Lecture 4 SUSY searches at the LHC. Giacomo Polesello. INFN, Sezione di Pavia

Transcription

1 . Lecture 4 SUSY searches at the LHC Giacomo Polesello INFN, Sezione di Pavia

2 Introduction:LHC and new physics With LHC open the TeV scale to experimentation From theoretical speculations expect to find signals for physics beyond SM In the past years many studies of possible extensions of SM For many models studied, large production cross-section, expect enough statistics for discovery in few weeks of data taking In the initial phase long time and large amount of work in order to: Master the performance of very complex detectors Understand and Control Standard Model backgrounds I will illustrate these issues applied to SUSY, leading new physics candidate, then to an interesting class of models with additional extra dimensions

3 Why physics beyond the Standard Model? Gravity is not yet incorporated in the Standard Model Hierarchy/Naturalness problem Standard Model only valid up to scale Λ < M pl (ex: M H =115 GeV Λ < 10 6 GeV ) Higgs mass becomes unstable to quantum corrections: from sfermion loops, M H (GeV) Landau pole δm 2 H λ 2 fλ 2 Potential bounded from below Λ (GeV) Additional problems: Unification of couplings, Flavour/family problem Need a more fundamental theory of which SM is low-e approximation

4 Naturalness problem and SUSY Problem: correction to higgs mass from fermion loop (coupling -λ f H ff): H f f H m 2 H λ2 f 4π 2(Λ2 + m 2 f) +... Where Λ is high-energy cutoff to regulate loop integral. If Λ M P lanck GeV radiative corrections explode Correction from scalar f, loop with coupling λ 2 fh 2 f 2, is m 2 H λ2 f 4π 2 (Λ 2 + m 2 f) +... Corrections have opposite sign, and cancel each other Full cancellation of divergences if for N f fermionic degrees of freedom one has N f scalars such that: λ 2 f = λ 2 f and m f = m f Achieved in theory where lagrangian is invariant under transformation Q: Q boson = fermion Q fermion = boson SUSY General class of theories, specialise studies to minimal model: MSSM

5 Minimal Supersymmetric Standard Model (MSSM) Minimal particle content: A spin J = ±1/2 superpartner for each Standard Model particle Two higgs doublets with v.e.v s v 1 and v 2 and superpartners. After EW symmetry breaking: 5 Higgs bosons: h, H, A, H ± If SUSY is unbroken, same mass for ordinary particles and superpartners No superpartner observed to date SUSY explicitly broken by inserting in the lagrangian all soft breaking terms The model has 105 free parameters (!) Additional ingredient: R-parity conservation: R = ( 1) 3(B L)+2S : Sparticles are produced in pairs The Lightest SUSY Particle (LSP) is stable

6 Impose phenomenological constraints (e.g FCNC suppression) to reduce SUSY breaking parameters. End up with parameters Soft parameters are three gaugino masses (M 1, M 2, M 3 ), higgsino mass (µ), tan β v 1 /v 2, sfermion masses, tri-linear couplings A. Resulting physical spectrum: quarks squarks q L, q R leptons sleptons ll lr W ± winos χ ± 1,2 charginos H ± charged higgsinos χ ± 1,2 charginos γ photino χ 0 1,2,3,4 neutralinos Z zino χ 0 1,2,3,4 neutralinos g gluino g For each fermion f two partners f L and f R for the two helicity states Charginos and neutralinos from the mixing of gauginos and higgsinos Measure masses of all neutralinos/charginos to reconstruct gaugino soft breaking parameters

7 Models of SUSY breaking Spontaneous breaking not possible in MSSM, need to postulate hidden sector Supersymmetry breaking origin (Hidden sector) Flavor-blind interactions MSSM (Visible sector) Phenomenological predictions determined by messenger field: Three main proposals, sparticle masses and couplings function of few parameters Gravity: msugra. Parameters: m 0, m 1/2, A 0, tan β, sgn µ Gauge interactions: GMSB. Parameters: Λ = F m /M m, M m, N 5 (number of messenger fields) tan β, sgn(µ), C grav Anomalies: AMSB. Parameters: m 0, m 3/2, tan β, sign(µ)

8 SUSY breaking structure SUSY breaking communicated to visible sector at some high scale m 0, m 1/2, A 0, tan β, sgn µ (msugra) Evolve down to EW scale through Renormalization Group Equations (RGE) M 1, M 2, M 3, m( f R ), m( f L ), A t, A b, A τ, m(a), tan β, µ From soft terms derive mass eigenstates and sparticle couplings. m( χ 0 j), m( χ ± j ), m( q R ), m( q L ), m( b 1 ), m( b 2 ), m( t 1 ), m( t 2 )... Structure enshrined in Monte Carlo generators (e.g ISAJET) Task of experimental SUSY searches is to go up the chain, i.e. to measure enough sparticles and branching ratios to infer information on the SUSY breaking mechanism

9 SUSY at the LHC: general features Sparticles have same couplings of SM partners production dominated by colored sparticles: squarks and gluinos Squark and gluino production cross-section only function of squark and gluino mass χ 2o χ 1+ ν ν t 1t 1 χ 2o q χ 2o g Prospino2 (T Plehn) σ tot [pb]: pp g g, q q, t 1t 1, χ 2 o χ 1+, ν ν, χ 2o g, χ 2o q q q g g S = 14 TeV NLO LO m [GeV] Production cross-section independent from details of model: σ SUSY 50 pb for m q, g 500 GeV σ SUSY 1 pb for m q, g 1000 GeV

10 Features of SUSY events at the LHC Broad band parton beam: all processes on at the same time: different from e + e colliders where one can scan in energy progressively producing heavier particles Bulk of SUSY production is given by squarks and gluinos, which are typically the heaviest sparticles If R p conserved, complex cascades to undetected LSP, with large multiplicities of jets and leptons produced in the decay. Both negative and positive consequences: Many handles for the discovery of deviations from SM, and rich and diverse phenomenology to study Unraveling of model characteristics will mostly rely on identification of specific decay chains: difficult to isolate from the rest of SUSY events SUSY is background to SUSY!

11 A SUSY event in ATLAS Multi-jet event in Bulk Region 6 jets 2 high-pt muons Large missing E T

12 vs. L1 jet P T Triggering on SUSY We do not know how SUSY will appear, use very simple, inclusive triggers The main features for RPC SUSY will be: high multiplicity of high P T jets, /E T +jets /E T might require time to be understood: in early running (10 31 cm 2 s 1 ) select SUSY with low-threshold multijet triggers Very useful to collect control samples with unbiased /E T p Bulk region SUSY after offline jet selection (jet1>100 GeV, jet2-4>50 GeV) Example: SUSY point m( q/ g) 600 GeV Require four jets with LVL1 Threshold > 25 GeV Efficiency close to one w.r.t to the offline selection. Absolute efficiency on signal 50 60%, rate 10 Hz (preliminary) ATLAS Preliminary Single jet trigger, to catch low multiplicity decays LVL1 Thresh:115 GeV, ɛ 90%, Rate 6 Hz Examples under discussion, need to match them with overall ATLAS trigger strategy

13 E miss T trigger /E T single cut with largest rejection power for SUSY Run the lowest /E T threshold compatible with budgeted rate Needed to achieve high efficiency for lowest mass points, and to put on tape control samples necessary for background evaluation At cm 2 s 1 for LVL1 /E T > 50 GeV rate 20 Hz (ATLAS) Validate /E T trigger by requiring additional high E T jet. Rejects instrumental and machine background CMS plot for cm 2 s 1 : evolution of /E T +jet rate for increasing value of jet E T CMS 1jet+Etmiss rate L1 Etmiss > 30 GeV ATLAS: /E T > 70 GeV, 1 Jet with E T > 70 GeV. Rate 20 Hz at cm 2 s 1. Etmiss (GeV)

14 SUSY discovery: basic strategy Basic assumption: discovery from squark/gluinos cascading to undetectable LSP Details of cascade decays are a function of model parameters. Focus on robust signatures covering large classes of models and large rejection of SM backgrounds g ~ ~ g t q ~ ~ t 1 q ~ χ2 0 ~ l + q ~ χ 1 ~ χ 1 0 l - l + t W W + ν l + /E T : from LSP escaping detection High E T jets: guaranteed if squarks/gluinos if unification of gaugino masses assumed. Multiple leptons (Z): Charginos/neutralinos in cascade from decays of b q q b Multiple τ-jets or b-jets (h): Often abundant production of third generation sparticles Define basic selection criteria on these variables for RPC SUSY with χ 0 1 LSP Optimisation of criteria on parameter space ongoing, will define set of topologies, and for each define sets of cuts aimed respectively at high and low SUSY masses Alternative LSP options with different signatures also under study

15 Inclusive signatures in msugra parameter space τ LSP 1 tanβ = 10, A = 0, µ > 0 0 no systematics m h = 120 GeV CMS fb (GeV) m 1/ m h = 114 GeV jet+met µ+jet+met SS 2 µ OS 2 l 2 τ Higgs 0 Z top trilept Multiple signatures over most of parameter space Dominated by /E T +jets 100 m χ = 103 GeV NO EWSB m 0 (GeV) Robust search, if signal observed in a channel, can look for confirmation in other channels ATLAS (preliminary) tried also scan in model with non universal higgs masses, with in principle different decay patterns, and result are very similar

16 M 1/2 (GeV) Discovery reach as a function of luminosity 100 pb 1 : few days of running 10 fb -1 g(2000) g(2500) E miss T Signature tan(β) = 10, µ > 0, A 0 = 0 q(2500) 1 fb 1 : first run (1-2 months) 10 fb 1 : first year at cm 2 s 1 Fast discovery from signal statistics, BU T Before we discover SUSY we need to: fb g(1500) 0.1 fb -1 q(1500) Understand early detector performance: q(2000) /E T tails, lepton id, jet scale Collect sufficient statistics of SM control samples: g(500) g(1000) q(1000) q(500) W, Z+jets, tt Two main background classes: Instrumental /E T Real /E T from neutrinos M 0 (GeV)

17 Instrumental backgrounds to /E T + jets analysis /E T from mismeasured multi-jet events: Populated by detector and machine problems Example of /E T cleaning in D0 Reject runs with detector malfunctioning Reject events with noise in the detector Remove bad cells Z( ee)+multi-jets ATLAS Preliminary ATLAS example: assume a few HV channels dead in calorimeters Tools being prepared to monitor and correct eventby-event Etmiss (GeV)

18 Instrumental background: Cleaning the sample 1 Next step is rejection of topologies which likely to yield instrumental /E T One jet is undermeasured, expect that /E T be aligned with its p T. If two-jet events, this will be measured as the second jet in the event If one jet overmeasured jet energy measurement: /E T back to back with respect to it From CMS TDR: φ jet2 φ /E T vs. φ jet1 φ /E T Left plot: Signal Right plot: QCD

19 Instrumental background: cleaning the sample 2 Scan fully simulated jet events in ATLAS (P T (jet) > 500 GeV) with /E T > 250 GeV (F. Paige, S. Willocq) /E T from: Jet leakage from cracks, Fake muons from cracks, Jet punch-through ATLAS Atlantis Event: JiveXML_5015_45309 Run: 5015 Event: ATLAS Atlantis Event: JiveXML_5015_09184 Run: 5015 Event: ρ (m) ρ (m) 10 0 Z (m) Z (m) 5 Problematic events characterised by large occupancy in muon chambers. Use this variable to clean the sample

20 MonteCarlo estimate of QCD background hard Instrumental backgrounds: data-driven estimate Requires complete understanding of detector Develop data-driven estimate (ATLAS preliminary) 1: Measure jet smearing function: Select events with: /E T > 60 GeV, φ( /E T, jet) < 0.1 jets fluctuating jet E T miss ATLAS Preliminary ETreco/ETtrue, est. Estimate true E T of jet closest to /E T as: E true T = p reco T + /E T 2: Take low /E T jet events and smear jets with measured smearing function 3: Normalize to data QCD /E T tail nicely reproduced ATLAS Preliminary All QCD Z--> SU3 Estimate (QCD)

21 Control of /E T from Standard Model processes Real /E T from ν production in SM: SUSY selection: /E T > 100 GeV At least 1 jet with p T > 100 GeV At least 4 jets with p T > 50 GeV Events/100GeV/1fb -1 ATLAS Preliminary ttbar+jets Z ( )+jets W+jets QCD jets All backgrounds SUSY signal (SU3) Plot M eff = 4 i=1 p T (jeti ) + E miss T M eff (GeV) SU3 benchmark Point: m 0 = 100 GeV, m 1/2 = 300 GeV, tan β = 6, A = 300 GeV, µ > 0 Comparable contributions from: tt+jets W +jets Z+jets Counting experiment: need precise estimate of background processes in signal region Complex multi-body final states: can not rely on MonteCarlo alone. Need both data and MonteCarlo

22 SM backgrounds: Monte Carlo issues SUSY processes: high multiplicity of final state jets from cascade decays Require high jet multiplicity to reject backgrounds: 4 jets Additional jets in tt, W, Z, production from QCD radiation Two possible way of generating additional jets: Parton showering (PS): good in collinear region, but underestimates emission of high-p T jets Matrix Element (ME): requires cuts at generation to regularize collinear and infrared divergences Optimal description of events with both ME and PS switched on Need prescription to avoid double counting, i.e. kinematic configurations produced by both techniques Additional issue: normalisation (no NLO calculation possible)

23 Final number of jets in event complicated convolution of ME, PS, and experimental definition of jets Jet was emitted collinearly Z(-> +4jets parton generated with ME jet activity after shower One parton divided into 2 jets Jet map in (, ) plane Contributions from Z+1,2,3,4,5.. jets to experimental 4-jet sample Prescriptions available (MLM, CKKW) to obtain MC predictions for experimental Z + 4 jets sample as a combination of all the exclusive Z+ n jets sample Very active field, experimental effort to see how well different prescriptions match Tevatron data BUT, tuning of matching valid for Tevatron might not be valid in LHC regime. A detailed comparison of W/Z+jets MonteCarlos with LHC data is necessary prerequisite for SUSY searches At the LHC Develop strategies based on the combined use of MC and data to correctly predict the backgrounds

24 Data driven estimates: Z νν+ jets Select a sample of Z µµ+multijets from data Same cuts as for SUSY analysis (4 jets+etmiss), throw away µ s and calculate /p T of events from µ momenta (normalized to 1 fb 1 ) Z --- data (pseudo) --- estimated By comparing Z µµ sample from data and MC, calculate normalisation factor to apply to MC Z νν sample ATLAS Preliminary Etmiss (GeV) Combination of data and MonteCarlo needed for estimate Good absolute prediction of background

25 Normalisation needs to be multiplied by BR(Z νν)/br(z ee) 6 Assuming SUSY signal Z νν bg, evaluate luminosity necessary for having N SUSY > 3 σ bg Stat error on background: σ bg = N(Z ee) BR(Z νν) BR(Z ee) fb -1 For each bin where normalisation required, need 10 reconstructed Z ll events. Need to consider acceptance/efficiency factors as well From M. Mangano Several hundred pb 1 required. Sufficient if we believe in shape, and only need normalisation. Much more needed to perform bin-by-bin normalisation M eff

26 Additional inclusive signatures /E T +jets signature is most powerful and least model-dependent BUT control of SM and instrumental backgrounds might require long time before discovery Optimize search strategy by tackling in parallel all of the inclusive discovery channels Example: single lepton + jets + /E T Smaller number of backgrounds: tt dominant,easier to control Shoulder m M eff might be observable If adequate lepton ID achieved, background dominated by SM events with Events/100GeV/1fb -1 ATLAS Preliminary ttbar ( l l )+jets ttbar ( l qq)+jets W+jets All backgrounds SUSY signal (SU3) neutrinos: tt, W +jets M eff (GeV)

27 1-lepton inclusive analysis. Control of top background Try to develop method to use top data to understand top background Preliminary ATLAS exercise (Dan Tovey) Standard semileptonic top analysis: P t (lep) > 20 GeV, /E T > 20 GeV 4 jets with P T > 40 GeV 2 b-tagged jets Very similar to cuts for SUSY analysis with looser /E T requirement If harden /E T cuts, sample contaminated with SUSY Possible approach: Select semi-leptonic top candidates (standard cuts: what b-tag available?) Fully reconstruct top events from /E T and W mass constraint obtain pure top sample with no SUSY contamination Apply SUSY selection criteria to pure top sample, and plot /E T distribution normalize pure top sample to data at low /E T obtain prediction of amount of top background at high /E T

28 Top mass reconstruction Reconstruct semi-leptonic top mass from lepton + /E T and W mass constraint ATLAS Preliminary Reduce jet combinatorics by selecting highest p T candidate ATLAS Preliminary T1 /E T and reconstructed top mass reasonably uncorrelated no bias on /E T distribution from selection on m(top) Subtract W +4 jets background under top peak using side-band Analysis based on two MC samples: T 1 (inclusive), T 2 (P top T > 500 GeV)

29 Normalising the estimate Estimate : fully reconstructed top sample after side-band subtraction Normalise estimate to SUSY selection sample, to account for relative efficiency of top selection Reminder: SUSY Selection sample: tt events with no top mass constraint /E T > 20 GeV (to be hardened later) At least 4 GeV with p T > 40 GeV Exactly 1 lepton with p T > 20 GeV ATLAS Preliminary Estimate SUSY selection T1 In low /E T region (100 GeV-200 GeV): SUSY signal expected to be small Assume low available statistics (0.5 fb 1 ) of fully simulated top Obtain scaling factor of 4

30 Background estimates Verify if method works on sample T 2 (P T (top) > 500 GeV) Compare number of events with /E T > 500 GeV in SUSY selection sample to background estimate Estimate SUSY selection ATLAS Preliminary T2 With 44 fb 1 : Found 174 ± 13 Ev (stat) Expected 198 ± 38 (stat) 20% Statistical error mainly from sideband subtraction Negligible contribution from normalisation

31 SUSY What happens if SUSY signal present? Study effect by mixing inclusive top sample and SUSY SU3 sample: Squark-gluino mass scale 600 GeV. Repeat previous steps Estimate SUSY selection (top) SUSY selection (total) ATLAS Preliminary T1 + SU3 Estimate SUSY selection ATLAS Preliminary SU3 Normalisation procedure OK for SU3 and GeV window Sideband subtraction seems to work Example of possible approach, work in progress

32 2-leptons + /E T + jets inclusive search Significantly lower reach than other channels, but also lower backgrounds Different topologies, corresponding to different SM background sources p Opposite-Sign Same-Flavour (OSSF) Opposite-Sign Opposite-Flavour (OSOF) Same-Sign Same-flavour (SSSF) Same-sign Opposite-Flavour (SSOF) OS-OF can show flavour-correlated signal. q L χ 0 2 q l± R(L) l Entries / 2 GeV - eµ ee + µµ (Flavour subtracted) fb Co-annihilation point Red: Atlas fullsim Blue: MC truth L Black: MC truth R χ 0 1 l ± M ll (GeV) Only Z/γ e + e, µ + µ has same-flavour leptons, strongly reduced by /E T +jets requirement Background dominated by tt, can be exactly flavour-subtracted Clear kinematic signature: if we are lucky best chance for early discovery

33 Inclusive analysis: critical reassessment I have shown how LHC experiment will try to discover RP conserving SUSY A certain number of generic assumptions: Detection through discovery of squark and gluino production Squark and gluino decay to jets + some kind of SU(2) U(1) gaugino/higgsino Mass difference between squark/gluino and gauginos with dominant BR such as to yield high p T jets. More or less guaranteed in case of gluino accessible and gaugino mass unification Gauginos will decay into something and finally into an invisible LSP Searches are therefore: 2 to 4 jets, depending on relation between gluino and squark masses + /E T + something Examples of something : nothing, 1,2,3 leptons (e, µ) τ (hadronic), b-jets, Z, h Generic variables: P T /η of ingredients + estimator of mass of system. Canonically: M eff = i p T (i) + E miss T

34 How generic? Typically reach shown on msugra plane (to fix the something ), but shown to cover other χ 0 1 LSP scenarios e.g NUHM Will also e.g. cover most cases in GMSB (gravitino LSP) NLSP is χ 0 1. If long lived: phenomenology as for msugra If short lived: add photons to the something If medium lived (decay inside the detector), discovery OK, need care to figure out photons NLSP is slepton/stau. If short lived OK, additional leptons in the something. If long-lived need detector-specific studies Specific searches for cases where assumption of accessible squarks/gluino breaks: light stops direct gaugino/higgsino search in 3-lepton channel long lived heavy particles (staus or R-hadrons) Also cases with very degenerate spectra need attention (see talk this week)

35 Light stops Search for direct t 1 t 1 production in models with stop lighter than top Model proposed in Les Houches 2005, compatible with CDM and baryogenesis: m( q) = m( l) = 10 TeV, and M 1 : M 2 = α 1 : α 2 M 1 = 60.5 GeV µ = 400 GeV tan β = 7 M 3 = 950 GeV m(q 3 ) = 1500 GeV m( t R ) = 0 GeV m( b R ) = 1000 GeV A t = GeV Masses: m( t 1 ) = 137 GeV, m( χ ± 1 ) = 111 GeV, m( χ 0 1) =58 GeV. BR( t 1 b χ ± 1 bw χ 0 1) = 100% Signature as for t t: 2 b-jets, /E T and either 2 l (e, µ) (4.8% BR) or 1 l+ 2 jets (29% BR) σ(pp t 1 t1) = 412 pb (CTEQ5L NLO Prospino) Consider semi-leptonic channel for analysis Perform ATLFAST analysis on 2 fb 1. ATLAS full simulation study ongoing Consider backgrounds from t t and from W bbjj production Case where m( t 1 ) < m( χ ± 1 ) much more difficult: either t 1 c χ01 or 4-body decay. First one impossible, second one still to be studied (MC generator)

36 Event selection One and only one isolated lepton (e, µ), p l T > 20 GeV. At least four jets P T (J 1, J 2 ) > 35 GeV, P T (J 3, J 4 ) > 25 GeV, E miss T > 20 GeV. Exactly two b-tagged jets with p T > 20 GeV. Assume ɛ b = 60%, R j = 100 Consider m(jj) min minimum invariant mass of any non-b jets in event with p T > 25 GeV for signal and ttbar. m(jj) min < m(χ ± ) m( χ 0 1) for signal Events/5 GeV Events/5 GeV m(jj) min m(jj) min By requiring m(jj) < 60 GeV, Achieve S/B=1/10 for signal eff. of 0.24%

37 The m(bjj) min variable Define m(bjj) min as the minimum invariant mass for the combination a b-tagged jet and the two non-b jets. The end-point of the m(bjj) min distribution should be m( t 1 ) m( χ 0 1) = 79 GeV for stop and 175 GeV for top Events/8 GeV Events/8 GeV m(bjj) min m(bjj) min Clear shoulder from stop, need precise determination of shape for top to see it An equivalent variable (m(bl) min ) can be built on the lepton side

38 Data-driven background subtraction Define a pure top control sample: require that on the lepton side m(lbν) compatible with top mass within 15 GeV. Reduce signal contamination by requiring m(lb) min > 60 GeV No signal in sample for m(bjj) min >80 GeV, can use this region for normalisation Control sample (points with errors) matches well distribution for top 300 Events/8 GeV Events/8 GeV m(bjj) min m(bjj) min After subtracting m(bjj) min for normalised top control sample from m(bjj) min for data, reproduce the distribution for signal

39 SUSY mass scale from inclusive analysis Start from multijet + /E T signature. Simple variable sensitive to sparticle mass scale: M eff = i p T (i) + E miss T where p T (i) is the transverse momentum of jet i -1 Events/50 GeV/10 fb M eff distribution for signal (red) and background (brown) (msugra m 0 = 100 GeV, m 1 /2 = 300 GeV, tan β = 10, A = 0, µ > 0) M eff (GeV) A cut on M eff allows to separate the signal from SM background The M eff distribution shows a peak which moves with the SUSY mass scale.

40 Define the SUSY mass scale as: M eff susy = M susy M χ 2 M susy, with M SUSY i M i σ i i σ i M SUSY (GeV) msugra : 5 parameters Estimate M eff peak by a gaussian fit to background-subtracted signal distributions Test the correlation of M eff with M eff susy on random sets: msugra and MSSM M SUSY 15 parameters MSSM M eff (GeV) Excellent correlation in msugra, less good for MSSM Can one think of a variable (on x or y) which gives better correlation for MSSM? M eff (GeV)

41 p Precision of mass scale estimate % precision on M SUSY vs M SUSY Evaluate uncertainty in mass scale from spread in correlation plots. 10 fb 1 - stars 100 fb 1 - open circles M SUSY (GeV) 1000 fb 1 - filled circles 10% precision on SUSY mass scale for one year at high luminosity Needs to be updated with more precise background estimates and their systematic uncertainties

42 What might we know after inclusive analyses? Assume we have a MSSM-like SUSY model with m q m g 600 GeV Observe excesses in /E T + jets inclusive, and in some of the /E T + jets + something channels. Null results in specialised searches Undetectable particles in the final state: /E T. Stable or ling-lived? Primary particles with mass 600 GeV (M eff study) Assigning spin hypotheses to produced sparticles can get an idea of couplings (exp. difficulty: need some assumption on gaugino spectrum to evaluate selection efficiency) Many more things depending on the excesses observed for the different something. Examples: Excess of of same-sign lepton pairs: some of the primary particles are Majorana See same number of leptons and muons: lepton flavour conserved in first two generations...

43 How can we use it? Too little information to zoom into a model Probably with guess the composition of the produced primary particles One can exclude detailed implementations of model (GeV) m 1/ τ LSP 1 = 114 GeV m h m χ = 103 GeV tanβ = 10, A = 0, µ > 0 0 no systematics = 120 GeV CMS fb jet+met µ+jet+met SS 2 µ OS 2 l 2 τ Higgs 0 Z top trilept NO EWSB m 0 (GeV) m h Ex. in msugra for each point one has different inclusive signatures, and one can compare observed and predicted relative rates Already quite a few theoretical attempts in this direction, e.g. LHC Olympics However, more detailed info can be extracted from the data

44 What kind of info for establishing SUSY? Long lists of requests. Need to demonstrate that: Every particle has a superpartner Their spin differ by 1/2 Their gauge quantum numbers are the same Their couplings are identical Mass relations predicted by SUSY hold Available observables: Sparticle masses, BR s of cascade decays Production cross-sections, Angular decay distributions Measurements of observables depends on detail of model and requires development of ad-hoc techniques. Over last ten years strategy based on detailed MC study of reasonable candidate models Did we focus too much on a too restricted class of models?

45 What path from the observables to the model? The problem is the presence of a very complex spectroscopy due to long decay chains, with crowded final states. Many concurrent signatures obscuring each other General strategy: Select signatures identifying well defined decay chains Extract constraints on masses, couplings, spin from decay kinematics/rates Try to match emerging pattern to template models, SUSY or anything else Having adjusted template models to measurements, try to find additional signatures to discriminate different options This is the painstaking work waiting for us if an excess is seen in inclusive analyses