在质子-质子碰撞中，在四个b夸克最终状态下寻找衰变成一对希格斯玻色子的共振。资源-CSDN文库

Open

Access

98 浏览量 2020-03-21 06:03:07 上传评论收藏 1.49MB PDF 举报

资源推荐

资源详情

资源评论

Physics Letters B 781 (2018) 244–269

Contents lists available at ScienceDirect

Physics Letters B

www.elsevier.com/locate/physletb

Search for a massive resonance decaying to a pair of Higgs bosons in

the four b quark ﬁnal state in proton–proton collisions at

√

s = 13 TeV

.The CMS Collaboration



CERN, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history:

Received

13 October 2017

Received

in revised form 23 March 2018

Accepted

29 March 2018

Available

online 4 April 2018

Editor: M.

Doser

Keywords:

CMS

Physics

Extradimensions

Graviton

Radion

di-Higgs

boson resonance

A search for a massive resonance decaying into a pair of standard model Higgs bosons, in a ﬁnal state

consisting of two b quark–antiquark pairs, is performed. A data sample of proton–proton collisions

at a centre-of-mass energy of 13 TeV is used, collected by the CMS experiment at the CERN LHC in

2016, and corresponding to an integrated luminosity of 35.9 fb

−1

. The Higgs bosons are highly Lorentz-

boosted

and are each reconstructed as a single large-area jet. The signal is characterized by a peak in

the dijet invariant mass distribution, above a background from the standard model multijet production.

The observations are consistent with the background expectations, and are interpreted as upper limits on

the products of the s-channel production cross sections and branching fractions of narrow bulk gravitons

and radions in warped extra-dimensional models. The limits range from 126 to 1.4 fb at 95% conﬁdence

level for resonances with masses between 750 and 3000 GeV, and are the most stringent to date, over

the explored mass range.

(http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP

1. Introduction

In the standard model (SM), the pair production of Higgs

bosons (H) [1–3]in proton–proton (pp) collisions at

√

s = 13 TeV

a rare process [4]. However, the existence of massive resonances

decaying to Higgs boson pairs (HH) in many new physics mod-

els

may enhance this rate to a level observable at the CERN LHC

using the current data. For instance, models with warped extra

dimensions (WED) [5] contain new particles such as the spin-0 ra-

dion [6–8] and

the spin-2 ﬁrst Kaluza–Klein (KK) excitation of the

graviton [9–11], which have sizeable branching fractions to HH.

The WED models have an extra spatial dimension compactiﬁed

between two branes, with the region between (called the bulk)

warped via an exponential metric κl, κ being the warp factor and

l the coordinate of the extra spatial dimension [12]. The reduced

Planck scale (M

≡ M

/8π, M

being the Planck scale) is con-

sidered

a fundamental scale. The free parameters of the model are

κ/M

and the ultraviolet cutoff of the theory 

≡

√

−κl

[6].

In pp collisions at the LHC, the graviton and the radion are pro-

duced

primarily through gluon–gluon fusion and are predicted to

decay to HH [13].

Other scenarios, such as the two-Higgs doublet models [14]

(in

particular, the minimal supersymmetric model [15]) and the



E-mail address: cms -publication -committee -chair @cern .ch.

Georgi–Machacek model [16]predict spin-0 resonances that are

produced primarily through gluon–gluon fusion, and decay to an

HH pair. These particles have the same Lorentz structure and effec-

tive

couplings to the gluons and, for narrow widths, result in the

same kinematic distributions as those for the bulk radion. Hence,

the results of this paper are also applicable to this class of models.

Searches for a new particle Xin the HH decay channel have

been performed by the ATLAS [17–19] and CMS [20–24] Collabo-

rations

in pp collisions at

√

s = 7 and 8TeV. More recently, the

ATLAS Collaboration has published limits on the production of a

KK bulk graviton, decaying to HH, in the bbbb ﬁnal state, using pp

collision

data at

√

s = 13 TeV, corresponding to an integrated lumi-

nosity

of 3.2 fb

−1

[25]. Because the longitudinal components of the

W and Z bosons couple to the Higgs ﬁeld in the SM, a resonance

decaying to HH potentially also decays into WW and ZZ, with a

comparable branching fraction for X → ZZ, and with a branching

fraction for X →WW that is twice as large. Searches for X → WW

and

ZZ have been performed by ATLAS and CMS [26–35].

This letter reports on the search for a massive resonance de-

caying

to an HH pair, in the bbbb ﬁnal state (with a branching

fraction ≈33% [36]), performed using a data set corresponding to

35.9 fb

−1

of pp collisions at

√

s = 13 TeV. The search signiﬁcantly

improves upon the CMS analysis performed using the LHC data

collected at

√

s = 8TeV[24], and extends the searched mass range

to 750–3000 GeV. This search is conducted for both the radion

https://doi.org/10.1016/j.physletb.2018.03.084

0370-2693/

SCOAP

The CMS Collaboration / Physics Letters B 781 (2018) 244–269 245

and the graviton, whereas the earlier search only considered the

former.

In this search, the X → HH decay would result in highly

Lorentz-boosted and collimated decay products of H → bb, which

are referred to as Hjets. These are reconstructed using jet sub-

structure

and jet ﬂavour-tagging techniques [37–39]. The back-

ground

consists mostly of SM multijet events, and is estimated

using several control regions deﬁned in the phase space of the

masses and ﬂavour-tagging discriminators of the two Hjets, and

the HH dijet invariant mass, allowing the background to be pre-

dicted

over the entire range of m

explored. The signal would

appear as a peak in the HH dijet invariant mass spectrum above

a smooth background distribution.

2. The CMS detector and event simulations

The CMS detector with its coordinate system and the relevant

kinematic variables is described in Ref. [40]. The central feature

of the CMS apparatus is a superconducting solenoid of 6m in-

ternal

diameter, providing a magnetic ﬁeld of 3.8 T. Within the

ﬁeld volume are silicon pixel and strip trackers, a lead tungstate

crystal electromagnetic calorimeter (ECAL), and a brass and scin-

tillator

hadron calorimeter (HCAL), each composed of a barrel and

two endcap sections. The tracker covers a pseudorapidity η range

from −2.5to 2.5 with the ECAL and the HCAL extending up to

|η| = 3. Forward calorimeters in the region up to |η| = 5pro-

vide

almost hermetic detector coverage. Muons are detected in

gas-ionization chambers embedded in the steel ﬂux-return yoke

outside the solenoid, covering a region of |η| < 2.4.

Events

of interest are selected using a two-tiered trigger sys-

tem [41].

The ﬁrst level (L1), composed of custom hardware pro-

cessors,

uses information from the calorimeters and muon detec-

tors

to select events at a rate of around 100 kHz. The second level,

known as the high-level trigger (HLT), consists of a farm of pro-

cessors

running a version of the full event reconstruction software

optimized for fast processing, and reduces the event rate to around

1kHz before data storage. Events are selected at the trigger level

by the presence of jets of particles in the detector. The L1 trigger

algorithms reconstruct jets from energy deposits in the calorime-

ters.

At the HLT, physics objects (charged and neutral hadrons,

electrons, muons, and photons) are reconstructed using a particle-

ﬂow

(PF) algorithm [42]. The anti-k

algorithm [43,44]is used to

cluster these objects with a distance parameter of 0.8 (AK8 jets) or

0.4 (AK4 jets).

Bulk

graviton and radion signal events are simulated at leading

order using the MadGraph5_amc@nlo 2.3.3 [45]event generator

for masses in the range 750–3000 GeV and widths of 1 MeV (nar-

row

width approximation). The NNPDF3.0 leading order parton

distribution functions (PDFs) [46], taken from the LHAPDF6 PDF

set [47–50], with the four-ﬂavour scheme, is used. The showering

and hadronization of partons is simulated with pythia 8.212 [51].

The herwig++ 2.7.1 [52] generator is used for an alternative model

to evaluate the systematic uncertainty associated with the parton

shower and hadronization. The tune CUETP8M1-NNPDF2.3LO [53]

used for pythia 8, while the EE5C tune [54]is used for her-

wig++.

The background is modelled entirely from data. However, sim-

ulated

background samples are used to develop and validate the

background estimation techniques, prior to being applied to the

data. These are multijet events, generated at leading order using

MadGraph5_amc@nlo,

and tt + jets, generated at next-to-leading

order using powheg 2.0 [55–57]. Both these backgrounds are inter-

faced

to pythia 8 for simulating the parton shower and hadroniza-

tion.

Studies using simulations established that the multijet com-

ponent is more than 99% of the background, with the rest mostly

from tt + jets production.

All generated samples were processed through a Geant4-based

[58,59] simulation of the CMS detector. Multiple pp collisions may

occur in the same or adjacent LHC bunch crossings (pileup) and

contribute to the overall event activity in the detector. This effect

is included in the simulations, and the samples are reweighted to

match the number of pp interactions observed in the data, assum-

ing

a total inelastic pp collision cross section of 69.2 mb [60].

3. Event selection

Events were collected using several HLT algorithms. The ﬁrst

required the scalar p

sum of all AK4 jets in the event (H

) to

be greater than 800 or 900 GeV, depending on the LHC beam in-

stantaneous

luminosity. A second trigger criterion required H

≥

650 GeV, with a pair of AK4 jets with invariant mass above

900 GeV and a pseudorapidity separation |η| < 1.5. A third set

of triggers selected events with the scalar p

sum of all AK8 jets

greater than 650 or 700 GeV and the presence of an AK8 jet with

a “trimmed mass” above 50 GeV, i.e. the jet mass after removing

remnants of soft radiation using jet trimming technique [61]. The

fourth triggering condition was based on the presence of an AK8

jet with p

> 360 GeV and trimmed mass greater than 30 GeV.

The last trigger selection accepted events containing two AK8 jets

having p

> 280 and 200 GeV with at least one having trimmed

mass greater than 30 GeV, together with an AK4 jet passing a loose

b-tagging criterion.

The

pp interaction vertex with the highest



of the as-

sociated

clusters of physics objects is considered to be the one

associated with the hard scattering interaction, the primary vertex.

The physics objects are the jets, clustered using the jet ﬁnding al-

gorithm [43,44]with

the tracks assigned to the vertex as inputs,

and the associated missing transverse momentum, taken as the

negative vector sum of the p

of those jets. The other interaction

vertices are designated as pileup vertices.

mitigate the effect of pileup, particles are assigned weights

using the pileup per particle identiﬁcation (PUPPI) algorithm [62],

with the weight corresponding to its estimated probability to orig-

inate

from a pileup interaction. Charged particles from pileup ver-

tices

receive a weight of zero while those from the primary vertex

receive a weight of one. Neutral particles are assigned a weight be-

tween

zero and one, with higher values for those likely to originate

from the primary vertex. Particles are then clustered into AK8 jets.

The vector sum of the weighted momenta of all particles clustered

in the jet is taken to be the jet momentum. To account for detec-

tor

response nonlinearity, jet energy corrections are applied as a

function of jet η and p

[63,64]. In each event, the leading and the

subleading p

AK8 jets, j

and j

, respectively, are required to have

> 300 GeV and |η| < 2.4.

The removal of events containing isolated leptons (electrons or

muons) with p

> 20 GeV and |η| < 2.4helps suppress tt + jets

and

diboson backgrounds. The isolation variable is deﬁned as the

scalar p

sum of the charged and neutral hadrons, and photons

in a cone of R = 0.3for an electron or R = 0.4for a muon,

where R ≡



(η)

+(φ)

, φ being the azimuthal angle in ra-

dians.

The energy from pileup deposited in the isolation cone, and

the p

of the lepton itself, is subtracted [65,66]. The isolation re-

quirement

removes jets misidentiﬁed as leptons. Additional quality

criteria are applied to improve the purity of the isolated lepton

samples. Electrons passing combined isolation and quality criteria

corresponding to a selection eﬃciency of 90% (70%) are designated

“loose” (“medium”) electrons. For the “loose” (“medium”) muons,

the total associated eﬃciency is 100% (95%). The probability of a

jet to be misidentiﬁed as an electron or a muon is in the range

246 The CMS Collaboration / Physics Letters B 781 (2018) 244–269

0.5–2%, depending on p

, η, and the choice of medium or loose se-

lection

criteria. Events containing one medium lepton, or two loose

leptons of the same ﬂavour, but of opposite charge, are rejected.

The H → bbsystem is reconstructed as a single high-p

AK8

jet, where the decay products have merged within the jet, and

the two highest p

jets in the event are assumed to be the Higgs

boson candidates. The jet is groomed [67]to remove soft and

wide-angle radiation using the soft-drop algorithm [68,69], with

the soft radiation fraction parameter z set to 0.1 and the angu-

lar

exponent parameter β set to 0. The groomed jet is used to

compute the soft-drop jet mass, which peaks at the Higgs boson

mass for signal events and reduces the mass of background quark-

and

gluon-initiated jets. Dedicated mass corrections [70], derived

from simulation and data in a region enriched with ttevents with

merged W → qqdecays, are applied to the jet mass in order to

remove residual dependence on the jet p

, and to match the jet

mass scale and resolution observed in data.

The soft-drop masses of j

and j

are required to be within the

range 105–135 GeV, with an eﬃciency of about 60–70%, for jets

arising from a signal of mass m

in the range 750–3000 GeV. The

“N-subjettiness” algorithm is used to determine the consistency of

the jet with two subjets from a two-pronged H → bbdecay, by

computing the inclusive jet shape variables τ

and τ

[71]. The

ratio τ

≡ τ

/τ

with a value much less than one indicates a jet

with two subjets. The selection τ

< 0.55 is used, having a jet

-dependent eﬃciency of 50–70%, before applying the soft-drop

mass requirement.

For background events, j

and j

are often well separated in

η, especially at high invariant mass (m

) of j

and j

. In contrast,

signal events that contain a heavy resonance decaying to two en-

ergetic

Hjets are characterized by a small separation of the two

jets in η. Events are therefore required to have a pseudorapidity

separation |η(j

, j

)| < 1.3.

The

eﬃciency of the trigger combination is measured in a sam-

ple

of multijet events, collected with a control trigger requiring

a single AK4 jet with p

> 260 GeV, and with the leading and

the subleading p

AK8 jets, j

and j

, respectively, passing the

above selections on p

, η, and the soft-drop mass. The eﬃciency

is greater than 99% for m

≥ 1100 GeV, and in the range 40–99%

for 750 < m

< 1100 GeV. The trigger eﬃciency of the simulated

samples is corrected using a scale factor to match the observed ef-

ﬁciency

in the data. This scale factor is applied as a function of

|η(j

, j

)| because it has a mild dependence on this variable.

The main method to suppress the multijet background is btag-

ging:

since a true H → bbjet contains two b hadrons, the Hjet

candidates are identiﬁed using the dedicated “double-b tagger” al-

gorithm [72].

The double-b tagger exploits the presence of two

hadronized bquarks inside the Hjet, and uses variables related

to b hadron lifetime and mass to distinguish between Hjets and

the background from multijet production; it also exploits the fact

that the b hadron ﬂight directions are strongly correlated with the

axes used to calculate the N-subjettiness observables. The double-b

tagger

is a multivariate discriminator with output between −1

and 1,

with a higher value indicating a greater probability for

the jet to contain a bbpair. The double-b tagger discriminator

thresholds of 0.3 and 0.8 correspond to Hjet tagging eﬃcien-

cies

of 80 and 30% and are referred to as “loose” (L) and “tight”

(T) requirements, respectively. Events must have the two leading

AK8 jets satisfying the loose double-b tagger requirement. The

data-to-simulation scale factor for the double-b tagger eﬃciency

is measured in an event sample enriched in bbpairs from gluon

splitting [72], and applied to the signals to obtain the correct sig-

nal

yields.

The main variable used in the search for a HH resonance is

the “reduced dijet invariant mass” m

jj,red

≡ m

− (m

− m

) −

Fig. 1. The soft-drop mass (upper), the N-subjettiness τ

(middle), and the double-b

tagger

discriminator (lower) distributions of the selected AK8 jets. The multijet

background components for the different jet ﬂavours are shown: jets having two

B hadrons (bb) or a single one (b), jets having a charm hadron (c), and all other

jets (light). Also plotted are the distributions for the simulated bulk graviton and

radion signals of masses 1400 and 2500 GeV. The number of signal and background

events correspond to an integrated luminosity of 35.9 fb

−1

. A signal cross section

σ (pp → X → HH → bbbb) = 20pb is assumed for all the mass hypotheses. The

events are required to have passed the trigger selection, lepton rejection, the AK8

jet kinematic selections p

> 300 GeV and |η| < 2.4, and |η(j

, j

)| < 1.3. The

reduced dijet invariant mass m

jj,red

is required to be greater than 750 GeV. The N-

subjettiness

requirement of τ

< 0.55 is applied to the upper and lower ﬁgures.

The soft-drop masses of the two jets are between 105–135 GeV for the middle and

lower ﬁgures.

− m

), where m

and m

are the soft-drop masses of the

leading and subleading H-tagged jets in the event, and m

125.09 GeV [73,74]is the Higgs boson mass. The quantity m

jj,red

is used rather than m

since by subtracting the soft-drop masses

of the two H-tagged jets and adding back the exact Higgs boson

mass m

, ﬂuctuations in m

and m

due to the jet mass resolu-

The CMS Collaboration / Physics Letters B 781 (2018) 244–269 247

Fig. 2. The jet separation |η(j

, j

)| (left) and the reduced dijet invariant mass m

jj,red

(right) distributions. The multijet background components for the different jet ﬂavours

are shown: events containing at least one jet with two B hadrons (bb) or a single one (b), events containing a jet having a charm hadron (c), and all other events (light). Also

plotted are the distributions for the simulated bulk graviton and radion signals of masses 1400 and 2500 GeV. The numbers of signal and background events correspond to

an integrated luminosity of 35.9 fb

−1

. The signal cross section σ(pp → X → HH → bbbb) is assumed to be 20 pb for all the mass hypotheses. The events are required to have

passed the online selection, lepton rejection, the AK8 jet kinematic selections p

> 300 GeV, |η| < 2.4. The soft-drop masses of the two jets are between 105 and 135 GeV,

and the N-subjettiness requirement of τ

< 0.55 and m

jj,red

> 750 GeV are applied. The m

jj,red

distributions (right) require |η(j

, j

)| < 1.3.

tion are corrected, leading to 8–10% improvement in the dijet mass

resolution. A requirement of m

jj,red

> 750 GeV is applied for select-

ing

signal-like events.

The soft-drop mass, τ

, and double-b tagger discriminator dis-

tributions

of the two leading p

jets are shown in Fig. 1 for sim-

ulated

events after passing the online selection, lepton rejection,

kinematic selection, and the requirement m

jj,red

> 750 GeV. Also,

the N-subjettiness requirement of τ

< 0.55 is applied for the

soft-drop mass and the double-b tagger distributions, while the

soft-drop mass requirement is applied to the τ

, and double-b

tagger

discriminator distributions. Since some of the triggers im-

pose

a trimmed jet mass requirement, this affects the shape of the

oﬄine soft-drop jet mass, resulting in a steep rise above ∼20 GeV.

The distributions of the |η(j

, j

)| and the m

jj,red

variables are

shown in Fig. 2. In these ﬁgures, the multijet background is shown

for different jet ﬂavour categories: jets having two B hadrons (bb)

or a single one (b), jets having a charm hadron (c), and all other

jets (light).

The double-b tagger discriminator of the two leading AK8 jets

must exceed the loose threshold. In addition, if both discrimina-

tor

values also exceed the tight threshold, events are classiﬁed in

the “TT” category. Otherwise, they are classiﬁed in the “LL” cate-

gory,

which contains events with both j

and j

failing the tight

threshold as well as events with either j

or j

passing the tight

threshold while the other passes the loose threshold only.

The

backgrounds are estimated separately for each category,

and the combination of the likelihoods for the TT and LL categories

gives the optimal signal sensitivity over a wide range of resonance

masses, according to studies performed using simulated signal and

multijet samples. The TT category has a good background rejection

for m

up to 2000 GeV. At higher resonance masses, where the

background is small, the LL category provides better signal sensi-

tivity.

The full event selection eﬃciencies for bulk gravitons and

radions of different assumed masses are shown in Fig. 3. The ra-

dion

has a smaller eﬃciency than the bulk graviton because its

|η(j

, j

)| distribution is considerably wider than that of a bulk

graviton of the same mass, as shown in Fig. 2 (left).

4. Signal and background modelling

The method chosen for the background modelling depends on

whether the resonance mass m

is below or above 1200 GeV, since

at low masses the background does not fall smoothly as a func-

Fig. 3. The signal selection eﬃciencies for the bulk graviton and radion models for

different mass hypotheses of the resonances, shown for the LL and the TT signal

event categories. Owing to the large sample sizes of the simulated events, the sta-

tistical

uncertainties are small.

tion of m

jj,red

, because of the trigger requirements, while above

1200 GeV it does. The background estimation relies on a set of

control regions to predict the total background shape and nor-

malization

in the signal regions. The entire range of the m

jj,red

distribution above 750 GeV is used for the prediction.

For

signals with m

≥ 1200 GeV, the underlying background

distribution falls monotonically with m

jj,red

, thus allowing the

background shape to be modelled by a smooth function, above

which a localized signal is searched for. This smooth background

modelling helps to reduce uncertainties in the background estima-

tion

from local statistical ﬂuctuations in m

jj,red

, thereby improving

the signal search sensitivity. The parameters of the function and

its total normalization are constrained by a simultaneous ﬁt of the

signal and background models to the data in the control and the

signal regions. For m

≥ 1200 GeV, the m

jj,red

distributions for the

signal are modelled using the sum of a Gaussian and a Crystal Ball

function [75], as shown in Fig. 4 for one signal category. The same

modelling is used for the other signal categories, with different pa-

rameters

for the Gaussian and the Crystal Ball functions.

The

signal and control regions are deﬁned by two variables re-

lated

to the leading p

jet j

: (i) its soft-drop mass m

and (ii) the

value of the discriminator of the double-b tagger. The background

剩余25页未读，继续阅读

评论收藏

内容反馈

weixin_38732519

粉丝: 2
资源: 951

在质子-质子碰撞中，在四个b夸克最终状态下寻找衰变成一对希格斯玻色子的共振。

在质子-质子碰撞中搜索sphaleron

在s = 8 $$ \ sqrt {s} = 8 $$ TeV的质子-质子碰撞中寻找强子最终态中衰变成希格斯玻色子和W或Z玻色子的大规模共振

在s = 13 $$ \ sqrt {s} = 13 $$ TeV的质子-质子碰撞中，使用大面积射流搜索四夸克最终态中希格斯玻色子对的产生

搜索在s = 8 $$ \ sqrt {s} = 8 $$ TeV的质子-质子碰撞中衰变成魅力和底部夸克的带电希格斯玻色子

搜索在质子-质子碰撞中，在$$ \ sqrt {s} = 13 \，\ text {Te} \ text {V} $$ s = 13Te的质子-质子碰撞中与希格斯玻色子衰变成一对底夸克相关的暗物质的搜索

寻找退化的希格斯玻色子

寻找类似希格斯玻色子的轻质风味的衰变

使用ATLAS探测器在s = 13 TeV时使用36 fb-1质子-质子碰撞数据搜索重共振衰减为玻色子和轻子最终状态的搜索组合

在13 TeV的质子-质子碰撞中寻找重共振衰变成两个希格斯玻色子或希格斯玻色子和W或Z玻色子

在13 TeV质子-质子碰撞中搜索希格斯玻色子衰变到底部夸克-反夸克对的共振对产生

在质子与质子碰撞的电子或介子碰撞下，以s $$ \ sqrt {\ mathrm {s}} $$ = 13 TeV搜寻带电的希格斯玻色子衰变成上下夸克

搜索质子-质子碰撞中质子-质子碰撞的矢量样夸克衰变到顶夸克和$$ \ mathrm {W} $$ W玻色子的单次生成= 13 \，\ text {TeV} $ $ s = 13TeV

搜索衰变为b夸克和希格斯玻色子的类矢量夸克的单项生成

搜索质子-反质子碰撞中与底夸克相关的希格斯粒子

使用ATLAS探测器在pp碰撞下pp碰撞的bb¯τ+τ-衰减通道中搜索共振和非共振希格斯玻色子对产生

搜索在s = 13 TeV的质子-质子碰撞中衰减成夸克对的成对共振

使用ATLAS探测器在质子-质子碰撞的完全轻子最终状态下寻找共振产生

在s = 13 TeV的质子-质子碰撞中搜索窄Hγ共振

最新资源