Search for Supersymmetry Using Rare B 0 s(d)  m + m - Decays at CDF Run II

1. Search for Supersymmetry Using Rare B0s(d)m+m-Decays at CDF Run II Vyacheslav Krutelyov Department of Physics Texas A&M University HEP Seminar Dec 20, 2005 Fermilab

2. Motivations Standard Model Supersymmetry Br(Bsμ+μ-) in SM. Br(Bsμ+μ-) in SUSY Loops: large tanb: mSUGRA,SO(10) Tree: R-parity violating models Analysis Strategy CDF Run II Dimuon Trigger Sample Br(Bsμ+μ-) measurement. Ingredients. Likelihood discriminant Background estimate. Signal efficiency × acceptance Optimization. Results SUSY implications. mSUGRA SO(10) RPV Prospects for Bsμ+μ- Summary. Outlines V. Krutelyov Search for SUSY using Bsmm

3. The SM describes the matter in terms of elementary particles Quarks of 6 flavors Leptons: charged (massive) and neutral (massless neutrinos) and interactions are mediated by force carriers Strong ↔ gluons Electroweak  broken (by Higgs)  weak (W+ and Z0) and electromagnetic (photon) with Masses defined by interaction with Higgs scalar Quarks (except t) form hadrons: qqq (baryons) q-anti-q (mesons) Bs (b anti-s), Bd (b anti-d) and B+ (u anti-b) are b-mesons with mass ~5.3GeV and lifetime ~1.6 ps will be often referred later Hadrons change flavor only via W+ exchange Standard Model + Higgs V. Krutelyov Search for SUSY using Bsmm

4. Standard Model (status) + Higgs • SM describes all collider physics observations • (has been for the past 20+ years) BUT: • neutrinos have masses • most of the matter in the Universe is not in SM • SM does not include gravity • SM has naturalness (hierarchy) problems • Need for physics beyond the SM  Supersymmetry is (one of) the best choice solves gauge hierarchy problem gives a candidate for dark matter can explain neutrino masses … V. Krutelyov Search for SUSY using Bsmm

5. Supersymmetry • SUSY requires for each particle to be a superpartner • SM superpartners: squarks and sleptons (s=0); gauginos and higgsinos (s=1/2) • SUSY is broken at low energy • MSSM: minimal SUSY SM extension (with 2 Higgs doublets) • EWSB w Higgs give masses 2 Higgs Doublets: tanb=vevu/vevd SUSY Charged Current Stop L-R mixing LSP– Cold Dark Matter Candidate V. Krutelyov Search for SUSY using Bsmm

6. Importance of Bs(d)mm • Bs(d)mm is a Flavor Changing Neutral Current process (FCNC) • Quarks b and s are of the same electric charge  hadronic state flavor change (b anti-s)(mm) has no charge transfer • FCNC processes are the benchmarks of the theory • Defined the construction of the SM • Constrain substantially any New Physics • bsg has been a golden FCNC mode for years • Bsmm is suggested as a new golden FCNC mode V. Krutelyov Search for SUSY using Bsmm

7. Bsmm in Standard Model • In the Standard Model, the FCNC decay of B m+m- is heavily • suppressed SM prediction  (Buchalla & Buras, Misiak & Urban) • Bdmm is further suppressed by CKM coupling (Vtd/Vts)2≈40 • Br(Bsmm) < 4.1×10-7 @ 90% CL ;D0 PRL 94 (2005) 042001 (240pb-1) • SM prediction is below the sensitivity of current experiments • SM  Expect to see 0 events at the Tevatron Any signal would indicate new physics!! V. Krutelyov Search for SUSY using Bsmm

8. Bmm in SUSY: large tanb • In many SUSY models, the BR could be enhanced by many • orders of magnitude: • For example: • - MSSM: Br(Bmm) is • proportional to tan6b • - GUT SO(10) models prefer • tanb≈50 (Yukawa coupling unification) • -BR could be as large as • 10-1000 times the SM prediction Could be observable at the Tevatron V. Krutelyov Search for SUSY using Bsmm

9. Bmm in SUSY: RPV tree • Another example: R-Parity violating (RPV) SUSY • R-parity ↔ SUSY particles come in pairs • - Tree level diagram is allowed in • R-parity violating (RPV) SUSY • models. • - No significant tanb dependence • - Enhancement depends strongly • on coupling constants (l , l’) • correspond to lijkqiqjnk l’ijkliljnk interactions m b ~ n l’i23 l i22 m s Could also be observable at the Tevatron V. Krutelyov Search for SUSY using Bsmm

10. Bsmm and Bdmm Monte Carlo • New physics may enhance Bs and • Bdmm differently • Minimal-flavor-violation (MFV) • assumption in SUSY yields SM • relations between Bs and Bdmm: • Br(Bs):Br(Bd) ≈ (Vts/Vtd)2≈40 • Can observe both Bs and Bd: unique to • Tevatron (Bd only at B-factories) • CDF has the mass resolution to • distinguish two decays, s(Mmm)≈23MeV : • unique to CDF • Either observation or null search, will provide important • clues about possible scenarios of new physics beyond SM M(Bs)-M(Bd)~90MeV V. Krutelyov Search for SUSY using Bsmm

11. Bs(d)mm Strategy _ p _ ppbb • Use p anti-p collisions at 1.96 TeV at Tevatron to produce Bs(d) mesons • B-hadrons are produced as a result of (p anti-p)(b anti-b) • Hadrons evolve out of b with N(B+):N(Bd0):N(Bs0):N(Lb)=fu:fd:fs:fbar~4:4:1:1 • N(Bsmm) = fs N(Hb) BR(Bsmm) • Use CDF detector to detect the muons from Bs(d) • of all Bs [N(Bsmm)] detect only n(Bsmm)≡ (ae) N(Bsmm) • (ae) – acceptance × efficiency • Hide signal data while choosing the best selections • Define N(Hb) from normalization mode N(B+J/yK+mmK+) = fu N(Hb) BR(B+mmK+) BR(B+mmK+) ≈ 6×10-5 p _ V. Krutelyov Search for SUSY using Bsmm

12. CDF II Components relevant to the analysis are highlighted 33o to 53o from ┴ Central Muon Extension: CMX(0.6< |h| < 1) Central Drift Chamber (COT) (|h| < 1) 0o to 53o from ┴ Silicon Vertex Detector Superconducting Solenoid (1.4T) Central Muon Chambers: CMU, CMP(|h| < 0.6) 0o to 33o from ┴ V. Krutelyov Search for SUSY using Bsmm

13. Typical Dimuon Event V. Krutelyov Search for SUSY using Bsmm

14. Data Sample • Using 364pb-1 of data (Feb02 – Aug04) from Rare B di-muon triggers: • - CMUCMU 2 muons 0o to 33o from ┴ • - CMUCMX 1 muon 0o to 33o from ┴ 1 muon 33o to 53o from ┴ • CMUCMU and CMUCMX channels treated independently • in this analysis (background and efficiencies are different) Background is huge Rare B di-muon triggers require additional cuts to reduce background relative to inclusive J/y di-muon trigger Search region V. Krutelyov Search for SUSY using Bsmm

15. Bsmm Measurement: Ingredients Overall picture: - Reconstruct di-muon events in the B mass window - Measure the branching ratio or set a limit - Normalize to B+J/y K+ decays Nobs=Nbg+NBs • Key elements in the analysis: • - Construct discriminant to select Bs signal and suppress bgd • -MC simulation for signal and mass sidebands for bgd estimate • - understanding the background • - accurately measure the acceptance and efficiency ratios • Analysis optimization: • Figure of merit  expected 90% C.L. upper limit • Perform unbiased optimization V. Krutelyov Search for SUSY using Bsmm

16. Normalization mode: B+J/yK+ Count the # of B+m+m-K+ candidates with |Mmm-3097|<50 MeV/c2 MJ/y • Selection Requires: • pT(B)>4 GeV && |y(B)|<1 • pT(K+)>1GeV • l/s(l) > 2 • well-measured displaced vertex • l = proper decay length • [l (B+) ≈502mm] CMUCMU (mmmK) B+mmK+ in rare B trigger Sample: N(CMUCMU) = 1767±59 N(CMUCMX) = 698±39 V. Krutelyov Search for SUSY using Bsmm

17. Bmm Signal Mode: “baseline selections” • Pre-selection requires: • pT(B)>4 GeV && |y(B)|<1 • l/s(l) > 2 • well-measured displaced vertex Bd Bs • [l (Bs) ≈468mm] CMUCMU Bsmm Search Sample: N(CMUCMU) = 22459 N(CMUCMX) = 14305 (completely Bgd dominated) Background shapes are linear for both channels “Baseline” cuts selectquality di-muons that should have passed the trigger, and have awell-measured vertexconsistent with long-lived b-hadron decay to reject events that are clearly background (non-trigger or mis-reconstructed) V. Krutelyov Search for SUSY using Bsmm

18. Signal vs Background Gluon Splitting g bb mm  Pmm y Bs mm DR < 1 (a <57o) m- m+ _ Da _ n c  c  Pm Pm n y _ b b g x  g L3D _ x q q z z Backgrounds are random combinations of (fake)muons The best look-alike is the “gluon splitting” V. Krutelyov Search for SUSY using Bsmm

19. Discriminating Variables To select Bmm events and suppress background use: • Invariant m+m- mass, M: |M-MB|<60 MeV/c2 (2.5s) sidebands: [4669, 5169] U [5469, 5969] MeV/c2 (0.50.5 GeV/c2) signal: |M-5279|<60 MeV/c2 (Bd0) or |M-5279|<60 MeV/c2 (Bs0) • Proper decay-length (l): displacement from production vtx in B rest frame • Isolation (Iso): (fraction of pT from Bmm within DR=(Dh2+Df2)1/2 cone of 1) • “pointing (Da)”: (3D opening angle between Bs momentum and decay axis) V. Krutelyov Search for SUSY using Bsmm

20. Discriminating Variables cut cut To further reduce Bgd, apply the additional cuts: Da<0.70 rad && Iso > 0.50 eff(signal) ≈ 92% This leaves in m+m- data: (4.7GeV < m < 6 GeV) N(CMUCMU) = 6242 N(CMUCMX) = 4908 ~x3 down in bkg but still… ~1.3K in signal window V. Krutelyov Search for SUSY using Bsmm

21. Likelihood Ratio Discriminant analog of • Use a likelihood ratio method: Ps/b is the probability for a given sig/bgd to have a value of x, where i runs over all discriminating variables. • The chosen variables (xi) are: • - isolation (iso) • - 3D pointing (Da), • - proper decay length probability [P(l)=exp(-l/lBs)] • Ps/b (PDFs) are constructed from the data sideband for background • and Pythia MC for signal • LH has among the best discriminating power if xi are uncorrelated V. Krutelyov Search for SUSY using Bsmm

22. Correlations between variables for Bgd Correlations between discriminating variables are negligible allows to reduce Stat uncertainty for expected bgd estimate V. Krutelyov Search for SUSY using Bsmm

23. Likelihood PDFs CMU-CMU Signal and background PDFs for: Isolation Pointing angle Proper decay length probability * Similar distributions for CMU-CMX V. Krutelyov Search for SUSY using Bsmm

24. Likelihood Ratio: Signal vs Bgd Likelihood ratio has strong discriminating power between signal and background V. Krutelyov Search for SUSY using Bsmm

25. Estimate Background Extrapolate number of events in the side-bands to the signal region to estimate expected background Assumes bgd events inside signal window look the same as outside x-check this using MC and control samples Bhh (h=K,p) decays are negligible • Nbg = #events in signal region surviving all requirements • NSB = #events in mass sidebands surviving pre-LH cuts • Rmass = WidthSignal / WidthSideBand = 0.12 • Assumes linear mass distribution • RLH = fraction of bgd events expected to survive LH cut V. Krutelyov Search for SUSY using Bsmm

26. Estimate Bgd Rejection by LH cut (RLH) Since discriminating variables are uncorrelated, use toy MC to estimate RLH based on input distributions from data SB Likelihood Ratio Data vs Toy MC Likelihood Ratio Rejection from Toy MC RLH*103 Cut CMU-CMU CMU-CMX LH>0.85 24.5±0.5 22.6±0.5 LH>0.92 13.0± 0.4 12.0±0.3 LH>0.99 1.4±0.1 1.5±0.1 (Errors are stat only) LH strongly suppresses bkg KS-Prob(CMUCMU)=11% KS-Prob(CMUCMX)=5% V. Krutelyov Search for SUSY using Bsmm

27. Acceptance and Efficiency • a(B+/Bs)= 0.297 ± 0.008 (CMUCMU) • = 0.191 ± 0.006 (CMUCMX) • etrig(B+/Bs) = 0.9997 ± 0.0016 (CMUCMU) • = 0.9986 ± 0.0014 (CMUCMX) Pythia MC J/ymm Data • ereco-mm(B+/Bs) = 1.00 ± 0.03 (CMUCMU/X) • evtx(B+/Bs) = 0.986 ± 0.013 (CMUCMU/X) • ereco-K(B+) = 0.938 ± 0.016 (CMUCMU/X) DataMC ≈1 var in optimization V. Krutelyov Search for SUSY using Bsmm

28. Efficiency of LH for Bsmm Signal • determined from Bsmm MC • MC modeling checked by comparing eLH(B+) • between MC and sideband subtracted Data eLH(Bs) cut CMU-CMU CMU-CMX LH>0.90 (68±1)% (66±1)% LH>0.92 (65±1)% (65±1)% LH>0.95 (59±1)% (60±1)% LH>0.98 (45±1)% (48±1)% LH>0.99 (35±1)% (39±1)% (stat uncertainties only) V. Krutelyov Search for SUSY using Bsmm

29. Optimization and Limit Setting • To set a limit given observation nobs • For a given BR and nbg the observed events nobs is given by Poisson distribution • with nobs=k BR + nbg • given actual nobs can use Bayesian method to extract P(BR| nobs,nbg,k) • This properly accounts for uncertainties in nbg and k • Defines BRCL(nobs,nbg,k) by CL = BR P(BR) d(BR) • Given nbg and k can definea priori expected 90% CL upper limitby • To optimize: Vary event requirements to minimize BR90%CL V. Krutelyov Search for SUSY using Bsmm

30. Optimization Result Fragmentationratio (Bs/B+) • For optimization, scan: LH>0.90-0.99, pT(B)>4-6 GeV • Assume 1 fb-1 of data •  Optimal cuts: LH>0.99 and pT(B)>4GeV • Uncertainties are included in the limit calculation • Dominant uncertainty is fs/fu from PDG ~ 15%(rel) Bsmm Summary SES = BR for Nsignal=1 V. Krutelyov Search for SUSY using Bsmm

31. Open Box For optimized cuts of LH >0.99 and pT(B) > 4GeV and a  60 MeV window around world avg Bs(d) mass CMU-CMX Channel CMU-CMU Channel Observed 0 event in the signal region! V. Krutelyov Search for SUSY using Bsmm

32. Limits Summary Bs: 0 events observed  yields a combined limit of: 1.5×10-7 @ 90% CL 2.0×10-7 @ 95% CL Bd: 0 events observed  yields a combined limit of: 3.9×10-8 @ 90% CL 5.1×10-8 @ 95% CL Compare to: Br(Bsmm) < 4.1×10-7 @ 90% CL ; D0 PRL 94 (2005) 042001 (240pb-1) Br(Bsmm) < 5.8×10-7 @ 90% CL ; CDF PRL 93 (2003) 032001 (171pb-1) Br(Bdmm) < 8.0×10-8 @ 90% CL ; BaBar PRL 94 (2005) 221803 (111fb-1) Both CDF Bs and Bd results are ×2 better than the best published result V. Krutelyov Search for SUSY using Bsmm

33. mSUGRA Dedes, Dreiner, Nierste, PRL 87(2001) 251804 • For mh~115GeV implies • 10-8<Br(Bsmm)<3×10-7 Excluded M0 [GeV] Excluded by this new result need ×3-5 improvement to exceed bsg exclusion Excluded Solid red = excluded by theory or experiment Dashed red line = light Higgs mass (mh) Dashed green line = (dam)susy (in units of 10-10) Black line = Br(Bsmm) V. Krutelyov Search for SUSY using Bsmm

34. mSUGRA scan Kane, Kolda, Lannon hep-ph/0310042 2.1×10-7 Run II Tevatron 95% CL • Within mSUGRA, if Bsmm is observed  tanb is large. V. Krutelyov Search for SUSY using Bsmm

35. SO(10) SUSY h2>0.13 mh<111GeV m+<104GeV • tan(b)~50 constrained by • unification of Yukawa coupling • White region is not excluded • Unification valid for small M1/2 • (~500GeV) R. Dermisek et al., hep-ph/0304101 • New Br(Bsmm) limit strongly • disfavors this solution for • mA< 500 GeV Excluded by this new result Red regions are excluded by either theory or experiments Green region is the WMAP preferred region Blue dashed line is the Br(Bsmm) contour Light blue region excluded by old Bsmm analysis V. Krutelyov Search for SUSY using Bsmm

36. RPV SUSY Exclusion B. Dutta et al, PLB 538 (2002) 121 m b R-parity violating SUSY ~ n l’i23 l i22 m s • Possible to exclude phase space • ~ independent of tan b • Exclusion strongly depends • on the coupling. Excluded V. Krutelyov Search for SUSY using Bsmm

37. Bsmm Prospects • Simplistic: no improvement to analysis • scale Nbg and NB+ linearly with Lumi  recalculate <BR>  at best~3×10-8 at 90% CL • Optimistic: <BR>~ 1/Lumi • Additional handles on bgd exist: • tighter muon ID (require CMP) • calorimeter isolation • additional 2D pointing • use mass resolution model in LH • Combine with D0 • BR(Bsmm) ≈ 1×10-8at 90% CL • is possible within Run II (by 09?) • Can be measured at SM level by CMS at LHC • after 2-3 years of data taking • (by 10?) V. Krutelyov Search for SUSY using Bsmm

38. MSSM Kane, Kolda, Lannon hep-ph/0310042 Dedes, Huffman PLB600 (2004) 261 • It is possible to constrain mA or/and tanb • given an observation of Bsmm • using BR(Bsmm)~ tan6b/mA4 MFV tanb=50 • Given Minimal Flavor Violation ↔ flavor change from CKM only • Substantial enhancement is possible in non-MFV case V. Krutelyov Search for SUSY using Bsmm

39. Summary • Bsmm is a powerful probe of new physics. Could potentially • provide the first hint of SUSY at the Tevatron • Using 364 pb-1 of data, CDF has obtained world best limits on • Bs and Bd channels (submitted to PRL): • Br(Bs) < 1.6×10-7 @ 90% CL • < 2.1×10-7 @ 95% CL • Br(Bd) < 4.2×10-8 @ 90% CL • < 5.5×10-8 @ 95% CL • The limits are now starting to constrain interesting regions • of SUSY parameter space • An order of magnitude has been covered since Run I result. Will • cover at least another order of magnitude before the end of RunII • Hint of SUSY may just be around the corner!! V. Krutelyov Search for SUSY using Bsmm

40. V. Krutelyov Search for SUSY using Bsmm

41. Backup Slides BACKUP SLIDES V. Krutelyov Search for SUSY using Bsmm

42. Check Bgd Estimate Using Control Samples LH CMU-CMU CMU-CMX cut pred obsv pred obsv OS- >0.90 37±1 32 33±1 36 >0.99 2.8±0.2 2 3.6±0.2 3 SS+ >0.90 0.25±0.03 0 0.44±0.04 0 >0.99 <0.10 0 <0.10 0 SS- >0.90 0.35±0.03 0 0.63±0.06 0 >0.99 <0.10 0 <0.10 0 FM+ >0.90 14.2±0.4 10 3.9±0.2 3 >0.99 1.0±0.1 2 0.41±0.03 0 1) OS- : opposite-charge dimuon, l < 0 2) SS+ : same-charge dimuon, l > 0 3) SS- : same-charge dimuon, l < 0 4) FM+: fake muon sample (at least one muon failed quality cut) Predictions are consistent with observations V. Krutelyov Search for SUSY using Bsmm

43. Check MC Modeling of Signal LH Compare B+ Data and MC • For CMU-CMU: • MC reproduces Data e(LH>x) to 6% or better • After correction DaMC DaMC+0.01 • Assign 6% (relative) • systematic • For CMU-CMX • MC vs Data • agreement is better • No correction to MC is needed V. Krutelyov Search for SUSY using Bsmm

44. CDF+D0 combination • CDF-D0 working group is formed to combine the • Bmmlimits from both experiments: • D0 Preliminary : Br(Bsmm) < 3.0×10-7 @ 90% CL (D0 note 4733, ~300pb-1) • CDF Preliminary : Br(Bsmm) < 1.6×10-7 @ 90% CL • Two independent groups cross-checking each other’s • combined results. Aim to release preliminary combined • results for LP05 • Combined CDF and D0 results is expected to improve the • limit by ~20% V. Krutelyov Search for SUSY using Bsmm

45. CDF Run2. |h|=0.6 |h|=1 2 ft steel 2 ft steel CMP muon det (drift cham + scint) pT>2.5GeV |h|<0.6 CMU muon det (drift cham) pT>1.5GeV |h|<0.6 SVX II (5layer) silicon strip det ssvx+IP~40mm COT tracker drift chamber dpT/pT2~0.1% dpT/pT2~1.5%(L1/XFT) |h|=1 • s=1.96TeV σ(Inelastic)~60mb. ℒ ~71031cm-2s-1 (~400pb-1 good recorded) • Plan: ~81031cm-2s-1 (Run2a); ~21032cm-2s-1 (4-8/fb Run 2) • 1.7MHz collision20kHz L1 trigger350Hz L260Hz L3/logging rate. • ~80% L1 (~1/3 at L3) of the trigger (bandwidth) are B physics |h|=0.6 CMX muon det (drift cham + scint) pT>2GeV 0.6<|h|<1 • Better silicon coverage (2), muon detection, improved tracking. • Better triggers: lower track pT, higher efficiency. Run2 vs Run1 V. Krutelyov Search for SUSY using Bsmm

46. Dimuon trigger pT>1.5GeV, |h|<0.6 pT>2GeV, 0.6<|h|<1 pT, f, muon ID used to cut on tracks Used for y,U,Bmm(+X) Two Track Trigger pT>2GeV, |h|<1 pT, f, d0 info used to cut on 2 tracks Used for: B,Dhadrons ; Dmm Triggers used ~ 1 mm primary vertex b, c decays secondary vertex impact parameter All are input to the various Level-3 triggers That use the offline quality information • Semileptonic trigger • pT, f, d0, muon ID used to cut on tracks • Used for y,D,Bm(+X) • no results presented in this talk V. Krutelyov Search for SUSY using Bsmm

47. Dimuon trigger pT>1.5GeV, |h|<0.6 pT>2GeV, 0.6<|h|<1 pT, f, muon ID used to cut on tracks Used for y,U,Bmm(+X) Triggers used V. Krutelyov Search for SUSY using Bsmm

48. Two Track Trigger pT>2GeV, |h|<1 pT, f, d0 info used to cut on 2 tracks @L2 Used for: B,Dhadrons ; Dmm Triggers used ~ 1 mm primary vertex b, c decays secondary vertex impact parameter V. Krutelyov Search for SUSY using Bsmm

49. Br(Bs(d)μ+μ-) measurement. CL=90% upper limit on <nsig> for nobs and nbg • Expect to detect at most few events that might only look like Bs(d)μ+μ– • SM predicts 0 events  really a “search” • Don’t look at the data signal region (blind search) • Signal inside 5.169<Mmm<5.469 GeV/c2 (3s window) • demonstrate understanding of background events • accurately estimate a (acceptance) and e (efficiency) • intelligently optimize cuts 171pb-1 or 10 trillion collisions only Run1 sBs=0.9mbfor pTBs>6 GeV/c, |y|<1 (use this as a baseline selection) 3% (CMU&CMP Run1) 25% Run1 Dimuon trigger and pT>6GeV baseline sample: 2940 events V. Krutelyov Search for SUSY using Bsmm

50. Optimization Results • Optimize on (Mmm, ct, DF, Iso) to get the best expected limit • (ct,DF,Iso) = (>200 mm, <0.10 rad, >0.65) and mass window  80 MeV around world avg Bs(d): 5.369 GeV (5.279 GeV) • Bs(d): aetotal = (2.0  0.2)% (6.6%, etotal30%) • Accepted bgd  = (6  2) fb • Expected bacground <Bgd> in 171pb-1 = 1.1  0.3 events Poisson prob V. Krutelyov Search for SUSY using Bsmm