Observation of Higgs boson decay to bottom quarks with CMS (and ATLAS) data

Size: px

Start display at page:

Download "Observation of Higgs boson decay to bottom quarks with CMS (and ATLAS) data"

Blake Dickerson
5 years ago
Views:

B 786 (2018) 59 Seminars for Particle physics Oxford, Nov.

1 Observation of Higgs boson decay to bottom quarks with CMS (and ATLAS) data CMS: PRL 121, (2018) ATLAS: Phys. Lett. B 786 (2018) 59 Seminars for Particle physics Oxford, Nov. 20 th 2018 Luca Perrozzi (ETH Zurich) on behalf of the CMS Collaboration

2 Warning This seminar might feel like the Pitt Rivers Museum Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 2

3 Outline The LHC and its multi-purpose detectors From the Higgs discovery (back) to H bb VH(bb) analysis with 2017 data in CMS (and ATLAS) Combination with previous results Prospects Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 3

4 the Mt. Blanc (1/fb or 1k Higgs per running day!) Geneva lake

The Large Hadron Collider After successful Run 1, LHC has produced >3 years of 13 TeV data with stunning performance Run2 delivered integrated luminosity >150 fb -1 DESIGN peak luminosity

5 The Large Hadron Collider After successful Run 1, LHC has produced >3 years of 13 TeV data with stunning performance Run2 delivered integrated luminosity >150 fb -1 DESIGN peak luminosity exceeded by a factor of 2 Average pileup ~38 in 2017 and 2018 Incredible machine availability, >50% of time in stable operation Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 5

6 The Large Hadron Collider Run1 Run2 Run3 NOW High Luminosity >x20 data to come! (in the next 20y) Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 6

7 A ToroidaL ApparatuS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 7

8 Compact Muon Solenoid Multipurpose detector at the LHC: silicon tracking, electromagnetic & hadronic calorimeters, a 3.8 T superconducting solenoid, & muon tracking chambers Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 8

9 Compact Muon Solenoid Multipurpose detector at the LHC: silicon tracking, electromagnetic & hadronic calorimeters, a 3.8 T superconducting solenoid, & muon tracking chambers Pixel detector upgraded for 2017 data taking Large impact on b-tagging performance (as discussed later) Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 9

10 FROM THE SM TO THE HIGGS DISCOVERY bb TO H BB Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 10

11 Standard Model of Particle Physics Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 11

12 T-shirt available also in this form If you want to keep it in mind, you can print it on a t-shirt Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 12

13 At the end what (Theo) physicists want is to put an equation on a t-shirt Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 13

14 G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 14

15 G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 15

16 it works Yesterday s discovery is today s calibration, and tomorrow s background. V. L. Telegdi [ ][ ] Inelastic collisions: ~ pb Eight orders of magnitude of EWK and top Physics Higgs W and Z bosons Top quarks Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 16

17 on both sides of the ring Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 17

18 The Higgs sector: a whole uncharted territory Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 18

19 Higgs boson: pheno profile in 1976 = 100 GeV Slide courtesy of A. David Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 19

20 The Higgs sector: a whole uncharted territory adapted from G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 20

Higgs boson discovery July 4th 2012 CMS and ATLAS reported independently the first observation of the Higgs boson Standard Model picture complete Nobel prize awarded to P.

21 Higgs boson discovery July 4th 2012 CMS and ATLAS reported independently the first observation of the Higgs boson Standard Model picture complete Nobel prize awarded to P. Higgs and Englert in 2013 Result driven by γγ and ZZ decay modes High mass resolution High signal to background ratio High signal trigger efficiency ATLAS: PLB 716 (2012) 1-29 CMS: PLB 716 (2012) 30 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 21

22 The Higgs sector: a whole uncharted territory adapted from G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 22

23 Mass: Established Higgs properties ATLAS+CMS: PRL 114 (2015) Spin/Parity: 0 + ATLAS: EPJC 75 (2015) 476 CMS: PRD 92 (2015) Width: < 1 GeV (direct) < GeV (indirect) Observed direct coupling to: Vector bosons CMS: JHEP 11 (2017) 047 ATLAS: arxiv: submitted to PLB ATLAS: PLB 716 (2012) 1-29 CMS: PLB 716 (2012) 30 t leptons top quarks ATLAS: ATLAS-CONF CMS: PLB 779 (2018) 283 ATLAS: PLB 784 (2018) 173 CMS: PRL 120 (2018) All measurements compatible with SM predictions Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 23

24 adapted from G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 24

$H bb : motivations H bb has the largest branching fraction (58%) for m H =125 GeV Unique final state to measure coupling with down-type quarks Drives the$

25 H bb : motivations H bb has the largest branching fraction (58%) for m H =125 GeV Unique final state to measure coupling with down-type quarks Drives the uncertainty of the total Higgs boson width Limits the sensitivity to BSM contributions Not yet GeV Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 25

First H bb searches started at LEP m H > 114.

26 First H bb searches started at LEP m H > %CL Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 26

27 and continued at Tevatron nb: after Higgs discovery Significance 2.8σ 125 GeV Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 27

S/B Primary decay mode for searches at LEP and Tevatron More difficult at LHC due to increased

28 Challenges of H bb H bb compared with the ZZ channel H bb search needs: Highly efficient b-jets identification Excellent resolution on m(bb) Use of full event information to increase S/B Primary decay mode for searches at LEP and Tevatron More difficult at LHC due to increased background S/B ~2.5x worse than Tevatron Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 28

29 H bb searches continue(d) at the LHC Very large datasets at LHC give access to several production modes to search for H bb CMS: PRL 120 (2018) ATLAS: Phys. Rev. D 98, (2018) ATLAS: JHEP 11 (2016) 112 CMS: HIG CMS: PRD 92 (2015) ATLAS: JHEP 05 (2016) 160 ATLAS: PRD 97, (2018) CMS: JHEP 09 (2014) 087 CMS: arxiv: submitted to JHEP CMS: JHEP 06 (2018) 101 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 29

VH(bb ) results at LHC as of Aug 18 VH(bb) evidence at LHC established with 2016 data by both ATLAS and CMS Detectors clearly demonstrated ability to deal with very high pile-up for such complex

30 VH(bb ) results at LHC as of Aug 18 VH(bb) evidence at LHC established with 2016 data by both ATLAS and CMS Detectors clearly demonstrated ability to deal with very high pile-up for such complex analysis Signal strength uncertainty ~40% signal strength significance (exp) significance (obs) ATLAS Run 1 [1] 2.6σ 1.4σ CMS Run 1 [2] 2.5σ 2.1σ ATLAS+CMS Run 1 [3] 3.7σ 2.6σ ATLAS [4] 3.0σ 3.5σ CMS 2016 [5] 2.8σ 3.3σ [1] JHEP 01 (2015) 069 [2] JHEP 08 (2016) 045 [3] JHEP 08 (2016) 045 [4] JHEP 12 (2017) 024 [5] PLB 780 (2018) 501 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 30

31 VH(BB) bb WITH 2017 DATA IN CMS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 31

Analysis strategy: VH(bb): Analysis strategy 3 channels with 0, 1, and 2 leptons and 2 b-tagged jets To target Z(νν)H(bb), W(lν)H(bb)and Z(ll)H(bb) processes Signal region designed to increase S/B

32 Analysis strategy: VH(bb): Analysis strategy 3 channels with 0, 1, and 2 leptons and 2 b-tagged jets To target Z(νν)H(bb), W(lν)H(bb)and Z(ll)H(bb) processes Signal region designed to increase S/B Large boost for vector boson Multivariate analysis exploiting the most discriminating variables (m bb, ΔR bb, b-tag) Control regions to validate backgrounds and control/constrain normalizations Z+bb tt 0-lepton (MET) 1-lepton [e,μ] 2-leptons [ee,μμ] W+bb normalization from data, shapes from MC Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) single t 32

discriminating variables (m bb, ΔR bb, b-tag) Control regions to validate backgrounds and control/constrain normalizations 0-lepton (MET)

33 Analysis strategy: VH(bb): Analysis strategy 3 channels with 0, 1, and 2 leptons and 2 b-tagged jets To target Z(νν)H(bb), W(lν)H(bb)and Z(ll)H(bb) processes Signal region designed to increase S/B Large boost for vector boson Multivariate analysis exploiting the most discriminating variables (m bb, ΔR bb, b-tag) Control regions to validate backgrounds and control/constrain normalizations 0-lepton (MET) 1-lepton [e,μ] 2-leptons [ee,μμ] Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) and di-boson, of course used to validate the analysis strategy 33

What makes it possible: Particle Flow Described in arxiv:1706.

34 What makes it possible: Particle Flow Described in arxiv: Reconstruct all particles: muons, electrons, photons, neutral hadrons, charged hadrons Combine info from all relevant sub-detectors in a Global Event Description Trigger: first step is regional reco, plus optimized version of PF run at HLT critical for (b-)jets, ET miss, lepton isolation Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 34

35 More about physics objects Lepton isolation: consider particles in a cone around lepton removing charged hadrons associated to pileup vertices and correct for neutral particles on average Jets: use anti k T (R = 0.4) clustering with all particle-flow objects except charged hadrons associated to pileup vertices response + data-simulation corrections E T miss : use negative vectorial sum of all particle-flow objects jet-based corrections and datasimulation corrections based on hadronic recoil Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 35

Event Selection/Categorization Control region example Selections (jets, leptons, b-tagging) optimized separately by channel 4 analysis categories: 0-lepton: p T (Z) > 170 GeV 1-lepton: p T (W) > 150

36 Event Selection/Categorization Control region example Selections (jets, leptons, b-tagging) optimized separately by channel 4 analysis categories: 0-lepton: p T (Z) > 170 GeV 1-lepton: p T (W) > 150 GeV 2-lepton: p T (Z) > 150 GeV 2-lepton: 50 < p T (Z) < 150 GeV [*] For 2-lepton channel use M Z veto Control regions designed to map closely each signal region Inverted selections to enhance purity in targeted backgrounds: tt, V+light flavor, and V+heavy flavor Signal Region [*] Number of additional jets in the event Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 36

37 n. events Fit strategy in a nutshell Signal regions Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 37

38 n. events Fit strategy in a nutshell Signal regions V+Heavy flavor control regions + control regions obtained inverting selections on number of jets, b-tagging, m(ll) Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 38

39 n. events Fit strategy in a nutshell Simultaneous fit to extract: Data/MC background normalization factors Signal strength, significance + control regions obtained inverting selections on number of jets, b-tagging, m(ll) Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 39

40 VH(bb) in 2017: main features Improved mass resolution from: Better b-jet identification New b-jet energy regression Kinematic fit in 2-lepton channel FSR jet recovery Use of deep neural network (DNN) to discriminate: Signal from background, in Signal Regions Background components among each other, in Control Regions Combined effect: O(5-10%) increase of the analysis sensitivity wrt 2016, depending on channel Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 40

41 A couple of words on Machine Learning Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 41

42 Wait, this is a fit! Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 42

43 It can solve easy problems h1 h2 y Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 43

44 It got Deep Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 44

45 It s often used to classify Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 45

46 But it can do much more Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 46

47 It got everywhere Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 47

48 Warning: it should not be used as a black box Andrew Ng Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 48

b-jet identification Continuous effort to improve b-tagging at CMS New pixel detector (4 layers) DNN algorithm (DeepCSV) with additional per-track information Contamination from q/g < 1%

49 b-jet identification Continuous effort to improve b-tagging at CMS New pixel detector (4 layers) DNN algorithm (DeepCSV) with additional per-track information Contamination from q/g < 1% for efficiency ~70% MC corrections derived on data with tt events Good agreement between data and MC verified in all analysis regions Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 49

50 b-jet identification Continuous effort to improve b-tagging at CMS New pixel detector (4 layers) DNN algorithm (DeepCSV) with additional per-track information Contamination from q/g < 1% for efficiency ~70% MC corrections derived on data with tt events Good agreement between data and MC verified in all analysis regions 0-lep Z+HF 1-lep ttbar 2-lep Z+HF Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 50

$flavor (µ/e), jet mass, fragmentation-like variable, energy fractions in DR rings Significant m(bb) resolution improvement without sculpting of the background σ/peak down to 11.9% in 2017 wrt 13.$

51 b-jet energy regression Regression mainly recovers missing energy in the jet due to neutrino Switch from Boosted Decision Trees to DNN algorithm Extended set of input variables now including lepton flavor (µ/e), jet mass, fragmentation-like variable, energy fractions in DR rings Significant m(bb) resolution improvement without sculpting of the background σ/peak down to 11.9% in 2017 wrt 13.2% in 2016 Higgs mass in 2-lep SR Top mass in 1-lep tt CR Z(ll)H(bb) Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) Stat. unc. only 51

52 Kinematic fit in 2-lepton channel No intrinsic missing energy in the Z(ll)H(bb ) process Improve jet p T measurement through kinematic fit procedure Constrain dilepton system to Z mass Balance the ll+bb+j system in the (p x,p y ) plane Improvement up to 36% on m(bb ) resolution w/ kin fit Stat. unc. only Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 52

Signal vs Background discriminator DNN discriminator used to extract signal 15 input variables: b-jet properties, di-jet kinematics, event topology, carefully validated through data/mc comparison

53 Signal vs Background discriminator DNN discriminator used to extract signal 15 input variables: b-jet properties, di-jet kinematics, event topology, carefully validated through data/mc comparison Trained separately in each channel Performance optimization with blind analysis Df(jj) in 0-lep Z+HF p T (W) in 1-lep ttbar N add-jets in 2-lep Z+HF Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 53

54 Signal vs Background discriminator Just an example plot Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 54

through data/mc comparison Trained separately in each channel Performance optimization

55 Signal vs Background discriminator DNN discriminator used to extract signal 15 input variables: b-jet properties, di-jet kinematics, event topology, carefully validated through data/mc comparison Trained separately in each channel Performance optimization with blind analysis 0-lepton 1-lepton 2-lepton Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 55

Heavy Flavor control region discriminators Reminder: leading systematic

pure Fit b-tag shape (DeepCSV) to discriminate processes 0- and 1-lepton

among background components Use same input variables as Signal vs Background

56 Heavy Flavor control region discriminators Reminder: leading systematic uncertainty from normalization of V+(b)b 2-lepton channel control region very pure Fit b-tag shape (DeepCSV) to discriminate processes 0- and 1-lepton channel control regions less pure Fit DNN multi-categorizer to distinguish among background components Use same input variables as Signal vs Background discriminator 2-lep 1-lep 0-lep Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 56

57 Fit setup and background normalization Simultaneous fit of Signal and Control Regions to extract signal and background normalizations Fitted variables: DNN or b-tagging shapes, or yields depending on the region MC shapes and normalizations floated within constraints from systematic uncertainties through nuisance parameters Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 57

58 Systematic uncertainties Total uncertainty ~34%, statistically dominated Major sources of systematic uncertainties from background normalization and modeling, b-tagging, MC sample size ` Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 58

59 Validation: VZ(bb) VZ analysis using Z(bb) standard candle next to H(bb) peak Same technology used for VH(bb) fit Same DNN inputs (but dedicated training), same Control Regions, VH(bb) normalized to SM and left free to float Larger m(bb) window in Signal Region to fully include Z(bb) peak Significance 5.0σ expected 5.2σ observed Signal strength µ = 1.05 ± 0.22 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 59

60 VH(bb) Results with 2017 data Results with 2017 data compatible with SM expectations Observed significance 3.3σ, signal strength 1.08 ± 0.34 O(5-10%) increase in analysis sensitivity wrt 2016, depending on channel Remarkable channel compatibility ± 0.4 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 60

61 Combining DNN distributions DNN distributions sorted into bins of similar S/B ratio and combined Excess well compatible with SM Higgs boson signal hypothesis 2-e, high p T (V) 2-µ, high p T (V) 2-e, low p T (V) 2-µ, low p T (V) 1-lepton (e) 1-lepton (µ) 0-lepton Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 61

m(jj) distributions combined and weighted by S/(S + B) Signal strengths compatible with main analysis

62 Visualizing the excess: m(jj) analysis Fit to the m(jj): lower sensitivity but direct visualization of the Higgs boson signal Events categorized in DNN sensitivity after removing correlations with m(jj) m(jj) distributions combined and weighted by S/(S + B) Signal strengths compatible with main analysis Excess compatible with the sum of the two peaks Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 62

63 COMBINATION OF VH(BB) bb RESULTS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 63

Combination of VH(bb) results 2017 Run 2 (2016+2017) Run 1 +

64 Combination of VH(bb) results 2017 Run 2 ( ) Run 1 + Run 2 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 64

Combination of VH(bb) results 2017 Run 2 (2016+2017) Run 1 + Run 2 NB: 5σ observation of VH production in reach if VH(tt) from HIG-18-007 is added.

65 Combination of VH(bb) results 2017 Run 2 ( ) Run 1 + Run 2 NB: 5σ observation of VH production in reach if VH(tt) from HIG is added. This result is not contained in the paper as VH probes the HVV coupling, already established in Run 1 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 65

66 COMBINATION OF H BB bb RESULTS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 66

67 Combination of H bb measurements Combination of CMS H bb measurements : VH, boosted ggh, VBF, tth Most sources of systematic uncertainty are treated as uncorrelated Theory uncertainties are correlated between all processes and data sets Measured signal strength is µ = 1.04 ± 0.20 Significance 5.5σ expected 5.6σ observed Observation of the H bb decay by the CMS Collaboration Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 67

68 VH(bb) Run2 result in ATLAS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 68

69 Observation of H bb in ATLAS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 69

70 FUTURE PROSPECTS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 70

71 G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 71

72 June 18 adapted from G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 72

73 August 18 Yukawa coupling of second generation leptons and quarks challenging but already promising adapted from G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 73

74 Conclusions CMS and ATLAS has independently reached >5σ observation of the H bb decay, Combination of several production channels, dominated by VH(bb) Result contained in CMS: PRL 121, (2018) ATLAS: Phys. Lett. B 786 (2018) 59 Standard Model assumption on Yukawa coupling to b s confirmed within the present O(20%) uncertainty This result is the culmination of H bb searches that started at LEP, continued at Tevatron and at the LHC Achievement possible only thanks to the fantastic run of the LHC, and the CMS detector performance But is only a step towards the ultimate H bb precision at LHC Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 74

75 BACKUP SLIDES Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 75

76 b-tagging performance in Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 76

77 b-tagging performance in Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 77

78 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 78

79 b-jet energy regression Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 79

80 b-jet energy regression Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 80

81 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 81

82 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 82

83 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 83

84 Kinematic fit in 2-lepton channel Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 84

m(bb) improvements: 2016 vs 2017 2017 2016 Luca

85 m(bb) improvements: 2016 vs Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 85

86 2-lepton heavy flavor control regions Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 86

87 0- and 1-lepton heavy flavor control regions Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 87

88 Signal and Control region definitions Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 88

89 SF comparison: 2016 vs 2017 Note: change in PDF, UE tune, generator versions, b-tagging algorithm, fit binning and 1-lepton pt(v) increase from 100 to 150 GeV between 2016 and 2017: no direct comparison of SF possible Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 89

Signal vs Background discriminator To increase sensitivity, use DNN discriminator to extract signal DNN outperforms BDT due to network depth Same input variables as 2016 (b-jet properties, di-jet

90 Signal vs Background discriminator To increase sensitivity, use DNN discriminator to extract signal DNN outperforms BDT due to network depth Same input variables as 2016 (b-jet properties, di-jet kinematics, event topology) Validated through data/mc comparison Trained separately in each channel to discriminate VH(bb ) from the weighted sum of all backgrounds Parameters optimized to maximize sensitivity Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 90

91 0- and 1-lepton signal regions DNN Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 91

92 2-lepton signal regions DNN Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 92

93 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 93

94 Jet energy scale: Split into 27 independent uncertainty sources Jet energy resolution: Systematic uncertainties 10% uncertainty on regressed b-jets from dedicated study Decorrelated for signal to avoid any possible constraining, covers any uncertainties from PS. Standard JER uncertainty for additional jets. B-tagging: Split into independent uncertainty sources Further de-correlated based on jet pt/η, as in 2016 analysis Background normalizations: Derived from fit to data for backgrounds with floating normalisation (V+udcsg, V+b, V+bb, tt) 15% uncertainty on VV and single top cross section. Monte Carlo statistics QCD scales and PDF variations Acceptance as well as overall cross section Lepton efficiency, pile-up re-weighting, luminosity Residual data/mc discrepancies Δη(jj) LO to NLO re-weighting in V+jets Full correction taken as uncertainty. p T (W) linear re-weighting for tt (all channels) and W+jets, single top (1-lepton channel only) Statistical uncertainty band from fit to derive corrections Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 94

similar signal-to-background ratio, and combined

95 VH(bb) DNN distributions 0-lepton 1-lepton 2-lepton DNN distributions can also be sorted into bins of similar signal-to-background ratio, and combined Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 95

96 Candidate event for Z(ee)H(bb) Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 96

97 H bb combination: syst unc correlations Combination of published Run1 and Run2 CMS measurements on H bb: VH, boosted ggh, VBF, tth Most sources of systematic uncertainty are treated as uncorrelated The dominant jet energy scale uncertainties correlated between processes at the same energy Theory uncertainties are correlated between all processes and data sets Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 97

98 Luminosity accumulation in CMS for 2018 J. Butler - 25th Rencontres du Vietnam 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 98

99 CMS pileup profiles during Run 2 σ inel = 80 mb Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 99

100 CMS design and 2017/18 Evolution CMS Design Large solenoid 6m diameter x 13 m long Tracking and calorimetry fit inside Very strong field 3.8T Excellent momentum resolution Chambers in the return iron track and identify muons, leading to a very compact system A lead tunstenate crystal calorimeter (~76K crystaks) for photon and electron reconstruction Hadron calorimeters for jet and missing E T reconstruction up to h~5 Charged Particle Tracking with all-silicon components A silicon pixel detector out to radius ~20 cm A silicon microstrip detector from there out to 1.1m Weigh, dominated by steel, is Tonnes Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 100

101 Evolution of Analysis Techniques J. Butler - 25th Rencontres du Vietnam 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 101

102 The LEP+Tevatron legacy adapted from C. Vernieri - SSI 2018 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 102

103 H bb searches continue(d) at the LHC Very large datasets at LHC give access to several production modes to search for H bb Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 103

104 J. Butler - 25th Rencontres du Vietnam 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 104

105 G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 105

106 J. Butler - 25th Rencontres du Vietnam 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 106

107 G. Salam, LHCP 18 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 107

108 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 108

109 Talk at the 12 th LHC Higgs XSWG Oct Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 109

110 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 110 Talk at the VH LHC Higgs XSWG soubgroup Apr. 2018

111 Talk at the Hbb Evidence Oct Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 111

112 Higgs at the LHC Run1 legacy Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 112

113 C. Vernieri - SSI 2018 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 113

114 C. Vernieri - SSI 2018 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 114

115 LHC Physics Evolution in 2008 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 115

116 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 116

117 ATLAS VS CMS Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 117

118 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 118

119 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 119

120 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 120

121 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 121

122 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 122

123 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 123

124 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 124

125 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 125

126 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 126

127 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 127

128 Luca Perrozzi - Observation of Hbb with CMS (and ATLAS) 128

HIGGS STUDIES IN ATLAS AND CMS

HIGGS STUDIES IN ATLAS AND CMS Nicola De Filippis Politecnico & INFN, Bari and LPC-FNAL, Batavia On behalf of the ATLAS and CMS collaborations N. De Filippis Aug. 31 - Sept. 9, 2018 1 Outline SM Higgs