark decay products with other partons, which is not allowed in the default tune. The uncertainty in the b fragmentation function is taken from measurements at LEP experiments [53–55] and SLD [56], and the parametrization in PYTHIA8 is changed accordingly. Finally, the semileptonic branching fractions [41] of b hadrons are varied within their measured uncertainties. The effect due to the uncertainty in the top quark mass is estimated using simulations with altered top quark masses. We quote as the uncertainty the cross section differences observed for a top quark mass variation of 1 GeV around the central value of 172.5 GeV used in the default simulation. For the PDF uncertainty only the variation in the acceptance is taken into account, while variations due to migrations between bins can be neglected. It is calculated according to the uncertainties in the NNPDF30_nlo_ as_0118 [31] parametrization. In addition, the uncertainties obtained using the PDF sets derived with the strong coupling strength set at αs ¼ 0.117 and 0.119 are considered. 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 R el at iv e un ce rt ai nt y (13 TeV)-135.8 fb CMS parton level +jetsμe/ Combined Stat Jet energy Luminosity b tagging FSR scale Tune Lepton tm damph Color rec. ISR scale Others 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 R el at iv e un ce rt ai nt y (13 TeV)-135.8 fb CMS particle level +jetsμe/ Combined Stat Jet energy Luminosity b tagging Lepton tm damph Tune ISR scale Color rec. FSR scale Others 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 R el at iv e un ce rt ai nt y (13 TeV)-135.8 fb CMS parton level +jetsμe/ Combined Stat Jet energy Tune FSR scale tm Color rec. damph ISR scale Others 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 R el at iv e un ce rt ai nt y (13 TeV)-135.8 fb CMS particle level +jetsμe/ Combined Stat Jet energy damph tm Tune Color rec. ISR scale FSR scale Others FIG. 10. Relative uncertainties due to the individual sources in the absolute (upper) and normalized (lower) measurement of pTðthÞ at the parton level (left) and particle level (right). Sources whose impact never exceeds 1% are summarized in the category “Others.” The combination of the individual sources of jet energy uncertainty is labeled “Jet energy.” The combined uncertainty is the sum in quadrature of the statistical and all the systematic uncertainties. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-13 As an example, the sources of systematic uncertainty in the measurements of pTðthÞ, as well as the statistical and total uncertainty, are shown in Fig. 10. Among the experimental uncertainties, the dominant sources are the jet energy scale; lepton triggering, reconstruction, and iden- tification; and the b identification. In the parton-level measurement, the FSR scale is typically an important contribution to the systematic uncertainty. As an additional consistency test, we subtract the tt̄ background and unfold the data using the reweighted simulations that include all the differences in the measured distributions at detector level described in Sec. VIII. The differences between these unfolded distributions and the one obtained with the unmodified simulation are small compared to the uncertainties evaluated by the variations described above. X. DIFFERENTIAL CROSS SECTIONS AS FUNCTIONS OF OBSERVABLES OF THE TOP QUARK AND THE tt̄ SYSTEM The cross section σ in each bin is calculated as the ratio of the unfolded signal yield and the integrated luminosity. These are further divided by the bin width or the product of the two bin widths to obtain the single- or double- differential cross section results, respectively. To allow for a 2−10 1−10 1] -1 [p b G eV ) h (t T dp σd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y 5−10 4−10 3−10 2−10 ] -1 [G eV ) h (t T dp σd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 1 1.2 D at a T he or y 2−10 1−10 1] -1 [p b G eV ) l (t T dp σd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] l (t T p 0.8 1 1.2 1.4 D at a T he or y 5−10 4−10 3−10 2−10 ] -1 [G eV ) l (t T dp σd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] l (t T p 0.8 1 1.2 D at a T he or y FIG. 11. Absolute (left) and normalized (right) differential cross sections at the parton level as a function of pTðthÞ (upper) and pTðtlÞ (lower). The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx, and the NNLO QCDþ NLO EW calculations. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-14 precise comparison of the measured shapes with theoretical predictions, irrespective of the integrated cross section and its uncertainty, we also present normalized differential cross sections. For this purpose the absolute differential cross sections are divided by the normalizing cross sections σnorm, which are obtained for each measurement from the integration of the cross section over the measured one- or two-dimensional range. The uncertainties in the normalized distributions are evaluated using error propagation and include the correlations between uncertainties in the individual measurements and σnorm. For the statistical uncertainty the covariances are taken directly from the unfolding procedure. For each of the studied systematic uncertainties we assume a full correlation among all bins, while the various sources are assumed to be uncorrelated. The same assumptions about correlations of uncertainty sources are made for the calculation of the normalized theoretical predictions. The measured differential cross sections are compared to the predictions of POWHEG, combined with the PS simu- lations of PYTHIA8 and HERWIG++, and the tt̄ multiparton simulation of MG5_aMC@NLO+PYTHIA8 FxFx. In addition, several parton-level results are compared to calculations of tt̄ production with NNLO QCDþ NLO EW [1] accu- racy, where a top quark mass of 173.3 GeV [57] is used. For the calculations of the theoretical cross sections as functions of Mðtt̄Þ and rapidities the scales are set to μr ¼ μf ¼ ð1=4ÞðmTðtÞ þmTðt̄ÞÞ and the scales for the pT )| h |y(t 20 40 60 80 100 120 140 160 180 200 [p b] hσd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 h 0.9 1 1.1 D at a T he or y )| h |y(t 0.1 0.2 0.3 0.4 0.5 0.6 0.7 hσd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 h 0.95 1 1.05 D at a T he or y )| l |y(t 20 40 60 80 100 120 140 160 180 200 [p b] l σd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 l 0.9 1 1.1 1.2 D at a T he or y )| l |y(t 0.1 0.2 0.3 0.4 0.5 0.6 0.7 l ) d y( t ) d y( t ) d y( t ) d y( tσd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 l )y(t )y(t )y(t )y(t 0.9 0.95 1 1.05 1.1 D at a T he or y FIG. 12. Absolute (left) and normalized (right) differential cross sections at the parton level as a function of jyðthÞj (upper) and jyðtlÞj (lower). The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx, and the NNLO QCDþ NLO EW calculations. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-15 calculations are selected as ð1=2ÞmTðtÞ or ð1=2ÞmTðt̄Þ depending on the variable under consideration. The PDF parametrizations LUXqed_plus_PDF4LHC15_nnlo_100 [58] are used for these calculations. The uncertainties consider variations of the scales μr and μf . The particle- level results are compared to a prediction obtained with the Monte Carlo generator SHERPA [59] (v2.2.3) in combi- nation with OPENLOOPS [60]. The processes of tt̄ produc- tion with up to one additional parton are calculated at NLO QCD accuracy, and those with up to four additional partons are calculated at LO. These processes are merged and matched with the Catani–Seymour PS [61] based on the SHERPA default tune. For the scales we select μr ¼ μf ¼ 1 2 � mTðtÞ þmTðt̄Þ þ X partons pT � ; ð3Þ where the summation over partons includes the pT of all partons obtained from the fixed-order calculation. The NNPDF30_nlo_as_0118 [31] PDF parametrizations are used. Uncertainties in the predictions of SHERPA are evalu- ated by halving and doubling the scales of renormalization, factorization, resummation, and the initial- and final-state PS. In addition, the PDF uncertainties are taken into account. For the predictions of POWHEG+PYTHIA8 we evaluate all the theoretical uncertainties described in Sec. IX. Although the ) [GeV] h (t T p 3−10 2−10 1−10 1 ] -1 [p b G eV ) h (t T dp σd (13 TeV)-135.8 fb particle level +jetsµe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y ) [GeV] h (t T p 5−10 4−10 3−10 2−10 ] -1 [G eV ) h (t T dp σd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsµe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.9 1 1.1 1.2 1.3 D at a T he or y ) [GeV] l (t T p 3−10 2−10 1−10 1 ] -1 [p b G eV ) l (t T dp σd (13 TeV)-135.8 fb particle level +jetsµe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] l (t T p 0.8 1 1.2 1.4 D at a T he or y ) [GeV] l (t T p 4−10 3−10 2−10 ] -1 [G eV ) l (t T dp σd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsµe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] l (t T p 1 1.2 D at a T he or y FIG. 13. Absolute (left) and normalized (right) differential cross sections at the particle level as a function of pTðthÞ (upper) and pTðtlÞ (lower). The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++) and the multiparton simulations MG5_aMC@NLO (MG5)+PYTHIA8 FxFx and SHERPA. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-16 SHERPA and the POWHEG+PYTHIA8 simulations are normal- ized to the NNLO tt̄ production cross section, we use their intrinsic scale uncertainties. The comparisons between the measurements and the theoretical predictions as a function of the top quark pT and jyj are shown in Figs. 11–12 and 13–14 for the parton and particle level, respectively. At parton level, the kinematic properties of th and tl are identical, and we measure the differential cross section as a function of the top quark pT using the higher- and lower-pT values in the tt̄ pair, as shown in Fig. 15. The measured pT spectra of th and tl are consistently softer than predicted by all the simulations using the PYTHIA8 PS generator at both the parton and particle levels. Also the NNLO QCDþ NLO EW calcu- lation predicts a slightly harder pT spectrum than observed in the data. The POWHEG+HERWIG++ simulation describes the data well at the parton level. However, at the particle level, the pTðthÞ distribution is noticeably softer than in the data. In Figs. 16 and 17, the cross sections as a function of kinematic variables of the tt̄ system are compared to the same theoretical predictions. In general, the predictions are in agreement with the measured distributions. The NNLO QCDþ NLO EW calculation predicts a higher-average Mðtt̄Þ spectrum than observed in the data. For POWHEG +HERWIG++ a similar behavior as for the pTðthÞ distribu- tions is observed—while the parton-level distribution is )| h |y(t 10 20 30 40 50 60 70 80 [p b] hσd (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 h 0.8 1 1.2 D at a T he or y )| h |y(t 0.2 0.4 0.6 0.8 1 hσd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 h 0.95 1 1.05 1.1 D at a T he or y )| l |y(t 10 20 30 40 50 60 70 [p b] l σd (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 l 0.8 1 1.2 D at a T he or y )| l |y(t 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 l σd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.5 1 1.5 2 2.5 l ) d y( t )y(t ) d y( t )y(t ) d y( t )y(t ) d y( t )y(t 0.9 1 1.1 D at a T he or y FIG. 14. Absolute (left) and normalized (right) differential cross sections at the particle level as a function of jyðthÞj (upper) and jyðtlÞj (lower). The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++) and the multiparton simulations MG5_aMC@NLO (MG5)+PYTHIA8 FxFx and SHERPA. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-17 well described, a softer spectrum is observed at the particle level. The measurement of double-differential cross sections allows for the study of correlations between kinematic proper- ties of the top quarks and provides insights into extreme regions of the phase space. The most fundamental double- differential distribution is that of the top quark properties jyðthÞj vs. pTðthÞ. The absolute double-differential cross sections are shown in Figs. 18 and 20, and the normalized inFigs. 19and21at thepartonandparticle levels, respectively. The observation of a softer pTðtÞ spectrum is persistent in all rapidity regions. In Figs. 22–25, the corresponding measure- ments as a function of Mðtt̄Þ vs. jyðtt̄Þj are shown. This distribution is sensitive to the PDFs [11]. AsMðtt̄Þ increases, the simulations overestimate the cross sections at high jyðtt̄Þj. Finally,wemeasurepTðthÞvs.Mðtt̄Þ, as showninFigs.26–29. For these distributions the simulations of POWHEG+PYTHIA8, MG5_aMC@NLO+PYTHIA8 FxFx, and SHERPA predict similar shapes, which differ substantially from the data. The precision of the differential cross section measure- ments is limited by the systematic uncertainty, dominated by jet energy scale uncertainties on the experimental side and PS modeling and scale uncertainties on the theoretical side. As expected, the theoretical uncertainties are reduced in the particle-level measurements since they are less dependent on theory-based extrapolations. (t) [GeV] T high p 2−10 1−10 1] -1 [p b G eV ) hi gh (t T dp σd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] high (t T p 0.8 1 1.2 1.4 D at a T he or y (t) [GeV] T high p 4−10 3−10 2−10 ] -1 [G eV ) hi gh (t T dp σd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] high (t T p 0.8 1 1.2 D at a T he or y (t) [GeV] T soft p 3−10 2−10 1−10 1] -1 [p b G eV ) lo w (t T dp σd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] low (t T p 0.8 1 1.2 1.4 D at a T he or y (t) [GeV] T soft p 5−10 4−10 3−10 2−10 ] -1 [G eV ) lo w (t T dp σd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] low (t T p 0.8 1 1.2 1.4 D at a T he or y FIG. 15. Absolute (left) and normalized (right) differential cross sections at the parton level as a function of the transverse momentum of the top quark with the higher and lower pT. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), and the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-18 ) [GeV]t(t T p 3−10 2−10 1−10 1 10 ] -1 [p b G eV )t (t T dp σd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 900 1000 ) [GeV]t(t T p 1 1.5 D at a T he or y ) [GeV]t(t T p 5−10 4−10 3−10 2−10 ] -1 [G eV )t (t T dp σd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 900 1000 ) [GeV]t(t T p 0.8 1 1.2 1.4 D at a T he or y )|t|y(t 20 40 60 80 100 120 140 160 180 200 220 240 [p b] t) d y( tσd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.9 1 1.1 1.2 D at a T he or y )|t|y(t 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 t) d y( tσd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t)y(t )y(t 0.9 1 1.1 D at a T he or y ) [GeV]tm(t 3−10 2−10 1−10 1 ] -1 [p b G eV )t dM (tσd (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 500 1000 1500 2000 2500 ) [GeV]tM(t 0.6 0.8 1 1.2 1.4 D at a T he or y ) [GeV]tm(t 5−10 4−10 3−10 2−10 ] -1 [G eV )t dM (tσd no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP NNLO QCD+NLO EW H++OWHEGP MG5 P8 [FxFx] 500 1000 1500 2000 2500 ) [GeV]tM(t 0.6 0.8 1 1.2 D at a T he or y FIG. 16. Absolute (left) and normalized (right) differential cross sections at the parton level as a function of pTðtt̄Þ (upper), jyðtt̄Þj (middle), and Mðtt̄Þ (lower). The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx, and the NNLO QCDþ NLO EW calculations. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-19 ) [GeV]t(t T p 3−10 2−10 1−10 1 ] -1 [p b G eV )t (t T dp σd (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 900 1000 ) [GeV]t(t T p 0.5 1 1.5 D at a T he or y ) [GeV]t(t T p 5−10 4−10 3−10 2−10] -1 [G eV )t (t T dp σd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 900 1000 ) [GeV]t(t T p 1 1.5 D at a T he or y )|t|y(t 20 40 60 80 100 [p b] tσd (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8 1 1.2 1.4 D at a T he or y )|t|y(t 0.2 0.4 0.6 0.8 1 1.2 tσd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t ) d y( t )y(t ) d y( t )y(t 0.9 1 1.1 1.2 D at a T he or y ) [GeV]tm(t 3−10 2−10 1−10 1 ] -1 [p b G eV )t dM (tσd (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 500 1000 1500 2000 2500 ) [GeV]tM(t 0.8 1 1.2 D at a T he or y ) [GeV]tm(t 5−10 4−10 3−10 2−10 ] -1 [G eV )t dM (tσd no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 500 1000 1500 2000 2500 ) [GeV]tM(t 0.8 1 1.2 D at a T he or y FIG. 17. Absolute (left) and normalized (right) differential cross sections at the particle level as a function of pTðtt̄Þ (upper), jyðtt̄Þj (middle), and Mðtt̄Þ (lower). The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++) and the multiparton simulations MG5_aMC@NLO (MG5)+PYTHIA8 FxFx and SHERPA. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-20 We evaluate the level of agreement between the mea- sured differential cross sections and the various theoretical predictions using χ2 tests. In these tests, we take into account the full covariance matrices of the measured differential cross sections. For the POWHEG+PYTHIA8 and the SHERPA predictions we consider the theoretical uncer- tainties and their correlation among the bins. In addition, we perform the χ2 tests without any uncertainties in the theoretical models. We do this since the generation of simulated events does not include any of these theoretical uncertainties; the simulated distributions of the various kinematic quantities correspond exactly to the central predictions. Therefore, it is worthwhile to compare how well the central predictions agree with the data, independent of the theoretical uncertainties. From the χ2 values and the numbers of degrees of freedom, which correspond to the numbers of bins in the distributions, the p-values are calculated. The results are shown in Table I for the parton-level and in Table III for the particle-level absolute measurements. The corresponding χ2 tests for the normalized distributions, for which the numbers of degrees of freedom are reduced by 1, are given in Tables II and IV for the parton- and particle-level measurements, respectively. The χ2 tests show that the measurements are largely compatible with the POWHEG+PYTHIA8 and SHERPA pre- dictions if the uncertainties in the simulations are taken into account. Some tension is observed between the data and the y binning 2−10 1−10 1 ] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 D at a T he or y y binning 2−10 1−10 1] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y y binning 3−10 2−10 1−10 1] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y y binning 3−10 2−10 1−10 1 ] -1 [p b G eV ) h (t T h ) d p d y( t ) < 0.5 0 < y(t ) d p d y( t ) < 1 0.5 < y(t ) d p d y( t ) d p d y( t σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS h ) < 1.5 1 < y(t ) < 2.5 1.5 < y(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y FIG. 18. Double-differential cross section at the parton level as a function of jyðthÞj vs. pTðthÞ. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), and the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-21 predictions of pTðtÞ and related distributions like pTðthÞ vs. Mðtt̄Þ. Comparisons of the p-values at the parton and particle level obtained for the central predictions, ignoring their theoretical uncertainties, show a similar performance. For all tested models we obtain p-values below 1% for at least two distributions. These are typically distributions related to pTðtÞ and pTðtt̄Þ. XI. MEASUREMENTS OF MULTIPLICITIES AND KINEMATIC PROPERTIES OF JETS In the following, we discuss the measurements involving the multiplicities and kinematic properties of jets in tt̄ events. These are performed at the particle level only. In the POWHEG simulations, all jets beyond one additional jet are described by the PS simulation and, hence, their description is subject to PS tuning. In the SHERPA simulation, the production of up to one additional jet is calculated at NLO accuracy, and up to four jets at LO. However, these LO calculations are very sensitive to the choice of the scales. Since in the MG5_aMC@NLO+PYTHIA8 FxFx simulation up to two additional jets are calculated at NLO, it is expected to be more accurate at high jet multiplicities. The absolute and normalized differential cross sections as a function of pTðthÞ, Mðtt̄Þ, and pTðtt̄Þ for different numbers of additional jets are shown in Figs. 30–35. These distributions are helpful to estimate the tt̄ background contribution in searches for physics beyond the standard y binning 5−10 4−10 3−10 2−10 ] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.9 1 1.1 1.2 D at a T he or y y binning 5−10 4−10 3−10 2−10] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 1 1.2 1.4 D at a T he or y y binning 5−10 4−10 3−10 2−10 ] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 D at a T he or y y binning 6−10 5−10 4−10 3−10 ] -1 [G eV ) h (t T h ) d p d y( t ) d p d y( t ) d p d y( t ) d p d y( t σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS h ) < 0.5 0 < y(t ) < 1 0.5 < y(t ) < 1.5 1 < y(t ) < 2.5 1.5 < y(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y FIG. 19. Normalized double-differential cross section at the parton level as a function of jyðthÞj vs. pTðthÞ. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), and the multiparton simulation MG5_aMC@NLO (MG5) +PYTHIA8 FxFx. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-22 model that are looking for signatures with high jet multiplicities. The observation that the pTðtÞ distribution is softer in data than in the simulations is mainly true for events with zero or one additional jet. We also measure the properties of the individual jets in tt̄ events. The absolute and normalized differential cross sections as a function of the pT of jets in the tt̄ system and of the four leading additional jets are shown in Figs. 36 and 37, respectively. The trend of a softer pT spectrum of the top quark is also visible for all jets of the tt̄ system. From these pT distributions we calculate the jet multiplic- ities with minimum pT thresholds of 30, 50, 75, and 100 GeV shown in Fig. 38, and gap fractions [16,17]. The gap fraction fnðpTÞ is the fraction of unfolded events that contain less than n additional jets above the given pT threshold. It is shown for n ¼ 1 and 2 in Fig. 39. In the calculations of jet multiplicities and gap fractions, we take into account the small fraction of jets above the displayed y binning 3−10 2−10 1−10 1 ] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y y binning 3−10 2−10 1−10 1 ] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y y binning 3−10 2−10 1−10 ] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.8 1 1.2 1.4 D at a T he or y y binning 4−10 3−10 2−10 1−10 ] -1 [p b G eV ) h (t T h σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS h ) dp d y( t ) < 0.5 0 < y(t ) d p d y( t ) < 1 0.5 < y(t ) d p d y( t ) < 1.5 1 < y(t ) dp d y( t ) < 2.5 1.5 < y(t Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.5 1 1.5 D at a T he or y FIG. 20. Double-differential cross section at the particle level as a function of jyðthÞj vs. pTðthÞ. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++) and the multiparton simulations MG5_aMC@NLO (MG5)+PYTHIA8 FxFx and SHERPA. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-23 pT ranges. The uncertainties are obtained by error propa- gation using the full covariance matrices. The jet multi- plicities and gap fractions are reasonably described by most of the simulations. However, the central predictions of SHERPA and POWHEG+HERWIG++ show noticeable devia- tions in the gap fraction. In Figs. 40–45, the absolute and normalized distribu- tions of jηj, ΔRjt , and ΔRt are shown for the jets in the tt̄ system and the additional jets. The differential cross section as a function of jηj is well described by most of the simulations, while POWHEG+HERWIG++ overestimates the radiation of additional jets close to the jets in the tt̄ system. In the predictions, such collinear radiation is mainly described by the PS model. Since the parton-level prediction is not affected by the simulation of the final- state PS, this overestimation of radiation may explain the discrepancies between the parton- and particle-level predictions of POWHEG+HERWIG++ in the pTðthÞ and Mðtt̄Þ distributions. Table V presents the results of the χ2 tests comparing the absolute measurements involving multiplicities and kin- ematic properties of jets to the simulations. The corre- sponding results for the normalized measurements are shown in Table VI. Most of the kinematic distributions and multiplicities of the additional jets are reasonably well modeled by POWHEG+PYTHIA8. Inconsistencies with the data are observed for pT and η of jets, and pTðtt̄Þ for different jet multiplicities. The POWHEG descriptions of y binning 4−10 3−10 2−10] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.9 1 1.1 1.2 D at a T he or y y binning 5−10 4−10 3−10 2−10] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.9 1 1.1 1.2 1.3 D at a T he or y y binning 5−10 4−10 3−10 2−10 ] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS h Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 1 1.2 D at a T he or y y binning 6−10 5−10 4−10 3−10] -1 [G eV ) h (t T h σ2 d no rm σ 1 (13 TeV)-135.8 fb particle level +jetsμe/CMS h ) d p d y( t ) < 0.5 0 < y(t ) d p d y( t ) < 1 0.5 < y(t ) d p d y( t ) < 1.5 1 < y(t ) d p d y( t ) < 2.5 1.5 < y(t Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 100 200 300 400 500 600 700 800 ) [GeV] h (t T p 0.5 1 1.5 D at a T he or y FIG. 21. Normalized double-differential cross section at the particle level as a function of jyðthÞj vs. pTðthÞ. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++) and the multiparton simulations MG5_aMC@NLO (MG5)+PYTHIA8 FxFx and SHERPA. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-24 additional jets rely on phenomenological models of the PS and are substantially different for PYTHIA8 and HERWIG++. With the selected settings SHERPA fails to describe most of the kinematic distributions and multiplicities of the jets. Comparisons of the measurements to the central predic- tions, ignoring their theoretical uncertainties, show that the p-values are typically below 1% for all models. Here the multiparton simulation MG5_aMC@NLO+PYTHIA8 FxFx performs best. All cross section values, together with their statistical and systematic uncertainties, are listed in Appendices A and B for the parton- and particle-level measurements, respec- tively. In addition, the corresponding normalized cross sections are provided in Appendices C and D. y binning 0.1 0.2 0.3 0.4 0.5 0.6 ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 450 GeVt300 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8 0.9 1 1.1 D at a T he or y y binning 0.1 0.2 0.3 0.4 0.5 ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 625 GeVt450 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 1 1.2 1.4 D at a T he or y y binning 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 850 GeVt625 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8 1 1.2 1.4 1.6 D at a T he or y y binning 2 4 6 8 10 12 14 3−10× ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 2000 GeVt850 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t ) ) d y( t )y(t ) ) d y( t )y(t ) ) d y( t )y(t ) ) d y( t )y(t 1 1.5 2 D at a T he or y FIG. 22. Double-differential cross section at the parton level as a function ofMðtt̄Þ vs. jyðtt̄Þj. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), and the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. MEASUREMENT OF DIFFERENTIAL CROSS SECTIONS … PHYS. REV. D 97, 112003 (2018) 112003-25 y binning 0.5 1 1.5 2 2.5 3−10× ] -1 [G eV t t dM (t σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 450 GeVt300 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.9 0.95 1 1.05 D at a T he or y y binning 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 3−10× ] -1 [G eV t t dM (t σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 625 GeVt450 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.9 1 1.1 1.2 1.3 D at a T he or y y binning 0.1 0.2 0.3 0.4 0.5 0.6 3−10× ] -1 [G eV t t dM (t σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 850 GeVt625 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8 1 1.2 1.4 D at a T he or y y binning 10 20 30 40 50 6−10× ] -1 [G eV t t dM (t σ2 d no rm σ 1 (13 TeV)-135.8 fb parton level +jetsμe/CMS ) < 2000 GeVt850 < M(t Data stat⊕Sys Stat P8OWHEGP H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t ) ) d y( t )y(t ) ) d y( t )y(t ) ) d y( t )y(t ) ) d y( t )y(t 1 1.5 D at a T he or y FIG. 23. Normalized double-differential cross section at the parton level as a function of Mðtt̄Þ vs. jyðtt̄Þj. The data are shown as points with light (dark) bands indicating the statistical (statistical and systematic) uncertainties. The cross sections are compared to the predictions of POWHEG combined with PYTHIA8 (P8) or HERWIG++ (H++), and the multiparton simulation MG5_aMC@NLO (MG5)+PYTHIA8 FxFx. The ratios of the various predictions to the measured cross sections are shown at the bottom of each panel. A. M. SIRUNYAN et al. PHYS. REV. D 97, 112003 (2018) 112003-26 y binning 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS ) < 450 GeVt300 < M(t Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8 1 1.2 D at a T he or y y binning 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS ) < 625 GeVt450 < M(t Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8 1 1.2 1.4 1.6 D at a T he or y y binning 10 20 30 40 50 60 70 80 90 3−10× ] -1 [p b G eV t t dM (t σ2 d (13 TeV)-135.8 fb particle level +jetsμe/CMS ) < 850 GeVt625 < M(t Data stat⊕Sys Stat P8OWHEGP CSHERPAS H++OWHEGP MG5 P8 [FxFx] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 t 0.8