Observation of a new boson with mass near 125 GeV in pp collisions at sqrt(s) = 7 and 8 TeV

1 Introduction

• In the Standard Model, the Higgs field gives mass to W and Z bosons and fermions while leaving the photon massless, but the Higgs boson mass itself is experimentally determined.Earlier precision data, direct searches, and collider exclusions narrowed the allowed mass region before the observation near 125 GeV.Source: §1 · p. 3
• The CMS search observed a new heavy boson near 125 GeV after combining five decay channels in 7 and 8 TeV collision data.The channels were H →γγ, H →ZZ →4ℓ, H →WW →ℓνℓν, H →ττ, and H →bb.Source: §1 · p. 3
• The γγ and ZZ →4ℓ channels provide the highest discovery sensitivity because they produce narrow mass peaks, while the other channels offer complementary information.The WW channel has the largest signal-to-background ratio among the three lower-resolution modes.Source: §1 · p. 3, §2 · p. 4
• The paper gives an in-depth description of the five analyses and compares their channel measurements with Standard Model predictions through Higgs coupling fits and ratios.The analysis includes couplings to fermions and vector bosons and various coupling ratios.Source: §2 · p. 4
• The paper proceeds from detector and event reconstruction through data and simulation, five decay-channel analyses, statistical combination, and first coupling measurements.The organization reflects the progression from analysis inputs to combined results and interpretation.Source: §2 · p. 4

2 The CMS experiment

• The CMS detector uses a 3.8 T, 13 m superconducting solenoid enclosing silicon tracking and electromagnetic and hadronic calorimetry, with muon detectors outside.Barrel, endcap, and forward systems provide broad angular coverage.Source: §2 · p. 4
• CMS uses pseudorapidity η rather than polar angle to describe angular separation because colliding partons receive boosts along the beam direction.The coordinate system defines η = −ln[tan(θ/2)].Source: §2 · p. 4
• The tracker covers |η| < 2.5 and provides high-efficiency charged-particle tracking, transverse-momentum measurement, vertex resolution, and b-jet identification.Muon tracking efficiency exceeds 99% for pT > 1 GeV, while b-jet tagging exceeds 50% efficiency with a light-jet rejection factor of approximately 200.Source: §2 · pp. 4–5
• The ECAL is central to H →γγ and electron measurements, achieving a barrel diphoton mass resolution estimated at 1.1 GeV for a 125 GeV Higgs boson.Its fine-grained lead-tungstate crystals measure electromagnetic shower shapes and support particle identification and isolation.Source: §2 · p. 5
• The trigger system reduces collision data to an offline-storage rate of approximately 500 Hz while maintaining negligible deadtime up to a 100 kHz first-level trigger rate.The high-level trigger reconstructs full events and applies global selection algorithms.Source: §4 · p. 6
• CMS analyzes recorded data through a worldwide grid linking CERN with Tier-1, Tier-2, and Tier-3 computing centers for reconstruction, analysis, and detector simulation.The infrastructure includes one Tier-0, seven Tier-1, more than 50 Tier-2, and over 100 Tier-3 centers.Source: §4 · p. 6

3 Event reconstruction

• Pileup averaged 9 interactions in 2011 and 19 in 2012, so CMS separates vertices along the beam and selects the hard-scattering vertex using the largest associated track-pT sum.In H →γγ, a dedicated algorithm assigns photons to a vertex because the photons carry much of the event transverse momentum.Source: §3 · p. 6
• CMS identifies and reconstructs charged hadrons, neutral hadrons, photons, muons, and electrons with a particle-flow algorithm that also supplies inputs to jets, missing energy, τ decays, and isolation.The method combines information from all CMS subdetectors and all reconstructed vertices.Source: §3 · p. 6
• Photon and electron reconstruction combines ECAL superclusters with tracker information, while photon-energy corrections use multivariate regressions for shower containment and conversion losses.A second BDT estimates the uncertainty of the corrected photon energy.Source: §3 · pp. 6–7, §6 · p. 8
• The ECAL response is stabilized with laser monitoring and collision-based equalization, achieving residual channel-response miscalibration from 0.5% in the central barrel to a few percent in the endcaps.The 7 and 8 TeV dielectron mass resolutions at the Z mass were measured separately in barrel and endcap regions.Source: §3 · p. 7, §6 · p. 8
• Jets are reconstructed with anti-kT and corrected for pileup, underlying-event contributions, detector response, and residual data-simulation differences; pileup jets are removed with a BDT.The residual jet-energy calibration factor is about 5% across the detector acceptance.Source: §6 · pp. 8–9
• Missing transverse energy is the negative vector sum of reconstructed-particle transverse momenta, with component resolution parameterized as σ(Emiss_x,y) = 0.5 × √ΣET.For the 2012 analyses, ΣET was approximately 600 GeV at an average pileup of 19 interactions.Source: §6 · p. 9
• Specialized algorithms identify b jets, hadronic τ leptons, and isolated electrons and muons using displaced vertices, track impact parameters, decay topologies, and pileup-corrected activity.The τ isolation discriminator uses transverse-momentum sums in rings around the candidate and accounts for charged and neutral pileup contributions.Source: §6 · p. 9, §8 · p. 10

4 Data sample and analyses performance

• The data sample contains approximately 5.1 fb−1 at 7 TeV from 2011 and 5.3 fb−1 at 8 TeV from 2012.The analyses cover multiple final states with different mass resolutions and production-mode targets.Source: §4 · p. 10

4.1 Simulated samples

• Simulated signal and background events are processed through GEANT4 detector simulation and the same reconstruction algorithms used for data, with pileup reweighted to observations.The samples use specified PDF sets, hadronization models, and τ-decay simulation packages.Source: §4.1 · p. 11

4.2 Signal simulation

• Higgs signals are simulated for gluon-gluon fusion, vector-boson fusion, associated vector-boson production, and tt-associated production, with cross sections normalized using higher-order QCD and electroweak calculations.Branching fractions are calculated with PROPHECY4F for four-fermion final states and HDECAY for the other final states.Source: §4.2 · p. 11
• Signal-theory uncertainties include PDF variations and renormalization- and factorization-scale variations, while the simulated gluon-fusion transverse-momentum spectrum is matched to higher-order predictions through reweighting or tuning.The treatment differs between energies and analyses, with H →ZZ at 7 TeV exempt from the stated reweighting.Source: §4.2 · p. 11

4.3 Background simulation

• Background simulations use process-appropriate generators including POWHEG, MADGRAPH, PYTHIA, GG2VV, and related higher-order cross-section normalizations for diboson, vector-boson, top-quark, and QCD processes.The generator choices are summarized by analysis in Table 2.Source: §4.3 · pp. 11–12

4.4 Search sensitivities

• Search sensitivity is quantified with expected 95% CL upper limits on σ/σSM and with local p-values and significances as functions of the hypothesized Higgs mass.A channel with an expected limit below unity would be expected to exclude the SM Higgs boson at 95% CL or more in the absence of signal.Source: §4.4 · p. 12
• The statistical methodology follows the approach developed by the ATLAS and CMS LHC Higgs Combination Group and is summarized for the combined search later in the paper.This framework supports the combination of the channel-specific analyses.Source: §4.4 · p. 12

5.1 Diphoton trigger

• Trigger efficiency exceeded 99.5% for events passing the offline H →γγ selection.The diphoton triggers used transverse-energy thresholds plus isolation and cluster-shape requirements, set at least 10% below final-selection thresholds.Source: §5.1 · p. 14

5.2 Interaction vertex location

• A vertex BDT combines track-based, diphoton-balance, and conversion information to select the interaction point, achieving 83.0 ± 0.4% efficiency at 7 TeV and 79.0 ± 0.2% at 8 TeV.The efficiency denotes locating the vertex within 10 mm of the true position for simulated 120 GeV Higgs events.Source: §5.2 · pp. 14–15

5.3 Photon selection

• Photon selection requires two candidates with mass-scaled transverse-momentum thresholds within the ECAL fiducial region, while a photon BDT suppresses nonprompt photons using shower and isolation variables.A loose photon-BDT requirement greater than −0.2 retains more than 99% of signal photons and is calibrated with Z →ee data.Source: §5.3 · pp. 15–16

5.4 Event classification

• Events are categorized by mass resolution and signal-to-background using a mass-independent multivariate classifier, with four classes per energy and dedicated VBF dijet categories.Photon-ID and energy-resolution uncertainties dominate classifier mismodelling; the former nearly covers the data–simulation difference, and both are propagated to the final result.Source: §5.4 · pp. 16, 18, 20

5.5 Signal and background modelling

• Signal and background modelling calibrates photon energy scale and resolution with Z →ee events, assigning 0.2%–0.9% uncertainties to the resulting corrections.The final signal model combines correct- and incorrect-vertex submodels, while parametric and binned mass-shape approaches give consistent results; reported results use the parametric model.Source: §5.5 · pp. 21–22
• The signal model reproduces simulated events, with effective diphoton-mass resolution improving from σeff = 2.77 GeV at low BDT values to 1.34 GeV at high values.The compared 8 TeV BDT classes use BDT values above 0.88 and between −0.05 and 0.50, respectively.Source: §5.5 · pp. 22–23, Figure 10 · p. 23
• Systematic uncertainties include production-process weighting, photon-energy scale effects, and a larger 7 TeV photon-identification uncertainty caused by poorer data–simulation agreement.Weighting uncertainties range from several percent for associated production to almost 20% for gluon-gluon fusion, while correlated additional energy-scale uncertainty is 0.47%.Source: §5.5 · p. 23, Table 4 · p. 24
• The data-driven background is fitted separately in each class with a polynomial selected through pseudo-experiments to keep bias below one-fifth of the statistical uncertainty.Four truth-model families are tested using approximately 40 000 pseudo-data sets per model; the selected polynomial orders range from 3 to 5.Source: §5.5 · pp. 23, 25, 26
• At mH = 125 GeV, the simultaneous signal-plus-background fit gives σ/σSM = 1.56 ± 0.43 across the eleven event classes.The mass hypothesis is scanned from 110 to 150 GeV in 0.5 GeV steps using the 7 and 8 TeV data samples.Source: §5.5 · p. 26, Figure 11 · p. 27, Figure 12 · p. 28

5.6 Alternative analyses

• The alternative analyses replace the baseline multivariate categorization or parametric background model with cut-based classes or six sidebands surrounding a ±2% signal region.The sideband method derives background BDT distributions from data, excludes adjacent sidebands to avoid signal contamination, and assigns a 1% systematic uncertainty for fit-form bias.Source: §5.6 · pp. 26, 30, Figure 14 · p. 30

6.1 Event selection and kinematics

• The analysis improves sensitivity through better muon reconstruction, lepton identification and isolation, FSR-photon recovery, and a kinematic discriminant based on signal and background decay probabilities.The discriminant uses dilepton masses and angular variables through KD = Psig/(Psig + Pbkg).Source: §6.1 · pp. 32, 36
• Candidate events are reconstructed from two same-flavour, opposite-charge lepton pairs, with mass, transverse-momentum, isolation, impact-parameter, and four-lepton-mass requirements designed to retain high-efficiency signal events.Lowering lepton pT thresholds provides no expected-significance gain because the accompanying Z+jets background increases substantially.Source: §6.1 · pp. 32, 34
• At mH = 125 GeV, combined signal reconstruction and selection efficiencies are 18% for 4e, 40% for 4µ, and 27% for 2e2µ, with per-event mass resolutions of 2.2%, 1.1%, and 1.6%, respectively.The three subchannels are analyzed separately because their reducible background rates and mass resolutions differ, then combined statistically.Source: §6.1 · pp. 32, 34

6.2 Background estimation and systematic uncertainties

• The irreducible ZZ(Zγ*) background is estimated with NLO MC simulation, while reducible Z+X backgrounds are estimated from data control regions using lepton misidentification probabilities.The two complementary reducible-background methods agree within uncertainties, whose difference is included in the final estimate.Source: §6.2 · p. 36

6.3 Results

• The four-lepton result uses observed and expected yields in the 110–160 GeV mass range, including a signal region from 121.5 to 130.5 GeV centered on the most significant excess.The yield table combines statistical and systematic uncertainties and separates the 4e, 4µ, and 2e2µ final states.Source: Table 5 · p. 37, §6.3 · p. 37
• The observed four-lepton mass distribution shows an excess near 125 GeV above background, with events clustering where the kinematic discriminant is high and the signal expectation is strongest.The clustering occurs around m4ℓ = 125 GeV with KD ≥ 0.7, and is also visible for KD > 0.5.Source: §6.3 · pp. 37–38

7 H →WW

• The search selects two oppositely charged isolated high-pT leptons, missing transverse momentum above 20 GeV, and mutually exclusive 0-, 1-, and 2-jet categories separated by dilepton flavor.The Higgs mass cannot be reconstructed because of neutrinos, so sensitivity comes from kinematic distributions such as dilepton mass and lepton–missing-momentum variables.Source: §7 · p. 39, §7.1 · pp. 39, 42

7.1 WW event selection

• 1594, 1186, and 1295 events remain in the 0-jet, 1-jet, and 2-jet categories, respectively, after WW selection; Drell–Yan rejection reduces background by three orders of magnitude while retaining over half the signal.The selected sample is dominated by nonresonant WW events in the 0-jet category and has similar WW and top fractions in the other categories.Source: §7.1 · p. 42

7.2 H →WW

• The final search uses mass-dependent cut-based selections for the 0- and 1-jet categories, while the 2-jet category targets VBF through forward jets, a central-jet veto, and large dijet separation and mass.The VBF cross section is roughly ten times smaller than gluon-gluon fusion but provides an independent production mechanism and additional sensitivity.Source: §7.2 · pp. 44–45

7.3 Background predictions

• The largest backgrounds are estimated from data, including W+jets and QCD with a tight-loose method, top with tagged-event control samples, and low-mass nonresonant WW with a high-dilepton-mass control region.The quoted uncertainties are about 36% for the tight-loose estimate, 20% in the 0-jet and 5% in the 1-jet top estimate, and 10% for low-mass nonresonant WW.Source: §7.3 · p. 45

7.4 Efficiencies and systematic uncertainties

• Signal efficiencies include gluon-gluon fusion, VH, and VBF production, with data-to-simulation lepton scale factors typically 0.9–1.0 and about 2% uncertainty per lepton leg.Systematic effects include reconstruction, momentum and jet-energy modeling, pileup, background normalization, and a 4.4% integrated-luminosity uncertainty.Source: §7.4 · pp. 46–47
• About 20% is the overall signal-efficiency uncertainty, dominated by missing higher-order corrections and PDF uncertainties.Theoretical variations include PDFs, αs, higher-order corrections, parton showers, and underlying-event modeling; jet-category uncertainties range from 10% to 30%.Source: §7.4 · p. 47
• About 15% is the total background-estimation uncertainty in the H →WW signal region, dominated by statistical uncertainty in background-control-region event counts.Source: §7.4 · p. 48
• Upper limits on the Higgs boson production cross section are interpreted in Section 10.Source: §7.4 · p. 48

8 H →ττ

• The search studies eτh, µτh, eµ, and µµ channels, using mττ to separate H →ττ from backgrounds, especially irreducible Z →ττ.Events are reconstructed from visible τ-decay products and the missing transverse momentum vector.Source: §8 · p. 48

8.3 Event categories

• Events are divided into VBF, 1-jet/high-pT, 1-jet/low-pT, 0-jet/high-pT, and 0-jet/low-pT categories to enhance production-mode sensitivity and signal purity.VBF selections require two separated jets, while jet categories and τ-decay-product pT target different production mechanisms and signal-to-background ratios.Source: §8 · p. 48, §8.3 · pp. 52–53

8.1 Trigger and inclusive event selection

• Trigger and inclusive selections combine lepton and τh triggers with isolation, kinematic, transverse-mass, Dζ, and dimuon-BDT requirements to suppress backgrounds.τh identification achieves 60–65% efficiency with a 2–3% jet misidentification probability.Source: §8.1 · pp. 48–49

8.2 The ττ invariant-mass reconstruction

• SVFit reconstructs a physical mττ solution for every event, achieves about 20% resolution, and separates SM Higgs signal from Z →ττ better than mvis.The method combines visible τ-decay products with missing transverse momentum in a maximum-likelihood fit.Source: §8.2 · pp. 50–51, Figure 26 · p. 52

8.4 Background estimation and systematic uncertainties

• Backgrounds are constrained using embedding for Z →ττ, control regions for W+jets and tt, and simulation or control samples for other processes, with observed distributions generally agreeing with predictions.The W+jets extrapolation to the signal region carries a 30% systematic uncertainty, while ℓ→τh misidentification uncertainties reach 30% for electrons and 100% for muons.Source: §8.4 · pp. 53, 55
• Backgrounds are estimated using charge-control samples, relaxed lepton selections, and simulation, with a 10% uncertainty on the QCD rescaling factor.In the VBF category, QCD yields instead use an inclusive estimate and data-measured VBF efficiency; diboson and single-top backgrounds use simulation.Source: §8.4 · p. 56
• The dominant experimental signal-yield uncertainties are τh identification (7%), missing-transverse-momentum scale (5%), luminosity (5%), and jet energy scale below 4%.Additional uncertainties affect lepton efficiencies, production-process theory predictions, and the mττ shape through τh and electron energy scales.Source: §8.4 · p. 56

8.5 Results

• The search combines five categories across four final states and both collision energies as 40 independent channels in a binned likelihood based on mττ distributions.Normalization uncertainties use log-normal priors, while τh and electron energy-scale uncertainties continuously deform the mass-spectrum shape [147].Source: §8.5 · p. 56
• A background-only maximum-likelihood fit adjusts nuisance parameters to match the data, with high-mττ VBF events and Z →ττ-rich categories constraining key background and τh-related uncertainties.Observed mττ distributions in all channels and categories are compared with fitted background predictions; upper-limit interpretation is given in Section 10.Source: §8.5 · pp. 56–57

9 H →bb

• The analysis searches for H →bb in WH and ZH events with leptonic vector-boson decays, using boosted kinematics and control regions to suppress and model backgrounds.The included final states are W(µν)H, W(eν)H, Z(µµ)H, Z(ee)H, and Z(νν)H.Source: §9 · p. 57, §9.1 · p. 60

9.1 Event selection

• The event selection reconstructs the vector boson and two b-tagged jets, then separates low- and high-pT(V) ranges because their signal and background compositions differ.Additional requirements suppress QCD multijet, top-quark, and WZ backgrounds, while the selected dijet mass resolution is approximately 10%.Source: §9.1 · pp. 60–61
• BDT regression improves the bb invariant-mass resolution by approximately 15%, increasing analysis sensitivity by 10–20% depending on the channel.A separate mass-hypothesis BDT is fitted rather than using a simple event count, improving sensitivity by approximately 20%.Source: §9.1 · pp. 61–62

9.2 Background control regions

• Background control regions provide channel-specific scale factors for tt and V+jets processes through simultaneous fits, with fitted values consistent with the previous analysis.These scale factors correct simulation yields and incorporate statistical and systematic uncertainties from the control-region fits.Source: §9.2 · p. 63

9.3 Systematic uncertainties

• Systematic uncertainties include luminosity, lepton efficiencies, jet calibration, b tagging, theoretical production corrections, finite simulation samples, and background modeling.The total uncertainty in data-driven background-yield predictions is approximately 10%, while boosted VH modeling receives additional electroweak and QCD uncertainties.Source: §9.3 · pp. 63, 65

9.4 Results

• Maximum-likelihood fits use channel- and mass-specific BDT output distributions, allowing signal and background shapes and normalizations to vary with correlated nuisance parameters.The analysis fits hypotheses from 110–135 GeV using the systematic and statistical uncertainties described for the search.Source: §9.4 · p. 66
• The BDT discriminant analysis is more sensitive than a restrictive counting analysis based on the dijet invariant mass alone.The counting selection uses tighter b tagging, no additional jets, and a large azimuthal opening angle, but is not pursued further because it is significantly less sensitive.Source: §9.4 · p. 66

10.1 Combination methodology

• The combination simultaneously analyzes decay-mode data, correlations, and statistical and systematic uncertainties using asymptotic likelihood methods and mass steps matched to the Higgs width and detector resolution.Limits use the CLs criterion with a profile-likelihood test statistic, nuisance parameters for systematic uncertainties, and 95% exclusion when CLs(µ) ≤ 0.05.Source: §10.1 · p. 70, §10.1.1 · p. 70

10.1.2 Characterizing an excess of events: p-values and significance

• The local significance is obtained from the background-only likelihood-ratio statistic, while scanning mass hypotheses and correcting the minimum local p-value for the look-elsewhere effect yields global significance.Global p-values are evaluated with pseudo-experiments that model statistical and systematic variations using specified probability distributions.Source: §10.1.2 · pp. 70–71

10.1.3 Extracting signal-model parameters

• Signal-model parameters are estimated by scanning profile-likelihood ratios, with 68% (95%) one-dimensional intervals defined by q = 1 (3.84) and two-dimensional contours by q = 2.3 (6.0).Other unconstrained model parameters are treated as nuisance parameters when deriving each interval.Source: §10.1.3 · p. 71

10.2.1 Results of searches in the five decay modes

• The γγ search observed exclusions at 113–122, 128–133, and 138–149 GeV, while an excess near 125 GeV weakened limits across three consistent analyses.The ZZ →4ℓ search observed exclusions at 130–164 and 170–180 GeV, whereas WW, ττ, and bb showed excess or insufficient exclusion sensitivity.Source: §10.2.1 · pp. 72, 75, Figure 32 · p. 74

10.2.2 Combined results

• The combined search excludes an SM Higgs boson at 95% CL for 110–121.5 GeV and 128.0–145 GeV, while a 121.5–128.0 GeV excess prevents exclusion.At 99% CL, excluded ranges are 110.0–111.5, 113.5–121.0, and 128.5–145.0 GeV; the corresponding CLs and limit representations agree.Source: §10.2.2 · p. 75, Figure 33 · p. 76

10.3.1 Results of searches in the H →γγ and H →ZZ →4ℓ decay modes

• The γγ channel reaches 4.1σ local significance at 125.0 GeV, while ZZ →4ℓ reaches 3.2σ at 125.5 GeV using the two-dimensional fit.The γγ median expected significance is 2.7σ; the ZZ →4ℓ median expected significance is 3.8σ for the 2D fit and 3.2σ for the 1D fit.Source: §10.3.1 · p. 76, Figure 34 · p. 77, Figure 35 · p. 78

10.3.2 Combined results

• 5.0σ local significance occurs at mH = 125.5 GeV from the five-channel combination, with global significance 4.6σ (4.5σ) over 115–130 GeV (110–145 GeV).The largest contributions arise from γγ and ZZ →4ℓ; the combined excess implies a new boson near 125 GeV, whose diphoton decay excludes spin 1.Source: §10.3.2 · pp. 78, 80

10.4 Mass of the observed state

• 125.3 ± 0.4 (stat.) ± 0.5 (syst.) GeV is the measured mass of the observed state from the combined γγ and ZZ →4ℓ modes.The statistical and systematic uncertainties are determined from likelihood scans with all nuisance parameters fixed or profiled, respectively.Source: §10.4 · p. 82, Figure 39 · p. 82

10.5.1 Measurement of the signal strength

• 0.87 ± 0.23 is the measured signal-strength modifier at mH = 125.5 GeV, consistent with the SM expectation µ = 1.The combination uses all search channels and includes statistical and systematic uncertainties.Source: §10.5.1 · p. 83, Figure 40 · p. 83
• The signal-strength measurements are consistent across decay-mode and production-mechanism-enriched channel combinations.For γγ, WW, and ττ decay modes, the SM point (µggH+ttH, µVBF+VH) = (1, 1) lies within each 95% CL interval.Source: §10.5.1 · p. 83, Figure 42 · p. 85

10.5.2 Consistency of the data with the SM Higgs boson couplings

• The coupling consistency test fits scale factors κ because partial widths are proportional to squared effective Higgs couplings, while loop-induced Γgg and Γγγ remain sensitive to new physics.Possible BSM decays are accommodated through an additional partial width ΓBSM contributing to Γtot.Source: §10.5.2 · p. 85
• The data cannot measure all eight independent coupling parameters, so unmeasured parameters are fixed to SM expectations or treated as unconstrained nuisance parameters.The framework includes seven relevant partial widths and the total width.Source: §10.5.2 · p. 85

A. Test of custodial symmetry

• Custodial symmetry protects the tree-level relation between W/Z mass and Higgs-coupling ratios against large radiative corrections, while BSM theories can violate it.CMS introduces κW and κZ and tests λWZ = κW/κZ against unity.Source: A. Test of custodial symmetry · pp. 85–86
• RWZ = 0.9 +1.1−0.6 in the combined untagged H → WW and inclusive H → ZZ searches, with measurements consistent with custodial symmetry.The event-rate ratio is related to λWZ^2, with a small 0.02 bias from VBF and VH contributions.Source: A. Test of custodial symmetry · p. 86, Figure 43 · p. 86
• λWZ = 1.1+0.5−0.3 from all search channels, consistent with the custodial-symmetry expectation λWZ = 1.The fit assumes ΓBSM = 0 and treats κZ and κF as nuisance parameters; using all channels increases sensitivity.Source: A. Test of custodial symmetry · pp. 86–87

B. Test of the couplings to vector bosons and fermions

• The SM point (κV, κF) = (1, 1) lies within the 95% CL contour, although the global (+, −) minimum is driven by γγ interference and is not statistically distinct from (+, +).A negative relative sign makes W/top-loop interference positive and increases the H → γγ branching fraction.Source: B. Test of the couplings to vector… · p. 87
• A fermiophobic Higgs boson at (κV, κF) = (1, 0) lies just outside the 95% CL contour and is excluded for mH = 125.5 GeV.The contour comparison combines the individual decay modes and restricts the allowed sign quadrants in the displayed fits.Source: B. Test of the couplings to vector… · p. 87
• κV = 1.00 ± 0.13 and κF = 0.5 ± 0.2, with 95% CL intervals [0.7; 1.3] and [0.2; 1.0], respectively.Each one-dimensional scan fixes the other coupling modifier to its SM value of 1; uncertainties combine statistical and systematic contributions.Source: B. Test of the couplings to vector… · p. 87

C. Test for the presence of BSM particles

• The κγ–κg fit is consistent with the SM point (1, 1), with best-fit values (κγ, κg) = (1.5, 0.75).The fit assumes ΓBSM = 0 and leaves tree-level production and decay partial widths unaltered.Source: C. Test for the presence of BSM… · p. 87
• The likelihood contours compare κV and κF with the SM point (1, 1), while one-dimensional scans quantify their confidence intervals.The displayed likelihood test statistics use 68%, 95%, and 99.7% confidence contours or intervals.Source: Figure 44 · p. 88, Figure 45 · p. 88, Figure 46 · p. 89
• BRBSM < 0.89 at 95% CL when κg and κγ are included as unconstrained nuisance parameters.Here BRBSM is defined as ΓBSM/Γtot, and the tree-level partial widths are assumed unchanged.Source: C. Test for the presence of BSM… · pp. 87, 89

D. Test for differences in the couplings to fermions

• Both |λdu| and |λℓq| are consistent with 0 and 1, with a 95% CL upper limit of 1.5 for each ratio.The fits treat the relevant vector and fermion coupling modifiers as unconstrained nuisance parameters and assume ΓBSM = 0.Source: D. Test for differences in the… · p. 89
• The λdu likelihood is nearly left-right symmetric, while λℓq is perfectly left-right symmetric because no measured loop-induced process probes the lepton–quark relative sign.The small λdu asymmetry reflects the nonvanishing b-quark contribution to the gluon coupling.Source: D. Test for differences in the… · p. 89, Figure 49 · p. 91
• The best-fit values near zero mainly reflect the lack of an event excess in H → ττ, while H → ττ and H → bb lack sufficient sensitivity for strong coupling constraints.These channels had not reached sufficient sensitivity to constrain the corresponding fermion couplings strongly.Source: D. Test for differences in the… · p. 89

• The data were collected at √s = 7 and 8 TeV, with integrated luminosities up to 5.1 fb−1 and 5.3 fb−1, respectively.Source: §11 · p. 91
• 5.0σ local significance was observed across five decay modes, signaling production of a new particle near 125 GeV.The expected significance for a Standard Model Higgs boson of that mass was 5.8σ.Source: §11 · p. 91
• The observed excess appears in the γγ, ZZ, WW, ττ, and bb decay modes.Source: §11 · p. 91
• 125.3± 0.4 (stat.)± 0.5 (syst.) GeV is the mass obtained by fitting the γγ and ZZ →4ℓ invariant-mass peaks.These two channels provide the best mass resolution.Source: §11 · p. 91
• The measurements of couplings to vector bosons, fermions, gluons, and photons are consistent within uncertainties with Standard Model Higgs expectations.More data are needed to determine whether the new state implies physics beyond the Standard Model.Source: §11 · p. 91

Yerevan Physics Institute, Yerevan, Armenia

• The roster includes CMS contributors affiliated with institutions in Armenia, Belarus, Belgium, Brazil, Bulgaria, China, Colombia, Croatia, Cyprus, and the Czech Republic.Examples include the Yerevan Physics Institute, Belgian universities, Brazilian research centers, and institutes in Sofia and Beijing.Source: Yerevan Physics Institute, Yerevan,… · p. 105, National Centre for Particle and High… · p. 105, Universiteit Antwerpen, Antwerpen,… · p. 105

DSM/IRFU, CEA/Saclay, Gif-sur-Yvette, France

• The affiliations continue across Estonia, Finland, France, Germany, Greece, Hungary, India, Ireland, and Italy, including universities, national laboratories, and INFN sections.The listed institutions include Helsinki and Aachen universities, French laboratories, German institutes, Indian centers, University College Dublin, and multiple Italian INFN sites.Source: National Institute of Chemical Physics… · p. 106, Helsinki Institute of Physics,… · p. 106, DSM/IRFU, CEA/Saclay, Gif-sur-Yvette,… · p. 106

Kyungpook National University, Daegu, Korea

• Additional affiliations represent CMS contributors in Korea, Lithuania, Mexico, New Zealand, Pakistan, Poland, Portugal, and Russia.The roster names institutions including Korean universities, Mexican research centers, the University of Canterbury, Pakistani and Polish institutes, and Russian nuclear-physics facilities.Source: Kyungpook National University, Daegu,… · p. 110, Vilnius University, Vilnius, Lithuania · p. 110, Centro de Investigacion y de Estudios… · p. 110

Institute for Nuclear Research, Moscow, Russia

• The Institute for Nuclear Research, Moscow, Russia, is represented by ten listed contributors.The names are Yu. Andreev, A. Dermenev, S. Gninenko, N. Golubev, M. Kirsanov, N. Krasnikov, V. Matveev, A. Pashenkov, D. Tlisov, and A. Toropin.Source: Institute for Nuclear Research,… · p. 111

Institute for Theoretical and Experimental Physics, Moscow, Russia

• The Institute for Theoretical and Experimental Physics, Moscow, Russia, is represented by eleven listed contributors.The names are V. Epshteyn, M. Erofeeva, V. Gavrilov, N. Lychkovskaya, V. Popov, G. Safronov, S. Semenov, A. Spiridonov, V. Stolin, E. Vlasov, and A. Zhokin.Source: Institute for Theoretical and… · p. 111

• The listed contributors include V. Andreev, M. Azarkin, I. Dremin, M. Kirakosyan, A. Leonidov, G. Mesyats, S.V. Rusakov, and A. Vinogradov.Source: P.N. Lebedev Physical Institute,… · p. 111
• The Skobeltsyn Institute of Nuclear Physics is identified as part of Lomonosov Moscow State University in Moscow, Russia.Source: P.N. Lebedev Physical Institute,… · p. 111

State Research Center of Russian Federation, Institute for High Energy Physics, Protvino, Russia

• Contributors are listed from the State Research Center of the Russian Federation, Institute for High Energy Physics, Protvino, Russia.Source: State Research Center of Russian… · p. 111

Universidad Aut´onoma de Madrid, Madrid, Spain

• Spanish contributors are affiliated with universities and institutes in Madrid, Oviedo, and Santander.Source: State Research Center of Russian… · p. 112, Universidad Aut´onoma de Madrid,… · p. 112, Universidad de Oviedo, Oviedo, Spain · p. 112

CERN, European Organization for Nuclear Research, Geneva, Switzerland

• European institutions represented include CERN, Paul Scherrer Institut, and ETH Zurich’s Institute for Particle Physics.Source: CERN, European Organization for… · p. 112, Paul Scherrer Institut, Villigen,… · p. 112, Institute for Particle Physics, ETH… · p. 112

National Central University, Chung-Li, Taiwan

• The collaboration includes institutions in Taiwan, Thailand, Turkey, and the United Kingdom.Source: National Central University, Chung-Li,… · p. 113, National Taiwan University (NTU),… · p. 113, Chulalongkorn University, Bangkok,… · p. 113

Imperial College, London, United Kingdom

• Numerous universities and laboratories across the United States contribute to the collaboration, including institutions in California, Massachusetts, Illinois, and New York.Source: Imperial College, London, United Kingdom · p. 113, Boston University, Boston, USA · p. 114, Fermi National Accelerator Laboratory,… · p. 115

University of Wisconsin, Madison, USA

• The affiliation list extends to institutions in Texas, Tennessee, Virginia, and Wisconsin, with additional cross-institution appointments recorded in the footnotes.Source: University of Tennessee, Knoxville, USA · p. 117, University of Virginia,… · p. 117, University of Wisconsin, Madison, USA · pp. 117–118