Supersymmetry: Experimental Status

4.2 MB·

by  Keith A. Ulmer

Supersymmetry: Experimental Status
Keith A. Ulmer,
on behalf of
6
1 the CMS and ATLAS Collaborations
0
2
Texas A&M University
n
Department of Physics and Astronomy
a
J College Station, TX, 77843, USA.

4
1 January 18, 2016

]
x
e
Abstract
-
p
This talk presents results from the CMS and ATLAS Collaborations
e
fromsearchesforphysicsbeyondtheStandardModelmotivatedbysuper-
h
symmetryfromRun1oftheLHC.Representativesearchesaredescribed
[
toillustratethediversenatureofthesearchprograminbothbackground

1 estimationtechniquesandfinalstatetopologies. Thestatusofpreparation
v
for Run 2 searches at 13 TeV is also presented.
4
7
7 1 INTRODUCTION
3
0
. The Standard Model of particle physics accurately describes the interactions of
1
all known fundamental particles in the universe, and has remained the prevail-
0
ing paradigm in the field for over forty years. Despite its success, the Standard
6
1 Model remains an incomplete theory of fundamental particles and interactions.
: It does not include a description of gravity, nor does it explain the compelling
v
astronomicalevidencefordarkmatter. OftheproposedextensionstotheStan-
i
X dardModel, supersymmetry(SUSY)hasremainedamongthemostpopularfor
r decades. It provides exactly the needed compensation to stabilize the Higgs
a
mass, while additionally providing an ideal candidate for dark matter with a
stable weakly interacting lightest supersymmetric particle (LSP).
The CMS [1] and ATLAS [2] experiments at the CERN Large Hadron Col-
lider are general purpose detectors built to explore the fundamental nature of
the universe. Among the results from the two experiments are many searches
for supersymmetry, which have thus far yielded null results [3, 4]. The search
programs in both experiments are based on a wide arrange of techniques to
measure standard model background contributions as well as a diverse range
of possible final states. In this talk, a sample of results are shown to illustrate
1
techniques deployed in these searches. By no means are all relevant results
discussed or presented.
In the first section, a series of general searches in different final states are
described. The second section contains a discussion of more targeted searches
focused on dedicated final state topologies, while the third section discussed
difficult to reach signatures. The forth section attempts to put the full set of
searches performed into a global context, while the final section shows progress
toward new searches in the LHC Run 2 with 13 TeV proton-proton collisions.
2 INCLUSIVE SEARCHES
Unlike the Standard Model Higgs boson, supersymmetry has many free param-
eters,whichcangiverisetoagreatvarietyofsignatures. Further,theunknown
mass spectrum can also give rise to a great variety of production cross sections
and final state kinematics. With such a broad range of possible signatures, a
fruitful class of supersymmetry searches is performed with inclusive sensitivity.
Here I describe four such examples in different final states.
The classic jets plus missing energy signature is searched for in a three
dimensional binned analysis taking advantage of sensitivity in different bins
of missing energy (MET), the sum of jet transverse momenta (HT), and the
number of jets tagged as bottom quarks [5]. Events are selected by removing
those with an identified electron or muon and then requiring at least three jets
and at least one tagged b-quark jet. The main Standard Model backgrounds
derive from tt¯, W + jets, Z + jets, and QCD multijet events. The contribution
from each category of background is measured from data control samples to
minimizetherelianceonaccuratesimulation. Inparticular,singleleptonevents
are used to predict the tt¯and W + jets backgrounds, and dilepton events are
usedtopredicttheZ +jetsbackgroundswithZ decaystoneutrinos. TheQCD
multijet contribution is predicted by utilizing a kinematic sideband enriched in
QCD events where a jet and MET are aligned.
No significant excess of events above the Standard Model predictions is ob-
served. Fig. 1 (left) shows the data compared to the expected Standard Model
contribution after selecting events with at least 3 b-quark jets. The search re-
sults are interpreted in several benchmark SUSY models, including gluino pair
productionwitheachgluinodecayingtotwob-quarksandtheLSP.Fig.1(right)
shows that such models are excluded for gluino masses as high as around 1.2
TeV.
Asimilarlybroadsearchwasperformedwithcomplementaryeventsselected
with exactly one muon or electron in [6]. Minimal requirements on MET and
the transverse mass (MT) of the lepton and MET are used to select SUSY-
like events. Sensitivity to a variety of models is obtained by classifying search
regions into large and small jet multiplicity. Backgrounds arise predominately
from W + jets and tt¯events. The size of the Standard Model contributions
are predicted by identifying data control regions enriched in each background.
MC simulation is then used with the overall normalization taken from the data
2
CMS, L = 19.4 fb-1, s = 8 TeV
CMS , L = 19.4 fb-1, s = 8 TeV
Events / 12 GeV110023 ZL, Nb-jet ‡ 3 D(T(TtWSt6111ia02n+bbt02gjabbe,5le bbt5,s b bt01o05p)0 )G GeVeV (GeV)m~0c111802000000pp fiOE x~bgp~gsee, cr ~gvt eefidd – –b 1b1 s s~ce10txh pe eoNrriymLeOnt+NLL exclusion 1 s section (pb)
QCD s
o
Z+jets cr
10 Diboson 600 10-1mit on
400 er li
1 10-2upp
Data/MC 21 150 200 250 300 350 4E0mT0iss 4(5G0eV5)00 200 5% CL
0 10-39
0 400 600 800 1000 1200 1400
150 200 250 300 350 400 450 500
EmTiss (GeV) m~g (GeV)
Figure 1: MET distribution in hadronic search with at least 3 b-jets (left) and
limV its from the search for gluino tVo bbLSP production (right). Taken from
RcFoEvents / 50 Geiegnf.t.T11100r02[23ho5le](Ah3(n.sa-Tlrjo=ereMLd 8etuA Tf1pTgSRLpteE eVei)r,/ o µE2T Tm0nwi.s3s) ifdbtt-1oihstpdrriaebtduaiDSTVFD(g~1tg~oat+cai0ab kpjt12icenaoe -t5tQds ossL,aoo tu5eerna4dpnpsrt5m t,kM o,ms hn6o(s5dg~)fep, e Glχo∼e±1,aV rχ∼b01r)=aeacdksitgg~or) [GeV]tm(nEvents / 50 Geo1tau67890h00000111l1n00000000e0234r~dg-e~gb AAh3(pgnHAslia-lTr asjo=oalerni=LiTmd 8drtu8dcoA OE u iLW1ptT sxcbT1SkLp penAetsaR-eie eeloVteeV crrgaS9nv,p/t ,e 5µ,Ee t2 2d%owr dc~gTm0 0 nl→ liioC.s imh m3fsobL) ii ittntu-f ct 1(bl(+±y±-sh1χ1∼n1 01σ σi,e tSgdhxmUpeSofi)(rYnyχ∼~)g01 ve)a→= t sec∼mχl0 tf(1~otr)irob DSTVFD(g~i-1dmg~oat+ ai0edab2 kpjtre12enaoe0n -5tQdgs ss L,aoG tua5eernami4dppesr5 t,tkoVM o,ms n6oi(s5nodg~o)e, Glχ∼r.ne±1,Ve χ∼.01)=jNetos sisigsnhifiowcanint
J
signal is observed in any of the search500regions and 95% CL upper limits aJre set H
on a1variety of simplified models. Fig.12 (right) shows the limits for gluinHo pair E
pbreData / SMoydo2nu1dcti1o.n3 wTeitVhfboorthligghltuiLnSoPssd.ecayiData / SMn34g000021intotwotopquarkswithlimitsreaEPching P0
1050 200 250 300 350 400 450 500ETmis5s5 [0GeV60]0 2004100500 205000250603000730500 480000450900500E1Tm0is50s5 0[0Ge1V610]000 1200m1(30~g0) 0[Ge1V4]00 4(
Events / 50 GeV 11110000234 Ah5(nsa-Tjo=erLd 8tuA T1pTSRLpe eVer,/ µE2 Tm0i.s3s) fb-1 DSTVFD(g~1g~oat+ai0ab kpjt12enaoe -5tQds ssL,ao tu5eerna4dppsr5 t,kM o,ms n6o(s5dg~)e, Glχ∼e±1,V χ∼01)= ∼0) [GeV]m(χEvents / 50 GeV1156789000000011100000010000234~g-~g HAApAh5(slnalsra- =TToljor=ier8mdLdd 8Ltu OEuiA T 1t W1psxTcAb-eS pLpltsaeeViReeotS eVp ce,r9nr vtt,2/e5 o,e µE d%02dn~g Tm 0l → lfioCimibm.s3nsL -i1)ittl t ty f(χ∼(b±±01-11,1~ gσ σme→tShx( Upe~tttSo))∼rχY 0y> )1f>or bmidd(~gen)DSTVFD(g~1g~oat+ai0ab kpjt12enaoe -5tQds ssL,ao tu5eerna4dppsr5 t,kM o,ms n6o(s5dg~)e, Glχ∼e±1,V χ∼01)= 4(2015) 2015)116
400
1 3001 1
SM 2 SM2002 1
Data / 1 Data / 1001 6
1050 200 250 300 350 400 450 500 550 600 5100500 206000250730000 385000400904050 150000055011060001200 1300 1400
ETmiss [GeV] ETmiss [GeV] m(~g) [GeV]
F(utTsFbhpeieogaipagtuekbtrruoraecemcrEnhke)8Tgmrt.fi(2ts¯orslu:D(oenlcefimduMsftttt)r),oiEfRbaianuniTnnedtdtiifdoecW.ndraet[ris+eo6stedfjt]seit.rstbuhsiye6bl(6trtmu%sihgiethsf(istr7aoi)no2rnrcg%moow)tnfrto,faFtta9drtohon5rouiihvre%sgedlsiurvsetettrtincCreomheuhaeorLgayetspstelh,1ileeeoiet83wxfioana.-Wceeimpptjlsdx9uherpps5ostmuqc%etilmvuhrisohteei+aeeCdmte¯ldrdyenlLkeil(fvg,ntiWewiojhiutngn˜xaaetimcrtl→t˜)hl(tttuay→hhsgsgnectEeillt¯deooulclχ˜osncTmnχihu˜bw01net01olaiaoi)(sircm(-rbsdnnokmbtdoiogilatittseoprnntdftsor)rordriiueomaosnoittngmtrbhgi)deoilnwuodle-t.deohhtnrtin-Tieetc3oloelraheptyh-aentjgepaLniesttsnrtqhhditd.dgouSeeaoosnart(g8ikPrnhrtnTlyk1ueocgbih%hplupnl±efuepnrao)-o1oelcsne(σdreparpda4nruortanvut5ceeoasatdtr%dlnhirai.inhoiestcatu)nd5syhitTuee,-ialcosfmjwihno.iennntineherteTdneoeislrhnothecieonotntlhodwhtubeehescse(mseattoh(rryievmpedvieegxig˜esdaeq,xcmnhlunpuaoet˜e1st(rmcxiχ˜kvt)Lp01ieen).eidl)scyatplaltielsmlhdaismrnuioltemiisumtfgeaoiidhsttr
the5-jettt¯(W)controlregion; the5-jetWshowcnonbtyraolsorliedgdioanrkirsedclriones,s-wciothntthaemdianraktreedddboyttetdt¯lienveesnintdsicaattingthe 1σvariationonthis
±
thelevelof40%. The“Data/SM”plotsslhimowitdtuheetorathteiothoeofredtaictaalstcoaletahnedsPuDmFmunecdertSatinatniedsaorndthMesoidgneallcrosssection.
expectation, which is derived from the fit described in section 9. The uncertainty band on the
Standard Model expectation shown here combines the statistical uncertainty on the simulated
event samples with the relevant systematic uncertainti3es (see text). The last bin includes the
overflow. The “Top Quarks” label includes all top-quark-related backgrounds, while “V+jets”
includesW+jets,Z+jetsandotherDrell-YanbackgroundssuchasZ→τ+τ−–a4n5d–γ∗/Z outside
the Z pole region. For illustration, the expected signal distributions are shown for gluino pair
productionwithmg˜=1025GeV,mχ˜±1 =545GeVandmχ˜01=65GeV.
–24–
Next, a search was performed based on events with two electrons or muons
with the same electric charge [7]. Such events are rare in the Standard Model,
butcanoccurreadilyinmanynewphysicssignatures. Theleptonsarerequired
to be well isolated to select prompt leptons from W or Z decays and remove
those associated with jets, for example from semi-leptonic b decays. The main
backgroundsarisefromeventswithanon-promptleptonthatmistakenlypasses
the isolation criteria in addition to another prompt lepton or from events with
two true prompt leptons arising from such rare processes as diboson produc-
tion. The background from non-prompt leptons is determined by measuring
the so-called “fake rate” of the likelihood of a non-prompt lepton to pass the
isolation criteria in a data control sample, as shown in Fig. 3 (left). The con-
tributions from rare backgrounds are taken from simulation. As with the other
inclusive searches, signal regions are defined in a number of bins of MET, HT,
and number of b-tagged jets to ensure sensitivity to a variety of possible signal
models and parameter space. No significant excess of events is observed above
the expected Standard Model backgrounds. Fig. 3 (right) shows the observed
limit for sbottom pair production with each sbottom decaying to t, W, and the
LSP.
CMS s = 8 TeV, L = 19.5 fb-1
L ratio0.04.55 M u o n s CMS s = 8 TaawweVaayy, Ljjeeintt tpp =T >>1 02400.5 GG feebVV-1 (GeV)mLSP334050000 ppfiOE~bxb1p~bse1e*,cr ~bvte1efidd – –t W 11 s~c s01et xh peNeorrLiymOen+tNmLLc~01 i/nemtxc~c+1l u=s 0io.5n 103s section (fb)
T 0.4 away jet pT > 60 GeV 250 os
0.03.53 T 200 102mit on cr
0.02.52 150 per li
p
0.15 100 u
L
0.1 C
50 %
0.05 5
05 10 15 20 25 30 35 350 400 450 500 550 600 650 10 9
p (GeV) m (GeV)
T sbottom
Figure 3: Tight to loose isolation ratio used to measure non-prompt lepton
fake rate (left) and results from same-sign dilepton search for sbottom pair
production (right). Taken from Ref. [7].
The final inclusive search described in this talk is based on events with two
high p photons [8]. Such a signature is common in gauge mediated (GMSB)
T
SUSY scenarios where the LSP is a gravitino and the lightest neutralino decays
into a photon and the gravitino. Sensitivity to strong and electroweak produc-
tionisachievedwithsearchbinsinlowandhighMETandjetmultiplicity. The
mainbackgroundsarisefromcombinatorialdiphotonproductionandfrompho-
ton + jets events where one of the jets is mistaken as an isolated photon. The
missingenergyinsuchbackgroundeventsisgenerallytheresultofmismeasured
jets. Since the MET does not come from the photons themselves, data control
4
samplescomposedofeventsfromthephotonisolationsidebandscanbeusedto
measure the expected MET shape. The MET shape is then normalized to the
low MET region in the true two photon sample to predict the background in
the signal region. Fig. 4 (left) shows the MET distribution for a signal region
with MET > 200 GeV where the signal extends to higher MET than the re-
maining backgrounds. No significant excess of signal events is observed. Fig. 4
(right) shows the upper limits on gluino production in a GMSB scenario with a
beVino10-4likAeTLAnSeutras l=i 8n TeoV,. 20.3 fb-1 Data 2012 eV104 ATLAS s = 8 TeV, 20.3 fb-1 Data 2012
Events / 25 GeVEvents / 25 G1111111011000000-0011-2234231 ATLAS s = 8mmmm TWWWW~~~~e ====V SS6666,0000 RR20000γγWW0,,,,γγ mmmm.--HH3∼∼∼∼χχχχ 10101010 f====b -1515100000000 ZZDeeγγγγssWWγjγjtt→→γγ aaaγγ++γγtttγγaγγ ..jj jj ⊕⊕mm2 0ssiiss1yy--2ssIIDDtt.. [GeV]m~Events / 25 GeVEvents / 25 Gg11111134561111101100000000000000-0011-2234231 GAGMTL: AASbAslRiTnl S =oγSlLiγ-m l 8iAaki etTnsS ean aVleytu,s t92risa50li .n%3so ,f = bCg -l8L1ummmm inTWWWW~~~~oe ====pV SSr6666,o0000 RRd20000uγγWW0,,,,γγ cmmmm.--tLL3i∼∼∼∼χχχχo 10101010n f====b -1515100000000 EEOZZDeeγγγγssWWγjγjtt→→xxγγ abaaγγ++γγpptttsγγaγγ ..eejj e jj ⊕⊕mm2ccr 0vttssiissee1eyy--dd2ssIIdDDtt ..llliiimmmiiittt (((±±± 121 σσσEETSxxhUppeS..))oY ry)
ata/SM1100--2211 ata/SM111121000000--2421 m ∼0>χ 1m~g
D 0 D 0
ata/SM 210 100 200 300 EmT4is0s0 [GeV50]0 ata/SM10002400 201000 400200600300800EmT4i1s0s00 0[0GeV1520]000 mχ∼110 4[G0e0V]
FoFafinDgitguhdreuerrS50e0e:RsD�Wu�4islH:ttri1s(blM0ue0ftfiotoE)rnaTong2fd0ltuhd0SeRiinms�Wt�oi3srLs0iipb0n(rgruigoEttrhdamTit4n)oius0ssnsv0 ce[elGtFaCrenfisiLcgedoeotuiVi5tnrrmnohe0]etnho80Sie:GmenRaExer�ScxMec��ncaLeltlDubupuaSesmntsilaooBtilnwhyEvse0lTtmiie0hsmmiEesfisotuTmdsornfioibmsindsrrp�˜ot101etrhkhhe0ee<leqn0og8us(clu0tauiri0rrmonevioGgemnp2-.eblh0eVTeinn0.htsoseCt)eumooi.ramtbvasssrbiee3Tsvirlcn0vifpaanh.e0ltadgiknoOleaniEesm,slvilunmTioeg4tsrfsriies0nlngasfqag0r lriuue[atnGiohienlrexoemeahSmraVri5eRbne0ei]d�StRtn��eg0hbdHtieisnefaooonfrnam.tlhyae[ss(i8snsloa]femro.erfinetmxa)�lc˜01lSuU�deS8dY0a0mtG9o5de%Vel
expectedSM� backgroundsas�afunctionofcrEosTms-issse,ctsioenpaexrapetectdatiionnt,oasthweelvlaarsifooursacSoUnStYribcruotsisnsgecstioounricnecrse.asAedlsaonddecreasedbyonestandard
FosmhfiogotduwherenelsS5a.:rRTeD�ht�eihseltorisw(bilgueentrfitaop)lnlaoeontxsfdpstehShceoRtawm�t�iiotshnseisn(rrfgaiotgtrirhoatthn)oesfsvoe(emlbdarenessWc˜vdeeti,ri±amvomt2ienoo�˜sdn01mtea)odnxefad=cntatheatreup(d6t-mtcod0reot0thvEsh,iseaTem-1tsiieE0coscs0onmtif)mroioasnGsnbrgsiteryenheVsseeqteodusmafaiSntarhmdteMeicmpe(uxlmeeenpxnWces˜ptecu,retetirmcatdvsitn�ia˜elvit01tlmy)iif.on.itA=ng.lO.sa(oFlv6lose0rhrr0eoltqw,ah5iunen0isir0seea)tmrhpeGeleoetnehxtVtspes,ectedlimit,aswellasthe±1
theinnerbWanHdrepresentstheWranLgeofstatisticaluncertaintywhilTetheouterbandrepresentsthecombined
expectedSM� backgroundsas�afunctionofEmiss,separatedintothevariouscontributingsources. Also
s3tatisticaTlanAdsyRstemGaticEunTcertEaintDy.EveSntsEoTutAsideRtheCrangHeofEtheSdisplayedregionareincludedinthe
smhhiogodhweenlssta.-vrTeahltueheeldoswbigiennr.aplloetxspsehcotawtiothnesrfaotriothoefo(mbsW˜e,rvme�˜d01)da=ta(6to00th,e10c0o)mGbienVedaSndM(mexW˜p,emct�a˜01t)io=n.(F6o0r0t,h5e0s0e)pGloetVs,
theinnerbandrepresentstherangeofstatisticaluncertaintywhiletheouterbandrepresentsthecombined
In addition to generic inclusive searches, some SUSY signatures are sufficiently
statisticalandsystematicuncertainty.Eventsoutsidetherangeofthedisplayedregionareincludedinthe
hwigehlelstm-vaoluteidvbaint.ed to demand dedicated searches targeting a more specific model.
Heretwosuchtargetedsearchesaredescribed. Thefirstisforstaupairproduc-
tion, while the second targets direct stop production.
Direct stau pair production is well motivated, in part3i0cular by its potential
connection to cosmological scenarios to describe the early evolution of the uni-
verse. While generic dilepton searches are often sensitive to stau production
through the stau decays to electrons or muons, a dedicated search is required
26
to capture sensitivity to hadronic stau decays which have the largest branching
fraction. In [9] events with two hadronic tau candidates with opposite charge
areselected. Z bosoncandidatesar2e6vetoedtorejectZ toττ events,andevents
withab-taggedjetarerejectedtoremovett¯events. Theremainingbackground
is dominated by QCD multijet events. To select signal from this background, a
multivariate boosted decision tree (BDT) is trained and only events with high
BDT score are retained. After such selection, the main backgrounds remaining
are W + jets and diboson events. The W + jets background is measured by
identifyingadatacontrolsampleenrichedinW +jetsandnormalizingtheMC
simulation prediction to the yield in this control sample. Fig. 5 (left) shows a
plot of the MT2 distribution in this control sample. No significant excess of
5
events is observed about the background predictions and 95% CL upper limits
areset. Evenafterstringentselectionandsignificantbackgroundrejection,only
thelighteststausareexcludedwithmassesaround100GeV,asshowninFig.5
(right).
Figure5: StausearchW +jetsvalidationregion(left)andstauexclusionlimits
(right). Taken from Ref. [9].
AnotherverywellmotivatedSUSYscenarioisdirectstopproduction,asthe
stop plays a key role in the cancelation of quadratic divergences to the Higgs
mass from top quark loops. The search in [10] utilizes the all hadronic final
statetotargetstoppairproductionwithstoptotop,LSPdecayswithbothtops
decaying hadronically. The analysis uses a customized jet algorithm to identify
two hadronic top decay candidates. Events with an isolated electron, muon, or
tau are removed. The most significant background arrises from tt¯events with
METfromaleptonicW decaywherethechargedleptonislost. Theseparation
ofsignalandbackgroundisachievedwithaBDTtrainedtoselectsignalevents.
KinematicvariablessuchastheanglebetweentheMETandthejetsinthesub-
leading top candidate as shown in Fig. 6 (left) are used in the BDT. After the
selection, MC simulation is used to estimate the total background contribution
in each signal region. The MC is corrected to achieve good agreement with
data in several key kinematic distributions and the background prediction is
validated in the BDT sidebands. No significant excess of data over background
is observed and limits are set on direct stop production. As shown in Fig. 6
(right) stop masses up to around 800 GeV are excluded for light LSPs.
4 EXPLORING GAPS
As more and more SUSY searches have yielded null results, an increasing effort
has been placed on considering where a signal may yet be hiding in space ac-
cessible with current LHC data. This section describes four such searches that
6
ppfi ~t ~t*; ~tfi t + c~ 0 18.9 fb-1 (8 TeV)
Events / 0.2233505000 DMMM1a(((~~~ttt8t)))a ===. 9367P057 005rf,,,e bMMMl(((-iccc~~~1m101010C ))) (===i8 n722M575 a5 T( x(rSx1ey100)V)) [GeV]mLSP233450500000 O E xbpseecrvteedd,, –– 11 ssethxpeoe1rriyment PrelimCinMarSy 111002s95% CL limit on
tt and W+jets 200
200 ttZ and Z+jets
Other processes
150 150 10-1
100 100
50 50 10-2
00 0.5 1 1.5 2 2.5 3 2000 300 400 500 600 700 800 900
min[|D f (jets ˛ top,pmiss|] m~t [GeV]
2 T
Figure 6: Minimum ∆φ distribution between MET and subleading top candi-
datejets(left)andstoptotop,LSPproductionlimits(right)fromthehadronic
stop search. Taken from Ref. [10].
explore regions not covered by more conventional SUSY searches.
OnesuchgapinSUSYsensitivityoccurswhenthestophasamassveryclose
to that of the top and LSP is very light. In this scenario, stop pair production
looks very similar kinematically to top pair production and the signal can be
very difficult to dig out. One approach is to use a precision measurement of
the top cross section and compare it to the theoretically predicted cross section
fromtheStandardModel. Ifexcesseventsexist,theycouldbefromthepresence
of stops. Additionally, the spin correlations of the scalar stops are somewhat
differentfromthatofthespin1/2tops. Thesearchin[11]exploitsthisdifference
togainsensitivitytostopproductioninthisdifficultregion. Dileptonictt¯events
are used to compare the observed angular difference between the leptons with
that expected from tt¯and stop pair production, as shown in Fig. 7 (left). No
deviationfromtheexpectedStandardModeldistributionisobservedandlimits
are set on stop pair production, as shown in Fig. 7 (right).
Another difficult to access region occurs when SUSY particle masses are
nearly degenerate. These so called “compressed” spectra can result in SUSY
decays with little missing energy if the LSP is close in mass the parent particle.
Asindirectdarkmattersearcheswiththemonojettopology,compressedSUSY
canbesearchedforineventswheretheSUSYsystemrecoilsagainstanISRjet.
The compressed spectrum then produces missing energy when it is boosted.
Such a technique is employed in [12] where the ISR jet and missing energy are
searchedforincombinationwithoneortwolowp leptons,whichcanoriginate
T
from stop or chargino decays. The resulting lepton p spectrum is soft, as
T
shown in Fig. 8 (left) for compressed decays. After selecting only events with
low p isolated leptons much of the background is removed and sensitivity to
T
this difficult region is obtained, as shown in Fig. 8 (right).
7
3
Process Yield 0.116000 Data ATLAS
Z/�t⇤t¯+jets 542080000±+�33346000000 Events/1124000000 Stt M(A t=t0) s = 8 TeV, 20.3 fb-1
tV (singletop) 2600 180
tt¯V 80±11 10000 ~B~ackground
WW,WZ,ZZ 180±65 8000 t1t1 , 180 GeV
Fakeleptons 780±780 6000 Fit
Totalnon-tt¯ 6400±860 4000
±
Expected 60000+3500 2000
3700
Observed 60�424 0
t˜t¯˜ 7100 1100 1.2
1 1 ± 1.1
(mt˜1=180GeV,m�˜01=1GeV) atio 1
R
0.9
TABLEI. Observeddileptonyieldindataandtheexpected
SUSYandtt¯signalsandbackgroundcontributions. Systematic 0.8
0 0.2 0.4 0.6 0.8 1
uncertaintiesduetotheoreticalcrosssectionsandsystematicun-
certaintiesevaluatedfordata-drivenbackgroundsareincludedin ∆φ [rad] / π
theuncertainties.
FthFIrGeie.g1du.ilerpetoRn7ecc:ohnasnAtnruenlcst.gedlTe�he�bpderietsdtwricibteiuoetnionnforftowbratcohkegrlsoeuumpndtoo(fbntlhusee in dileptonic tt¯ cross section measure-
likelihoodfitisusedtoextractthespincorrelationfrom hmistoegnratm)cpolumsSpMartt¯epdrotduoctiSonU(SsoYlidbsltacokphisstioggrnama)la(nldeft)andresultingexclusionlimitforstop
the��distributionindata. Thisisdonebydefininga backgroundplustt¯predictionwithnospincorrelation(dashed
coefficientfSMthatmeasuresthedegreeofspincorrela- bplacrkohdisutocgrtaimo)nis(croimgpharte)d.toTthaekdeatnaafnrdotmotheRreesfu.lt[o1f1].
tionrelativetotheSMprediction.Thefitincludesalinear thefittothedata(reddashedhistogram)withtheorangeband
superpositionofthe��distributionfromSMtt¯MCsim- representingthetotalsystematicuncertaintyonfSM. Boththe
uwewtniic+looiiaattrnelthhmi�poofaa,rrnuleeµticdezw+soiapcwimtµttiihinoi�motnhnccoaooonanoerbdfrffitvefiaetxalhci±aleneiutdeµeieonpd⌥bntrooabcffwdcyhSkufiaMttgcSnhhtr,Mneioocea,ufionlnnsletdedcfaaafirrfvognecrisrooinefisermgenmtststeetatchdw(hlt1ieeiiszotia�htmtntt¯t¯ituo[fhns7lnSeito1.maMr]ntmhu)Tew.elaohaoiluettTirihzseohliattnynet¯-- Stltdtrsoohhehuevents / 5 GeVMcgieocsttawtihN11rictoo00eint¯rnLt23nohCsa(ONsmMsn(s[oSNmd1c slcPiLr0ett˜rodohe11cOrlseit]grmsie=orinpnccesnaorlteeruioy1ncioss ss1t8npsi9h0noSs.i7inonss Mf[Gbtet7oc-ic1ed1n ogattVi,ictrrSt¯ sorali7pgseunmpan 2ll=daanra ((]l)8i22io:ytd .n n22miT(de55ogect(T~,,utdmhV n)l21,uc)h en14m.�t55dt˜ee(i))tt¯~c01 ioxThGG10ln)ntpoeeehg-=VVwroteeoprendy-pl1ruleirecxsuepatGtdnilddWtDZsmV-booticiaet VYniucfia ontte+ngVlattc li-srsegjinjeeknt)onThtt-aa sosge+nlopinro xrnjweeoofgtttso-ryusantmrnorotoihd-t˜artfl1hmoleit¯˜mi7saz1sae%del (GeV)mpidaicdznLSPlrooiesogtsroodn----334405050000 pCpM fiOESx b~tp s~ te,Pe ~ctr rvtfieeedd lbi –m– f 11if' ns s~ceat10hx per eoNyrriymLeOnt+N19L.L7 efbx-c1 l(u8s iToenV) 110023 ss section [pb]
tahnetiu-cnocrerretlaaitniotineso.fNtheegatotipveanvdaluanestitoofpfqSuMarckorsrpeisnpso.nAdvtoalaune tbraibcuktgiroonusn(dexpcreepdticftoironb.ackgroundonly)dividedbytheSMtt¯plus250 n cro
oufesfSoMff=SM0i>mpl1ieisntdhiactatteheaspdiengsreaereoufntct¯orsrpeilnatecdorarneldatviaoln- 10 200 mit o
larSgyersttehmanatpicreudnicceterdtaibnytitehseaSreMe.valuatedbyapplyingthefit Ttiohneumnocdeertlaiinstydedteuremtionethdebpyarctoonmpshaoriwngertawnodtht¯adsaromnpizleas-150 10 er li
p
psaromcpeldeusrmeotodipfiseedutdoo-reexflpeecrtimtheenstysscteremaatetidcfvraormiatisoimnsu.laTtehde gtheMC eneort12ah0teerdo20bnyeiA4n0tLerPf6Ga0cEeNd8,0wonit1eh00iHntE1e2R0rfWac1Im4Ge0ud.Pt[Tm1w6eh0dieituhmu1M8nPu0IcnYdeeTr2xt0]Ha0iInAtyanodn100 L up
fitfeetmreoanftcifecSMubnetciwsererteeapninettahyteeudmsitnoegadnteshteeorfnmoGimnaeuinstashileatneemffifeptsclatttooefst.heeaTchrheessudylistfs-- ttehhvdata/eeenas01itm..155ms,ouuslhnatotewodfetirt¯neidstaiamwl-pitalhendiPsfiYanTsasHel-IssAste,adtwebriytahdciovamatiropienadr(iInaSgmRAo/FuLSnPRtGs)EoiNnf 51000 150 200 250 300 350 400 1 95% C
ffireodmpmseaundyo-pdsaetuadios-teaxkpeenriamsetnhtessuyssintegmnaotimciunnaclearntadinmtyodoin- IcSoRmpa0an0tdibFl2eS0wR.i4t0Ahsthi6en0rRece80efn.t[14m020e],a1st2hu0eresm14i0zeent1osp60fTo(mtfh)1 ea8[G0dvedaVi2rt]0ii0oatniaolnjeist mstop (GeV)
fSM[102]. activityintt¯events[103]. TheWtnormalizationisvar-
Thevarioussystematicuncertaintiesareestimatedinthe ieFdiwgiuthrineth8e:thMeoruetoicnalpuncerdtaiisnttireisbouftthieocnros(sl-esefctt)ionand compressed stop search limits (right)
samewayasinRef.[42]withthefollowingexceptions: calculation[86],andtheseTnsitivitytotheinterferencebe-
sincethisanalysisemploysb-tagging,theassociatedun- twineensoWfttpmrouduocntioSnUanSdYtt¯psreoadurcctihon. aTtaNkLeOnisfsrtoudm- Ref. [12].
certainty is estimated by varying the relative normaliza- iedbycomparingthepredictionsofPOWHEG-BOXwith
tionsofsimulatedb-jet,c-jetandlight-jetsamples. The the diagram-removal (baseline) and diagram-subtraction
uncertaintyduethechoiceofgeneratorisdeterminedby schemes[85,104]. AsinRef.[42],theuncertaintydue
comparingthedefaulttoanalternativett¯samplegenerated totheTtophqeuasrekamracshsisineva[l9u]ateedxbtuetnndotsintchluededsoinftthelepton plus ISR topology even further in
withthePOWHEG-BOXgeneratorinterfacedwithPYTHIA. sysseteamracthiciunngceritnainteievse,snintcseiwtwitohuldthhavreeneosoigrnimficaonrte low pT leptons plus large MET. This
allowsforsensitivitytosuchSUSYsignaturesascharginoorneutralinoproduc-
tion decaying to a neutralino LSP with intermediate sleptons, which can give
up to four leptons in the final state. The Standard Model background for three
ormoreisolatedleptonspluslargeMETandahighp ISRjetisverylow. Fig-
T
ure 9 (left) shows the single observed signal event in one of the search regions
compared to the background prediction, while Fig. 9 (right) shows the results
8
of the search when combined with same-sign dilepton and high p multilepton
T
searches.
Figure9: AngularseparationbetweenleadjetandMETinthethreesoftlepton
+ISRsearch(left)andresultinglimitsonelectroweakSUSYproduction(right).
Taken from Ref. [9].
AnotheralternativetoISRtoboostthecompressedSUSYspectrumisvector
bosonfusion(VBF).ThepairofVBFjetsservesthesamepurposeofproviding
a boost to the SUSY system, which would otherwise have very low MET. The
search in [13] exploits the VBF topology to search for compressed SUSY with
complementarysensitivitytotheISRsearches. Asanadditionaldiscriminating
variable, the mass of the VBF dijet system can be utilized to select high mass
events more typical of signal. Figure 10 (left) shows the dijet mass distribution
forbackgroundcomparedtosignal. Theobserveddistributionisconsistentwith
the Standard Model expectation and no evidence for SUSY is found. Figure 10
(right) shows the search results interpreted as limits on compressed sbottom
pair production as well as direct dark matter production.
5 THE BROAD PICTURE
With the plethora of possible SUSY signatures and searches performed at the
LHC, it is important to put the entirety of the search program together to as-
sess where things stand. Many different searches can be sensitive to the same
model. Whenmutuallyexclusivefinalstatesprovidecomplementarysensitivity,
a combination of the results of the different relevant searches can extend the
overall reach. For example, in Fig. 11 (left) the results from searches using five
different final states are shown along with the combination of the five searches,
which extends the sensitivity beyond any of the individual searches alone. Al-
ternatively,differentsearchescanbedesignedtobesensitivetodifferentregions
of parameter space for a given model. When the exclusion regions for each in-
9
CMSPreliminary 18.5 fb-1 (8 TeV)
50 GeV120 dZa(fitann)+jets [fb] 104CMSPreliminary O– b1s se ervxepde c9t5e%d 9C5L% CL18.5 fb-1 (8 TeV)
Events / 21680000 Wopptpph( fiefifir s l b~cnb~c)* +jjjj,j,e Lmts b~= = 6 30000 G GeeVV, ,m mc c~=10 =1 0209 5G eGVe V s 11100023 –ss ((2b~c cb~s j jjej))x ((pLNeOLc)Ot,e L)d, D =9m 56% 0=0 mC GLb~e V- m~c01 = 5 GeV
1
40 100 200 300 400 500 M [Ge6V0]0
Obs./Prediction0122...00025551800100012001400160018002000220m0jj [2G4e0V0] L [GeV] 1122330505055000000000000000 DM1 EFT scalaE–OLLr x b1o<<ps s p2Mee eMcercrvtx/ac2eeptdpdoe rc99t55e%%d C9C5LL% CL10 102RL = 80ggg%eeeffffff === M124c [GeV]
Figure 10: Dijet mass distribution comparing signal and background (left) and
limitsforsbottompairanddarkmatterproduction(right)fromtheVBFSUSY
search. Taken from Ref. [13].
dividual search are overlaid, the total exclusion can show significant coverage.
For example, Fig. 11 (right) shows the exclusions from eight different searches
targetingdirectstopproduction. Intotal,theyexcludeaverysignificantregion
of the plane. Such summary plots also serve to highlight regions where gaps
exist in the current sensitivity and can motivate future efforts.
~g-~g production, ~gfi t t ~c 0
1
V]1000
Ge CMS SUS-13-012 0-lep (ET+HT) 19.5 fb-1
SP mass [ 890000 s OE=xbp s8eecr vtTeeddeV SSSSUUUUSSSS----11113344----000001117300 1232----lllleeeepppp ((((OnS3SljSe+t)+sb 1‡b)9 )61 .15)9 9 .1f5.b95 f.- 3b1fb -f1-b1-1
L 700 SUS-14-010 0+1+2(SS,OS)+>2-lep 19.5 fb-1
600
500
234000000 m(gluino) - m(LSP) = 2 m(top)
100
0
600 800 1000 1200 1400 1600
gluino mass [GeV]
Figure 11: Summary plots of various search results from gluino-mediated stop
production (left) from [3] and direct stop production (right) from [4].
An alternative approach to assess the overall state of the SUSY search pro-
gram is to consider full SUSY models. A popular approach is to utilize the
parameterized minimal supersymmetric standard model (pMSSM) which pa-
rameterizes SUSY with 19 free parameters after making several experimentally
10