Quark Excitations Through the Prism of Direct Photon Plus Jet at the LHC Satyaki Bhattacharya∗, Sushil Singh Chauhan†, Brajesh Chandra Choudhary‡ and Debajyoti Choudhury§ Department of Physics and Astrophysics, University of Delhi, Delhi 110007, India. Thequesttoknowthestructureofmatterhasresultedinvarioustheoreticalspeculationswherein additional colored fermions are postulated. Arising either as Kaluza-Klein excitations of ordinary quarks, or as excited states in scenarios wherein the quarks themselves are composites, or even in theories with extended gauge symmetry, the presence of such fermions (q∗) can potentially be manifested in γ+jet final states at theLHC. Using unitarized amplitudes and the CMS setup,we 9 0 demonstratethatintheinitialphaseofLHCoperation(withanintegratedluminosityof200pb−1) 0 one can discover such states for a mass upto 2.0 TeV. The discovery of a q∗ with a mass as large 2 as 5 TeV can be acheived for an integrated luminosity of 140fb−1. We also comment on the ∼ ∼ feasibility of mass determination. n a PACSnumbers: 12.60.Rc,13.40.-f,13.85.Qk J 5 2 I. INTRODUCTION with quarks and/or leptons. In many such models[3, 4], though not all, quarks and leptons share at least some ] h common constituents. Thereplicationoffermionfamilies,whilebeingofpro- p Accepting this hypothesis would naturally lead to the found significance in our understanding of fundamental - existence of excited states of fermions at a mass scale p issues such as CP-violation and baryogenesis, nonethe- e lessispuzzlinginitsownright. Despiteitsenormoussuc- comparable to the dynamics of the new binding force. h In the simplest phenomenologicalmodels [5], the excited cess in explaining all observed phenomena in the regime [ fermions are assumed to have both spin and isospin 1/2 of particle physics, the Standard Model (SM) has been andtohaveboththeirleft-andright-handedcomponents 1 entirelyunabletoprofferanyinsightintothisaspect. In- v deed, though the observed mass hierarchies and fermion in weak isodoublets (i.e. they are vector-like). Similar is 7 mixingscaneasilybeaccommodated,theSMframework, the case for, say higher-dimensional models wherein the 2 known universe is constrained to be on a 4-dimensional byitsverystructure,isunabletoevenasksuchquestions 9 subspace (a 3-brane) while the SM fields—in particular, ofitself. Thishasledtovariousspeculativeideasseeking 3 the fermions—live in all the dimensions. The analogues . to explain these ill-understood issues. Prominentamong 1 of the excited fermions would be the Kaluza-Klein ex- these are (i) models with extended (family) symmetry, 0 citations with the mass scale being identified with the (ii) constructions based on higher dimensional theory 9 inverse of the compactification scale. (withorwithoutastringtheorymotivation)and(iii)the 0 Giventhatthe“excitedstates”dosuffertheSMgauge : possibility of quark-lepton compositeness, namely that v interactions, these may be produced at high-energy col- the SM fermions are not elementary at all. i lidersand,subsequently,woulddecay,radiatively,intoan X Many of the ideas discussed in this article would be ordinary fermion and a gauge boson (photon, W, Z or r equally applicable—possibly with minor variations—to a gluon). At an e+e− collider, charged excited fermions theories belonging to any of these three classes. How- could be pair-produced via s-channel (γ and Z) ex- ever, for the sake of concreteness, we shall consider the- changes in collisions, while for excited neutrinos only Z ories of compositeness as the basic template for our dis- exchange contributes. Although t-channel diagrams are cussions. Part of the motivation lies in the fact of these also possible (W for ν∗ and γ/Z for e∗), such contri- theories having a more straightforward ultraviolet com- e butions to the overall pair production cross-section are pletion and, furthermore, suffering from a fewer number generally much smaller on account of the smallness of of extra channels,thereby reducing possible ambiguities. the cross-couplings [5] between the excited state, its SM Insuchtheories,the fundamental constituents ofmat- counterpart and a gauge boson. This very same cou- ter,veryoftentermedpreons[1],arepostulatedtoexperi- pling, on the other hand, may be used to singly produce ence anhitherto unknownforceonaccountofanasymp- such states (through both s- and t-channel diagrams). totically free but confining gauge interaction[2], which ThefourLEPcollaborationshaveusedthese(andother) becomes very strong at a characteristic scale Λ, thereby modestoessentiallyruleoutsuchexcitationsalmostupto leading to bound states (composites) to be identified the kinematically allowed range [6]. At the hera, on the other hand, both excited leptons and quarks may be produced singly through t-channel diagrams and these processes have been looked at without any positive re- ∗EmailAddress: [email protected] sults [7]. †EmailAddress: [email protected] ‡EmailAddress: [email protected] If quarksand leptons are composite particles made up §EmailAddress: [email protected] ofsmaller constituents, phenomenologicaleffects may be 2 observableatthe LHC andmightevenshow up atTeva- process gleaned from previous experiments and studies. tron once sufficiently large luminosities are accumulated The rest of the paper has been organized as follows. and analysed. If the compositeness/excitation scale (Λ) In section II we have discussed the effective Lagrangian is not too high, then excited quarks can be produced on for the theory andnew physics contribution. In Sections shell,whileatenergiessufficientlybelowΛsuchaneven- III and IV, we respectively discuss the backgrounds and tuality would manifest itself as an effective four fermion event generation. Section V describes the photon and contact interaction invariant under SM guage transfor- jet algorithms used for the analysis. In section VI we mations [8, 9]. Based on phenomenological studies of discuss the smearings due to detector resolution effects. flavour independent contact interactions for two photon SectionVIIgivesthedetailsofkinematicalvariablesused production, the lower bound has been estimated to be to separate the signalfromthe backgroundwhereas Sec- Λ >2.88(3.24)TeVat95%CL foranintegratedlumi- tion VIII deals with isolation study. The significance of ± nosity of 100 (200) fb−1 [10] at the LHC. signalanddiscoveryarediscussedinsectionIXandXre- Both the experiments at the Tevatron, the CDF and spectively. In section XI we have presented the result of DØ, have searchedfor excited quarks. The latter are as- the analysis. Systematics isdiscussedindetailin section sumed to couple to the SM particles primarily through XII followed by our conclusions and outlook. gauge couplings. Their most visible signature could be either pair production or single excited state production viaquark-gluonfusion,providedtheq∗qg couplingissuf- II. NEW PHYSICS CONTRIBUTION TO γ+Jet ficiently large. Enhanced dijet production rate with an PRODUCTION invariant mass peak above the SM continuum is one of the prominent signals, extensively searched for at the With our interest lying not in the pair-production of Tevatron and the DØ collaboration has excluded the such excited states, but rather on their contribution to mass range of 200-720 GeV[11]. Similarly, the CDF col- the photon plus single jet rates at a hadronic collider, laboration has excluded a mass range of 80-570 GeV we confine ourselves to examining only the relevant part [12, 13] based on searches with various final states. In of the Lagrangian, namely the (chromo-)magnetic tran- a similar vein, the CDF collaboration has put a lower sition between ordinary and excited states. In general, bound of Λ 2.81 GeV at 95% CL using the qq¯ eν these may be parametrized by ≥ → prcoess[14] whereas the DØ collaboration rules out Λ ≤ 2.0 TeV at 95% CL from diject mass peak searches[15]. 1 = q¯∗ σµν g b Ta Ga q +h.c., (1) AttheLHC,boththeATLASandtheCMScollabora- Lint 2Λ R " i i i iµν# L i tions have predicted the sensitivity in the dijet produc- X tion mode. The ATLAS collaboration has xlaimed that where the index i runs over the three SM gauge groups, the use nof dijet angular distributions would allow con- viz. SU(3), SU(2) and U(1) and g , Ga and Ta are i iµν i tact interactions to be probed upto Λ =10 TeV with a the corresponding gauge couplings, field strength ten- integrated luminosity of 700 pb−1. The CMS collabora- sors and generatorsrespectively. The dimensionless con- tion, onthe other hand, hasestimated thatΛ= 6.2TeV stants b are, a priori, unknown and presumably of or- i canbeexcludedat95%CLwithaluminosityof100pb−1 der unity. With these determining both the production and that 5σ sensitivity could be reached for Λ =8 TeV rates and the branching into various modes, clearly, the with just 1fb−1 of data[16]. Recently, the possibility of phenomenologywoulddependconsiderablyontheir(rel- top quark compositeness has been explored through the ative)sizes. Inthisarticle,weshallmakethesimplifying pp tt¯tt¯production process and it has been estimated assumption that the excited states do not couple at all → thata5σexcesscanbeobservedforanewstateof2TeV tothe weakgaugebosons,butdo sowiththe gluonsand [17]. the photon. At first glance, this might seem incompat- As an effective tool for the measurement of gluon ible with a SU(2) U(1) invariant structure. However, ⊗ density inside the colliding hadrons and for precision complicatedembeddingscouldbetheanswer. Morethan test of pQCD predictions, the isolated γ+jet final state this, since the assumption would not change the results has been studied with great detail at the Tevatron qualitatively, it, at least, has the merit of reducing the collider[18, 19, 20] and fixed target experiments[21]. number of possible couplings and hence simplifying the Since γ + jet final state will be one of the key back- analysis. grounds for the H γγ search at the LHC, extensive A further point needs to be noted here. With the La- → isolation studies addressing all known issues have been grangian of eq.(1) being a higher dimensional operator, performed with this process both theoretically and with the cross sections would typically grow with the center detailed detector simulations. It is also an important of mass energy, consequently violating unitarity. This backgroundformanynewphysicsscenarioe.g. LargeEx- is not unexpected in an effective theory as the term in tra Dimensions[22, 23], Randall Sundrum Gravitons[24] eq.(1) is only the first term and the loss of unitarity, to etc. It will, thus, be very interesting to look at γ+jet as a given order, is presumably cured once suitable higher a probe of excited quarks in view of the unprecedented dimensional operators are included. An equivalent way energy scale at LHC and the in-depth knowledge of this to achieve the same goal is to consider the b to be form i 3 γ factorsratherthanconstants. Tothisend,weshalldefine q γ q the q∗qγ and q∗qg vertices to be given by q q* ee f Q2 −n1 q∗qγ (p): q 1 1+ σ pν µ Λ Λ2 µν g q g q g f (cid:18) Q2 −(cid:19)n3 (2) (a) (b) q∗qg (p): s 3 1+ σ pν T µ Λ Λ2 µν α q γ q γ (cid:18) (cid:19) where Q denotes the relevant momentum transfer and fi 1aredimensionlessconstantsrelatedtobiofeqn.(1). q* q It ∼can be checked that, for Q2 = s, unitarity is restored as long as the constants n 1. From now on, eqn.(2) i definesourtheory[1]. Forthe≥restofouranalysis,weshall g q g q confine ourselves to a discussion of n = 1. While this (c) (d) i might seem to be an optimistic choice, it is not quite so. Infact,thecollidersearchlimitsintheliteratureactually FIG. 1: Production of γ +Jet final state through excited correspondton =0and,thus,ourlimitswouldbemore quark mediation (a & c) as well as SM processes (b & d). i conservative. WiththeaforementionedLagrangian,thewidthofthe q is given by contribution as portrayed in Fig.1(a). Adding this con- ∗ tribution to the pure QCD one, the ensuing differential Γ(q ) = Γq+g+Γq+γ cross section reads ∗ Γq+g = 2αsf32 Γ0 dσ = −πααse2q C +2f1f3C + f12f32C 3 dtˆ 3sˆ2 sm Λ2 I Λ4 Q Γ = e2qαf12 Γ (3) (cid:12)(cid:12)(cid:12)qg→qγ uˆ sˆ (cid:20) (cid:21) q+γ 2 0 (cid:12) Csm + ≡ sˆ uˆ 2 Γ0 ≡ MΛq23∗ 1− 4Mmq2∗2q! 1− Mmq22q∗! CI ≡ (sˆ−sˆ2M(sˆq2−∗)M2+q2∗Γ)2FMsq2∗ + uˆuˆ−2FMuq2∗ and can be very large for a heavy q∗ (see Table I). As a CQ ≡ sˆuˆ+Mq2∗t fat resonance is often difficult to observe, this will turn (cid:0) sˆ2(cid:1) 2 uˆ2 2 out to have profound consequences. Fs + Fu "(sˆ−Mq2∗)2+Γ2Mq2∗ (uˆ−Mq2∗)2 # TplAinBgLsEtreI:ngΓth(qs.∗)Baosthaαfusnacntidonαeomf Marqe∗(e=vaΛlu)afoterddaiffteMrenq∗t.cou- + 2Mq2∗uˆ−sˆtˆMuˆq2∗ (sˆ(−sˆ−MMq2∗)q22∗)+FΓs2FMuq2∗ Mq∗ Γ (GeV) s 1+sˆ/Λ2 −(n1+n3) F ≡ (TeV) f1 =f3 =1.0 f1 =f3 =0.5 (cid:0)1 tˆ/Λ2(cid:1)−(n1+n3) t 0.5 34.4 8.61 F ≡ − 1.0 63.6 15.9 (cid:0)1 uˆ/Λ2(cid:1)−(n1+n3) 2.0 118 29.6 Fu ≡ − 3.0 170 42.6 (4) (cid:0) (cid:1) 4.0 221 55.2 with the SM result being recoveredin the limit Λ . →∞ 5.0 271 67.6 Thenewphysicscontributiontothedifferentialcrosssec- 6.0 319 79.8 tion thus depends on four parameters, namely f ,f ,Λ 1 3 7.0 367 91.8 andthemassoftheexcitedstateMq∗. Forsimplicity,we assumethesetobeflavour-independent(withinagenera- tion, it obviouslyhas to be so). For eq.(1) to make sense With the introduction of these (flavour-diagonal) ver- as an effective Lagrangian, the masses have to be less tex as in eq.(2), the subprocess qg qγ acquires a new → than Λ (Ref.[25] requires that Mq∗ < Λ/√2). Note that aslongasΛ sˆ,oneoff canalwaysbeabsorbedinΛ. 1,3 ≫ In our analyses, we would be considering only moderate values for these parameters. [1]While a Lagrangian formulation leading to such vertices would For qq¯ gγ,the Feynmandiagramsare as in Fig.1(c- necessitate a seemingly non-local Lagrangian, this is not unex- → pectedinaneffective theory. d); the differential cross-sections are related to those in 4 eqn.(4) by crossing symmetry and are given by g γ dσ = 8πααse2q B 2f1f3B + f12f32B dtˆ 9sˆ2 sm− Λ2 I Λ4 Q (cid:12)qq¯→gγ (cid:20) (cid:21) (cid:12) g g (cid:12) uˆ tˆ (cid:12) B + sm ≡ tˆ uˆ tˆ2 uˆ2 FIG. 2: Type-I Background: Additional contribution from B Ft + Fu gluon fusion. Lowest order background emanates from (b) I ≡ tˆ−Mq2∗ uˆ−Mq2∗ and (d) in Fig.1. tˆ2 2 uˆ2 2 B tˆuˆ Ft + Fu Q ≡ "(tˆ−Mq2∗)2 (uˆ−Mq2∗)2# q g 2 q q tˆ uˆ + Mq2∗sˆ"tˆ−FMtq2∗ + uˆ−FMuq2∗# q g (5) q q g q III. BACKGROUNDS q q g g Theγ+jetfinalstatecanbemimickedbymanyknown processesoftheSM.Weconsideralltheleadingcontribu- g g tions(boththephysicsbackgroundsaswellasthe detec- tor ones) and broadly categorize these into three classes viz., q q g g Type-I:where a photonand a hard jet is produced FIG. 3: Type-II Background: QCD jet production where a • in the hard scattering. jet fakes a photon giving a γ+Jet final state. Type-II:QCDdijet,whereoneofthejetsfragments • into a high ET π0 which then decays into a pair of ina2 2hardscatteringprocess),the totalproduction overlapping photons and, hence, is registered as a crosss→ectionis 104 times largerthanthe Type-I back- singlephoton. Moreover,insomecasesthe electro- ground. Howeve∼r,the factofthe probabilityofa jet fak- magnetic fraction of a jet can mimic a photon in ingaphotoninthedetectorbeing 10−3 10−4reduces the detector. theType-IIbackgroundtothesam∼eorder−asthe Type-I. Moreover, for high transverse momenta the QCD dijet Type-III: Photon + dijet production, where one of • falls very rapidly ( P−4), thereby suggesting a simple the jets is either lost or mismeasured. This could ∼ T mechanism of suppressing this background proceed either from hard processes such as qq¯ → Although the total Type-III background is very small qq¯γ,ggγ (with all possible interchanges of initial comparedtotheothers,fortheP rangeunderconsider- and final state partons) or result from W/Z + γ T ationinthisanalysis,itturnsouttobeofthesameorder production with the heavy bosons decaying into a as gg γg. And while nonresonant subprocesses (such pair of jets. → as the (α2α) contributions to qq¯ qq¯γ or gg qq¯γ) O s → → While the leading contributions to the Type-I back- can,inprinciple,besubstantial,notethatthese,insome ground emanates from the SM amplitudes of Fig. 1, a sense are related to the much larger Type-I and Type-II further contribution is displayed in Fig.2. In Figs.3 & 4, backgrounds. Consequently,theformeraretypicallysup- we show the major contributing Feynman diagrams for pressed when appropriate phase space cuts are imposed the Types II & III backgrounds respectively. At the LHC, the Type-I background is dominated by q γ q γ the Compton process (qg γq), while the other two → subprocesses,namelyanihilation(qq¯ γg)andgluonfu- → sion (gg γg) contribute only a small fraction in the q q → low transverse momentum (P ) ranges. For higher P , T T the annihilation subprocess can contribute upto about 23% of the total Type-I background. q Z0 q’ W ∼ TheType-IIbackgroundaccruesmainlyfromqg qg, qq¯ qq¯andgg gg. ForPˆ 200GeV(herePˆ →isthe FIG.4: Type-IIIBackground: HereW/Z0 decaystotwojets T T P →oftheoutgoi→ngpartonsinc≥enterofmomentumframe and only oneof thejets passes the trigger threshold. T 5 TABLEII: Production cross section in different PˆT binsfor various Standard Model backgroundswith γ+Jet final state. Subprocess 50-100 GeV 100-200 GeV 200-400 GeV 400-600 GeV 600-1000 GeV 1000-1500 GeV 1500 GeV and above (pb) (pb) (pb) (pb) (pb) (pb) (pb) qg γq 7.22 103 569 36.3 1.53 2.22 10−1 1.19 10−2 7.6 10−4 qq¯→γg 65×2 65.3 5.56 3.18 10−1 5.67×10−2 3.76×10−3 2.8×10−4 gg→γg 1.79 8.6 10−2 3.1 10−3 7.04×10−5 6.32×10−6 1.75×10−7 5.8×10−9 → × × × × × × QCD Jet 1.71 107 9.70 105 4.44 104 1.39 103 171 8.19 5.34 10−1 × × × × × aZ(jj)+γ 5.08 8.49 10−1 9.50 10−2 6.23 10−3 1.16 10−3 8.48 10−5 6.46 10−6 aW(jj)+γ 4.80 6.93×10−1 6.19×10−2 4.16×10−3 7.39×10−4 4.67×10−5 2.99×10−6 × × × × × × aHerethebranchingfractionistakenintoaccount. to reduce the latter. here that in the final selection of γ, we have used the In Table-II we show the production cross section for fiducial volume of the electromagneticcalorimeter of the all backgrounds in different Pˆ ranges. CMS detector i.e. η 2.5 with 1.444 η 1.566 ex- T | |≤ ≤| |≤ cluded on account of the insensitive region between the barrel and the endcaps[28]. The jets were selected up to IV. EVENT GENERATION & CUTS η 3.0 only, because of poor resolution in the forward | |≤ calorimeter. The event generation for signal and different back- groundprocesseswasdonewithpythia-v6.325[26]. For Fig. 5 shows the deviation in the total cross-sectionof saingdna(l5)evwenertegiemnperleamtioenntetdheinmsiadteritxheelepmytenhtisaoffraEmqw.o(r4k). tqigon→isγw+eljl-eatpapsrfouxnimctaiotnedobfyMaq∗Λ(−=2Λco)n.tCrilbeuartliyo,ntshuepvearrimia-- We used the following common parameters and pythia posed upon a constant (the SM value). This is reflective switches: of the fact that, for large Λ, the new physics contribu- tion is dominated by the interference term in Eqs.(4,5) • Parton Distribution Function(PDFs): CTEQ ratherthanthe pureΛ−4 term. Only ifwe hadimposed 5L [27]; harder cuts on the photons, would the latter term have Q2 = sˆ; dominated (albeit at the cost of reducing event numbers • and hence the sensitivity). MultiParton Interaction(MPI): “ON”; • Initial State Radiation(ISR) and Final State Radi- • ation (FSR): “ON”. LHC s =14TeV 104 To get enough statistics for both the signal and the qg→ γ+jet backgrounds, we divided the whole analysis into three qq→ γ+jet phase space regions determined by the value of the P 103 T of the final state γ and the jet. For this purpose, the followingPˆT (ckin(3)parameterofpythia)criteriawere (fb)M102 used for different mass points of signal: S σ Pˆ 180GeV: 1.0–3.0 TeV, σ - 10 T • ≥ PˆT 450GeV: 3.5–4.5 TeV, 1 • ≥ Pˆ 950GeV: 5.0–6.0 TeV. T • ≥ 10-1 1 2 3 4 5 6 A total of 16 mass points, 11 for coupling strength M = Λ (TeV) q* f = f = 1.0 (with a step size of 0.5 TeV) and 5 for 1 3 f =f =0.5weregenerated. Thedifferentbackgrounds w1ere a3lso generated in various PˆT range. No pseudora- FatIG√.s5=: D14evTiaeVtio.n of cross section from SM with Mq∗(= Λ) pidityrestrictionwasappliedwhilegeneratingtheevents as the large Pˆ cut requirement naturally restricts the T events to well within η < 5.0. We must also mention | | 6 V. PHOTON AND JET CANDIDATES AT THE scheme( Pi). The computed direction is then used to GENERATOR LEVEL seed a new proto-jet. The procedure is repeated until P both the energy and the direction of the putative jet is Although a mass peak in the signal will appear as an stable between iterations. We quantify this by requiring excessofevents overthe continuumSMbackground,the that the energy should change by less than 1% and the significance for such an observation will depend on the direction by less than ∆R=0.01. When a stable proto- size of this continnum. Hence, to enhance the signal jetisfound,allobjectsintheproto-jetareremovedfrom peak,itisnecceassarytoreducethebackgroundasmuch the listofinputobjectsandthe stable proto-jetisadded as possible. For the signal under investigation, QCD di- to the list of jets. The whole procedure is repeated until jet is the largest background, as it mimics γ+jet final the list is bereft of objects with an ET above the seed state when one of the jet fakes a photon. To estimate threshold. The cone size and the seed threshold are the this backgroundreasonablyat the generatorlevel, it is a parameters of the algorithm. must to have a proper understanding of the reconstruc- tionalgorithmforaspecificdesignofthe detectorrather VI. SMEARING EFFECTS than limit ourselves to only partonic level photon and jets from the final state in an event. Taking this into consideration, we have used a clus- While a detailed and full-scale detector simulation is teringalgorithmtoaccountforfakephotonsarisingfrom beyond the scope of this work, realistic detector effects jets[29]. Theelectromganeticcalorimeter(ECAL)ofthe can easily be approximated. To this end, we smear the CMS detector is made up of PbWO crystals and each generatorlevelinformation with ECAL and HCAL reso- 4 crystal covers 0.0175 0.0175 (equivalently, 1◦) in the lutions of the CMS detector. ∆η ∆φspace(φbein×gtheazimuthalangle). Forphoton For the ECAL resolution function, we use the form − reconstruction, we have used the “hybrid” algorithm[30] δE a a where we consider only those final state electromagnetic = n C (6) particles (i.e., γ,e+ ande−) in the η φ space suchthat E √E ⊕ E ⊕ − neither ofthe distances∆η and∆φ fromthe seedobject whereadenotesthestochasticterm,a isthewhitenoise n exceeds 0.09. This extension is equivalent to a 10 10 × termandC istheconstantterm,withthethreecontribu- crystalsizeintheCMSdetector. Theseedforsuchaclus- tionsbeingaddedinquadrature. Foreachoftheseterms, teringmusthaveaminimumP of1GeV.Aphotoncan- T we use values identical to those for the electromagnetic didate is either a direct photon or other electromagnetic calorimeter of the CMS [30], namely obejcts such as π0 γγ,ρ0 γγ etc. The main con- tribution of fake ph→otons com→es from π0 γγ( 81%), C = 0.55% → ∼ η γγ( 12%) while other sources give only a small 2.1 10−3GeV η <1.5 → ∼ contribution. A detailed discussion of this reconstruc- a = × | | tion algorithm at the generator level can be found in a n ( 2.45 10−3GeV 1.5 η 2.5 × ≤| |≤ previous work [31]. 2.7 10−2GeV1/2 η <1.5 a = × | | For jet reconstruction, various algorithms have been ( 5.7 10−2GeV1/2 1.5 η 2.5 . usedbydifferentcolliderexperiments. Theseinclude the × ≤| |≤ MidpointCone[32],IterativeCone[33,34],andtheK al- t Theresolutionsfor∆ηand∆φweretakentobe0.02and gorithms [35, 36, 37]. The K and MidPoint algorithms t 0.001 respectively for both the barrel and endcap. are used mostly for offline analysis. Since we have used Forthehadroniccalorimeter,theresolutionswereonce the CMS setup in our analysis,we use the IterativeCone again assumed to be the same as those for the CMS algorithm to reconstruct jets at the generator level. Be- HCAL [39, 40], namely, ingmuchfaster,thisiscommonlyusedforsoftwarebased triggers. While the first algorithms for the jets at the Barrel: • hadroncollidersstartedwithsimpleconesinthe∆η ∆φ − δE 65% space [38], clustering techniques have greatly improved = 5%,∆η =0.04,∆φ=0.02 in sophistication over the last two decades [32, 35]. E E/GeV ⊕ Forarealdetector,thefirststepinthereconstruction, p before invoking the jet algorithm, is to apply noise and Endcaps: • pile-up suppression with a set of cuts on E . To make T δE 83% “perfect detector jets”, we used a seed P cut on the T = 5%,∆η =0.03,∆φ=0.02 P -ordered final state particles and selected only those E E/GeV ⊕ T which had a transverse momentum above the required p minimum[2] of P 1.0 GeV. Once the seed is se- Tseed ≥ lected, we search around for all the particles in a cone of ∆R 0.5. The objects inside the cone are used to [2]Theseedthresholdcanvaryfrom0.5to 2.0GeVdependingon ≤ calculate a proto-jet direction and energy using the E- theenergyofreconstructedjet. 7 Forwardregion: • TABLE III: Preselection-efficiency and geometrical accep- δE 100% tencefor various SMbackgrounds and few signal points. = 5%,∆η =0.04,∆φ=0.04 . E E/GeV ⊕ Selection Cut Signal γ+Jet QCD Z+γ W +γ % % % % % p Pγ,jet 200GeV 48.7 44.2 0.90 38.4 37.1 The four momenta of the photon and jet were recal- T ≥ [1 TeV] culated after applying these resolution effects using an appropriate Gaussian smeared function. In Fig. 6, we Pγ,jet 500GeV 40.2 39.8 0.42 50.4 50.6 show the effect of resolutionon the mass peak for a Mq∗ T ≥ [4 TeV] of 1 TeV. It should be noted that the ATLAS detector at the Pγ,jet 1TeV 47.4 46.0 0.51 58.8 59.9 LHChasabetterjetenergyresolutionwiththeconstant T ≥ [5 TeV] term being 2 % [41] compare to 5 % in the CMS ∼ ∼ detector. Onthe other hand, the CMS ECAL has a bet- ηγ 2.5, ηjet 3.0, 42.4 38.2 0.81 32.8 33.2 ter resolution than the ATLAS one owing to a smaller |ηγ|≤ [1.44|42,1|.≤5666] [1 TeV] | |6∈ constant term. However,with the resolving power being dominated by the jet energy resolution, ATLAS should 38.2 37.8 0.40 47.4 48.4 do somewhat better. In other words, our results corre- [4 TeV] spond to a conservative choice. 46.4 45.0 0.50 56.3 58.7 [5 TeV] 6000 Smeared Unsmeared 5000 bands. It may be noted that the QCD dijet background is more than 10 times as large as the signal, but falls s4000 nt very rapidly with Pγ/jet (the mistagging probability has e T Ev already been included). Fig. 7(f) shows the distribution 3000 in the subprocess center of mass scattering angle, with cosθ∗ = tanh[(ηγ ηjet)/2.0]. Note that major differ- − 2000 ences between the signal and background profile occur onlyforp andinvariantmassdistributions,whereasthe T otherphasespacevariablesarenotverysensitivediscrim- 1000 inants. 800 850 900 950 1000 1050 1100 1150 M (GeV) γ-jet Fig. 8 shows similar distributions as in Fig. 7 but for the Mq∗ = 5 TeV point instead. For these distributions FIG. 6: Effect of smearing on the mass peak for an excited we have used a Pγ,jet cut of 1 TeV at the pre-selection T quark of 1 TeV. level. With the P spectrum for the photon from QCD T background falling very rapidly, the signal dominates overthe backgroundabove 2 TeV even without isolation cuts. As for the corresponding invariant mass (M ) γ−jet VII. KINEMATICAL VARIABLES distribution—see Fig. 8(e)— a combination of the large natural width and smearing effects results in a broad InFig.7,weshowthekinematicaldistributionsforthe bump rather than a sharp one. Once again, the other leading photon and the leading jet for Pγ,jet 200 GeV distributions do not discriminate between the signaland T ≥ dataset for the background and the signal for Mq∗ =1 backgroundin any forceful manner. While the the slight TeV. For purpose of visual clarity, the distributions for dip in the central ηγ region for the W +γ process might Z +γ and W +γ backgrounds have been scaled up by seem intriguing (especially in the absence of any such a factor of 10. The bump in the transverse momentum dipin theZ+γ distribution), itis but astraightforward distributionsareprimarilydrivenbytheon-shellproduc- reflection of the well-known Radiation-Amplitude-Zero tion of the q∗ and, therefore, are centred slightly below (RAZ) present in the former [42, 43]. That the RAZ in Mq∗/2. As is evident from Fig. 7(e), an excess in the in- the angular distributuion is apparent only for the high variantmassspectrumwouldbequiteprominentforeven Pγ,jet cutoff case can be understood by realising that T L.dt =1pb−1. The t-channel contribution has been in- the rapidityofthe photonas measuredinthe laboratory cluded and manifests itself in the elongation of the side can be related to the rapidity (scattering angle) in the R 8 -1No of Events / 10.00 GeV / 1 pb11111000100----0432121 QSZγW i+C( gj( DjjnJ)j )ae + d+lt iγ jγe ( tx(x 1 100)) -1No of Events / 10.00 GeV / 1 pb111100010----014321 QSWγZ iC+( g(j DjjnJ)j) ae + d+lt iγ jγe ( t(xx 1 100)) 10-50a 200 400 600 800 1000 1200 1400 1600 10-50 200 400 600 800 1000 1200 1400 1600 (a) PγT (GeV) (b) PTjet (GeV) Signal Signal -11 pb 101 WγZQ C+( (j jDjJ)j) e+ d+t iγ jγe ( tx(x 1 100)) -11 pb 101 WγZQ C+( (j jDjJ)j) e+ d+t iγ jγe ( tx(x 1 100)) 12 / 10-1 12 / 10-1 0. 0. s / s / nt10-2 nt10-2 e e v v E E o of 10-3 o of 10-3 N N 10-4 10-4 10-5 -4 -2 0 2 4 10-5 -4 -2 0 2 4 (c) ηγ (d) ηjet Signal 10 -1V / 1 pb 101 QWγZ C+( (j DjjJ)j) e + d+t iγ jγe ( t(xx 1 100)) -1ns / 1 pb 1 No of Events / 10.00 Ge11110000----4321 No of Events / 0.01 radia11110000----4321 γZQS iC +(g j jDn J) a e+ dl t iγ j e ( tx 1 0 ) W (jj) + γ (x 10) 10-5 500 1000 1500 2000 2500 3000 3500 4000 10-50 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (e) Mγ-jet (GeV) (f) |Cosθ*|γ-jet FIG.7: Kinematicvariabledistributionsafter200GeVpre-selectioncutonPT (a)PTγ distribution(b)PTjet distribution(c)ηγ distribution (d) ηjet distribution (e) Mγ−jet distribution and (f) cosθ∗ γ−jet. The signal corresponds to Mq∗ =1TeV. | | partonic subprocess center of mass frame through geometricalacceptencesfortheCMSdetectorforvarious backgrounds and signal of Mq∗ =1, 4 and 5 TeV against 1 x η(γ)= ln 1 +η∗(γ) the total generated events. 2 x (cid:18) 2(cid:19) where x are the momentum fractions of the incom- i ing partons. For small sˆ values (hence lower ckin(3) VIII. ISOLATION VARIABLES cuts)thepartondensitiesaremaximizedwhenthe(anti- )quark acquire small(large) momentum fractions respec- In a detector, a photon candidate is reconstructed by tively. This leads to a considerably large contribution to summingtheelectromagneticenergydepositioninECAL ηγ from the boost, thereby smearing the original double towers in a limited region of space, with the sum being peakedηγ distributionintocentrallypeakedone. Onthe required to be above a certain ET threshold. For the contrary, for typically high sˆ (ckin(3) 1 TeV) values sakeofsimplicity,thislimitedregioncanbevisualizedas ≥ the x tend to be not too different thereby reducing the a cone in ∆η ∆φ space given by ∆R ∆φ2+∆η2, i − ≡ smearing on this account. andcontainingmostoftheenergyoftheelectromegnetic p In Table III we show the pre-selection efficiencies and object. 9 10-2 Signal 10-3 Signal QCD dijet QCD dijet -11 pb10-3 ZWγ +( j( jjJ)j )e+ +t γ γ ( (xx 1 100)) -11 pb10-4 WγZ +( (j jjJ)j) e+ +t γ γ ( (xx 1 100)) V / 10-4 V / Ge Ge10-5 0 10-5 0 0 0 s / 10.10-6 s / 10.10-6 nt nt Eve10-7 Eve10-7 of of No 10-8 No 10-8 10-9 1000 1500 2000 2500 3000 3500 4000 4500 10-9 1000 1500 2000 2500 3000 3500 4000 4500 (a) PγT (GeV) (b) PjTet (GeV) 10-2 10-2 Signal Signal QCD dijet QCD dijet -1Events / 0.12 / 1 pb11110000----6543 γZW +( j( jjJ)j )e+ +t γ γ ( x(x 1 100)) -1Events / 0.12 / 1 pb11110000----6543 WγZ +( (j jjJ)j) e+ +t γ γ ( (xx 1 100)) o of 10-7 o of 10-7 N N 10-8 10-8 10-9-3 -2 -1 0 1 2 3 10-9-3 -2 -1 0 1 2 3 (c) ηγ (d) ηjet 10-3 Signal 10-3 -1No of Events / 10.00 GeV / 1 pb1111100000-----87654 QWγZ C+( (j DjjJ)j) e + d+t iγ jγe ( t(xx 1 100)) -1No of Events / 0.01 radians / 1 pb1111100000-----87654 WSZQγ iC+ ( g (j jDjn J)j ) a e + d+l t iγ jγ e ( t(x x 1 1 0 0 ) ) 10-9 2000 3000 4000 5000 6000 7000 8000 10-90 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (e) Mγ-jet (GeV) (f) |Cosθ*|γ - jet FIG. 8: As in Fig.7 but with a 1 TeV pre-selection cut on PT and a signal corresponding to Mq∗ =5TeV instead. Ajetfragmentingintoneutralandchargehadronswith the scalar sum of transverse energy (E ) in- TSUM a π0 γγ (with overlapping photons) carrying max- • side a cone around the photon. Although, in a → imum momentum can also lead to fake single-photon full detector simulation the E is measured TSUM candidates. To remove such events, photons are re- separately for ECAL and HCAL, but working at quired not to have associated charged tracks within a the generator level, we combine them into a sin- cone of size R . This is implemented by requiring that gle variable with all kinds of electromagnetic and iso the scalar/vector sum of energy/transverse momentum hadronic objects around the photon being taken within R should be below a certain threshold. For ex- into account. iso ample, the DØ and the CDF experiments demand that the E due to chargedtracks within a cone of ∆R=0.4 T around the photon should be less than a certain value. A. Track Isolation In this anlaysis,we closely follow the CMS detector sim- ulation studies [44] and consider the following isolation Forthepurposeoftrackisolation,only‘stable’charged variables: particles e.g. π±, K±, e± and P± were considered. The other particles were found to have only a negligible con- thenumberoftracks(N )aboveacertainthresh- tribution. Indeed,π± alonecontributemorethan80%of trk • old inside a cone around the photon candidate. thechargedtracks. Fig 9showsthedistributionofnum- 10 ber of charged tracks (Ntrk) around the leading photon 1 awsitwheinllaascfoonrethoefstioztea∆lbRac≤kg0r.o3u5nfdo.rSainMceq∗th=e1leTadeVingsipghnoa-l 10-1 Riso≤ 0.35 SQiCgDna dl(i j1eTteV) tonisthetruephotonforsignalevents,mostofthemare associated with zero tracks (N =0) and the distribu- trk s10-2 tionfallsoffveryrapidlyforlargerNtrk values. Forback- nit U g8raonudndtheevnenftaslltshsolouwghly,.thTehdeisstmriablultriiosenapteaNktsrkat=N0triks∼du7e- bitrary 10-3 to the fact that γ+jet(SM) and W/Z +γ backgrounds Ar10-4 havetruephotonsastheleadingphotonintheeventand have no tracks around them, while the rising part along 10-5 withtheextendedtailismainlycontributedbytheQCD dijet events where the fake photon typically has a large 10-6 number of tracks around itself. In this study, we accept 0 5 10 15 20 25 30 35 40 45 (a) Highest Ptrk (GeV) a photon to be an isolated one if there is no track with T minimum transverse momentum (Ptrk ) within a given Tmin cone aroundit. It shouldbe notedthat comparativedis- tributions of signal and total background, as shown in 10-1 Riso≤ 0.35 Signal (5 TeV) Fig 9, is not overly sensitive to moderate changes in the QCD dijet PTtrmkin value (the exact values are discussed at a later 10-2 stage). nits U y 10-3 ar 102 R ≤ 0.35 Arbitr10-4 iso Signal (1 TeV) 10 Background 10-5 -1b p 1 1 vents / 10-1 10-60(b) 5 10 15Hig2h0est 2P5tTrk (G30eV)35 40 45 50 E er of 10-2 FIG.10: HighestPT trackaroundleadingphotonforthesig- Numb10-3 TnaelVa.nAdntihsoelaQtiCoDn cboancekgorfosuizned0a.3)5Mhaq∗s b=e1enTuesVedb.)BMotqh∗d=is5- tributionsarenormalizedtounityforthesakeofcomparison. 10-4 Note that the background differs between the two panels on account of thediffering requirements on PˆT (videSect.IV). 0 10 20 30 40 50 60 N trk FIG.9: Numberoftracks(Ntrk)forthesignal(Mq∗ =1TeV) To optimize the value of Ptrk , it is useful to exam- and the background eventsaround thephoton. Tmin ine both the signal and the QCD dijet background in termsofthedistributionforthehighest-P track. InFig T In pp collisions at the LHC, a large number of soft 10(a) and (b) respectively, we display this distribution tracks (in the range of a few MeVs to a few GeVs) will for the signal for Mq∗ = 1 TeV and 5 TeV. Accompany- be produced in each event. The main sources of such ing these, in each case, are the corresponding QCD dijet soft tracks are ISR, FSR, minimum bias and underly- background. For ease of comparision both the distribu- ing events. For a direct photon emerging from the hard tions are nomralized to unity. As is evident, any tracks interaction, such soft tracks could actually be in the accompanying the photon in a signal event tend to have near vicinity of the photon. Labelling such photons as a low P , whereas for the background events, the distri- T non-isolatedonescouldpotentiallyreducethesignaleffi- bution is a very wide one. An indicative value for the ciency,andmanyinterestingevents,suchasthoseinthis optimalPtrk isgivenbythepointofintersectionofthe Tmin study,couldbelost. Topreventsuchloss,tracksareusu- two normalised distributions (signal and background). ally required to pass certain minimum selection criteria, Theoptimalchoicedoesdependonthesignalprofile(de- withaminumumthresholdonthetransversemomentum termined, in a large measure by the typical momentum being a common requirement [29, 44, 45]. Adopting this transfers), as is evident from the crossover points being strategy, we investigate the dependence of the signal ef- 4(6)GeVforsignalscorrespondingtoMq∗ of1(5)TeV. ∼ ficiency and the signal/background (S/B) ratio on the Thus,characterizingonlythose tracks,aroundaphoton, chosenPT threshold(PTtrmkin),varyingthe latterbetween with a PT >∼ 4GeV as true tracks (or, in other words, 1-3 GeV. accepting photons with accompanying tracks satisfying