VH production

Chapter 5: First evidence of a Higgs boson decay to a pair of muons

5.7.2 VH production

Events considered in the V𝐻 category contain at least two muons passing the selection requirements listed in Sec.5.3and Sec.5.5. The VH category is required to be orthogonal to all other categories in the analysis. Events are also required to have at least one additional lepton (electron or muon), which is expected from the leptonic decay of the W or Z boson. Electrons and muons are required to pass

0 10 20 30 40 50 60 70 80 103

×

Events / GeV

(13 TeV) 137 fb-1 CMS Preliminary

ggH-cat1 = 125.38 GeV mH

Data S+B fit Bkg. component

σ

± 1 σ

± 2

110 115 120 125 130 135 140 145 150 (GeV) µ mµ

−400200

− 0 200 400

Data-Bkg.

0 10 20 30 40 50 103

×

Events / GeV

(13 TeV) 137 fb-1 CMS Preliminary

ggH-cat2 = 125.38 GeV mH

Data S+B fit Bkg. component

σ

± 1 σ

± 2

110 115 120 125 130 135 140 145 150 (GeV) µ mµ

−400200

− 0 200 400

Data-Bkg.

0 2 4 6 8 10 12 14 16 103

×

Events / GeV

(13 TeV) 137 fb-1 CMS Preliminary

ggH-cat3 = 125.38 GeV mH

Data S+B fit Bkg. component

σ

± 1 σ

± 2

110 115 120 125 130 135 140 145 150 (GeV) µ mµ

−200 0 200

Data-Bkg.

0 2 4 6 8 10 103

×

Events / GeV

(13 TeV) 137 fb-1 CMS Preliminary

ggH-cat4 = 125.38 GeV mH

Data S+B fit Bkg. component

σ

± 1 σ

± 2

110 115 120 125 130 135 140 145 150 (GeV) µ mµ

−200 0 200

Data-Bkg.

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 103

×

Events / GeV

(13 TeV) 137 fb-1 CMS Preliminary

ggH-cat5 = 125.38 GeV mH

Data S+B fit Bkg. component

σ

± 1 σ

± 2

110 115 120 125 130 135 140 145 150 (GeV) µ mµ

−−10050 0 50 100

Data-Bkg.

Figure 5.35: Comparison between the observed data and the total background extracted from a signal-plus-background fit performed across the𝑔𝑔 𝐻 categories.

First row, from left to right: 𝑔𝑔 𝐻-cat1,𝑔𝑔 𝐻-cat2, and𝑔𝑔 𝐻-cat3. Second row, from left to right: 𝑔𝑔 𝐻-cat4 and 𝑔𝑔 𝐻-cat5. The one (green) and two (yellow) standard deviation bands include the uncertainties in the background component of the fit.

The lower panel shows the residuals after background subtraction and the red line indicates the signal with𝑀_𝐻 =125.38 GeV extracted from the fit.

the medium WP of a multivariate discriminant developed to identify and suppress non-prompt leptons [164], with a selection efficiency of about 90 (85)% per prompt muon (electron).

Events containing exactly one additional lepton belong to the W𝐻 category. If the additional lepton is a muon, the two pairs of oppositely charged muons are required to have 𝑚_{𝜇 𝜇} >12 GeV to suppress background events from quarkonium decays.

Moreover, neither of the two oppositely charged muon pairs can have an invariant mass consistent with𝑚_𝑍 within 10 GeV. Finally, at least one of these two muon pairs must have 𝑚_{𝜇 𝜇} in the range 110–150 GeV. If both𝑚_{𝜇 𝜇} pairs satisfy this criterion, the highest-𝑝

T pair is considered as the Higgs boson candidate. If the additional lepton is an electron, the only requirement imposed is that 110 < 𝑚_{𝜇 𝜇} < 150 GeV.

The𝑍 𝐻 category targets signal events where the Higgs boson is produced in asso- ciation with a Z boson that decays to a pair of electrons or muons. Events in the

𝑍 𝐻category are therefore required to contain four leptons, with a combined lepton number and electric charge of zero. The invariant mass of each pair of same-flavour opposite-charge leptons is required to be greater than 12 GeV. An event is rejected if it does not contain exactly one pair of same-flavour and oppositely charged lep- tons with invariant mass compatible with the Z boson within 10 (20) GeV for muon (electron) pairs. In addition, each event must contain one oppositely charged muon pair satisfying 110< 𝑚_{𝜇 𝜇} < 150 GeV. For events with four muons, the muon pair with 𝑚_{𝜇 𝜇} closer to𝑚_𝑍 is chosen as the Z boson candidate, while the other muon pair is selected as the Higgs boson candidate. A summary of the selection criteria applied in the W𝐻and𝑍 𝐻production categories is reported in Table5.5.

Selection W𝐻 leptonic 𝑍 𝐻leptonic

𝜇 𝜇 𝜇 𝜇 𝜇e 4𝜇 2𝜇2e Number of loose (medium)𝑏-tagged jets ≤ 1(0) ≤ 1 (0) ≤ 1(0) ≤ 1(0)

N(𝜇) passing id.+iso. 3 2 4 2

N(e) passing id.+iso. 0 1 0 2

Lepton charge Í

𝑞(ℓ) =±1 Í

𝑞(ℓ) =0

Low mass resonance veto 𝑚_ℓℓ > 12 GeV

N(𝜇⁺𝜇⁻)pairs with 110< 𝑚_{𝜇 𝜇} <150 GeV ≥ 1 1 ≥ 1 1 N(𝜇⁺𝜇⁻)pairs with|𝑚_{𝜇 𝜇}−𝑚_𝑍|< 10 GeV| 0 0 1 0 N(e⁺e⁻)pairs with|𝑚

ee−𝑚_𝑍|< 20 GeV| 0 0 0 1

Table 5.5: Summary of the kinematic selection used to define the W𝐻 and 𝑍 𝐻 production categories.

The main backgrounds of the WH category are the WZ (off-shell Z boson decay), ZZ (one lepton is not reconstructed) and the DY process (with associated lepton production). The main backgrounds of the ZH category are the ZZ and ggZZ processes.

5.7.2.1 Multivariate discriminator

Two separate BDT discriminants are trained to discriminate between signal and background events in the W𝐻 and𝑍 𝐻categories. The BDT input variables are not significantly correlated with the𝑚_{𝜇 𝜇}of the Higgs boson candidate. During the BDT training, weights are applied to the signal events that are inversely proportional to the per-event uncertainty on the measured𝑚_{𝜇 𝜇}, as described in Section5.7.1.

The input variables to the W𝐻category BDT are :

• 𝑝

T𝜇 𝜇 of the Higgs boson candidate, the𝜂_𝜇 of the two muons, and the angular separationΔ𝑅_{𝜇 𝜇} between them.

• the flavour and the𝑝

Tof the additional leptonℓ

W.

• Δ𝜂(𝜇 𝜇, ℓ

W),Δ𝜙(𝜇 𝜇, ℓ

W),Δ𝜂(𝜇

1, ℓ

W)andΔ𝜙(𝜇

2, ℓ

W)

• The𝐻®^miss

T is defined as the negative vector sum of the𝑝

Tof all jets in the event with 𝑝

T > 30 GeV and|𝜂|< 4.7. The transverse mass and angular distances in𝜙and𝜂of the combinedℓ

Wand𝐻®^miss

T system are also considered.

The input variables to the𝑍 𝐻category BDT are :

• the mass 𝑚_{𝑙 𝑙}, 𝑝

T(𝑙 𝑙) , the 𝜂_{𝑙 𝑙} and the Δ𝑅(𝑙 𝑙) of the two leptons from the Z boson candidate.

• Δ𝜂(𝜇 𝜇, 𝑙 𝑙) and cos𝜃_{𝐶 𝑆}(𝜇 𝜇, 𝑙 𝑙)

• the flavour of the lepton pair associated to the Z boson decay

Figure5.36shows the output of the BDT classifiers in the W𝐻(left) and𝑍 𝐻(right) categories. Based on these outputs, events in the W𝐻 category are further divided into three subcategories termed W𝐻-cat1, W𝐻-cat2, and W𝐻-cat3. Similarly, events in the 𝑍 𝐻 category are divided into two subcategories, labelled 𝑍 𝐻-cat1 and 𝑍 𝐻-cat2. The boundaries of these categories, defined in terms of the BDT discriminant and indicated in Fig.5.36by black dashed vertical lines, are chosen via an optimization strategy analogous to that described in Section5.7.1 for the𝑔𝑔 𝐻 category. In this category, the BWZ function (Eqn. 5.11) is used to estimate the total background instead of the mBW (Eqn.5.10).

5.7.2.2 Signal extraction

The systematic uncertainties considered in this analysis are similar to the ggH category (Section 5.7.1). Figure 5.37show the 𝑚_{𝜇 𝜇} distributions in the W𝐻 (first row) and 𝑍 𝐻 (second row) event categories. The signal is extracted via a binned maximum-likelihood fit in each event category, where the signal is modelled with a DCB function and the background is modelled with the BWZGamma function in W𝐻-cat1, as defined in Eqn.(5.12) and the BWZ function in the remaining categories, as defined in Eqn.(5.11). A bias test on the choice of the background

−2 10

−1 10

1 10 102 103 104 105

Events

Data WZ

ZZ DY

Top quark Other bkg.

WH ZH

Other sig.

(13 TeV) 137 fb-1

CMS Preliminary

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1

WH BDT output

0.5 1 1.5

Data/Pred.

−2 10

−1 10

1 10 102 103

Events

Data qq → ZZ

→ ZZ

gg DY

Other bkg. qq → ZH

→ ZH gg

(13 TeV) 137 fb-1

CMS Preliminary

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1

ZH BDT output

0 0.5 1 1.5 2

Data/Pred.

Figure 5.36: The observed BDT output distribution in the W𝐻 (left) and 𝑍 𝐻 (right) categories compared to the prediction from the simulation of various SM background processes. Signal distributions expected from different production modes of the 125 GeV Higgs boson are overlaid. The description of the ratio panel is the same as in Fig.5.34. The dashed vertical lines indicate the boundaries of the optimized event categories.

modelling function is performed similar to Section 5.7.1.2 and is observed to be small and is therefore neglected in the signal extraction. Finally, Table5.6 reports the signal composition in the W𝐻and 𝑍 𝐻categories, along with the hwhm of the expected signal shape. In addition, the estimated number of background events, the S/(S+B) and S/√

B ratios, and the observation in data within the hwhm of the signal peak are also listed.

Category Sig. W𝐻 𝑞 𝑞 𝑍 𝐻 𝑔𝑔 𝑍 𝐻 𝑡 𝑡 𝐻+t𝐻 hwhm Bkg. S/(S+B)(%) S/√

B Data

(%) (%) (%) (%) (GeV) in hwhm in hwhm in hwhm in hwhm

W𝐻-cat1 0.82 76.2 9.6 1.6 12.6 2.00 32.0 1.54 0.09 34

W𝐻-cat2 1.72 80.1 9.1 1.5 9.3 1.80 23.1 4.50 0.23 27

W𝐻-cat3 1.14 85.7 6.7 1.8 4.8 1.90 5.48 12.6 0.35 4

𝑍 𝐻-cat1 0.11 — 82.8 17.2 — 2.07 2.05 3.29 0.05 4

𝑍 𝐻-cat2 0.31 — 79.6 20.4 — 1.80 2.19 8.98 0.14 4

Table 5.6: The total expected number of signal events with 𝑚_𝐻 =125.38 GeV, the hwhm of the signal peak, the estimated number of background events and the observed number of events within±hwhm, and the S/(S+B)and the S/√

B ratios computed within the hwhm of the signal peak for each of the optimized event categories defined along the W𝐻and𝑍 𝐻 BDT outputs.