The simulated samples used in the Jet Vertex Charge calibration analy- sis described in Chapter 4 are used in this analysis as well; for this rea- son, only additional samples and different treatments will be described
in the following. For further details, “my twenty-five readers”2are re-
ferred to Section 4.2.
All simulated events are processed through the full simulation of
the ATLAS detector based on GEANT4. Samples generated with a
fast simulation software, for which the calorimeter response is replaced by a parametrization of the shower shapes, are used to estimate mod- elling systematic uncertainties. Simulated events are subsequently re- constructed using the same software also used for data.
1Both figures of merit are useful in this context, as the usual significance expression, S/√B,
is only valid under the assumption that systematic uncertainties are small compared to the sta- tistical ones, which is not necessarily the case for regions containing a large number of events, where a small systematic uncertainty can have an effect on the total yields as big as the expected signal contribution.
The top quark mass is fixed to the value of mt =172.5 GeV and the EvtGen program is used to model the properties of bottom and charm
hadron decays for all the non-SHERPAsamples.
5.2.1 Signal samples
The t¯tH signal process is described via simulated samples produced
with MADGRAPH5_aMC@NLO, for the generation of the hard-scatter
event at NLO accuracy in QCD, interfaced with PYTHIA8 for the simu-
lation of the parton shower (PS) and hadronization processes.
The decay of the top quarks is done with the MADSPIN [165] soft-
ware in such a way that the spin correlations are preserved and the Higgs
boson mass in the simulation is set to be mH =125 GeV, with all of the
decay modes considered.
The signal cross section, σt¯tH, is equal to 507+35−50 fb, taken from cal-
culations up to NLO in QCD and including NLO electroweak correc- tions [166–171].
5.2.2 t¯t + jets background modelling
The nominal t¯t + jets process is generated inclusively in all its subcom-
ponents, namely t¯t + ≥ 1b, t¯t + ≥ 1c and t¯t+ light, using POWHEG
interfaced with PYTHIA8. Given that these subcomponents populate
different regions of the selected phase space, the inclusive t¯t + jets sam- ple is subdivided into these three main components, which are treated as separate samples.
The categorization is done according to the flavour of additional parti- cle jets not originating from the t¯t system. Particle jets are reconstructed
with the same anti-kt algorithm (R = 0.4) as calorimeter jets, but by
clustering stable truth particles3, with the exclusion of muons and neu-
trinos. The kinematic selection for particle jets requires them to have a
pT greater than 15 GeV and |η| less than 2.5.
If a jet is matched to exactly one b-hadron with pT >5 GeV, the jet
is labelled as single b-jet, whereas jets matched with more than one b- hadron are labelled B-jets, such as jets originating from gluon splitting
3Particles with a mean lifetime τ > 3 · 10−11s are considered as stable particles, as they are
into a b¯b pair with small angular separation. Single c- and C-jets are de- fined in a similar way, but considering jets that are not already matched to one or more b-hadrons.
Events that contain at least one single b- or B-jet, excluding the ones coming from the top or W boson decays, are labelled as t¯t + ≥ 1b; while events with no extra b-jets but with at least one additional c-jet are labelled t¯t + ≥ 1c. Events labelled as either t¯t + ≥ 1b or t¯t + ≥ 1c are generically referred to as t¯t+HF (HF stands for “heavy flavour”), whereas events without heavy flavour jets are labelled as t¯t+ light.
A finer classification is also provided: if an event has exactly two single b-jets, the event is labelled t¯t+b¯b, those with one single b-jet are called t¯t+ b and those with exactly one B-jet are labelled t¯t+ B events. Finally, remaining events enter in the t¯t + ≥ 3b events. Events with additional b-jets coming from multi-parton interactions (MPI) or final- state radiation (FSR), i.e. originated from gluon radiation from the top quark decay products are considered in a separate category.
A second t¯t sample, which has a great impact on the flow of the
analysis, is represented by the SHERPA+OPENLOOPS NLO t¯t + b¯b
sample [79, 147], referred in the following as SHERPA4F, as only the
lightest four flavour quarks are considered massless. It represents the state-of-the-art of the theoretical knowledge of the t¯t+ b¯b process and is expected to provide the most accurate estimate of this process. This sample is used to reweight the fractions of the various subcategories of
the t¯t + ≥ 1b background as predicted by POWHEG+PYTHIA8, in order
to improve its already good description of observed data.
This is possible because the description of the kinematics of the two additional b-jets is done at the NLO precision in QCD, taking the b- quark mass into account. As a matter of fact, considering massive b-quarks and massless light-quarks and gluons affects the balance be- tween the various t¯t + jets production modes: the gluon splitting g → b¯b dominates over the production of two b-quarks in the initial state.
In Figure 5.3 is shown a comparison of the predicted fractions of
the various sub-categories of the t¯t + ≥ 1b for the POWHEG+PYTHIA8
and SHERPA4F samples. The t¯t+ b MPI/FSR sub-category, accounting
for 10% of the events in POWHEG+PYTHIA8 t¯t + ≥ 1b sample, is not
uncertainties on the SHERPA4F prediction is derived by varying tune parameters, renormalization and factorization scales, as well as PDF
sets for the SHERPA4F sample. A detailed source of the uncertainties
entering the the band is given in Section 7 of Ref. [172].
tt+b tt+bb tt+B tt+≥3b Fraction of events 2 − 10 1 − 10 1 POWHEG+PYTHIA 8 4F HERPA S
ATLAS Simulation Preliminary
tt + b tt + bb tt + B tt + ≥3b 8 YTHIA +P OWHEG P 4F HERPA S 0.5 1 1.5 2
Figure 5.3: Comparison of the relative predicted fractions of the t¯t+b, t¯t+b¯b, t¯t + B and t¯t + ≥ 3b sub-categories, before any event selection, for the
POWHEG+PYTHIA8 sample and the SHERPA4F. The fractions are normal-
ized to the sum of the four contributions present in both generators, i.e. without considering the t¯t+ b (MPI/FSR) sub-category as part of the total. The uncertainty band is derived with the procedure described in Section 7 of Ref. [172].
Four additional t¯t samples are generated to assess the modelling of the
t¯t system. Three of them, namely the POWHEG+Herwig7 and the two
POWHEG+PYTHIA8 with increased and reduced radiation in the final state, have already been described in Section 4.2.1 and they will not be
described again. A sample generated with SHERPAand interfaced with
OPENLOOPS, which considers the bottom-quarks massless, referred to
as SHERPA5F in the following, is used to assess the matrix element
generator. It should not be confused with the SHERPA4F sample used
5.2.3 Other backgrounds
Other processes rather than t¯t + jets can enter in the analysis regions with different yields depending on the jet and b-jet multiplicity of the region. Such background processes are taken from simulation with cor- rections applied to them, with the exception of the fakes and non-prompt lepton contribution, which in the SL channel is estimated using the data- driven Matrix Method described in Section 4.2.1. For the paper analy- sis, in the three most sensitive SR in the SL channel, the expected fake lepton background represents a minor contribution, compatible with zero and hence neglected. In the DIL channel this background is esti- mated from simulation, but normalized to data in a dedicated same-sign lepton region.
The single top, W /Z+jets and the diboson backgrounds are estimated using the same MC samples described in Section 4.2.1. For Z+jets events, an additional correction is applied to the normalization of the heavy-flavour component by scaling up the contribution by a factor 1.3, extracted from dedicated control regions with a definition close to the signal regions, but requiring the two leptons with opposite charge and same flavour to have an invariant mass close to the Z mass.
Samples of t¯tW and t¯tZ, referred collectively as t¯tV , are generated
using MADGRAPH5_aMC@NLO interfaced with PYTHIA8.
The production of four top quarks in the final state, t¯tt¯t, and the pro- duction of a t¯t pair in association with a W boson pair, t¯tWW , was gen-
erated with MADGRAPH5_aMC@NLO with LO accuracy and inter-
faced with PYTHIA8. On the other hand, tZ events were still produced
with MADGRAPH5_aMC@NLO with LO accuracy, but interfaced with
the PYTHIA6. Finally, MADGRAPH5_aMC@NLO samples interfaced
with PYTHIA8 are used to describe the tZW process at NLO accuracy.
The associated production of the Higgs boson with a single top quark is very small in the SM, but is nevertheless included in the analysis and treated as background. The tW H production is modelled via samples
generated with MADGRAPH5_aMC@NLO interfaced with Herwig++,
whereas samples describing the tHqb production mode were produced
with MADGRAPH5_aMC@NLO interfaced with PYTHIA8 at LO accu-
racy. Other Higgs boson production modes are negligible and hence not considered in the analysis.