Invariant Mass Distribution of Jet Pairs Produced in                   Association                                    √ wi...
Theoretical interest of the final state                                                                    Measurements of ...
Diboson with jets at CDF        Diboson represent a Standard Model reference that can be used as a starting        point f...
CDF II        Silicon detectors (L00+SVX+ISL) and central drift chamber (COT) in 1.4 T        magnetic field        Calorim...
Strategy      Trigger on the charged lepton      Require presence of neutrino:      undetected neutrino manifests as an   ...
High pT Electron and Muon selection      The analysis uses 4.3 fb−1 of data collected by      CDF.V. Cavaliere (University...
High pT Electron and Muon selection      The analysis uses 4.3 fb−1 of data collected by      CDF.      Electrons → ET > 2...
High pT Electron and Muon selection      The analysis uses 4.3 fb−1 of data collected by      CDF.      Electrons → ET > 2...
High pT Electron and Muon selection      The analysis uses 4.3 fb−1 of data collected by      CDF.      Electrons → ET > 2...
Jet Definition       Jets are selected using the standard CDF       JETCLU algorithm       Cluster energy in cones of ∆R < ...
Jet Definition       Jets are selected using the standard CDF       JETCLU algorithm       Cluster energy in cones of ∆R < ...
Sample Composition        W → ν + jets (l = e, µ, τ ):           same signature as signal with a much higher cross section...
Cross sections measurement                                                             Use a fit to dijet mass to extrapola...
What did we learn?        CDF established Diboson production with jets in the final state        We can safely state that w...
Cross sections measurements                                                               Observed an interesting discrepa...
Moving to different kinematical region        Using exactly the same kinematical cuts as the diboson analysis        but: ...
Modeling of samples                  Process                              Model          σ (pb)               WW/WZ inclus...
Alpgen+Pythia        While simulating W/Z + N-jets, we need to get the inclusive cross section and        the relative cro...
QCD Multijet Model Modeled using data sidebands        Non isolated muons:             Using non-isolated events, events w...
QCD Multijet estimation        QCD multijet events are characterized by low &T , so &T distribution is                    ...
Putting ingredients together    1   Mjj of data compared to SM predictions    2   Want to investigate further → Need to ch...
Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   18 /...
Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   19 /...
Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   20 /...
Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   21 /...
QCD Multijet model                    Check shape and rate using                    independent ways:                     ...
QCD Multijet model                    Check shape and rate using                    independent ways:                     ...
¯  tt shape    1    ¯        tt production is characterized by a larger number of jets with respect to other        proces...
W + jets shape              Use Z + jets data to check W+jets shape              Require two leptons (one tight and one lo...
Fitting procedure        Combined χ2 fit to the dijet mass distribution in electron and muon samples.        5 templates:  ...
Gaussian Assumption        Chosen to estimate the significance of the excess assuming an additional gaussian        compone...
Fit to data with SM templates + gaussian     Events/(8 GeV/c2)                         700                     CDF data (4...
Fit to data with SM templates + gaussian     Events/(8 GeV/c2)                         180                                ...
Systematics        systematics affecting background shapes        Evaluated generating statistical trials with the varied ...
Shape systematic: W + jets        Alpgen MC depends on some parameters:            1   Factorization and normalization sca...
Significance        To evaluate the significance of the resonance, we apply a procedure called        “supremum p-value”    ...
W +jets (mis)modeling → ∆Rjj        In order to investigate possible mismodeling we consider two control regions, the     ...
Fit floating systematics                                                                     Events                        ...
NLO effects        In order to test Next to Leading Order contributions to the W +2 partons        prediction, we compare ...
Production Cross Section ?        We increase the jet ET threshold in steps of 5 GeV and check the fraction of        exce...
Flavour composition        Tried to look at the flavour composition of this jets.                Assuming a branching ratio...
Resonant production        Finally, to investigate the possibilities of a parent resonance or other quasi-resonant        ...
Conclusions        We study the invariant mass distribution of jet pairs produced in association with        a W boson.   ...
Conclusion II       Even if it is hardV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   38 ...
Conclusion II       Even if it is hard       Need to keep the focus on things we know:         1 CDF now sensitive to very...
Conclusion II       Even if it is hard       Need to keep the focus on things we know:         1 CDF now sensitive to very...
Acknowledgments        Would like to thank J.Campbell, E. Eichten, R. K.Ellis, C.Hill,        K.Lane, M. Mangano, A. Marti...
BackupV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   40 / 38
∆R studies        If we do reweigh to the sidebands we observe that the significance drops to 2.3σ        Does it really ma...
∆Rjj                                           DeltaR_jj_mjj                                     mjj                      ...
Fit requiring pT,W > 60 GeV/cV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar   04/06/2011   43 / 38
What happens if we change the Jet Energy scale? Result of the fit scaling JES up by 7%                                     ...
Results from D0 using 1.1 fb−1                                                                                      lin pa...
Upcoming SlideShare
Loading in …5
×

Invariant Mass Distribution of Jet Pairs Produced in Association with a W boson in p-pbar Collisions at sqrt(s) = 1.96 TeV

772 views

Published on

We report a study of the invariant mass distribution of jet pairs produced in association with a W boson using data collected with the CDF detector which correspond to an integrated luminosity of 4.3 fb^-1. The observed distribution has an excess in the 120-160 GeV/c^2 mass range which is not described by current theoretical predictions within the statistical and systematic uncertainties. In this letter we report studies of the properties of this excess.

Published in: Education, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
772
On SlideShare
0
From Embeds
0
Number of Embeds
140
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Invariant Mass Distribution of Jet Pairs Produced in Association with a W boson in p-pbar Collisions at sqrt(s) = 1.96 TeV

  1. 1. Invariant Mass Distribution of Jet Pairs Produced in Association √ with a W boson in pp Collisions at s = 1.96 TeV ¯ Viviana Cavaliere for the CDF Collaboration University of Illinois at Urbana-Champaign 04/06/2011
  2. 2. Theoretical interest of the final state Measurements of associate νℓ production of a W boson and jets are q ¯ p ¯ a fundamental test of the Standard ℓ+ W+ Model q p q′ g ν+jets signature shared by several q ¯ important processes W+ q ¯ W− 1 W W + W Z, single top, W H p ¯ q ¯ p ¯ γ, Z q′ p q p q W− W+ Essential starting point for physics W+ q ¯ Z beyond the SM p ¯ q′ ¯ p ¯ W+ q p q p Z q′ W+ Important to understand the tools used in high energy physicsV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 2 / 38
  3. 3. Diboson with jets at CDF Diboson represent a Standard Model reference that can be used as a starting point for searches. W W, W Z, ZZ observed in &T + jets mode at CDF in 2009 PRL 103, 091803 E (2009) W W, W Z observed in ν +jets final state. Two analysis: 1 Using matrix elements technique 2 Looking for the W → jj peak in the dijet invariant mass Phys. Rev. Lett. 104, 101801 (2010) Results have been updated with 4.3 f b−1V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 3 / 38
  4. 4. CDF II Silicon detectors (L00+SVX+ISL) and central drift chamber (COT) in 1.4 T magnetic field Calorimeters for electrons and jets Muon chambers up to |η| ≈ 1.4V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 4 / 38
  5. 5. Strategy Trigger on the charged lepton Require presence of neutrino: undetected neutrino manifests as an imbalance in transverse momentum: “missing” transverse energy → − →i −  T = − E ET i Get a sample of inclusive W Look for at least two jets and form the invariant mass Fit the expected W/Z and background dijet mass shape to data to extract the diboson fractionV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 5 / 38
  6. 6. High pT Electron and Muon selection The analysis uses 4.3 fb−1 of data collected by CDF.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 6 / 38
  7. 7. High pT Electron and Muon selection The analysis uses 4.3 fb−1 of data collected by CDF. Electrons → ET > 20 GeV/c2 (GeV/c) and |η| < 1.0 1 Require calorimeter showers consistent with electromagnetic interactions 2 Require that 90% of energy is deposited in the EM calorimeterV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 6 / 38
  8. 8. High pT Electron and Muon selection The analysis uses 4.3 fb−1 of data collected by CDF. Electrons → ET > 20 GeV/c2 (GeV/c) and |η| < 1.0 1 Require calorimeter showers consistent with electromagnetic interactions 2 Require that 90% of energy is deposited in the EM calorimeter Muons → pT > 20 GeV/c2 (GeV/c) and |η| < 1.0 1 Require high quality track and good matching between the track and the hit in the muon chambersV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 6 / 38
  9. 9. High pT Electron and Muon selection The analysis uses 4.3 fb−1 of data collected by CDF. Electrons → ET > 20 GeV/c2 (GeV/c) and |η| < 1.0 1 Require calorimeter showers consistent with electromagnetic interactions 2 Require that 90% of energy is deposited in the EM calorimeter Muons → pT > 20 GeV/c2 (GeV/c) and |η| < 1.0 1 Require high quality track and good matching between the track and the hit in the muon chambers Both are required to be isolated to reject leptons from semi-leptonic decays of heavy flavor hadrons W We further require & > 25 GeV and MT > 30 ET GeV to ensure the presence of a real WV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 6 / 38
  10. 10. Jet Definition Jets are selected using the standard CDF JETCLU algorithm Cluster energy in cones of ∆R < 0.4 Calorimeter signature must be inconsistent with electron signatures Select exactly two jets with ET > 20 GeV (corrected for detector inefficiencies) and |η| < 2.4 Jet Energy scale known at 3% level Independent check of the scale: W from top → pubic webpage Require pT,jj >40 GeV/c → in order to improve data/MC modelingV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 7 / 38
  11. 11. Jet Definition Jets are selected using the standard CDF JETCLU algorithm Cluster energy in cones of ∆R < 0.4 Calorimeter signature must be inconsistent with electron signatures Select exactly two jets with ET > 20 GeV (corrected for detector inefficiencies) and |η| < 2.4 Jet Energy scale known at 3% level Independent check of the scale: W from top → pubic webpage Require pT,jj >40 GeV/c → in order to improve data/MC modelingV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 7 / 38
  12. 12. Sample Composition W → ν + jets (l = e, µ, τ ): same signature as signal with a much higher cross section (2066 pb) ∼ 80% of the sample Z → ll + jets (l = e, µ, τ ): where one of the two leptons escapes detection and produces &T E cross section 187 pb ¯ tt + single top: final state similar to signal with at least one real W and two jets. ¯ σ(tt) = 7.5 pb and σ(single top) = 2.9 pb (assuming a mass of 172.5 GeV/c2 ) QCD Multijet: events without a real high pT lepton e.g a three-jet event in which one jet passes all lepton cuts and, simultaneously, the energies are so badly measured that a large &T is E reported. probability for a jet to mimic a lepton is very small, but QCD processes have very large cross sections estimated from data using orthogonal selectionV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 8 / 38
  13. 13. Cross sections measurement Use a fit to dijet mass to extrapolate the WW/WZ contribution We estimate 1582 ± 275(stat.) ± 107(syst.) events for a significance of 5.2σ. The resulting cross section is σ(W W/W Z) = 18.1±3.3(stat.)±2.5(syst.) pb that is in agreement with SM expectation (15.9 ± 0.9 pb).V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 9 / 38
  14. 14. What did we learn? CDF established Diboson production with jets in the final state We can safely state that we are able to observe dijet resonances Moreover in the W W/W Z → ν+jets analysis we learned: 1 W + jets normalization compatible with expectations scaled to NLO 2 Jet Energy Scale well under control → multiplicative factor to correct diboson position is found to be compatible with 1 3 ¯ Our mass fit is sensitive to tt + single top with an uncertainty of 10% → cross section compatible with expectationV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 10 / 38
  15. 15. Cross sections measurements Observed an interesting discrepancy in the Mjj Needed work in order to understand the nature and properties of the excess 1 an artifact of background subtraction 2 misunderstanding of one of the backgrounds 3 real physics?V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 11 / 38
  16. 16. Moving to different kinematical region Using exactly the same kinematical cuts as the diboson analysis but: We require both jets to have ET > 30 GeV 1 Energetic jets are measured with better accuracy. 2 Modeling in this region is expected to be more accurate 3 A possible heavier particle would be characterized by more energetic jets Sample modeling using same processes with different relative contribution All cuts chosen “a priori”V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 12 / 38
  17. 17. Modeling of samples Process Model σ (pb) WW/WZ inclusive P YTHIA 15.9 ± 0.9 Z → e, µ, τ +jets A LPGEN + P YTHIA 787 ± 85 ¯ tt P YTHIA 7.5 ± 0.83 single top M ADEVENT + P YTHIA 2.86 ± 0.36 W+jets A LPGEN + P YTHIA from data QCD multijet from data from data Correct MC for: Trigger Efficiencies: Data must pass the trigger to be selected → apply these efficiencies to the MC Lepton Energy Scale, Energy Resolution, and Identification. Luminosity Profile: not the same as for the data → reweight as a function of number of verticesV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 13 / 38
  18. 18. Alpgen+Pythia While simulating W/Z + N-jets, we need to get the inclusive cross section and the relative cross section for exclusive N-jets We simulate by pairing Alpgen (LO matrix elements) and Pythia (parton showers) 1 Matrix Elements: Fixed order. Gives an accurate description of the hard process. Needed for N-jet description 2 Parton showers: Needed for a realistic description of the final state in the detector Combine them using MLM scheme to avoid double counting. Cluster the showered partons into cone jets. Keep events only if each jet is matched to just one parton q q ¯ q q ¯ q q ¯ q q q g g ℓ+ ℓ+ ℓ+ W+ W+ W+ q′ ¯ νℓ q′ ¯ νℓ q′ ¯ νℓ ALPGEN produces events from diagrams (a) and (b). PYTHIA’s showering routine will sometimes takes events from diagram (b) and add a radiated gluon to produce diagram (c)V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 14 / 38
  19. 19. QCD Multijet Model Modeled using data sidebands Non isolated muons: Using non-isolated events, events which pass all selection criteria except the requirement of lepton isolation. Based on the rationale that non-isolated events are typically leptons contained in jets, and jets that contain energetic leptons are mostly non-W events. “AntiElectrons”: Some non-kinematic cuts for the electron are used to reject fake electrons. Model is constructed with events which fail at least two of the non-kinematic quality cuts but pass all the kinematic cuts of the electron.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 15 / 38
  20. 20. QCD Multijet estimation QCD multijet events are characterized by low &T , so &T distribution is E E completely different from W + jets Best solution → Fit the &T distribution on data in order to constrain multijet E normalization. Extract the fraction of QCD and knowing all the others contributions can extract also a preliminary W + jets normalization (left completely free in the final fit) Systematic associated with the normalization estimated using different models (25%)V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 16 / 38
  21. 21. Putting ingredients together 1 Mjj of data compared to SM predictions 2 Want to investigate further → Need to check every background carefully. Public Webpage 1 Look at Excess sidebands (115 > Mjj > 175 GeV/2 ) 2 QCD multijet shape modeling ¯ 3 tt production 4 W + jet shapeV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 17 / 38
  22. 22. Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 18 / 38
  23. 23. Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 19 / 38
  24. 24. Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 20 / 38
  25. 25. Control sample: 115 > Mjj > 175 GeV/2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 21 / 38
  26. 26. QCD Multijet model Check shape and rate using independent ways: 1 ∆φ(E T , Missing Pt) → large for & qcd events CDF Run II Preliminary ∫ L dt = 4.30 fb-1 Arbitrary Units W+jets CMUP+CMX 0.15 QCD 0.10 0.05 0.00 0 1 2 3 4 ∆ φmet, trkMetV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 22 / 38
  27. 27. QCD Multijet model Check shape and rate using independent ways: 1 ∆φ(E T , Missing Pt) → large for & qcd events 2 ∆φ(E T , closest jet) → & mis-measured jets tend to be aligned with the &T E CDF Run II Preliminary L = 4.30 fb-1 int 0.25 Arbitrary Units W+jets CMUP+CMX 0.20 QCD 0.15 0.10 0.05 0.00 0 1 2 3 4 5 min ∆ φmet, closest jetV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 22 / 38
  28. 28. ¯ tt shape 1 ¯ tt production is characterized by a larger number of jets with respect to other processes. 2 To test it, we exploit its higher jet multiplicity and look at events with at least 3 jets and events with at least 4 jets. No mismodeling is observed 3 Furthermore we consider events with exactly three jets passing selection criteria and we look at the distribution of the invariant mass system of the other combinations of jets, Mj1,j3 and Mj2,j3 .V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 23 / 38
  29. 29. W + jets shape Use Z + jets data to check W+jets shape Require two leptons (one tight and one loose) Selection on the hadronic side is the same Basic Idea : Similar kinematics Due the purity of the sample can be modeled by ALPGEN only Problem: 10 times less data Entries Entries -1 -1 Muon Data (4.3 fb ) 0.16 Electron Data (4.3 fb ) 0.14 MC Z+jet MC Z+jet 0.14 0.12 0.12 0.10 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00 0 50 100 150 200 250 300 0 50 100 150 200 250 300 Mj1,j2 [GeV/c2] Mj1,j2 [GeV/c2]V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 24 / 38
  30. 30. Fitting procedure Combined χ2 fit to the dijet mass distribution in electron and muon samples. 5 templates: 1 W + jets (uncostrained, normalization determined from the fit) 2 QCD (normalization constrained to its fraction with 25 % error) 3 Z + jets (normalization constrained to the measured cross section) 4 top & single top (normalization constrained to the theoretical cross section) 5 W W/W Z (normalization constrained to the theoretical cross section) Events/(8 GeV/c2) Events/(8 GeV/c2) CDF data (4.3 fb-1) 150 700 Bkg Sub Data (4.3 fb-1) WW+WZ 4.5% 600 W+Jets 80.2% WW+WZ (all bkg syst.) Top 6.5% 100 500 Z+jets 2.8% QCD 5.3% (b) 400 50 (a) 300 200 0 100 -50 0 100 200 100 200 2 Mjj [GeV/c ] Mjj [GeV/c2]V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 25 / 38
  31. 31. Gaussian Assumption Chosen to estimate the significance of the excess assuming an additional gaussian component. The gaussian assumption is a simplified model, since the exact shape would depend on the specific physics process and the heavy flavor content of the decay products. Since the excess looks narrow with respect to the detector resolution, we search for a peak compatible with the detector resolution for a given dijet mass value. Mjj σgaussian = σW = 14.3GeV MW Procedure: 1 Fit the data without the gaussian → evaluate χ2 2 Fit the data with the gaussian → evaluate χ2 3 We add 3 degrees of freedom to the fit (mass, separate e/µ yields) so the ∆χ2 should have the distribution of a χ2 with 3 degrees of freedom. 4 Verify the behaviour of the ∆χ2 with statistical trials with trial factor.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 26 / 38
  32. 32. Fit to data with SM templates + gaussian Events/(8 GeV/c2) 700 CDF data (4.3 fb-1) Gaussian 2.5% WW+WZ 4.8% 600 W+Jets 78.0% Top 6.3% Data fitted with SM 500 Z+jets 2.8% QCD 5.1% templates plus a gaussian. 400 ∆χ2 observed 20.31 that (c) 300 corresponds to a statistical significance of 3.7σ 200 (including trial factor) 100 0 100 200 Mjj [GeV/c2] Electrons Muons Excess events 156 ± 42 97 ± 38 Excess events / expected diboson 0.60 ± 0.18 0.44 ± 0.18 Mean of the Gaussian component 144 ± 5 GeV/c2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 27 / 38
  33. 33. Fit to data with SM templates + gaussian Events/(8 GeV/c2) 180 Bkg Sub Data (4.3 fb-1) 160 Gaussian 140 120 WW+WZ (all bkg syst.) Data fitted with SM 100 (d) templates plus a gaussian. 80 ∆χ2 observed 20.31 that 60 corresponds to a statistical 40 significance of 3.7σ 20 (including trial factor) 0 -20 -40 100 200 Mjj [GeV/c2] Electrons Muons Excess events 156 ± 42 97 ± 38 Excess events / expected diboson 0.60 ± 0.18 0.44 ± 0.18 Mean of the Gaussian component 144 ± 5 GeV/c2V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 27 / 38
  34. 34. Systematics systematics affecting background shapes Evaluated generating statistical trials with the varied templates and fitting with the standard ones. Affected Quantity Source Uncertainty (%) Number of Excess Events QCD shape ±1.9 Q2 ±6.7 JES ±6.1 Total ±9.3 0.1 Uncertainties on JES Quadratic sum of all contributions Absolute jet energy scale 0.08 Measurement affected by Jet Energy Scale: Out-of-Cone + Splash-out Relative - 0.2<|η|<0.6 0.06 1 Apply to all MC modeled processes at the same Underlying Event 0.04 time 0.02 QCD shape systematic evaluated using different Isolation ranges. 0 50 100 150 200 250 300 350 400 450 500 pcorr (GeV/c) TV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 28 / 38
  35. 35. Shape systematic: W + jets Alpgen MC depends on some parameters: 1 Factorization and normalization scale Q2 = MW +2 p2 which can be varied by a T,j constant factor on an event by event basis 2 kT Scale Factor: Alpgen’s scale factor for αs at each decay vertex. 3 Parton matching cluster pT threshold: the minimum pT for jet clusters that are used for matching procedure. 4 Parton matching clustering radius size: the size of the jet cone used when creating jet clusters for matching procedure → Alpgen authors recommend using the generator level cut. The only significant effect for this kind of selection is given by the Q2 Use standard CDF procedure: Double and halve it to obtain alternative templates This choice is motivated by standard practice based on extensive theoretical workV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 29 / 38
  36. 36. Significance To evaluate the significance of the resonance, we apply a procedure called “supremum p-value” If Nsyst is the number of systematics sources, we generate a toy MC sample for each combination of the Nsyst , i.e. in each sample, some of the systematics are varied. For each sample, we evaluate the corresponding p-value using the ∆χ2 between the background only and signal hypothesis as test statistics. The significance we quote for our final result is the largest among the p-values we obtain. To take into account the trial factor, in our toy experiments we scan the mass of the resonance in the search region [120 - 200] GeV/c2 using step of 4 GeV/c2 and evaluate, at each step, the corresponding χ2 : for each toy sample, the minimum χ2 of the scan is used in the ∆χ2 evaluation. The largest p-value is 7.6 × 10−4 , corresponding to a significance of 3.2 standard deviations.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 30 / 38
  37. 37. W +jets (mis)modeling → ∆Rjj In order to investigate possible mismodeling we consider two control regions, the first defined by events with Mjj < 115 and Mjj > 175 GeV/c2 and the second defined by events with pT < 40 GeV/c. We use these regions to derive a correction as a function of ∆Rjj to reweigh the events in the excess region. The reweighings change the statistical significance of the result by plus or minus one sigma. However, the ∆Rjj distribution is strongly correlated to Mjj and the control regions both have significantly different distributions of ∆Rjj . Reweighing our W +jets sample may be correct ∆Rjj mismodeling or may introduce bias in the Mjj distribution. In addition, the ∆Rjj distribution is consistent within the one sigma variation of the systematic uncertainties for events outside the excess mass region.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 31 / 38
  38. 38. Fit floating systematics Events 800 Events/(8 GeV/c2) CDF Data Bkg Sub Data (4.3 fb-1) WW+WZ 150 W+jets Gaussian QCD Multijet 600 WW+WZ Z+jets 100 (a) Top uncertainty 400 (b) 50 0 200 -50 0 100 200 0 1 2 3 4 5 Mjj [GeV/c2] ∆ Rjj Fit performed combination of systematic that fits data best: lowest χ2 Returns a p-value intermediate between the largest and statistical only ∆Rjj shown with the same combination of systematic. χ2 /ndf =26.6/18.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 32 / 38
  39. 39. NLO effects In order to test Next to Leading Order contributions to the W +2 partons prediction, we compare (private communication with J.Campbell, E. Eichten, K.Lane, A.Martin) A LPGEN and interfaced to P YTHIA for showering to a sample of W +2 partons simulated using the MCFM. We extract a correction as a function of Mjj that is applied to the A LPGEN + P YTHIA sample used in our background model. This procedure returns a statistical significance of 3.4σ. W+jj : comparing different µ, generators 2 Events/8 (GeV/c2) 200 ALP+Pyth (inc). µ 2= M2 + p2 vs. µ 2 = M2 Bkg Sub Data (4.3 fb-1) 1.8 W T,jj W ALP+Pyth (exc). µ 2= M2 + p2 vs. µ 2 = M2 W W T,jj 1.6 Gaussian ALP+Pyth (inc.) µ 2= M2 vs. MCFM LO µ 2 = M2 W 2 W 2 150 ALP+Pyth (exc.) µ 2= MW vs. MCFM NLO µ 2 = MW 1.4 WW+WZ 1.2 100 ratio 1 0.8 50 0.6 0.4 0 0.2 0 100 200 300 0 50 100 150 200 250 300 Mjj (GeV) Mjj [GeV/c2]V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 33 / 38
  40. 40. Production Cross Section ? We increase the jet ET threshold in steps of 5 GeV and check the fraction of excess events that are selected as function of the jet ET . The result is compatible with expectation from a Monte Carlo simulation of a W boson plus a particle with a mass of 150 GeV/c2 and decaying isotropically into two jets. In this model, we estimate a cross section times the particle branching ratio into dijets of the order of 4 pb → not compatible with Standard Model W H σ × BR(b¯ = 39 fb b)V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 34 / 38
  41. 41. Flavour composition Tried to look at the flavour composition of this jets. Assuming a branching ratio in b¯ of 100% and the tagging efficiency b from W H we expect to be able to see a much stronger signal in the tagged sample Compared the fraction of events with b-jets in the sidebands (120> Mjj >160 GeV/c2 ) to that in the excess region Tag requirement Excess region Sideband region Muons 1 tag 0.1027 ± 0.0112 0.0813 ± 0.0096 2 tag 0.0078 ± 0.0030 0.0084 ± 0.0030 Electrons 1 tag 0.0897 ± 0.0088 0.0945 ± 0.0087 2 tag 0.0110 ± 0.0030 0.0095 ± 0.0026 No significant difference is observedV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 35 / 38
  42. 42. Resonant production Finally, to investigate the possibilities of a parent resonance or other quasi-resonant behavior, we consider the M(lepton,ν,jj) and the M(lepton,ν,jj) -Mjj distributions for events with Mjj in the range 120-160 GeV/c2 and to investigate the Dalitz structure of the excess events, the distribution of M(lepton,ν,jj) -Mjj , in bins of Mjj . The distributions are compatible in shape with the background-only hypothesis in all cases.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 36 / 38
  43. 43. Conclusions We study the invariant mass distribution of jet pairs produced in association with a W boson. The best fit to the observed dijet mass distribution using known components, and modeling the dominant W +jets background using A LPGEN+P YTHIA Monte Carlo, shows a statistically significant disagreement. One possible way to interpret this disagreement is as an excess in the 120-160 GeV/c2 mass range. If we model the excess as a Gaussian component with a width compatible with the dijet invariant mass resolution we obtain a p-value of 7.6 × 10−4 , corresponding to a significance of 3.2 standard deviations, after accounting for all statistical and systematic uncertainties.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 37 / 38
  44. 44. Conclusion II Even if it is hardV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 38 / 38
  45. 45. Conclusion II Even if it is hard Need to keep the focus on things we know: 1 CDF now sensitive to very small cross sections and small differences between data and model also in complex final statesV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 38 / 38
  46. 46. Conclusion II Even if it is hard Need to keep the focus on things we know: 1 CDF now sensitive to very small cross sections and small differences between data and model also in complex final states Work to do: 1 Understand if the excess is truly new physics or simply mismodeling 2 If it is found to be mismodeling need to comprehend why we don’t understand this hard kinematical region as we would expect from theory 3 Need to carefully consider both possibilities as it is exciting both ways.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 38 / 38
  47. 47. Acknowledgments Would like to thank J.Campbell, E. Eichten, R. K.Ellis, C.Hill, K.Lane, M. Mangano, A. Martin for the useful suggestions and discussionsV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 39 / 38
  48. 48. BackupV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 40 / 38
  49. 49. ∆R studies If we do reweigh to the sidebands we observe that the significance drops to 2.3σ Does it really makes sense? 1 The two sidebands are qualitatively different (events are from two different kinematical regions, with different sample composition) → ∆Rjj is highly correlated (and the correlation is hard to understand) with the Mjj . 2 reweighing to the sum of the two sidebands, since the low one has more statistics, is not completely right because we artificially make the mjj look more like the low sideband.V. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 41 / 38
  50. 50. ∆Rjj DeltaR_jj_mjj mjj 400 350 300 250 200 150 100 50 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 DeltaR_jjV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 42 / 38
  51. 51. Fit requiring pT,W > 60 GeV/cV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 43 / 38
  52. 52. What happens if we change the Jet Energy scale? Result of the fit scaling JES up by 7% Sig always above 3σV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 44 / 38
  53. 53. Results from D0 using 1.1 fb−1 lin paperV. Cavaliere (University of Illinois at Urbana-Champaign) Seminar 04/06/2011 45 / 38

×