Global (SUSY) Fits in 2012 (Frequentist Approach)

Transcription

1 Global (SUSY) Fits in 2012 (Frequentist Approach) Oliver Buchmueller Imperial College London darkattack2012 1

2 Confronting a theory with data Recipe: Combine measurements Compare with predictions Constrain the parameters or exclude the model! Ingredients: Accurate set of predictions Consistent set of measurements 2

3 Why Global Fits?! Example: Wmap strips in the CMSSM [hep-ph] Wmap strips some years ago Wmap strips are the result of a 2D scan of m 0 and m 1/2 where the other two parameters in the CMSSM, i.e. A 0 and tanβ are fixed. Therefore, a 4D parameter space is collapsed to a 2D Model, WITHOUT taking into account the other parameters. 3

4 Why Global Fits?! Example: Wmap strips in the CMSSM Global Fit instead of 2D scan: Wmap strips today Carry out a simultaneous fit of all relevant NP and SM parameter to the experimental data/constraints. χ 2 The result of such a global fit will not suffer from the ambiguity symptoms of the scans and in addition will provide a well-defined statistical interpretation of the results Its proper execution is by far a non-trivial problem, though! 4

5 Confronting Theory with Data Global fits LEP EW FIT Comprehensive statistical confrontation of theory and data has always been a very important subject of particle physics phenomenology. Global Fits Prominent Examples are: LEP EW Fit Constraining the SM (EW part) Unitarity Triangle UT Fit vs. CKM Fitter UT Fit CKM Fitter 5

6 Global NP Fits in the LHC Era R. Lafaye, M. Rauch, T. Plehn, D. Zerwas H. Flächer, M. Goebel, J. Haller, A. Höcker, K. Mönig, J. Stelzer P. Bechtle, K. Desch, M. Uhlenbrock, P. Wienemann GFitter Sfitter Fittino O. Buchmueller, R. Cavanaugh, A. De Roeck, J.R. Ellis, H.Flacher, S. Heinemeyer, G. Isidori, K.A. Olive, F.J. Ronga, G. Weiglein L. Roszkowski, R. Ruiz de Austri, R. Trotta S.S. AbdusSalam, B.C. Allanach, M.J. Dolan, F. Feroz, M.P. Hobson MasterCode Superbayes 6 Slide from 2009

7 Global NP Fits in the LHC Era R. Lafaye, M. Rauch, T. Plehn, D. Zerwas H. Flächer, M. Goebel, J. Haller, A. Höcker, K. Mönig, J. Stelzer To excellent talks on global fits already on Monday: P. Bechtle, K. Desch, M. Uhlenbrock, P. Wienemann Charlotte Strege GFitter Constraints on Dark Matter and SUSY from LHC and Direct Detection Experiments Sfitter Nelly Nguyen AstroFit and Fittino: Results for the CMSSM Fit Fittino O. Buchmueller, R. Cavanaugh, A. De Roeck, J.R. Ellis, H.Flacher, S. Heinemeyer, G. Isidori, K.A. Olive, F.J. Ronga, G. Weiglein L. Roszkowski, R. Ruiz de Austri, R. Trotta S.S. AbdusSalam, B.C. Allanach, M.J. Dolan, F. Feroz, M.P. Hobson MasterCode Superbayes 7 Slide from 2009

8 Global NP Fits in the LHC Era R. Lafaye, M. Rauch, T. Plehn, D. Zerwas H. Flächer, M. Goebel, J. Haller, A. Höcker, K. Mönig, J. Stelzer To excellent talks on global fits already on Monday: P. Bechtle, K. Desch, M. Uhlenbrock, P. Wienemann Charlotte Strege GFitter Constraints on Dark Matter and SUSY from LHC and Direct Detection Experiments Sfitter Nelly Nguyen AstroFit and Fittino: Results for the CMSSM Fit Fittino O. Buchmueller, R. Cavanaugh, A. De Roeck, J.R. Ellis, H.Flacher, S. Heinemeyer, G. Isidori, K.A. Olive, F.J. Ronga, G. Weiglein L. Roszkowski, R. Ruiz de Austri, R. Trotta Disclaimer: Will mainly focus on MasteCode results in this MasterCodetalk simply because I am familiar with it. Results are similar to the one obtained from other frequentist groups. Superbayes S.S. AbdusSalam, B.C. Allanach, M.J. Dolan, F. Feroz, M.P. Hobson 8

9 SUSY (as an important example) Direct Searches SUSY? Indirect Searches 9

10 Direct Searches Putting it all Together Where are Toady? Flavour Physics EWK results SUSY 10 Low Energy Data e.g. g-2 Cosmology Dark Matter Searches

11 Example MasterCode Consistency Relies on SLHA interface Modularity Compare calculations Add/remove predictions State-of-the-art calculations Direct use of code from experts Confront prediction and data Form statistical metric (e.g.χ 2 ) and use as input to e.g. MCMC or Minuit 11

12 Example MaterCode Consistency Relies on SLHA interface Modularity Compare calculations Add/remove predictions State-of-the-art calculations Direct use of code from experts Confront prediction and data Form statistical metric (e.g.χ 2 ) and use as input to e.g. MCMC or Minuit 12

13 How to determine preferred regions of parameter space Frequentist approach Build up a Chi^2 for each point Feed to Markov Chain Monte Carlo, try to minimise Chi^2 Used as a sampling tool SUSY models multi-dimensional 13

14 Used Observables 14

15 Models typically considered Sampling large multi-dimensional spaces hard and computationally expensive Start with the simplest of constrained models defined at the GUT scale (focus of most groups) msugra and VCMSSM CMSSM: m0, m12, tanb, A0, sign(mu) NUHM1: break m0 = mh^2 degeneracy pmssm also under study but difficult to reduce free parameters Example MC: We have sampled O(10^8) points for each model 15

16 Models typically considered Sampling large multi-dimensional spaces hard and computationally expensive Start with the simplest of constrained models defined at the GUT scale (focus of most groups) msugra and VCMSSM CMSSM: m0, m12, tanb, A0, sign(mu) NUHM1: break m0 = mh^2 degeneracy Main focus on: pmssm also under study but difficult to reduce free parameters Example MC: We have sampled O(10^8) points for each model 16

17 The pre-lhc era CMSSM NUHM1 17 For references NDF ~ 22

18 The post-lhc era in 2011 CMSSM NUHM1 Chi^2 increases Shifting to higher masses, larger tanβ Plane relatively flat no real preferred minima anymore 18

19 DISCOVERY! Finally! Observation of a new Particle (Boson)! 19

20 First Mass Measurement First mass measurement from CMS (ATLAS still to come) 20 M(new particle) = 125 ±0.4(stat) ± 0.5(sys)

21 Higgs: SUSY vs. SM Global SUSY Fit MasterCode Collaboration OB (Exp), R. Cavanaugh (Exp), A. De Roeck (Exp), J. Ellis (Theo), H. Flaecher (Exp), S. Heinemeyer (Theo), G. Isidori (Theo), K. Olive (Theo), P. Paradisi, (Theo), F. Ronga (Exp), G. Weiglein (Theo) Pull for CMSSM fit [hep-ph] Example: redo SM fit in SUSY predicting the lightest higgs boson mass in the Constraint Minimal Supersymmeteric Standard Model (CMSSM) 21

22 SUSY: Light Higgs Predictions CMSSM NUHM1 Higgs important probe of SUSY Predictions above produced based on analogous method to SM best-fit plots No Higgs constraints imposed to make these plots!! 22

23 The post-higgs era CMSSM NUHM1 Assume a putative measurement of m H =125 +/- 1.5(theo) +/- 1.0 GeV Further reduction in potential phase-space!

24 CMSSM PRELIMINARY Post LHC&Higgs era in 2012 NUHM1 PRELIMINARY Updated with 5/fb direct search results Updated BR(Bs-> µµ) combination from the LHC (May 2012 Prospects look bleak for constrained models p-value ~10% (max)

25 CMSSM: Evolution with time 2008 Pre-LHC 2008 Pre-LHC post-lhc post-lhc+xenon Xenon post-lhc-discovery post-lhc post-lhc

26 Fittino: Time Evolution 26

27 Spin Independent XS vs. M LSP Pre-LHC 2008 CMSSM NUHM1 27

28 Spin Independent XS vs. M LSP Post-LHC (1/fb), Post-Xenon CMSSM NUHM1 28

29 Spin Independent XS vs. M LSP Post Discovery! assume m H =125 +/- 1.5(theo) +/- 1.0 GeV 29 CMSSM NUHM1

30 Spin Independent XS vs. M LSP Today CMSSM NUHM1 30

31 CMSSM: Evolution with time 2008 Pre-LHC 2011 post-lhc+xenon post-lhc-discovery 2012 post-lhc

32 SUSY on life support? The answer to this question is NO! See 2012 Experimental SUSY PDG review [OB & Paul De Jong]: 32

33 SUSY on life support? The answer to this question is NO! See 2012 Experimental SUSY PDG review [OB & Paul De Jong]: In general, the LHC does not (yet) place limits on parameter space with M LSP >~400 GeV Leaving a very large Region of the MSSM, even at the mass scale below 1 TeV, unexplored! 33

34 SUSY Coverage Msusy An illustration considering summer 2011 data 34

35 SUSY Coverage Msusy An illustration considering summer 2011 data Note: access are flipped 35

36 Need to find new SUSY benchmark models but how many parameters can we effort? [hep-ph] pmssm: ~20 NP parameter Wow! 36

37 Fitting the Soft Scale [hep-ph] Bayesian fit In general very strong prior dependence not surprising with 20+ parameters and only indirect constraints 37

38 Fitting the Soft Scale Frequentist fit to 18+ parameters but now assuming LHC input from 300/fb of data [hep-ph] MSSM18 LHC 300/fb 38 38

39 Fitting the Soft Scale Frequentist fit to 18+ parameters but now assuming LHC input from 300/fb of data [hep-ph] Fitting soft scale parameters that are much MSSM18 closer to the experimental measurements LHC 300/fb is an interesting approach but at least initially we need to reduce the set of parameters. Question: Can we define a meaningful set of 4 to 5 NP soft scale parameters? 39 39

40 Dark Matter Toy Model (Example) Excitation of u quark key feature of UED models: Heavy up-type quark, U 1 Decays into neutral stable scalar particle, A 1 (Dark Matter candidate) Results in jets and missing energy P 2 u U 1 A 1 Interpret using results from CMS alphat SUSY search 40 Simulate signal with MADGRAPH + PYTHIA Detector simulation with DELPHES Perform scan in mass and splitting variables P 1 A 1 Ū 1 ū

41 heavy quark mass Dark Matter Toy Model (Example) LHC 7 TeV 1.1/fb Numbers represent upper limit on signal strength multiplier NLO x-sec used We can apply the LHC limits to these models but 95% excluded 41

42 heavy quark mass Dark Matter Toy Model (Example) LHC 7 TeV 1.1/fb Numbers represent upper limit on signal strength multiplier NLO x-sec used 42 95% excluded We can apply the LHC limits to these models but for a global fit we also need precision calculations for indirect constraints (otherwise there is not fit!). Yet, this predictions are often not available or not precise enough a real problem.

43 Conclusions Constrained SUSY models are on life support but not yet ruled out! Low-energy SUSY in general, however, is still very much alive! Today no significant limits from the LHC for M LSP > ~ 400 GeV. Need to define other SUSY benechmarks beyond constaint models. Examples are soft-scale models like the pmssm but the number of parameters is an issue needs intelligent ideas! We also need go beyond SUSY but the lack of precision calculations for all relevant observables is an issue Expected the field of global fits to grow further in the future 43 after all, its all about putting it all together

44 MasterCode History Collaboration started in 2007 Main aim to prepare a tool for SUSY studies - other data constraints 10 th paper in preparation Prediction for the Lightest Higgs Boson Mass in the CMSSM using Indirect Experimental Constraints: arxiv: Predictions for Supersymmetric Particle Masses using Indirect Experimental and Cosmological Constraints: arxiv: Likelihood Functions for Supersymmetric Observables in Frequentist Analyses of the CMSSM and NUHM1. arxiv: Predictions for mt and mw in minimal supersymmetric models arxiv: Frequentist analysis of the parameter space of minimal supergravity: arxiv: Implications of Initial LHC Searches for Supersymmetry. arxiv: Supersymmetry and Dark Matter in Light of LHC 2010 and Xenon100 Data: arxiv: Supersymmetry in Light of 1/fb of LHC Data: arxiv: O. Higgs Buchmueller, and darkattack2012, Supersymmetry arxiv:

45 Gluino masses CMSSM NUHM1 PRELIMINARY PRELIMINARY --- 1/fb 5/fb CMSSM: 1500 GeV GeV NUHM1: >1500 GeV

46 Squark masses CMSSM NUHM1 PRELIMINARY PRELIMINARY --- 1/fb 5/fb And more results in upcoming paper: on stau, Bs->µµ, Higgs, σ SI p

47 Trend of best-fit points Many fitting groups, many different approaches Bayesian vs Frequentist Basic conclusion: direct searches pushing masses higher 47

48 Aside: Using experimental results SUSY results from CMS and ATLAS are usually presented in the m0 vs m12 plane with tanb=10, A0=0 How do we use the information for other values of these parameters? Use generator (PYTHIA) + fast-simulation code (DELPHES) to reproduce experiment signal yields and repeat interpretation If this works, vary parameter space and come up with generic scaling laws 48

49 Aside: Validation for ATLAS 5/fb search Reproduce 95% observed exclusion contour using DELPHES Use this to determine scaling law in the plane Mean: -2.8 Sigma: Observed Exclusion (Delphes,ATLAS) [%] 49

50 Aside: Validation for ATLAS 5/fb search Try to reproduce 95% observed exclusion contour Use this to determine scaling law in the plane ~ M^(-4) Exclusion Confidence m 1/2 = m 0 m 1/2 = 3m 0 m = 1 3 m 1/ R/R 95 50

51 Soft SUSY breaking MSSM L 51

52 Soft SUSY breaking MSSM L Gaugino s and their masses M 3, M 2, M 1 52

53 Soft SUSY breaking MSSM L Squarks and sleptons, and their masses 53

54 Soft SUSY breaking MSSM L Tri-linear couplings A 54

55 Soft SUSY breaking MSSM L Higgs sector: 2 complex doublets (1 for u-type, 1 for d-type) 55

56 [hep-ph] How Much Parameters are Reasonable? pmssm: 20 NP parameter Wow! 56

57 Kinematically Kinematically not not accessible accessible 57