Search for Quark Substructure in 7 TeV pp Collisions with the ATLAS Detector

Transcription

1 Search for Quark Substructure in 7 TeV pp Collisions with the ATLAS Detector Frank Berghaus SUPERVISOR: Michel Lefebvre University of Victoria June 12, 2012

2 Introduction ATLAS and the LHC The LHC provides counter-rotating proton beams each at 3.5 TeV ( ) Beams intersect in ATLAS causing collisions of quarks (and/or gluons) Decay products of collisions are recorded by ATLAS Pseudo-rapidity η is the measure of longitudinal angle: η = ln tan (θ/2) Differences in pseudo-rapidity are invariant under longitudinal boost Rapidity (y) of massless particles F. Berghaus (UVic) Quark Substructure June 12, / 16

3 Introduction Quark Compositeness Quark Compositeness Are quarks fundamental particles or composite objects? Constituents are generally called preons Preons may reveal themselves at an energy scale Λ Expect to see the effects of composite nature if ŝ is sufficiently high At lower energies quarks appear point-like Search for deviation in dijet cross-section from QCD prediction A 4-fermion contact interaction should become evident for an observation of quark compositeness If data agrees well with QCD set a limit on Λ Aside: Composite nature could also be observed by finding a quark resonance F. Berghaus (UVic) Quark Substructure June 12, / 16

4 Introduction Quark Compositeness Quark Contact Interactions (CI) Effects of contact interactions should appear if ŝ is sufficiently large If Λ > s interactions between constituents are suppressed, with quarks appearing point-like dominant contribution to cross section from 4-fermion contact term 4-fermion contact term in Lagrangian L qqqq (Λ) = ξg2 2Λ 2 Ψ L qγ µ Ψ L q Ψ L qγ µ Ψ L q g 2 /4π = 1, Ψ are the left handed quarks ξ = +1 ( 1) is destructive (constructive) interference with QCD Exclusion limits change by 1 % depending on the choice of ξ F. Berghaus (UVic) Quark Substructure June 12, / 16

5 Measurement Observable: χ Sensitive Observables QCD at LO looks like Rutherford scattering in the centre-of-mass frame: d ˆσ dω 1 sin 4 (Θ/2) Contact interactions are expected to yield a more isotropic spectrum Useful angular variable in hadron collider experiment is χ: χ = e y 1 y 2 = 1 + cosθ 1 cosθ y 1 is the rapidity of the leading (in p T ) jet y 2 is the rapidity of the sub-leading (in p T ) jet Invariant under Lorentz boosts along the beam (z) F. Berghaus (UVic) Quark Substructure June 12, / 16

6 Expected Distribution Measurement Observable: χ Rutherford scattering (QCD): d ˆσ dω 1 sin 4 (Θ/2) d ˆσ dχ 1 Isotropic scattering (CI): ( 1 / N ) dn / dχ QCD QCD+CI d ˆσ dω 1 d ˆσ 0.05 dχ 1 (1 + χ) χ Higher order QCD predicts more events at low χ Can test NLO QCD using events at low dijet mass F. Berghaus (UVic) Quark Substructure June 12, / 16

7 Data Distributions Measurement Distribution dn / dχ ( 1 / N ) ATLAS Work In Progress < m jj < 800 GeV ( ) 800 < m jj < 1200 GeV ( ) 1200 < m jj < 1600 GeV ( ) 1600 < m jj < 2000 GeV ( ) 2000 < m jj < 7000 GeV ( ) QCD Prediction QCI with Λ = 5000 GeV χ = e y -y data L = 36 pb 1 Low χ implies large scattering angle Data and MC prediction are normalized per m jj bin Offset each m jj bin for display QCD and CI prediction is corrected to NLO using k-factors derived from QCD NLO calculations for CI have recently become available [5] F. Berghaus (UVic) Quark Substructure June 12, / 16

8 Likelihood Function Analysis Methodology Counting events in bins of m jj and χ so binned Poisson Likelihood: L(n Λ) = N bins j=0 ( µj (Λ) n j n j! ) e µ j (Λ) n j is the number of recorded events in bin j µ j (Λ) is the number of events predicted for bin j at CI scale Λ Shape only analysis: The MC is normalized to contain the same number of events as the data in each m jj bin Since we only have MC simulation for some discrete points in Λ fit: µ j (Λ) a }{{} 0 + a 1 Λ }{{ 4 } QCD CI + a 2 Λ 2 }{{} Interference F. Berghaus (UVic) Quark Substructure June 12, / 16

9 Analysis Methodology Test Statistics Maximum likelihood ratio ( ) L(n Λ) Q(Λ) = 2 ln L(n ˆΛ) ˆΛ is the most likely value of Λ found by minimizing 2 ln L(n Λ) where n stands for the data recorded or pseudo-experiments F. Berghaus (UVic) Quark Substructure June 12, / 16

10 Analysis Methodology p-value for Q (Maximum Likelihood Ratio) Create pseudo-experiments by drawing Poisson random numbers in each bin centred on the MC-prediction for Λ = Fill likelihood distribution with Q from each pseudo-experiment Determine the fraction of pseudo-experiments with Q greater than the data p-value from ATLAS data: 0.95 ATLAS Work In Progress The p-value is the probability of observing a more extreme Q given QCD than the data: P QCD (Q( ) Q data ( )) F. Berghaus (UVic) Quark Substructure June 12, / 16

11 Analysis Methodology Definition of Confidence Limits Extend the idea of the p-value to be computed from our prediction at some Λ [6]: CL s+b (Λ) = P QCD+CI (Q(Λ) Q data (Λ)) Previous ATLAS exclusion limit was set at CL s+b < 0.05 (the 95 % exclusion limit) A more conservative approach corrects for the QCD likelihood distribution assuming Λ: CL s (Λ) = P QCD+CI(Q(Λ) Q data (Λ)) 1 P QCD (Q(Λ) Q data (Λ)) CMS used this definition in their recent paper F. Berghaus (UVic) Quark Substructure June 12, / 16

12 Λ Limit using CL s with Q Analysis Results CL s observed expected expected ± 1σ expected ± 2σ ATLAS Work In Progress Λ [TeV] Using 2010 ATLAS data L = 36 pb 1 Test statistic Q = 2 ln L(n Λ) L(n ˆΛ) Observed 95 % CL s limit = 5.36 TeV Expected 95 % CL s limit = 5.31 TeV F. Berghaus (UVic) Quark Substructure June 12, / 16

13 Analysis Errors Systematic Effects Experimental Jet Energy Scale ( 0.1% change to CL s ) Could cause excess of events at high jj Jet p T resolution ( 0.1% change to CL s ) Minimal bin-to-bin migration in m jj Theoretical: Factorization (µ f ) and renormalization (µ r ) scale choice ( 1.3% change to CL s ) Dominant effect Parton distribution function errors ( 0.2% change to CL s ) Angular observable minimizes the effect of all of these uncertainties Systematic effects are included through Bayesian integration F. Berghaus (UVic) Quark Substructure June 12, / 16

14 Effect on Limits Analysis Errors ATLAS Work In Progress CL s CL s+b Effect Obs [TeV] Exp [TeV] Obs [TeV] Exp [TeV] No Systematics MC Statistics Jet p T Resolution Jet Energy Scale µ f /µ r Scale Choice PDF Fit Errors All Assuming effects are independent F. Berghaus (UVic) Quark Substructure June 12, / 16

15 Previous Limits with ξ = 1 Analysis Errors Experiment Limits [TeV] s [TeV] L [fb 1 ] observed expected Stat ATLAS CL s+b CMS CL s D binned χ 2 CDF binned χ 2 All limits based on χ and m jj [1, 2, 3, 4] ATLAS also performed measurement using the centrality ratio finely binned in m jj to get a 95 % CL s+b exclusion of 7.6 TeV (expected 8.2 TeV) 2 CDF paper is from 1996, while D0 s is from 2009 F. Berghaus (UVic) Quark Substructure June 12, / 16

16 Summary Summary ATLAS Work In Progress An exclusion limit for quark compositeness has been obtained from the 2010 data-set Λ > 5.29 TeV at 95% CL s 7 TeV Collisions L = 36 pb 1 Accounted For: Jet energy scale uncertainty Jet p T resolution Factorization and renormalization scale uncertainty PDF fit errors Statistical limitations of simulated events F. Berghaus (UVic) Quark Substructure June 12, / 16

17 Appendix Data-set Event Cleaning and Selection GRL: Require LHC stable beam, good conditions for the Inner detector and calorimeters Using single jet triggers on efficiency plateau Cleaning cuts: One primary vertex with at least five tracks All jets pass quality cuts Selection Cuts Leading jet p T > 60 GeV Sub-leading jet p T > 30 GeV Dijet rapidity separation y = (y 1 y 2 )/2 < 1.7 Dijet boost ȳ = (y 1 + y 2 )/2 < 1.1 F. Berghaus (UVic) Quark Substructure June 12, / 16

18 Appendix Data-set Total Data For Analysis ATLAS Work In Progress m jj Bin [TeV] Periods A-E Periods F-I min max L1 Item L [pb 1 ] EF Item L [pb 1 ].52.8 J j50_jetnoef J j77_jetnoef J j95_noalg J j95_noalg J j95_noalg Trigger selection guarantees 100 % efficiency in each m jj bin Total 2010 data: 36 pb 1 F. Berghaus (UVic) Quark Substructure June 12, / 16

19 Test Statistics Appendix Exclusion Limits Likelihood ratio (reference to most probable) ( ) L(n Λ) Q(Λ) = 2 ln L(n ˆΛ) ˆΛ is the most likely value of Λ found by minimizing 2 ln L(n Λ) Likelihood ratio (reference to QCD) ( ) L(n Λ) q(λ) = 2 ln L(n ) Λ = implies the likelihood of QCD given the data where n stands for the data recorded or pseudo-experiments F. Berghaus (UVic) Quark Substructure June 12, / 16

20 Appendix Λ Limit using CL s+b with Q Exclusion Limits CL s+b observed expected expected ± 1σ expected ± 2σ ATLAS Work In Progress Λ [TeV] Using 2010 ATLAS data L = 36 pb 1 Using test statistic Q = 2 ln L(n Λ) L(n ˆΛ) Observed 95 % CL s+b limit = 5.60 TeV Expected 95 % CL s+b limit = 5.48 TeV F. Berghaus (UVic) Quark Substructure June 12, / 16

21 Appendix Exclusion Limits Λ Limit using CL s+b with q Using 2010 ATLAS data L = 36 pb 1 Using test statistic ATLAS Work In Progress q = 2 ln L(n Λ) L(n ) Observed 95 % CL s+b limit = 5.85 TeV Expected 95 % CL s+b limit = 5.70 TeV F. Berghaus (UVic) Quark Substructure June 12, / 16

22 Appendix Exclusion Limits Λ Limit using CL s with q Using 2010 ATLAS data L = 36 pb 1 Using test statistic ATLAS Work In Progress q = 2 ln L(n Λ) L(n ) Observed 95 % CL s limit = 5.54 TeV Expected 95 % CL s limit = 5.47 TeV F. Berghaus (UVic) Quark Substructure June 12, / 16

23 Previous Limits with ξ = +1 Appendix Review Experiment Limits [TeV] s [TeV] L [fb 1 ] observed expected Stat ATLAS 7 CL s+b CMS CL s D binned χ 2 CDF binned χ 2 ATLAS currently has no MC simulation to make this measurement 3 CDF paper is from 1996, while D0 s is from 2009 F. Berghaus (UVic) Quark Substructure June 12, / 16

24 Appendix QCI Simulation Fits Residuals PDF Fits MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ Index of α = Λ -4 Index of α = Λ -4 Index of α = Λ -4 Index of α = Λ -4 MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ Index of α = Λ -4 Index of α = Λ -4 Index of α = Λ -4 Index of α = Λ -4 MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ MC fit MC ( n (α) - n (α) ) / σ ATLAS Work In Progress Index of α = Λ -4 Index of α = Λ -4 Index of α = Λ -4 F. Berghaus (UVic) Quark Substructure June 12, / 16

25 r f Appendix Systematics Factorization/renormalization Scale Choice Large ( 20%) variation across χ Dominant effect Prior: µ r and µ f picked from 1 x between 0.5 and 2.0 each )/σ(nominal),µ σ(µ ( µ, µ ) = ( 2.0, 2.0 ) Q r f 2 ( µ, µ ) = ( 0.5, 0.5 ) Q r f 2 ( µ, µ ) = ( 1.0, 2.0 ) Q r f 2 ( µ, µ ) = ( 1.0, 0.5 ) Q r f 2 ( µ, µ ) = ( 2.0, 1.0 ) Q r f 2 ( µ, µ ) = ( 0.5, 1.0 ) Q r f 2 ( µ, µ ) = ( 1.0, 1.0 ) Q r f χ F. Berghaus (UVic) Quark Substructure June 12, / 16

26 Appendix Systematics PDF Fit Error Errors fully correlated across χ Large ( 20%) effect absorbed by normalization Variation of error across χ is 4% Prior: significance of PDF error follows Gaussian Use same sign error bars to absorb asymmetry Relative Error on d 2 σ/dχdm jj χ F. Berghaus (UVic) Quark Substructure June 12, / 16

27 Appendix Summary Table to Calculated Exclusion Limits Stat ATLAS Work In Progress Q = 2 ln L(n Λ) q = 2 ln L(n Λ) L(n ˆΛ) L(n Λ=inf) obs [TeV] exp [TeV] obs [TeV] exp [TeV] CL s+b CL s As presented above F. Berghaus (UVic) Quark Substructure June 12, / 16

28 Bibliography I Appendix References The ATLAS Collaboration Search for New Phenomena in Dijet Mass and Angular Distributions 47th Rencontres de Moriond on Electroweak Interactions and Unified Theories ATLAS-CONF S. Chatrchyan et al. [CMS Collaboration], Search for quark compositeness in dijet angular distributions from pp collisions at sqrt(s) = 7 TeV, JHEP 1205, 055 (2012) [arxiv: [hep-ex]] F. Abe et al. [CDF Collaboration], Measurement of dijet angular distributions at CDF, Phys. Rev. Lett. 77, 5336 (1996) F. Berghaus (UVic) Quark Substructure June 12, / 16

29 Bibliography II Appendix References V. M. Abazov et al. [D0 Collaboration], Measurement of dijet angular distributions at s**(1/2) = 1.96-TeV and searches for quark compositeness and extra spatial dimensions, Phys. Rev. Lett. 103, (2009) J. Gao et al Next-to-leading QCD effect on the quark compositeness search at the LHC Phys. Rev. Lett. 106, (2011) T. Junk Confidence level computation for combining searches with small statistics Nuclear Instruments and Methods 434, Pages (1999) F. Berghaus (UVic) Quark Substructure June 12, / 16

30 Bibliography III Appendix References M. Cacciari and G. P. Salam, Phys. Lett. B 641, 57 (2006) [hep-ph/ ]. S. Catani and M. H. Seymour, Nucl. Phys. B 485, 291 (1997) [Erratum-ibid. B 510, 503 (1998)] [arxiv:hep-ph/ ]. T. Carli, D. Clements, A. Cooper-Sarkar, C. Gwenlan, G. P. Salam, F. Siegert, P. Starovoitov and M. Sutton, Eur. Phys. J. C 66, 503 (2010) [arxiv: [hep-ph]]. F. Berghaus (UVic) Quark Substructure June 12, / 16