1. 8 TeV W+bb Cross Section Analysis Using the Compact Muon Solenoid Detector at the Large Hadron Collider 1 T. Mastrianni Perry Preliminary Exam 5 December 2013 -

2. Outline 2 T. Mastrianni Perry Standard Model of Particle Physics Motivations for studying W+bb cross section Large Hadron Collider and CMS Accelerator and Detector used in this study Observing Particles Monte Carlo Simulation Particle Reconstruction Analysis Procedure Signal/Background Extraction -

3. Standard Model Particles 3 T. Mastrianni Perry Fundamental Fermions (spin ½) Quarks (u, d, s, c, b, t) Leptons (e, μ, τ, ν e,ν μ,ν τ ) Fundamental Bosons Spin 1: Force Carriers Gluon: Strong Force massless color / anti-color Photon: Electroweak (EM) massless uncharged W ±, Z: Electroweak (Weak) 80.4 GeV, 91.2 GeV electric charge Spin 0: Higgs 125 GeV EWK Symmetry Breaking Mass to W ±, Z, quarks, leptons Mass Charge Spin

4. 4 T. Mastrianni Perry - Weak Force (W ±, Z) Maximally parity violating Mixing of quarks Nuclear decay EM Force (photon) Binds atoms together Binds molecules together Relative Strength Standard Model Particle Interactions Proton Parton Distribution Function Describes Components of Proton Strong Force (gluon) Binds the nucleus together Hadrons baryons (qqq) mesons (qq)

5. 5 E T miss + muon = W T. Mastrianni Perry Vector Sum of Energy Should Be Zero Not all Particles Interact with Detector E T miss = - Σ i p T (i) Our Signal has a neutrino in the final state: E T miss E T miss Missing Transverse Energ y (E T miss ) Vertices Primary Vertex (PV) Interaction point where proton beams collide Secondary Vertex (SV) Point from which particles emanate (not PV) results from secondary decays of unstable particles

6. Signal: W(μν)+bb 6 T. Mastrianni Perry W decays to a muon and neutrino Gluon Splits to bottom / anti-bottom pair b quarks decay to light quarks which hadronize At √s = 7 TeV Measured Cross Section = 0.53 ±0.05(stat.) ±0.09(syst.) ±0.01(lum.) pb - We See: Muon + Missing Energy Two Hadron Showers from Secondary Vertex

7. Single Top t-channel 2 real b Jets, Muon, E T miss 1 extra Jet not detected Major Backgrounds 7 T. Mastrianni Perry TTbar: 2 real b Jets, Muon, E T miss 2 extra jets not detected or lepton missing W+light quarks: 2 jets, Muon, E T miss but fake displaced SV

8. Motivation for W+bb Study T. Mastrianni Perry Searching at √s = 8 TeV 125 GeV Image from FermiLab D0 8 - Other Experiments See Tension With Simulaton and Observation for W+b(b) particularly in collinear region Important for Searches H(bb) has highest branching ratio E T miss + muon + heavy quark predicted in BSM models This is the only cro ss section measurement in this phase space - -

9. Outline 9 T. Mastrianni Perry Standard Model of Particle Physics Motivations for studying W+bb cross section Large Hadron Collider and CMS Accelerator and Detector used in this study Observing Particles Monte Carlo Simulation Particle Reconstruction Analysis Procedure Signal/Background Extraction -

10. LHC: General Information 10 T. Mastrianni Perry At CERN : Located Near Geneva, Switzerland Proton – Proton Collider 4.3 km radius 8 TeV CM Energy (2012) 13-14 TeV in 2015 100 m underground Four Detectors CMS, ATLAS General Purpose ALICE Heavy Ions LHC-B B Quark Physics

11. LHC: Accelerator Components, Acceleration Process 11 T. Mastrianni Perry CMS Proton Source 90 keV energy, pulsed every 1.2 s Radio Frequency Quadrupole 750 keV, pulsed every 1.2 s Linac 2 50 MeV, pulsed every 1.2s PS Booster 1.4 GeV, 1.2s cycle time Proton Synchrotron 25 GeV, 3.6 s cycle time Super Proton Synchrotron 450 GeV, 200 MHz Large Hadron Collider 3.5, 4 TeV, 89 μs orbit time Collisions Every 50 ns

12. 12 T. Mastrianni Perry Pileup: “extra” collisions in a bunch crossing LHC: Statistics of Collisions *using estimated xc -

13. CMS: Overview 13 T. Mastrianni Perry Electronic Calorimeter (ECAL) PbWO 4 Crystals Hadronic Calorimeter (HCAL) Brass with Plastic Scintillator Tracker Silicon Pixels and Microstrips Solenoid Magnet 3.8 Tesla Muon Barrel and Endcap Drift Tube (DT) (B) Resistive Plate Chamber (RPC) (B,E) Cathode Strip Chamber (CSC) (E) Iron Yoke Mass: 12 500 T Radius: 7.5 m Length: 21.5 m Proton Beam

14. CMS: Geometry 14 T. Mastrianni Perry +z -z pTpT ϕ η(θ) Pseudorapidity, η Lorentz Invariant angle down to beamline* η=-log[tan(θ/2)] Polar Angle, ϕ Transverse Momentum, p T Momentum in radial direction η = 0 (θ = π/2) η = 0.88 (θ = π/4) η = ∞ (θ = 0) * true to the extent mass is negligible Beam

15. CMS: Tracker 15 T. Mastrianni Perry Pixel Outer Barrel (TOB) Inner Barrel (TIB) Endcap (TEC) Radius: 1.2 m Length: 5.4 m All Silicon High p T Muons Hadrons with high momentum resolution Isolated Electrons Secondary Vertices Four Single-Sided Outer Barrel Layers Two Double-Sided Outer Barrel Layers Four Inner Barrel Layers 10 cm x 80 μm Three Pixel Layers 100 x 150 μm 2 |η| < 2.5 K Meson – 2-3 cm B Meson – 1-2 mm 25 cm x 180μm δp T /p T ≈ ( 15 x p T [TeV] + 0.5 ) %

16. CMS: ECAL 16 T. Mastrianni Perry 80,000 Lead-Tungstate Crystals attached to avalanche photodiodes σ/E[GeV] ≈ ( (2.83/√E ) + (0.124/E) + 0.3 ) % Crystals Radiation Length = 0.89 cm Length: 26 RL = 23 cm Molière Radius = 22 mm Cross Section: 22 mm x 22 mm Δη x Δϕ = 0.0175 x 0.0175 Preshower in Endcap: 1.65 < |η| < 2.6 lead converter – silicon strip detector EM Interacting Particles Electrons, Photons

17. CMS: HCAL 17 T. Mastrianni Perry HB + HE (|η| < 3) : σ/E[GeV] ≈ ( (115/√E ) + 5.5 ) % HF (3 < |η| < 5): σ/E[GeV] ≈ ( (280/√E ) + 11 ) % Barrel (HB): |η| < 1.4 Endcap (HE): 1.3 < |η| < 3.0 50 mm brass plates 4 mm scintillator sheets Tiles: Δη x Δϕ = 0.87 x 0.87 Forward (HF): 3.0 < |η| < 5.0 Steel + Quartz Fibers Neutral Hadrons Jets Total Particle Energy Missing Energy

18. CMS: Muon Spectrometer 18 T. Mastrianni Perry B = 2 Tesla External to Magnet Barrel DT, RPC: |η| < 1.3 Endcap CSC: 0.9 < |η| < 2.4 RPC: |η| < 1.6 DT: 40 mm x 11 mm Ar/CO 2 Mixture CSC: Crossed Wires (r) Cathode Strips (ϕ) RPC: Gasious Parallel Plate Plastic Anode + Cathode All Components are used in triggering Four Layers: Called Stations Endcap Disc Designed by UW

19. Level 1 Trigger 19 T. Mastrianni Perry 40 MHz bunch crossing rate to 100 kHz output Using Custom Hardware and Algorithms First Look at Data Keep “Interesting” Events 4 μs Triggers For: Muon, Electron/Photon, Jet, Tau, Missing Energy, Combinations

20. Upgraded Calorimeter Trigger 20 T. Mastrianni Perry I also work on Electron/Photon and Tau UCT In Technical Design Report (plot made by me) Electron/Photon Rate at Constant Fraction of Plateau Efficiency Electron/Photon Jet Tau

21. High Level Trigger 21 W+bb Analysis: Isolated Muon p T > 24 T. Mastrianni Perry Detailed Look at Objects 100 kHz L1 Output Rate to 300 Hz for Permanent Storage Custom Software on Commercial Processor Farm Algorithms Similar to Offline Reconstructions Uses Full Event Data Muon Trigger Path L1: Estimate Transverse Momentum Match DT/CSC with RPC Candidates HLT: L2 Find muon path segments Isolation: Sums of Transverse Energy HLT: L3 (If good path found in L2) Tracker: Match Tracks to L2 Candidate Isolation: Sums of p T of charged particle tracks in cone around muon Generated Sample from LHCC-G-134 -

22. Outline 22 T. Mastrianni Perry Standard Model of Particle Physics Motivations for studying W+bb cross section Large Hadron Collider and CMS Accelerator and Detector used in this study Observing Particles Monte Carlo Simulation Particle Reconstruction Analysis Procedure Signal/Background Extraction -

23. Event Simulation 23 T. Mastrianni Perry MadGraph – Matrix Element Calculator Uses Parton Distribution Function (PDF) Underlying Hard Interaction Pythia – Generates Events Showering, Hadronization, Final State Radiation (FSR) GEANT – Passage of Particles Through Matter Simulates CMS Detector Effects Using Monte Carlo (MC) Techniques CMSSW: Process Simulated and Real Data Identically Corrections Applied to MC Simulation: Scale Number of Events to Data Luminosity Pileup Identity - Different PU Distribution in MC vs. Data

24. Particle Flow 24 combines information from subdetectors to reconstruct particles with better resolution T. Mastrianni Perry Reconstructed Particles: Photons, Charged/Neutral Hadrons, Muons, Electrons Serves as an Input for Higher Level Reconstruction Algorithms Three Basic Elements: Charged Particle Tracks Calorimeter Clusters Muon Tracks Reconstruction Steps: Iterative Track Finding ID Tracks in Tracker, Direction of Particle at PV Calorimeter Clustering Energy/Direction of Hadrons, Separate Neutral/Charged Deposits ID Electrons/Bremsstrahlung Photons Linking Match Elements to form ‘blocks’ and avoid double counting Jets, b-Jets, Missing Energy, Taus

25. Electron Reconstruction 25 Match Superclusters to Track Seeds T. Mastrianni Perry Track Seed: Iteratively ID hits in tracker Expect helical path if no Bremsstrahlung radiation Kinks indicate emission of Bremsstrahlung Photons Search a straight line tangent to track for ECAL hit Supercluster: Group of Clusters of ECAL Energy Deposits Uses Strip in ϕ to Account for Bremsstrahlung Radiation HCAL / ECAL Deposit Ratio < 0.15

26. Muon Reconstruction 26 Tracker Track Reconstructed from inner tracker Standalone-Muon Track Using muon stations T. Mastrianni Perry Global-Muon Reconstruction (outside-in) For each standalone-muon track find matching tracker track + reconstruct Tracker-Muon Reconstruction (inside-out) For tracker track (p T > 0.5 GeV, p > 2.5 GeV) find standalone-muon track + reconstruct Tracker - better efficiency at low momentum: only requires single segment in muon system Global - better efficiency if multiple muon stations are hit Often muons qualify as both

27. Muon Selection and Corrections 27 Efficiency Corrections Muon ID (99%), Trigger (98%), Isolation (100%) η dependent, in p T range T. Mastrianni Perry Tight Selection Global Muon: χ 2 /d.o.f. of track < 10 ≥ 1 muon chamber used in fit Tracker Muon: matched to standalone using ≥ 2 stations ≥ 10 inner tracker hits ≥ 1 pixel hit track vertex ≤ 2 mm from primary vertex Isolation in cone ΔR = 0.3 around muon (excluding muon) Σ(p T tracker +E T (E/H)CAL )/p T muon < 0.12 “Rochester” Corrections corrects magnitude of momentum misalignments in CMS detector function of muon charge, η, ϕ

28. Jets Particle Shower from Hadronization 28 T. Mastrianni Perry Anti-k T Algorithm for Jet Clustering Highest p T track, i, called a jet For Subsequent Tracks, j: If d ij <d iB : combine with i Else j is a jet Soft particles cluster around hard ones Hard Shape Circular, clips from soft particles Collinear and Infrared Safe “footprint of a parton” “Loose ID” Requirements anti-k T using R = 0.5 No Single Hit in ECAL with 90% Deposition E T ECAL /(E T HCAL +E T ECAL ) > 1% Generated Sample

29. Jet Corrections 29 T. Mastrianni Perry Pileup Correction (MC and Data) Remove energy from pileup events – remove luminosity dependence Relative Jet Correction (MC and Data) Make Jet Response Independent of η - map to |η|< 1.3, dijet balance Absolute Jet Correction (MC and Data) Jet Response Independent of p T – corrected back to particle level Residual Data Calibration (Data) Relative Energy Scale (small η,p T dependence) GEANT isn’t Perfect Resolution Corrections (MC): MC Resolution (Z Peak) sharper than Data, smear out (η, p T dependent) Remove Leptons (MC and Data): Make sure “Jet” isn’t actually an electron or muon correct for nonlinearities in calorimeters Propagate Corrections to E T miss

30. Bottom Quark Identification “bTagging” – CSV Variable 30 Secondary Vertex (SV) – Displaced from Primary Vertex Result of Secondary Decay T. Mastrianni Perry Combined Secondary Vertex (CSV) Algorithm b hadrons have long life (1.5 ps, ct = 450μm) multivariate analysis / neural network displaced tracks, secondary vertices, soft leptons combine information into single variable Efficiency Correction for CSV bTag 93% at p T = 30 is a function of p T, η Light Jet Passes Tag: 0.5%

31. Outline 31 T. Mastrianni Perry Standard Model of Particle Physics Motivations for studying W+bb cross section Large Hadron Collider and CMS Accelerator and Detector used in this study Observing Particles Monte Carlo Simulation Particle Reconstruction Analysis Procedure Signal/Background Extraction -

32. Samples Used 32 T. Mastrianni Perry W + Jets Samples Labeled by Generated Quark Content in Jets TTbar: Fully Leptonic and Semi Leptonic Decay All MC Samples Labeled Using Generator Level Information* Single Top tW Channel Single Top s Channel Single Top t Channel Diboson Production *QCD From Data-Driven Technique Drell-Yan + 2 Jets

33. PreSelection 33 Muon Isolation Tight Selection p T > 25 GeV |η| < 2.1 No Electrons At Least 2 Jets p T > 25 GeV |η| < 2.4 No Jets |η| >2.4 T. Mastrianni Perry Basic Control Region from which we draw out Signal / Backgrounds Data: 6.9M Events MC: 6.0M Events Data / MC = 1.2

34. QCD Multijet 34 T. Mastrianni Perry Data-Driven Method: Invert Muon Isolation (Iso > 0.2) Subtract Data-MC Scale QCD to Fill Gap in Transverse Mass from 0 to 20 GeV require m T > 45 to remove QCD Multijet background m T 2 = E T 2 - p T 2 CUT Before Cut Data: 6.9M Events MC: 6.0M Events Data/MC = 1.2 After Cut: Data: 3.9M Events MC: 3.6M Events Data/MC: 1.1

35. Signal Cut: 2 bTags 35 T. Mastrianni Perry Cut on CSV Variable to Eliminate W+Light Jets, Drell-Yan, Dibosons, QCD CUT Apply bTag Efficiency Correction Data: 3.9M Events MC: 3.6M Events Data / MC = 1.1 Before Cut: Data: 20k Events MC: 18k Events Data / MC = 1.1 After Cut: CUT

36. Signal Cut: Mass of SV 36 T. Mastrianni Perry Require that SV has reconstructed mass Eliminate W+Charm with mis-bTag and W+Light Quark Before Cut: Data: 20k Events MC: 18k Events Data / MC = 1.1 After Cut: Data: 17k Events MC: 16k Events Data / MC = 1.1

37. Signal Cut: Jet Veto 37 T. Mastrianni Perry Number of Jets with p T > 25 GeV Require Exactly Two Jets with p T > 25 Cut away TTbar and Single Top CUT Before Cut: Data: 17k Events MC: 16k Events Data / MC = 1.1 After Cut: Data: 5.1k Events MC: 4.7k Events Data / MC = 1.1

38. Signal Region Summary 38 T. Mastrianni Perry CutData kEvents MC kEvents Preselection69006000 m T > 4539003600 2 bTags20.18. SV Mass > 017.16. Exactly 2 Jets (p T > 25) 5.14.7 MC Normalized To Integrated Luminosity Data / MC = 1.1

39. TTbar Control Region Cut: Jet Requirement 39 T. Mastrianni Perry CUT Preselection Cuts: 0 Electrons 1 Muon p T > 25 GeV, |η| < 2.1 > 1 Jet p T > 25 GeV, |η| < 2.4 (no cut on forward jets) Require > 3 Jets Eliminate W+bb Before Requiring >3 Jets Data: 21k Events MC: 20k Events Data / MC = 1.1 After Requiring >3 Jets Data: 7.7k Events MC: 7.0k Events Data/MC = 1.1

40. TTbar Control Region Summary 40 T. Mastrianni Perry Mass of Non-bTagged Jets Data / MC Agree Within Statistical Error Data: 7700 Events MC: 7000 Events Data / MC = 1.1

41. Z Control Region: DiMuon Mass Window 41 T. Mastrianni Perry Cuts: 2 Jets p T > 25 GeV, |η| < 2.4 2 Muons p T > 25 GeV, |η| < 2.1 Data Events: 47k Events MC Events: 45k Events Data / MC = 1.0

42. Results and Comparison to 7 TeV 42 T. Mastrianni Perry At √s = 7 TeV Measured Cross Section = 0.53 ±0.05(stat.) ±0.09(syst.) ±0.01(lum.) pb Post Fit* W+bb Signal Region: MC 10% low TTbar Control Region: MC 10% low Z DiMuon Control Region: MC 3% low Consistent Results Not Fitted * Fit Allowing Cross Sections to Float Within Systematic Uncertainties Normalized to Integrated Luminosity

43. 8 TeV W+bb Moving Forward 43 Short Term To Do: Understand Systematic Errors Perform Fitting to get Cross Section Long(er) Term To Do: W + single b boosted bb merge into single jet Investigate Vetos strengthen lepton veto? jet veto on 4 th jet instead of 3 rd ? vary MC generation algorithm proton PDF 4 vs. 5 flavors, other generators W(electron) T. Mastrianni Perry - -

44. 14 TeV Analyses 44 Future Projects Higgs Something with b quarks? T. Mastrianni Perry - W+bb Higher Cross Section: ~1.23 pb, More Luminosity From 8 TeV Analysis – improved modeling W+bb Has Been A Great Training Ground Analysis Framework and Procedures Data Acquisition and Triggering MC Simulation and Corrections -