Signals and Backgrounds for the LHC SUSY

Transcription

1 Signals and Backgrounds for the LHC SUSY Ian Hinchliffe LBNL July 27, 25

2 SUSY in hadron colliders Inclusive signatures provide evidence up to 2.5 TeV for squarks and gluinos. Everything is produced at once; squarks and gluinos have largest rates. Production of Sparticles with only E-W couplings (e.g sleptons, Higgs) may be dominated by decays not direct production. Must use a consistent model for simulation cannot discuss one sparticle in isolation. Makes studies somewhat complicated and general conclusions difficult to draw. LHC Strategies different from Tevatron where weak gaugino production probably dominates Studies shown here are not optimized Large event rates are used to cut hard to get rid of standard model background. Dominant backgrounds are combinatorial from SUSY events themselves. Studies shown here are not optimized; large event rates are exploited to cut hard to get rid of standard model background. Full program difficult to estimate, depends on masses and branching ratios Ian Hinchliffe Princeton July 25 1

3 General remarks Huge number of theoretical models Most general SUSY model has > 1 parameters Simulation has concentrated on cases that are qualitatively different; some examples were chosen in the expectation that they would be hard. Model determines the masses, decays and signals. A Model must be used for simulation in order to understand the problems of reconstruction Background to SUSY is SUSY itself Analysis does not depend on a particular model SUSY production is dominated by gluinos and squarks Other susy particles produced in their decays Therefore observation of direct production of sleptons and weak gauginos is difficult but not impossible Final states with e, µ, τ, γ, /E T studied Ian Hinchliffe Princeton July 25 2

4 Characteristic SUSY signatures at LHC Not all present in all models /E T High Multiplicity of large p t jets Many isolated leptons Copious b production Large Higgs production Isolated Photons Quasi-stable charged particles N.B.Production of heavy objects implies subset these signals Important for triggering considerations I show only examples, many cases have been studied ( SUGRA, GMSB, broken R parity...) Simplest models have few parameters; m 1/2 and m determine gluino and slepton masses. Ian Hinchliffe Princeton July 25 3

5 SUGRA Model Grandaddy of SUSY models Unification all scalar masses (m ) at GUT scale Unification all gaugino masses (m 1/2 ) at GUT scale Three more parameters tan β = v 1 /v 2 sign(µ) (superpotential has µh 1 H 2 ) and Trilinear term A, important only for 3 rd generation Full mass spectrum and decay table predicted Gluino mass strongly correlates with m 1/2, slepton mass with m. R parity good neutral LSP stable all events have 2 LSP s in them missing E T Gravitino has mass in TeV region: irrelevant to colliders Can relax unification assumption more parameters Ian Hinchliffe Princeton July 25 4

6 Contours of fixed gluino and squark mass Ian Hinchliffe Princeton July 25 5

7 Contours of fixed wino and slepton mass Ian Hinchliffe Princeton July 25 6

8 Where to look ' Take model seriously: really!?! Too much dark matter unless LSP light or annihilation enhanced usually by some almost degeneracy, but also by content of LSP If SUSY explains (g 2) µ, masses are small Coannihilation! "$# % & Bulk (*),+ -. / 2143 Funnel - ( )65 Focus point Ian Hinchliffe Princeton July 25 7

9 Nomenclature Bulk region: Masses low χ 1 χ 1 e+ e Connahilation region: enhanced Something almost degenerate with LSP: χ 1 e R eγ Higgs pole region: M A 2m eχ 1 Focus point region: Large m, LSP Higgsino, region is sensitive to m t op, may not exist if m top = 172. phenomenology of split SUSY is very similar to this region Caveat emptor: Don t take this too seriously Ian Hinchliffe Princeton July 25 8

10 Two typical spectra M (GeV) Point SU3 q L q R b 1 H, H+ A t 1 g χ 4, χ+ 2 χ 3 8 m [GeV] H, A H± χ χ 4 3 χ ± 2 g ũ L, d R ũ R, d L t 2 b2 b1 t 1 2 l L, τ 2 χ 2, χ+ 1 l R, τ R h 3 J = J = 1/2 χ ll ν l τ 2 lr τ 1 χ 2 χ ± 1 χ 1 Used in ATLAS studies: important decays shown SPS1a: used in many LHC/ILC comparisons h Ian Hinchliffe Princeton July 25 9

11 Several SUGRA cases studied in detail In one case unification assumptions were relaxed to investigate how signals changed (New signals appeared, old ones stayed) Some cases were restudied assuming that R-Parity was broken LSP decayed inside detector. Note typically large rates Table 1: SUGRA parameters for the six LHC points. Point m m 1/2 A tan β sgn µ σ (GeV) (GeV) (GeV) (pb) Ian Hinchliffe Princeton July 25 1

12 Point eg eχ ± eχ ± eχ eχ eχ eχ eu L eu R ed L ed R et et e b e b e L e R eνe eτ eτ eντ h H A H ± Ian Hinchliffe Princeton July 25 11

13 General Features In general m squark > m slepton, m gluino > m fw Splitting between m eel and m fer Stop is usually lightest squark Lightest SUSY particle (LSP) stable if R-parity good. LSP must be neutral if stable SUSY particles produced in pairs even if R-parity broken. SUSY production is dominated by gluinos and squarks. Not necessarily true for Tevatron. Stable LSP Missing E T Background for SUSY usually other SUSY, not Standard Model. Ian Hinchliffe Princeton July 25 12

14 Gauge Mediated Model Aims to solve FCNC problem by using gauge interactions instead of Gravity to transmit SUSY breaking Messenger Sector consists of some particles (X) that have SM interactions and are aware of SUSY breaking. Mi 2 = M 2 ± F A Simplest X is complete SU(5) 5 or 1 to preserve GUT Fundamental SUSY breaking scale F > F A, but F < 1 1 GeV or SUGRA breaking will dominate Gaugino masses at 1-loop Squark and Slepton masses at 2-loop M eg α s N X Λ M e α W NX Λ True LSP is a (almost) massless Gravitino Sparticles decay as in SUGRA, then NLSP decays to G lifetime model dependent NLSP does not have to be neutral Ian Hinchliffe Princeton July 25 13

15 6 parameters Λ, M, N 5, tan β, signµ 1 TeV < Λ F A /M < 4 TeV: Scale for SUSY masses. M > Λ: Messenger mass scale. N 5 1: Number of equivalent messenger fields. 1 < tan β < m t /m b : Usual ratio of Higgs VEV s. sgn µ = ±1: Usual sign of µ parameter. C grav 1: Ratio of M eg to value from F A, controls lifetime of NLSP. Point Λ M m N 5 tan β sgn µ C grav 1 σ (TeV) (TeV) (pb) G1a G1b G2a G2b Ian Hinchliffe Princeton July 25 14

16 Sparticle G1 G2 Sparticle G1 G2 eg eχ ± eχ ± eχ eχ eχ eχ eu L eu R ed L dr e et et e b e b e L e R eνe eτ eτ eντ h H A H ± Mass spectrum more spread out than in SUGRA m(squark)/m(slepton) bigger Ian Hinchliffe Princeton July 25 15

17 Anomaly mediated model Superconformal anomaly always present predicts sparticle masses in terms of m 3/2 Randall, Sundrum, Luty, Giudice, Wells, Murayama Simplest version predicts tachyonic sleptons! Some other SUSY breaking mechanism must be present to get realistic spectrum Add universal squark masses (mamsb) or new very heavy fields (DAMSB) (similar to gauge mediated), both variants are in ISAJET. AMSB only Most important feature M 3 > M 1 > M 2 LSP is a W and almost degenerate with χ + 1 χ + 1 χ 1 π+ with cτ < 1 cm DAMSB has very short lifetime and bigger mass difference Wells, Paige Sleptons are lighter than squarks q r χ 2 q and q l χ 1q, i.e. opposite to SUGRA and GMSB. Gravitino mass is TeV, irrelevant to LHC. Ian Hinchliffe Princeton July 25 16

18 AMSB has 4 parameters m, m 3/2, tan β, signµ DAMSB has 5 parameters M, m 3/2,,n tan β, signµ. n is the number of new fields at mass M. m 3/2 is the gravitino mass Sparticle AMSB DAMSB Sparticle AMSB DAMSB eg eχ ± eχ ± eχ eχ eχ eχ eu L eu R ed L dr e et et e b e b e L e R eνe eτ eτ eντ h H A H ± Ian Hinchliffe Princeton July 25 17

19 Start with SUGRA Look for characteristic signals jets and missing energy these should work anywhere Ian Hinchliffe Princeton July 25 18

20 Inclusive analysis Select events with at least 4 jets and Missing E T A simple variable M eff = P t,1 + P t,2 + P t,3 + P t,4 + /E T At high M eff non-sm signal rises above background note scale Peak in M eff distribution correlates well with SUSY mass scale M SUSY = min(m eu, M eg ) Will determine gluino/squark masses to 15% Events/5 GeV/1 fb M eff (GeV) Ian Hinchliffe Princeton July 25 19

21 m 1/2 (GeV) fb 1 TH 1 fb 8 1 fb 6 1 pb 4 1 pb 2 EX A =, tanβ = 35, µ > m (GeV) CMS Ian Hinchliffe Princeton July 25 2

22 Can also add lepton(s): More channels more robustness m 1/2 (GeV) msugra reach in various final states for 1 fb -1 Charged LSP q ~ (2) g ~ (3) l + 1l + 2l OS 3l 1l 2l OS l 2l SS g ~ (2) A =, tanβ = 35, µ > miss E T g ~ (25) h(123) g ~ (15) q ~ (25) h (114) mass limit q ~ (1) q ~ (15) g ~ (1) 2 1 q ~ (5) g ~ (5) Chargino Searches at LEP No symmetry breaking m (GeV) CMS Ian Hinchliffe Princeton July 25 21

23 Energy of LHC is most crucial: reach increases slowly with luminosity m 1/2 (GeV) miss msugra reach in E T + jets final state Charged LSP q ~ (2) g ~ (3) 1 year, high lumi (1 fb -1 ) 1 year, low lumi (1 fb -1 ) A =, tanβ = 35, µ > 3 years, high lumi (3 fb -1 ) g ~ (25) g ~ (2) h(123) q ~ (25) month, low lumi (1 fb -1 ) g ~ (15) h (114) mass limit q ~ (5) q ~ (1) q ~ (15) g ~ (1) g ~ (5) 1 Chargino Searches at LEP No symmetry breaking m (GeV) Should get nervous if no signal in few inverse fb Ian Hinchliffe Princeton July 25 22

24 How robust is this? Backgrounds based on showering MC may underestimate multi-jet final states. May lower reach slightly Plot shows all jet state Signal in Lepton+jets+missin more robust 15 Asai et al Ian Hinchliffe Princeton July 25 23

25 SUSY Peak in M eff distribution correlates well with SUSY mass scale the ge = min(m, M ~ q ~) (GeV) g M SUSY M SUSY = min(m eu, M eg ) Effective Mass (GeV) Use this and similar global distributions to establish that new physics exists and Correlation line from: determine its mass scale Hinchliffe, Paige et al., PRD55 (1997) 552; Method D. Tovey, is slightly Ph. Lett. model B498 dependent (21) 1 8 of 17 Ian Hinchliffe Princeton July 25 24

26 Generalizations to other models Similar method works in GMSB and MSSM In MSSM, 15 parameters were varied Events selected to have no isolated leptons, at least 4 jets, large missing E T More global variables were used; best is Error is bigger in MSSM jets E T or jets E T + /E T Ian Hinchliffe Princeton July 25 25

27 M eff susy (GeV/c 2 ) M eff susy (GeV/c 2 ) M eff susy (GeV/c 2 ) 15 (a) 1 5 msugra M est (GeV/c 2 ) 15 (b) 1 5 MSSM M est (GeV/c 2 ) 15 (c) 1 5 GMSB M est (GeV/c 2 ) Model Var x σ σ/ x Prec. (%) msugra MSSM GMSB σ(m susy < 13%) Not optimized Leptonic channels not used More work on global signatures needed Ian Hinchliffe Princeton July 25 26

28 What about 1fb 1 E How Useful Is This?? F, H, J, K, M Covers preferred region L G I B C D A Now for examples of specific final states... Ian Hinchliffe Princeton July 25 27

29 Digression on Simulations Varying degrees of sophistication Run an event generator: Makes a list of particles (e.g Pythia) Full simulation Full material description of detector Track particle: model its interactions, follow all the secondaries with energy above some threshold Translate hits into simulated electronic signals Now event looks like a real event: reconstruct it using same software Advantage: Full description Disadvantage: Slow ( 15 mins per event), can only be done by experimenter. Problems, Low energy hadronic interactions, geometry is hard Fast simulation (theorists version) Assume perfect detector: Apply jet finding algorithm Smear, electron, muon, jet momenta and missing ET. using some resolution function Ian Hinchliffe Princeton July 25 28

30 Advantage. Very fast (comparable to generation time), can by done by theorist Disadvantage: Only as good as resolution function: problems in tails Fast simulation (experimenters version) Response of individual particles is parametrized. Might use full for some particles (e.g. Photons) and parametrized for others (pions) May use full reconstruction. In practice all are used: Theorists version is often good enough for evaluating a model Ian Hinchliffe Princeton July 25 29

31 Characteristic SUSY Decays Illustrate techniques by choosing examples from case studies. Both q and g produced; one decays to the other Weak gauginos ( χ i, χ ± i ) then produced in their decay. e.g. q L χ 2 q L Two generic features χ 2 χ 1 h or χ 2 χ 1 l+ l possibly via intermediate slepton χ 2 l + l χ 1 l+ l Former tends to dominate if kinematically allowed. Use these characteristic decays as a starting point for mass measurements Many SUSY particles can then be identified by adding more jets/leptons Ian Hinchliffe Princeton July 25 3

32 Decays to Higgs bosons If χ 2 χ 1h exists then this final state followed by h bb results in discovery of Higgs at LHC. In these cases 2% of SUSY events contain h bb Event selection /E T > 3 GeV 2 jets with p T > 1 GeV and 1 with η < 2 No isolated leptons (suppresses tt) Only 2 b-jets with p T,b > 55 GeV and η < 2 R bb < 1. (suppresses tt) Clear peak in bb mass Very small standard model background (pale) Dominant background is other SUSY decays (dark) Ian Hinchliffe Princeton July 25 31

33 Generally applicable This method works over a large region of parameter space in the SUGRA Model Hatched region has S/ B > 5 Contours show number of reconstructed Higgs Channel is closed at low m 1/2 Over rest of parameter space, leptons are the key... Ian Hinchliffe Princeton July 25 32

34 Starting with Leptons Isolated leptons indicate presence of t, W, Z, weak gauginos or sleptons Key decays are χ 2 l + l and χ 2 χ 1 l + l Mass of opposite sign same flavor leptons is constrained by decay Decay via real slepton: χ 2 l + l Plot shows e + e + µ + µ e ± µ Decay via virtual slepton: χ 2 χ 1 l + l and Z from other SUSY particles Ian Hinchliffe Princeton July 25 33

35 Building on Leptons Decay q L q χ 2 q ll qll χ 1 Identify and measure decay chain 2 isolated opposite sign leptons; p t > 1 GeV 4 jets; one has p t > 1 GeV, rest p t > 5 GeV /E T > max(1,.2m eff ) Mass of qll system has max at (M 2 llq = [ eq L M 2 )(M 2 eχ 2 eχ 2 M max M 2 eχ 2 M 2 eχ 1) ] 1/2 = GeV and min at 271 GeV Ian Hinchliffe Princeton July 25 34

36 25 2 Events/2 GeV/1 fb M llq (GeV) smallest mass of possible lljet combinations Kinematic structure clearly seen Can also exploit ljet mass largest mass of possible lljet combinations Ian Hinchliffe Princeton July 25 35

37 Can now solve for the masses. Note that no model is needed Very naive analysis has 4 constraints from lq, llq upper, llq lower, ll masses 4 Unknowns, m ql, m er, m eχ, m 2 eχ 1 Errors are 3%, 9%, 6% and 12% respectively m l (GeV) correlations m er vs. m eχ 1 m 1 (GeV) dp/m 1 (/1GeV) LSP mass m 1 (GeV) Mass of unobserved determined Errors are strong and a precise in determination of o reduces the errors on Ian Hinchliffe Princeton July 25 36

38 Right squarks s-tranverse mass. Definition Select Events: pp X + q R q R X + q q χ 1 χ 1 2 Jets with P T >2GeV (j1-j2)>1 Missing E T >4GeV ~ χ 1 q~ R q~r Jet 2 Jet 1 ~ χ 1 Partition E T = E T,1 + E T,2 in all possible ways and compute: M 2 T = min ET,1, E T,2 [ max{m 2 T (P T,j1, E T,1, M χ 1 ), m2 T(P T,j2, E T,2, M χ 1 )} ] M 2 T depends on the choice of M( χ 1) Measured from dilepton and dilepton+jet edges (see next talk) Ian Hinchliffe Princeton July 25 37

39 s-tranverse mass for the SU5 point Fit the end-point with a straight line and extrapolate to the x-axis Use true value of M( χ 1) (Ola Kristoff Oye) L dt = 14.2 fb L dt = 18.1 fb Mass (GeV) From the fit: M( q R ) = 156 ± 181GeV Generator: M( q R ) = 119GeV Again Limited by LSP mass uncertainty Mass (GeV) From the fit: M( q R ) = 1113 ± 164GeV Generator: M( q R ) = 121GeV Ian Hinchliffe Princeton July 25 38

40 Final states with taus Large tan β implies that m( τ ) < m( µ) Taus may be the only produced leptons in gaugino decay. Leptonic tau decays are of limited use where did lepton come from? Use Hadronic tau decays, using jet shape and multiplicity for ID and jet rejection. Full simulation study used to estimate efficiency and rejection Rely on Jet and E t (miss) cuts to get rid of SM background Measure visible tau energy Event selection 4 jets, one has p t > 1 GeV, rest p t > 5 GeV No isolated leptons with p t > 1 GeV /E T > max(1,.2m eff ) Ian Hinchliffe Princeton July 25 39

41 Look at mass of observed tau pairs Events/2.4 GeV/1 fb Real signal visible above fakes (dashed) and SM (solid) M ττ (GeV) Can use peak position to infer end point in decay χ 2 τ τ χ 1 Estimate 5% error Large tan β light sbottom Look for these (61 GeV) Ian Hinchliffe Princeton July 25 4

42 g b b bbτ ± τ χ 1 Previous sample with 2 b jets having p t > 25 GeV Lots of missing E T : tau decays and χ 1 s Select 4 < m τ τ < 6 GeV Combine with b jets Look at τ τ bb and τ τ b: should approximate gluino and sbottom use partial reconstruction technique assuming mass of χ 1 Peaks are low; should be expected due to missing energy -1 2 Events/bin/ 1fb Plot of m( χ 2 bb) vs m( χ 2 bb) m( χ 2 b) 6 M(χ 2 bb) (GeV) M(χ 2 bb) -M(χ 2 b) (GeV) 2 Ian Hinchliffe Princeton July 25 41

43 Projections 4 Events/1 GeV/ 1 fb Events/35 GeV/1 fb M(χ 2 bb)-mχ 2 b) (GeV) gluino-sbottom should be 16 GeV gluino 54 GeV M(χ 2 bb) (GeV) Ian Hinchliffe Princeton July 25 42

44 Lepton Universality e/µ/τ universality must be tested Large β m τ < mẽ Expect larger rate of τ + τ and hence µ + e Plot of invariant mass distributions µ + e (dashed) and µ + µ (solid) from SUSY cascades vs. tan β µ + µ structure less distinct at large tan β. Denegri, Majerotto, Rurua Ian Hinchliffe Princeton July 25 43

45 Explicit flavor violation is also possible Neutrino oscillations imply lepton number is violated Atmospheric muon neutrino deficit implies ν mu ν τ with maximal mixing In a SUSY model, expect significant flavor violation in slepton sector Simplest model of lepton number violation involves addition of right handed neutrino N with SUSY conserving Majorana mass mnn and coupling to lepton left doublet and Higgs of the form LNH Including only µ τ mixing gives M 2 e l e l = M 2 L + D L M 2 L + D L M 2 µτ M 2 µτ M 2 τ L + D L m τ Ā τ M 2 R + D R M 2 R + D R m τ Ā τ M 2 τ R + D R Atmospheric neutrinos suggest maximal mixing i.e. δ = O(1) Ian Hinchliffe Princeton July 25 44

46 Two types of flavor violation production ( χ 2 τ µ) and decay ( τ χ 1 µ) Branching ratio for χ 2 χ 1 µτ and for χ 2 χ 1 µµ through an intermediate τ 1 Must use hadronic tau decays to distinguish BR µτ 1 5 µµ δ Signal for lepton number violation comes by comparing µτ h and eτ h final states Ian Hinchliffe Princeton July 25 45

47 3 Events/1 GeV/1 fb l ± τ h signal ( red ), l± τ ± h signal (blue), µ ± τ h from LFV decays with BR = 1% 1 (magenta), and Standard Model l ± τ h M lτ (GeV) Lepton number violating decay χ 2 µτ h χ 1 give harder µτ mass distribution than that from χ 2 τ τ χ 1 µτ h χ 1 Ian Hinchliffe Princeton July 25 46

48 Subtraction removes background 16 Events/1 GeV/1 fb l ± τ h l± τ ± h ( red ) and µ± τ h from LFV decays with BR = 1% (magenta) Signal is established from E = N(µ ± τ h ) N(e± τ h ) 1 fb 1 and 5σ implies BR=2.3% or δ.1 well within value needed for neutrino data M lτ (GeV) Sensitive provided that χ 2 production is large enough (large fraction of parameter space More sensitive than µ eγ Ian Hinchliffe Princeton July 25 47

49 R-parity broken Implies either Lepton number or Baryon number is violated and LSP decays Either χ 1 qqq, or χ 1 qql or χ 1 l+ l ν First two have no /E T, last 2 have more leptons and are straightforward First case is hardest, Global S/B is worse due to less /E T Example, SUGRA with χ 1 qqq Leptons are essential to get rid of QCD background 8 jets with p t > 5 GeV 2 OSSF isolated leptons. S T >.2, selects ball like events Σ jets+leptons E T > 1 TeV Ian Hinchliffe Princeton July 25 48

50 Dilepton mass still shows clear structure with small background from χ 2 l+ l χ 1 Events/3 GeV/3 fb M ll (GeV) As nothing is lost, should be possible to reconstruct χ 1 Difficult because jet multiplicity is very high and χ 1 mass is usually small, so jets are soft Ian Hinchliffe Princeton July 25 49

51 8 jets with p t > 17.5 GeV, 8 jets with p t > 25 GeV 2 jets with p t > 1(2) GeV and η < 2 1 or 2 leptons with p t > 2 GeV Sphericity cut combine 6 slowest jets into 2 sets of 3; require M(jjj) 1 M(jjj) 2 < 2 GeV Events/1 GeV/3 fb Nominal mass 122 GeV m jjj (GeV) Ian Hinchliffe Princeton July 25 5

52 Can cut around peak and combine with either leptons or quarks reconstruct q R q χ 1 ( qqq)) and χ 2 ll χ 1 Events/2 GeV/3 fb m(χ 2 ) = 212 GeV m(χ 2 ) = 252 GeV m jjjll (GeV) Plot shows χ 2 Note that tight cuts imply low event rate (analysis not optimized) Ian Hinchliffe Princeton July 25 51

53 New signals in GMSB Lightest superpartner is unstable and decays to Gravitino ( G) Either neutral χ 1 γ G : cτ C 2 (1 GeV/M χ ) 5 (Λ/18TeV) 2 (M M /18TeV) 2 mm 1 extra photons or similar signals to SUGRA depending on lifetime Or charged Almost always slepton: e R e G No Missing E T if cτ large, events have a pair of massive stable charged particles ( G2b ) Large lepton multiplicity if cτ small ( G2a ). Discovery and measurement in these cases is trivial In case G2b, every decay product can be measured In case G1a G momenta can be inferred and events fully reconstructed. Ian Hinchliffe Princeton July 25 52

54 GMSB case 1a: Event selection (not optimized) Decay χ 2 l+ l χ 1 l+ l γ G is key Lifetime of χ 1 is short Find jets Require M eff > 4 GeV; /E T >.1M eff. Looking for M eff /E T + p T,1 + p T,2 + p T,3 + p T,4. χ 2 l ± l χ 1 l± l Gγl ± l, Electrons and photons : p T > 2 GeV Muons : p T > 5 GeV. Require at least 2 photons and two leptons. Ian Hinchliffe Princeton July 25 53

55 Dilepton mass distribution, flavor subtracted e + e + µ + µ e ± µ 2 ee+µµ-eµ Events/2.5 GeV/1 fb End is at M eχ 2 ( MelR 1 M eχ 2 ) 2 1 ( ) 2 M eχ 1 = 15.1 M elr M ll (GeV) Ian Hinchliffe Princeton July 25 54

56 15 ee+µµ-eµ Events/2.5 GeV/1 fb Form l + l γ mass and take smallest combination. Linear vanishing at M 2 eχ 2 M 2 χ 1 = GeV, M llγ (GeV) Ian Hinchliffe Princeton July 25 55

57 ee+µµ-eµ Events/2.5 GeV/1 fb Form l ± γ mass also. Two structures at and M 2 l R M 2 χ 1 = GeV M 2 χ 2 M 2 l R = GeV M lγ (GeV) Ian Hinchliffe Princeton July 25 56

58 These four measurements are sufficient to determine the masses of the particles ( χ 2, l R, and χ 1 ) in this decay chain without assuming any model of SUSY breaking. Now use this to reconstruct the decay chain and measure the G momenta despite the fact that there are two in each event and both are invisible! Ian Hinchliffe Princeton July 25 57

59 Full reconstruction of SUSY events Know masses can calculate p assuming p 2 = : 2p k 2 p k = M 2 eχ 1 2p l 2 p l = M 2 e lr M 2 eχ 1 2k l 2p k 2 p q = M 2 eχ 2 M 2 e lr 2(k + l) q C fit with 2 2 solutions. Event has two of these decays so require 4 leptons and 2 gammas Ian Hinchliffe Princeton July 25 58

60 Calculate missing E T Form a χ 2 using measured missing E T to resolve ambiguities χ 2 = ( ) 2 /Ex p 1x p 2x + /E x ( ) 2 /Ey p 1y p 2y. /E y use /E x = /E x =.6 E T +.3E T. 15 Events/.2/1 fb Compare to generated G momenta Plot shows all solutions with χ 2 < 1 p = p G p reconst p / p 1% p/p Ian Hinchliffe Princeton July 25 59

61 Squark and Gluino Masses Use measured χ 2 momenta and combine with jets q gq χ 2 qqq Require at least 4 jets with p T > 75 GeV 15 Events/4 GeV/1 fb Figure shows mass of χ 2 +2 jets; peak is below gluino mass (747 GeV); no correction applied for small jet cone M χjj (GeV) Much easier than the SUGRA cases; masses measured directly Ian Hinchliffe Princeton July 25 6

62 Measuring the fundamental scale of SUSY breaking Lifetime of χ 1 G is important as it measures the fundamental scale of SUSY breaking Measure lifetime of χ 1 ( Gγ) using Dalitz decay χ 1 e+ e γ G Works for short lived χ 1 Statistics limited ( few-k events) Measure lifetime of χ 1 ( Gγ): photon pointing. Angular resolution of photons from primary vertex (ATLAS) θ 6mr/ E Detailed study of efficiency for non-pointing photons Important for long lived χ 1 Decays are uniformly distributed in the detector Cross check from time delay of decay Failure to see photons cτ > 1 km or F 1 4 TeV Ian Hinchliffe Princeton July 25 61

63 Mass measurement of quasi-stable sleptons ATLAS Sleptons are produced at the end of decay chains large velocity Most of these will pass the Muon Trigger Measure the velocity using TOF in Muon system, then infer mass Time resolution 65 ns M/M 3% for M = 1 GeV / 22 Constant Mean 12.2 Sigma 3.82 Events/.2/1 fb Events/.5 GeV/1 fb velocity β M (GeV) M/M Ian Hinchliffe Princeton July 25 62

64 How well does this work in CMS? Three cases studied m stau = 14, 33, 636 GeV 1/β particles / 2 GeV m(τ 1 )=114GeV; L=1/fb; eff=5% m(τ 1 )=33GeV; L=1/fb; eff=15% m(τ 1 )=636GeV; L=1/fb; eff=26% momentum (GeV) reconstructed mass (GeV) Ian Hinchliffe Princeton July 25 63

65 AMSB Look at paper of Wells and Paige Ian Hinchliffe Princeton July 25 64

66 Odd Ball: R-hadron (split susy) Long lived or quasi stable gluino produces cannon ball that charge exchanges as it PYTHIA R-hadron event from ATLSIM R-hadron Interactions of R-hadron giving off low energy pions R-hadron passes through detector Ian Hinchliffe Princeton July 25 65

67 Spin measurements at LHC? Conventional wisdom says that you need LC for this but... q L q L l R + (near) χ 2 l R - χ 1 l R - (far) Angle between q and e in χ 2 rest frame is sensitive to spin correlations. But effect washes out if we do not know which lepton comes out first: Use kinematics But effect washes out if average over q and q: LHC is a pp machine: more q than q Ian Hinchliffe Princeton July 25 66

68 Form an asymmetry from invariant mass distribution of lepton and jet A = (l+ q) (l q) (l + q)+(l q) A +-.5 Green:spin correlation off -.5 Yellow: No detector (.6) Needs at least 1fb Barr m lq / GeV Ian Hinchliffe Princeton July 25 67

69 Difficult cases I: Small mass gaps Co-annihilation region: Near degeneracy between LSP and sleptons.soft leptons and more messy decays. q L q χ 2 then χ 2 (26) l R (153)l ll χ 1 (136) and χ 2 l L (255)l ll χ 1 (136) Leptons can still be found despite small mass gaps Entries / 2 GeV - eµ ee + µµ ATLAS M ll (GeV) Ian Hinchliffe Princeton July 25 68

70 Conclusions Serious thinking has started about what might be done at 1 35 and what machine and detector upgrades are needed. Ian Hinchliffe Princeton July 25 69

71 Discovery cannot be far off The train is already late (Altarelli): Fine tuning is a problem already. We expected gauginos in the LEP range Tevatron window is small but low masses are more likely Ian Hinchliffe Princeton July 25 7

72 Discovery cannot be far off The train is already late (Altarelli): Fine tuning is a problem already. We expected gauginos in the LEP range Tevatron window is small but low masses are more likely An era is about to end Low energy SUSY has provided employment for > 2 years It will be discovered or die in the next 6 years. Ian Hinchliffe Princeton July 25 7