Search for Xtra DimensionsSearch for Xtra Dimensionswith the ATLAS detectorwith the ATLAS detector

LHCb Exotica WS – May 26th 2009 - Helenka Przysiezniak LAPP-CNRS/U.de Montreal

AntipastiTrou normand (ATLAS))Primi piatti (Black Holes)Secondi piatti (Dielectron RS resonance)Formaggi (Diphotons resonance and excess)Sgroppino e limoncello

New concepts relative to ED explored within a string theory framework to address the so called Hierarchy ProblemHierarchy ProblemBut is the HP really solved…

Planck 1019 GeV → EW (SB) 103 GeV

These models bring down gravity scale to TeV range → space-time geometrygeometry is responsible for the apparent hierarchy.

In approach developped by Arkani-Hamed Dimopoulos Dvali (ADD)Arkani-Hamed Dimopoulos Dvali (ADD)gravity fileld lines spread throughout the ED and gravity appears to be dilutedgravity appears to be diluted

These spatial ED have to be of finite size, or compactifiedcompactified to avoid any deviation from Newton’s law. Compactification radius RCompactification radius RC C (or scale 1/Rc) and number of EDs number of EDs are free parameters.

High dimensional space-time called bulk bulk populated by orbifoldsorbifoldse.g. small intervals formed by the new dimensions

and a 4-D sub-dimensional space-time called branebranelocalized at the end of these orbifolds.

Particles propagating inside bulk lead to appearance on 4-D brane oftower of Kaluza Klein (KK) particle excitationstower of Kaluza Klein (KK) particle excitations

with same properties as the original particle but with mmKKKK n/R n/R

→ → In the ADD modelIn the ADD model, only gravitons propagate in the bulk == Graviton KK towercontinuum of massive states due to the large size of ED e.g. eV-1 (direct or virtual production of gravitons)

→ → Models with TeV-1 sized EDModels with TeV-1 sized ED either all SM particles or just gauge bosons propagate into the bulk :KK excitations of SM particles (or of gauge bosons only) of mass ~ 1 TeV

Antipasti

??

→ → Randall Sundrum (the infamous RS)Randall Sundrum (the infamous RS)warped ED compactified on an orbifold holding 2 4-D branes at fixed points.

SM fields live on TeV brane. Gravity is everywhere (TeV, Planck branes; bulk).Graviton KK tower of well spaced states with mass 1st level excitation ~ TeV

(resonant Kk gravition production)

Trou normand

Thin Superconducting Solenoid (B=2T)

LAr Electromagnetic Calorimeter :L R = 13.3m2.25m3.2 (4.9)E /E = 10%/E0.7%

Hadronic Calorimeter :Endcaps LArgBarrel Scintillator-tileL R = 12.2m4.25mE /E = 50%/E3% (3)

Large SuperconductingAir-Core Toroids

Muon SpectrometerL R = 25 (46) m11m

Inner Detector :Semiconductor Pixel and StripsStraw Tube Tracking Detector (TRT)L R = 7m1.15mR=12-16m, Z=66-580m

Photons → jet rejection ~ few 10Photons → jet rejection ~ few 1033 for ~ 80% efficiency for ~ 80% efficiencyElectrons → jet rejection ~ few 10Electrons → jet rejection ~ few 1055 for ~ 80% efficiency for ~ 80% efficiencyJet and Etmiss reconstructionJet and Etmiss reconstruction

The ATLAS detector

N.B. ATLAS analyses covered here

- Most recent (Expected Detector and Physics Performance publication 12.2008)- Appropriate for early data

- Presented in ATLAS meetings

“Recent work” - not all topics covered here

• Tev gravity – Black Holes• Search for highly excited string states

• Properties of BSM (SUSY) particles : mass, spin, CP• New Physics with taus

• Top polarization in Randall Sundrum• KK gluon in variant ADD model

• ADD virtual graviton• ADD graviton decaying inside the detector

• RS graviton decaying• TeV-1 scale ED with KK gauge bosons

• TeV-1 scale ED a la Universal Extra Dimension• Distinguishing spins in decay chains with photons at the Large Hadron Collider (GMSB, UED)

• Graviscalar discovery potential• Xtra Dimensional scalars

• SUSY vs UED studies• etc.

Black Holes Black Holes predominent concern, although large uncertainties in the theory,since these high multiplicity events enable to test the software, trigger, electronics saturation, etc.

In this view interesting for first data.

Simple final states Simple final states such as dileptons or diphotons also interesting as early data analyses

In ED modelsIn ED modelsMMPlanck Planck ~ M~ MEWSBEWSB

→ coupling strength of gravity increased to size ~ the other interactions

→ unification of gravity and gauge interactions

→ quantum gravity effects could be observable at the LHC

→ → Black Holes (BH) could be producedBlack Holes (BH) could be produced

These would decay semi-classically by Hawking radiation emitting high energy particles.

BLACK HOLES

N.B.“A general formulation of black hole (BH) production is extremely complex.

Semi-classical assumptions,valid only above the Planck scale,

are necessary to enable a quantitative description and predictions. Consequently,

a minimum BH mass must be imposed in simulations,above which these conditions are satisfied.”

1. BH Formation : 1. BH Formation : By semi classical argumentsa BH is formed if impact parameter of head-on collision between 2 partons < Schwarzschild radius RS

Schwarzschild 1916 + generalization by Myers and Perry 1986

For D=4+n dimensions RFor D=4+n dimensions RS S (1/M (1/MDD) (M) (MBHBH/M/MDD))1/(n+1)1/(n+1)

which depends only on the number of dimensions and on the effective Planck scale MD

Exact Xsec needs QG theory → use quasi classical black disc approximation

= f = f R RSS22 (f=formation factor~1)(f=formation factor~1)

Parton level Xsec grows with energy, non perturbative

valid for Mvalid for MBH BH >> M >> MDD

Possible for any combination of quarks and gluons.All gauge and spin quantum numbers are allowed.

BH are charged and coloured.BH are charged and coloured.

2. Hawking radiation2. Hawking radiationStephen Hawking 1975Pairs of virtual particles appear at the event horizon with one particle escapingParticles have black body spectrum in D=4+n dimensions with

TTHawkingHawking=(n+1)/ (4=(n+1)/ (4RRSS) ) M MDD x (M x (MBHBH/M/MDD))1/(n+1) 1/(n+1) x (n+1)x (n+1)

3. BH decay3. BH decay1. Balding phase : Graviton radiation2. Evaporation phase : MBH>>MD Hawking radiation where most of initial energy is emitted mostly in SM particles3. Planck phase : MBH→MD QG regime : predictions “very difficult”…

BLACK HOLES

Charybdis MCCharybdis MC1. Balding phase : not simulated

2. Evaporation phase : only SM particles are generated, no gravitons. Approximated by democratic decay into SM particles.

3. Planck phase : only SM particles generated. Two body decays.

N.B. N.B. → → Current MCs reasonable for MCurrent MCs reasonable for MBHBH>>M>>MDD

→ → Total Xsec = convoluting parton-level Xsec with PDFs integrating over phase space, summing over parton types.Total Xsec = convoluting parton-level Xsec with PDFs integrating over phase space, summing over parton types.Transition from parton-level to hadron-level Xsec based on a factorization ansatz.

Validity of this formula for energy region above the Planck scale is unclear.Even if factorisation is valid, extrapolation of the PDFs into this transplanckian region is questionable.

BLACK HOLES

BH event simulation with CharybdisBH event simulation with Charybdis→ Semi classical model : MBH 5MD → Due to the high Hawking temperature and mass scale, semi-classical black holes tend to emit particles with very high E and pT→ High multiplicity and high sphericity events are produced.→ Democratic BH decay into all SM particles only loosely achieved because of charged and coloured input state.

Event selectionEvent selection

→ High-PT jet trigger >99% efficient for BH events→ Event shape variables not very useful in distinguishing BHs from SM background:

• large background cross-sections with tails overlapping BH distributions• different black hole samples can have widely varying multiplicities and event shape distributions

→ Two complementary signal selections with different sensitivity to experimental and theoretical uncertainties :1. (|Pt|) > 2.5 TeV and one lepton with pT lepton > 50 GeV.

Errors given in Table1 are statistical only. The theoretical uncertainties on their cross-sections are large. 2. 4 objects with pT >200 GeV with 1 lepton passing this cut

BLACK HOLES

Discovery reachDiscovery reach

Difficult to produce robust discovery potential using semi-classical assumptions valid only above the Planck scale.Threshold to Transplanckian region difficult to model.Apply and vary threshold cut on true BH mass → conservative estimate of discovery reach

If the semi classical cross-section estimates are valid,If the semi classical cross-section estimates are valid,BHs above a BHs above a 5 TeV threshold can be discovered with a few pb−1 of data,5 TeV threshold can be discovered with a few pb−1 of data,while 1 fb−1 would allow a discovery to be made even if the production threshold was 8 TeV.while 1 fb−1 would allow a discovery to be made even if the production threshold was 8 TeV.

BLACK HOLES

(|Pt|) > 2.5 TeV 4 objects with pT >200 GeV

- Systematic uncertainties mainly from Charybdis parameters which are varied.- BH mass distribution can be reconstructed despite that Etmiss resolution not perfect due to emitted gravitons not in simulation.- As well, Th and number of extra dimensions can be extracted from the data.

Di-electron RS resonance

New heavy states giving narrow resonance decaying into ℓ+ℓ - predicted by GUTs, Technicolor, little Higgs, XtraDs.Simplicity of the final state make it important to study with early ATLAS data 10 fb−1

Heavy neutral particles excluded from direct searches at the Tevatron for m < 1 TeV

Randall-Sundrum graviton : Randall-Sundrum graviton : Exponentially warped 5th dimension linking two branes (SM and Planck)

dsds2 2 = e = e -2kr-2krcc

|y| |y| dx dx dxdx – r– rcc22 dy dy22

rc = compactification radiusMPlanckbar = effective Planck scalek/MPlanckbar ~< 0.06

SM fields live on the TeV brane (y=)Gravity lives everywhere: TeV (y=), Planck (y=0) branes and in the bulk

Tower of KK excitations of the graviton → resonances decaying into lepton (or photon) pairs at the LHCTower of KK excitations of the graviton → resonances decaying into lepton (or photon) pairs at the LHCCurrent limits depend on parameters of RS model : from several hundred GeV to 1 TeV. Here RS graviton decay to electron pairs.e+e - = 10−4 to a few 10−3 times the mass

Di-electron RS resonanceIdentification and reconstruction of very high transverse momentum (pT ) electrons:Identification and reconstruction of very high transverse momentum (pT ) electrons:Low background → only minimal selection criteria, in ATLAS “loose” electron selection.Reconstructed clusters :

• ||< 2.5• associated with track reconstructed in the inner detector

Two selections studied:• Loose selection (efficiency close to 1):

hadronic leakage and shower shape variables. High efficiency and rejection against highly energetic pions with wide showers.

• Medium selection (efficiency between 65 and 70%):Further requirements for better rejection against 0→ using fine granularity of 1st e.m. calo. compartmentTighter requirements on the associated track.

Reconstruction efficiency <80% dominated by cluster to track association but improved in more recent software version.

(E)electron ~ 1% at high pT ; except in crack region between forward and central calorimeters (E)electron ~ 5%Probability to assign the wrong charge to an electron : 1% to 5% for pT = 100 GeV to 1 TeVFor a 1 TeV Z’ the dielectron mass resolution = (0.80±0.02)%

Di-electron RS resonanceTrigger:Trigger:Reduce rate of events down to 200 Hz without throwing away the signal…Four triggers were studied:

• e55 - one electron with pT ≥ 60 GeV, • e22i - one isolated electron with pT ≥ 25 GeV, • 2e12 - two electrons with pT ≥ 15 GeV, • 2e12i - two isolated electrons with pT ≥15 GeV.

Efficiency at the three ATLAS trigger levels (L1, L2, EF) for a sample of graviton events

BackgroundsBackgrounds

Di-electron RS resonanceSignal samplesSignal samples

G→eG→e++ee−− resonance reconstruction : efficiency and leftover DY background resonance reconstruction : efficiency and leftover DY background• two electrons – no charge requirements – with pT ≥ 65 GeV • loose electron selection criteria • e55 single electron trigger• two electrons are roughly back-to-back in with cos e1e2 < 0

The efficiency decreases at high graviton masses, due to the track match requirement.

Di-electron RS resonanceDiscovery potential:Discovery potential:5 discovery and 3 evidence reach in cross-section and k/ MPlanckbar coupling constant as a function of graviton mass(determined using parametrized fit approach) estimated for 1 fb−1

Sources of systematic uncertainties:Sources of systematic uncertainties:• NLO EW and QCD corrections• uncertainty in

- efficiency of electron identification 1%- energy scale 1% for electrons- resolution of electrons 20 %- luminosity, assumed to be 20% with an integrated luminosity of 100 pb−1 of data and 3% for 10 fb−1

Combined effect of the systematics is to increase the amount of integrated luminosity needed for discoveryCombined effect of the systematics is to increase the amount of integrated luminosity needed for discoverybetween 10 and 15 % for the different parameter sets.between 10 and 15 % for the different parameter sets.

Diphoton RS resonanceEnergy resolution of the ATLAS e.m. calorimeter provides a narrow invariant mass distribution. Tevatron excludes MG < 240 GeV/c2 with coupling k/ MPlanckbar = 0.01

Signal and background samples :Signal and background samples :G→ with MG = 500 GeV/c2 , 1 TeV/c2 with coupling k/ MPlanckbar = 0.01Dijets in two pT ranges, +jets in two pT ranges, and samples

Trigger: Trigger: g60 one photon with pT > 60 GeV/c.With no reconstruction selection and cutstrigger efficiencies = 93.0±0.3% for MG = 500 GeV/c2 and = 96.5±0.3% for MG = 1 TeV/c2.With the requirement of having two identified photons, the relative trigger efficiencies become99.86±0.06% and 99.92±0.06% for MG = 500 GeV/c2 and MG = 1 TeV/c2 respectively.

Event selection : Event selection : trigger cuts + 2 photons + optimized high pT photon isolation + photon identification cuts + pT1,2 > 80 GeV/c

SignificanceSignificance = 2(√(s+b)−√b)

Two main systematic uncertainty sources:Two main systematic uncertainty sources:- PDF uncertainty is ~5% for a 1 TeV mass.- Luminosity 20% (3%) with100 pb−1 (10fb-1)

55 discovery with 102 pb discovery with 102 pb−1−1 (1140 pb (1140 pb−1−1))k/ Mk/ MPlanckbarPlanckbar = 0.01 = 0.01

MMGG = 500 (1000) GeV/c = 500 (1000) GeV/c22

Diphoton RS resonance

Universal Extra Dimension diphoton excess“Universal” == ALL SM particles propagate into the XtraD(s)

n=1,2,3,… Kaluza Klein (KK) excitations for each SM particle, of mass

mmnn22=n=n22/R/R2 2 + m+ mSMSM

22

n=0 corresponds to the SM particleR : compactification scale

Λ : cutoff scale above which the theory is no longer valid

In 3D (3D+t), conservation of the KK number→ never a vertex with only 1 KK excitation, KK particles always produced in pairs

KK number is conserved at tree level, but can be violated in first order loops.Tree level radiative corrections :

~20% for strongly interacting particles (heaviest being the gluon)<10% for leptons and EW bosons (lightest being the photon)

Quark and gluon KK excitations cascade decay down to theLightest Kaluza Klein Particle (LKP) :Quark and gluon KK excitations cascade decay down to theLightest Kaluza Klein Particle (LKP) :the excited photon the excited photon **

UED diphotonUED model - Minimal scenario : UED model - Minimal scenario :

One extra dimension: fermions and bosons live in a 4+1 ( R ~ TeV -1 ) dimensional “thick” brane+ Gravity mediated decays : + Gravity mediated decays :

“thick brane” embedded in larger space of (4+N) dimensions (size~eV-1) where only gravitons propagateGraviton field appears as a massless particle with an

infinity of excited modes whose masses differ by 1/R~eV(infinite KK tower).

Graviton couples to all KK particles, which decay through gravitymediated KK number violating interactions by emitting a graviton

e.g. e.g. * * GG

If (mass splitting) (gravity mediated)gluon and quark excitations cascade decay down to the *

which in turn will decay * GFinal state : 2 high pT photons + Etmiss + soft jets final stateFinal state : 2 high pT photons + Etmiss + soft jets final state

Signal Xsec for different 1/R valueswith R=20 and N=6 at LHC Ecm=10TeV

UED diphotonSignal, background and significance Signal, background and significance {2[(S+B)ln(1+S/B)-S]}{2[(S+B)ln(1+S/B)-S]}Simple cuts: ||<2.5, m12>100 GeV/c2, |cos 12|<0.95 1a,b) pT1,2>100 GeV/c, Etmiss > 100 GeV, 200 GeV 2a,b) pT1>200 GeV/c, pT2>100 GeV/c, Etmiss > 100 GeV, 200 GeV 3a,b) pT1,2>200 GeV/c, Etmiss > 100 GeV, 200 GeV

<50 pb-1 for 1/R=1000 GeV; <700 pb-1 for 1/R=1500 GeV<50 pb-1 for 1/R=1000 GeV; <700 pb-1 for 1/R=1500 GeVFast simulation analysis. First look at fully simulated events show that even “loose” photon cuts can still be loosened.Fast simulation analysis. First look at fully simulated events show that even “loose” photon cuts can still be loosened.Etmiss tails have to be well understood.Etmiss tails have to be well understood.

Di jet angular distributionsPreliminary presentation only

Dijet angular distributions Dijet angular distributions allow for general search beyond the SM, by looking at deviations from QCD predictionsIn large extra dimensions (ADD model), if MPlanck ~ 1TeV, can possibly observe ggravitational scattering ravitational scattering through the exchange of virtual Kaluza-Klein (KK) modes, and/or black holesblack holes

Different model parametersDifferent model parameters Discovery potentialDiscovery potential

OUTLOOK - Sgroppino e limoncello

First data First data will have surprises.surprises.Calorimetry, tracking, Etmiss. SM. Hopefully new physics.Hopefully new physics.Whatever is seen in the first data will sweep away many models.Start with resonances or indubitable deviations from SM.Start with resonances or indubitable deviations from SM.

Search by topologies would be wiser.Search by topologies would be wiser.But need a good understanding of the SM.Tevatron is only “now” publishing topological searches deviating from SM

• Tev gravity – Black Holes• Search for highly excited string states

• Properties of BSM (SUSY) particles : mass, spin, CP• New Physics with taus

• Top polarization in Randall Sundrum• KK gluon in variant ADD model

• ADD virtual graviton• ADD graviton decaying inside the detector

• RS graviton decaying• TeV-1 scale ED with KK gauge bosons

• TeV-1 scale ED a la Universal Extra Dimension• Distinguishing spins in decay chains with photons at the Large Hadron Collider (GMSB, UED)

• Graviscalar discovery potential• Xtra Dimensional scalars

• KK excitations of the W boson• SUSY vs UED studies

• etc.

BACKUP SLIDES

AntipastiEspecially for a low number of extra-dimensions (n), the minimum Planck scale in ADD models is highly constrained.Constraints on these models come from a variety of sources, from the Tevatron data, tabletop experimentsand astrophysical measurements from supernovae and neutron star cooling, cosmic and gamma-ray data.

BH backup slides

If the effective number of large extra dimensions is n = D−4, the inverse-square law would smoothly change from the 1/r2 form for r R to a 1/r2+n form for rR.

Direct searches for black holes at collider experiments have not yet been performed.

The latest direct search result from the CDF collaboration hasset lower bounds with 1.1 fb−1 of Run II data on M_D of 1.33 TeV for n=2 to 0.88 TeV for n=6.

Indirect searches by the DØ collaboration set lower limits around 1.28 TeV.

The ATLAS working model for black holes uses the black disk cross-section, which depends only on the horizon radius r_h. The (4+n)-dimensional Myers-Perry solution, similar to the 4-dimensional Schwarzschild radius,is chosen for the horizon radius r_h.It depends only on the number of dimensions and the Planck scale.

The classical black hole cross-section at the parton level is_hat ab → BH = r 2

h

where a and b are the parton types.

In most cases, we work with initial black hole masses at least five times higher than the Planck scale at which the expression for the cross section should be valid.

The total cross-section is obtained by convoluting the parton-level cross-section with the parton distribution functions (PDFs),integrating over the phase space, and summing over the parton types.The transition from the parton-level to the hadron-level cross-section is based on a factorisation ansatz.The validity of this formula for the energy region above the Planck scale is unclear.Even if factorisation is valid, the extrapolation of the parton distribution functions into this transplanckian regionbased on Standard Model evolution from present energies is questionable,since the evolution equations neglect gravity and possible KK states in the proton.

The details of horizon formation, and the balding and spin-down phases have been ignored. Important effects of angular momentum in the production and decay of the black hole in Eds are not accounted for in the MC. The BHs are considered as D-dimensional Schwarzschild solutions.Only the Hawking evaporation phase is generated by the simulation.

According to the quantum mechanical uncertainty principle, rotating black holes should create and emit particles.The Hawking radiation process reduces the mass of the black hole and is therefore also known as black hole evaporation.

We can view the Hawking evaporation phase as consisting of two parts:determination of the particle species and assigning energy to the decay products.A particle species is selected randomly with a probability determined by its # of degrees of freedom and the ratio of emissivities(emissivity, a dimensionless value, is a property of the body surface and is dependent on the T of the bodyand on the of the emitted radiation).The degrees of freedom take into account are polarisation, charge and colour.The emitted charge is chosen such that the magnitude of the black hole charge decreases.All SM particles are considered, including a Higgs boson. All particles are treated as massless.

Gravitons have not been included in the simulation, which is one of the drawbacks of the current model.Because the graviton lives in the bulk, the # of degrees of freedom of the graviton becomes significant for high numbers of EDs.In addition, the graviton emissivity is highly enhanced as the space-time dimensionality increases. Therefore the black hole may lose a significant fraction of its mass into the bulk, resulting in missing transverse energy.

The energy assignment to the decay particles in the Hawking evaporation phase has been implemented as follows : the particle species is given an energy randomly according to its extra-dimensional decay spectrum.A different decay spectrum is used for scalars, fermions and vector bosons, i.e. the spin statistics factor is taken into account. Grey-body spectra are used without approximations. The grey-body factors depend on the number of dimensions.The Hawking temperature is updated after each decay.It is assumed the decay is quasi-stationaryin the sense that the black hole has time to come into equilibrium at each new temperature before the next particle is emitted.The energy of the particle given by the spectrum must be constrained to conserve energy and momentum at each step.

The evaporation phase ends when the chosen energy for the emitted particle is ruled out by the kinematics of a two-body decay.At this point an isotropic two-body phase-space decay is performed.In our simulation, the decay is performed totally to Standard Model particles and no stable exotic remnants survive.

Baryon number, colour and electric charge are conserved in the black hole production and decay.Etmiss in the generator comes only from the neutrinos, while in reality missing transverse energy is also possible due to the lost Ein inel. prod., graviton emission, a non-detectable BH remnant and the possibility that the BH can leave the SM brane.For the BHs we consider, only a small amount of energy, on average, is lost due to neutrinos.If gravitons were considered, the average energy loss would be approximately 9%.

EVENTS PROPERTIES:The high mass scale, and the thermal nature of the decay process, result in black hole events beingcharacterised by a large number of high-pT final state particles, including all the Standard Model fields.Graviton emission is also expected, but is not simulated in CHARYBDIS. Of the final state particles, thedetector can measure jets, electrons, muons and photons well, and will be able to reconstruct some of theZ and W bosons. The missing transverse energy, produced mainly by neutrino and graviton emission,can also be measured. In this section, the data sample with two extra dimensions and black hole massesabove 5 TeV is used as the reference signal sample.A key feature of black hole decays is that the Hawking temperature is higher for larger n, for a givenblack hole mass. A higher temperature produces higher energy emissions, with the consequence that theenergy is shared between fewer particles. This has a significant effect on the multiplicity and event shapedistributions. Similarly, the samples with a higher black hole low-mass cutoff produce more high energyfinal state particles.PARTICLE TYPES AND MULTIPLICITIESFigure 2 shows the types of particles produced directly by black hole decay. The vertical axis showsthe average number of particles per black hole decay. From this figure, we see that a heavier black holehas more decay products. The particle-antiparticle balance is broken by the initial state of two protonscolliding. Moreover, due to conservation of energy and momentum, colour connection etc., a perfectdemocratic decay cannot be achieved, e.g., the number of top quarks is smaller than that of ligher quarks.The possibility of identifying fermions and bosons and determining their branching ratios in black holedecays was studied in [40].

Figure 3 shows pT and pseudorapidity (h) distributions of particles produced directly from blackhole decays. As expected, the shape depends little on particle type. Figure 4 shows the reconstructedmultiplicity of final-state jets, leptons and photons. Four signal samples are shown for n = 2, 4 and 7with a minimum black hole mass of 5 TeV, and for n = 2 with a minimum black hole mass of 8 TeV. Thefigure also compares the reference signal to the backgrounds. The multiplicity in the signal falls as nrises, because the black holes decay at a higher temperature.

EVENT SHAPES:At first sight, one would expect black hole events to be very different from the background in eventshape variables such as sphericity, because of the high multiplicity thermal decay. However,the event shape of the black hole events varies considerably with n, making such variables less usefulthan could be hoped. Though the background distributions show less variation, when these are scaledby their large cross sections, there is a large degree of overlap, disfavouring their use as a cut variable.Additionally, our ignorance of the decay modes of the final black hole remnant introduces a significantsystematic effect. In our version of CHARYBDIS, once the mass of the black hole has dropped belowthe Planck scale, the remnant decays to either 2 or 4 bodies. We have selected the two-body option forour standard samples. This means that at high n, where events can reach this stage after few emissions,the circularity of the events is reduced, and the thrust increased.

The distinguishing power between signal and backgrounds of a selection of event shape variables wasstudied; Figure 5 shows the circularity distribution for the same samples as Figure 4; similarly Figs. 6and 7 show their thrust distributions, sphericity and aplanarity. The expected bias towards more “jetlike”events is clearly seen at high n. For this reason, we choose not to use event shape variables as adiscriminant in this analysis.

TRIGGER:The ATLAS trigger and data-acquisition system consists of three levels (L1, L2, EF) of online eventselection. Each subsequent trigger level refines the decisions made at the previous level and mayapply additional selection criteria. The ATLAS trigger is described in detail in ref. [44].The total trigger efficiencies are listed in Table 4 for three signal samples and demonstrated in Figure9 for the signal sample with n = 2 and m > 5 TeV. The highest efficiency is provided by the single-jettrigger, which we consider to be the master trigger for the black hole events. The presence of multiplehigh-pT jets per event, each of which is likely to pass the trigger, results in very high total efficiencies.Setting this trigger threshold at 400 GeV will provide greater than 99% efficiency at all trigger levels.The Standard Model process rate at this threshold is expected to be less than 0.1 Hz at an instantaneousluminosity of 10^31 cm^−2 s^−1, which should allow this trigger to run at this threshold without prescalingfor the first few years of LHC data taking. The rate of black hole events is expected to be less than 5 mHzat the 10^31 cm^−2 s^−1 luminosity. For the start-up running at the luminosity of 10^31 cm^−2 s^−1, it is plannedto set the highest threshold for the single-jet trigger at 120 GeV, guaranteeing an efficiency of almost100% for black hole events.

EVENT SELECTION:Since all types of Standard Model particles are produced from black hole decay, we make full use ofparticle identification information (PID) from our detectors. First we select muons, electrons, photonsand jets, which are called objects in this section. Table 5 shows the details of their selection criteria.The identification of objects is sometimes ambiguous: e.g., an electron could be simultaneouslyreconstructed as a jet. To resolve this, we apply PID to each object, selecting muons, electrons, photonsand jets in that order of priority. Once an object passes the PID criteria in a given category, any remainingambiguous assignments are removed if they match the chosen object within a DR of less than 0.1.Next we select black hole events using these objects as described below. Then we reconstruct ablack hole from all the identified objects for the selected event. The mass of the black hole in an eventis calculated from the four-momenta of the reconstructed final state objects and missing ET , which isincluded in the calculation to improve the reconstructed mass resolution.

We present to select black hole events. One is based on the scalar summation of pTand the other on the multiplicity of high-pT objects. Both make use of the characteristic of a black holehaving large mass. After that, we require a high-pT lepton to reject backgrounds further.

BH mass reconstructionBH mass reconstructionfrom the four-vectorsof all final statereconstructed particlesand missing ET

SCALAR SUMMATION OF PTFigure 10 shows the scalar summation of the pT of each object, å|pT |, which demonstrates good background discrimination and high signal efficiency for all black hole samples. We require å|pT | to be larger than 2.5 TeV to reject backgrounds. This requirement is relatively unaffected by changes in the model, in particular by changes to the number of extra dimensions n. Figure 11 shows mBH distributions after this requirement. The QCD dijet background is already well suppressed, but we also investigated the effect of a further selection, requiring a lepton with a pT > 50 GeV. This resulted in the QCD dijet background being rejected by a factor greater than 10^6 as shown in Table 6, which summarises the event numbers for an integrated luminosity of 1 fb−1. Though the high statistics QCD samples used were generated with PYTHIA, a leading order generator, there were also pT -sliced small ALPGEN multijet samples available. When investigated using the Sum(|pT |) and lepton cut method, a very similar, marginally lower number of background events was predicted according to the very limited statisticsavailable. Larger scale studies would be needed to conclude anything more concrete. Poisson confidence limits areused for samples where fewer than 20 events passed the requirements. Signal cross-section errors arestatistical only; the theoretical uncertainties are large as discussed in Section 2.

MULTIPLICITY OF HIGH PT OBJECTSAn alternative selection procedure was also used. Figure 12 shows the pT distributions of the leading,2nd-, 3rd- and 4th-leading objects out of all the selected objects. The 4th-leading object still has larger pTin the signal events than in the background events. We require the number of objects with pT > 200 GeVto be equal to or greater than four. Figure 14 (left) shows mBH distributions after this requirement. SinceQCD processes still remain large, a lepton requirement is again used to decrease it. Figure 13 shows thedistribution of the highest pT lepton (muon or electron). As expected, the number of leptons from QCDprocesses is small. Requiring the number of leptons (muons or electrons) with pT >200 GeV to be equalto or greater than one results in the mBH distributions shown in Figure 14(right).

BH DISCOVERY REACH:Making a robust discovery potential for black hole events is difficult, because the semi classical assumptionsused to model them are only valid well above the Planck scale. Close to the Planck scale, eventsmay occur due to gravitational effects with lower multiplicities, but without the signatures anticipatedby our event selections. As the energy rises above the threshold needed for black hole creation, ourrequirements should become more efficient. Lack of theoretical understanding makes it impossible tomodel this threshold region.To account for this, we impose a lower requirement on the true mass of black holes created in oursimulated samples, BH_thresh, normally set at 5 TeV, and we do not attempt to account for any additionalsignal from lower masses. In order to estimate the discovery potential, two methods have been considered.

1. we keep our signal selection requirements constant, and increase the value of BH_thresh. Since theanalysis requirements are unchanged, the background remains constant, while the signal drops asthe production of events occurs at higher mass. We then evaluate the luminosity required to detecta minimum of 10 signal events, with S/sqrt(B) > 5, assuming the production cross-section is as highas predicted. Such a study is shown in Figure 17, using the Sum(|pT |) and lepton requirements. Thismethod produces conservative limits, taking some account of the uncertainty in the productioncross-section near the threshold.

2. We keep the production model unchanged with BH_thresh = 5 TeV, but apply an additional requirementon the reconstructed black hole mass. This requirement reduces substantially backgroundevents, while allowing the higher mass signal to pass unchanged. This is less conservative, since itallows black hole signal events to be produced at low mass, but to migrate above the reconstructedmass requirement because of the detector mass resolution, hence increasing the signal. As before,we use the nominal value of the production cross-section, and evaluate the luminosity neededto meet our discovery criteria, this time as a function of reconstructed mass. A study using thismethod is shown in Figure 18 using the 4-object and lepton requirements.

The two approaches are complementary and illustrate the uncertainties in different ways. We observethat the search reach is limited eventually at high mass by the falling production cross-section, reflectingthe falling parton luminosity and the limited energy of the LHC. We conclude that, if the semi classicalcross-section estimates are valid, black holes can be discovered above a 5 TeV threshold with a few pb−1of data, while 1 fb−1 would allow a discovery to be made even if the production threshold was at 8 TeV.

SYSTEMATIC ERRORS:There are a number of theoretical parameters associated with CHARYBDIS which can generatesystematic errors in the estimates of the acceptance for signal events. These are:

• The kinematic cutoff. This parameter is normally true, and causes the generator to end thermalemission if an unphysical emission is randomly selected. The generator moves immediately to thefinal remnant decay phase. This approximation deteriorates at high numbers of extra dimensionsbecause of the high temperature and emitted particle energies. We have investigated the alternative,where a new emission is selected until a physical one is chosen. In this case, thermal emission willcontinue until the black hole mass falls below MDL.

• Temperature variation. The Hawking temperature of the black hole is normally allowed to increaseas its mass decreases, as expected if the black hole has time to equilibrate between decays. Wehave investigated the alternative of keeping the temperature fixed at the initial value, as would bethe case if the black hole decayed very quickly or “suddenly” (see Table 8).

• Number of extra dimensions. In addition to our full simulation samples with n = 2, 4 and 7, wehave simulated n = 3 and 5 with the fast simulation. As noted above, the events become morejet-like at high n and the particle multiplicity drops (see Figure 20), due to the increased Hawkingtemperature. Our signal selection remains robust, as shown in Table 8.

• Planck scale. We have investigated changing the Planck scale from its default value of MDL =1 TeV to 2 TeV. We note that, since the model is only valid for black hole masses much larger thanthe Planck scale, this scenario is not well modelled in the range of masses accessible at the LHC.

• Remnant decay. We have investigated changing the remnant decay model from a two-body to afour-body mode (see Figure 20).

CONCLUSIONS:The studies presented show that the ATLAS detector is capable of discovering the production of blackholes up to the kinematic limit of the LHC, assuming that the signal is correctly modelled. This conclusionis largely based on the predicted huge production cross-section, and the small background fromQCD dijets at very high-pT , especially when the presence of a high energy lepton is required. However,both of these assumptions are suspect. The high production cross-section is subject to considerablediscussion in the literature, as discussed in Section 2. Moreover, until the LHC has measured the QCDcross-section at 14 TeV, we cannot be certain of the tails of the QCD distributions. The Monte Carlosimulations of these tails are working at the limit of their validity, given the high energies and largemultiplicities involved.

For these reasons, we prefer not to place too much weight on the detailed search reach limits. Instead,we confine ourselves to the statement that, with current understanding, the black hole signatureconsidered should be clearly visible if it exists.

We have considered the possibility of extracting model parameters from the data, should a signal beobserved. There are two key parameters: the Planck scale M_D (or MDL depending on the convention)and the number of extra dimensions n. In Ref. [46, 47] a method was proposed to extract MD from thecross-section data, which fixes the Planck scale (within the model assumptions), and from events withhigh energy emissions.

The Hawking temperature T_H of the black hole depends on n. If we detect events with emissionsnear m_BH/2, the energy of those emissions is a measure of the initial T_H. Hence, over the sample ofblack holes, the probability of such emissions is a measure of the characteristic temperature, and canbe used to extract n.

Di lepton resonances in RS GravitonBackup slides

The Randall-Sundrum model addresses the hierarchy problem by adding one extra-dimension linking two branes,the Standard Model brane and the Planck brane.The hierarchy is solved by assuming for the fifth dimension a warped geometryin which the size of the ordinary coordinates decreases exponentially from the Planck scale to the TeV scale.The Randall-Sundrum model predicts the existence of a tower of Kaluza-Klein excitations of the graviton.These should be observable as resonances which decay into lepton pairs at the LHC.The current limits depend on the parameters of the model, and range from several hundred GeV to one TeV.

We consider the observability of a Randall-Sundrum graviton decaying into electron pairs.The width of the graviton resonance would be very small.For the parameters considered here it ranges from 10−4 to a few 10−3 times the mass.

In this section we present a sensitivity study for the Randall-Sundrum G→ee final state.In this channel, it is assumed that there is no interference between the G and the dilepton background.The table shows the parameters of the different G samples used in this analysis.Gamma_G is the simulated graviton resonance widthand sigma_m stands for the width of the observed resonance after convolution with detector resolutions.For k/ Mplanckbar < 0.06 the resonance is narrow compared to the experimental resolution.The main Standard Model background is neutral Drell-Yan production.Other backgrounds such as dijets with both jets misidentified as electronsare expected to be small and neglected at this time.

Di lepton resonances in RS Graviton

Parameters of the G → ee samples used: natural width (Gamma_G), Gaussian width after detectoreffects (sigma_m) and leading order cross-section.

Pair of electrons with pT>=65 GeV using loose electron criteria described in section 2.1.The evts must pass the e55 single electron trigger (see section 3.1.).Require that the 2 electrons are roughly back to back in phi with cos delta phi_ee <0 between the 2 electrons.The table shows the remaining Xsec at each stage of the selection and total efficiency for different mass points.The efficiency decreases at high graviton masses, due to the track match requirement.

Di-electron RS resonanceSearch using the parameterized fit approachSearch using the parameterized fit approachThe data distributions are compared to two models :

a background-only model (null hypothesis H0)and a signal-plus-background model (test hypothesis H1)

The input signal and background shapes are given to the fitting algorithmseither as histograms in the non-parameterized approachor as functions in the parameterized approach.

For each of the models a likelihood or a 2 distribution is computedand the log of the ratio of the two likelihoods (LLR) or the difference of two 2

are estimated and used to compute the confidence levels. Either CLb = CLH0 alone, or CLs = CLH1/CLH0 can then be used to compute the significance S :

Discovery potential:Discovery potential:Pseudo-experiments are generated from both the null and test hypothesis. Each pseudo-experiment is fit twice. The first fit assumes the data are described by the SM using a specific function.The second fit assumes the data are described by the sum of a Gaussian and the shape describing the DY background. The Gaussian is allowed to float throughout the entire mass region considered,and the width is fixed to the detector resolution.

The distribution of the logarithm of the likelihood ratio between H0 and H1 is constructed. The average expected discovery potential is then calculated from the fraction of the likelihood ratio distribution for bgd-onlypseudo-experiments that extends beyond the mean of the distribution for S+B experiments.Figure 13 shows the 5 discovery and 3 evidence reach in cross-section and k/ M_planckbar coupling constant as a function of graviton mass, estimated for an integrated luminosity of 1 fb−1.

Di-electron RS resonanceDiscovery potential:Discovery potential:

The DY bgd distribution after this evt sel. Is shown in Fig 12 along with signal at m_G = 1Tev and coupling k/Mplbar = 0.02.

An excess of evts is seacrhed for in the mass range from 300 GeV up to 2 TeVAnd the signal sensitivity is studies using an “extended max. lik” fitting method.2 hypotheses are considered: the null hyp. H0 where the data are described by the SM,And the test hyp. H1 where the data are described by the sum of the bgd and a narrow Gaussian resonance.Pseudo expts are generated from both pypotheses. Each expt is fit twice.The first fit assumes the data are described by the Standard Model using the function described in Section 5.1.The second fit assumes the data are described by the sum of a Gaussian and the shape describing the DY bbg.During this second fit the mean of the Gaussian is allowed to float throughout the entire mass region considered,and the width is fixed to the detector resolution.

We can then compare the likelihood of the signal and background hypotheses. The distribution of the logarithm of the likelihood ratio between H0 and H1 is constructed, and shown for one signal point in Fig. 12. Based on this, we calculate the average expected discovery potential from the fraction of the likelihood ratio distribution for background-only pseudo-experiments that extends beyond the mean of the distribution for signal plus bgd expts.Figure 13 shows the 5sigma discovery and 3sigma evidence reach in cross-section and k/ Mpl bar coupling constant as a function of graviton mass, estimated for an integrated luminosity of 1 fb−1.

The LO cross-sections are multiplied by the K-factors discussed in section 4.2.2 for both signal and DY bgd.Various sources of systematic uncertainties for siigand bgd are considered in the evaluation of the experimental sensitivity, including luminosity, energy scale, energy resolution, electron identification efficiency and DY bgd uncertainties as listed in section 4.4.The combined effect of the systematic uncertainties is to increase the amount of integrated luminosityneeded for discovery between 10 and 15 percent for the different parameter sets.