Constraints on Higgs boson properties using $WW^{*}(\rightarrow e\nu\mu\nu) jj$ production in 36.1 fb$^{-1}$ of $\sqrt{s}=13$ TeV $pp$ collisions with the ATLAS detector
This article presents the results of two studies of Higgs boson properties using the $WW^*(\rightarrow e\nu\mu\nu)jj$ final state, based on a dataset corresponding to 36.1 fb$^{-1}$ of $\sqrt{s}=13$ TeV proton-proton collisions recorded by the ATLAS experiment at the Large Hadron Collider. The first study targets Higgs boson production via gluon-gluon fusion and constrains the CP properties of the effective Higgs-gluon interaction. Using angular distributions and the overall rate, a value of $\tan(\alpha) = 0.0 \pm 0.4 (\mathrm{stat.}) \pm 0.3 (\mathrm{syst.})$ is obtained for the tangent of the mixing angle for CP-even and CP-odd contributions. The second study exploits the vector-boson fusion production mechanism to probe the Higgs boson couplings to longitudinally and transversely polarised $W$ and $Z$ bosons in both the production and the decay of the Higgs boson; these couplings have not been directly constrained previously. The polarisation-dependent coupling-strength scale factors are defined as the ratios of the measured polarisation-dependent coupling strengths to those predicted by the Standard Model, and are determined using rate and kinematic information to be $a_\mathrm{L}=0.91^{+0.10}_{-0.18}$(stat.)$^{+0.09}_{-0.17}$(syst.) and $a_{\mathrm{T}}=1.2 \pm 0.4 $(stat.)$ ^{+0.2}_{-0.3} $(syst.). These coupling strengths are translated into pseudo-observables, resulting in $\kappa_{VV}= 0.91^{+0.10}_{-0.18}$(stat.)$^{+0.09}_{-0.17}$(syst.) and $\epsilon_{VV} =0.13^{+0.28}_{-0.20}$ (stat.)$^{+0.08}_{-0.10}$(syst.). All results are consistent with the Standard Model predictions.
28 September 2021
Table 01
Definition of the three benchmark scenarios used in the ggF + 2 jets analysis. The parameter settings correspond to a CP-even (i.e. the SM hypothesis), a CP-odd, and a CP-mixed scenario.
png (8kB)
pdf (42kB)
Table 02
Overview of the simulation tools used to generate signal and background processes, and to model the underlying event and parton shower. The PDF sets are also summarised. The perturbative accuracy (in QCD and if relevant in EW corrections) of the total cross section is stated for each process. The configuration marked with
(*) is used for the signal samples in the studies of the Higgs boson couplings to longitudinally and transversely polarised W and Z bosons, while the configurations marked with
(**) are used as background in the ggF + 2 jets analysis. Alternative event generators and configurations used to estimate systematic uncertainties are shown in parentheses. Further information about the alternative event generators and their configurations is given in Section 7, where MG5 is again used as an abbreviation for MadGraph5.
png (423kB)
pdf (68kB)
Table 03
Event selection criteria used to define the signal regions for the ggF + 2 jets and VBF event categories.
png (32kB)
pdf (69kB)
Table 04
Event selection criteria used to define the various control regions for the ggF + 2 jets and VBF event categories. Only the changes relative to the signal region definitions are stated.
png (16kB)
pdf (57kB)
Table 05
Post-fit NFs and their uncertainties for the Z + jets, top and WW backgrounds. Both sets of normalisation factors differ slightly depending on which (B)SM model is tested, but are consistent within their total uncertainties.
png (8kB)
pdf (45kB)
Table 06
Post-fit event yields in the signal and control regions obtained from the study of the signal strength parameter μ
ggF + 2 jets. The quoted uncertainties include those from theoretical and experimental systematic sources and those due to sample statistics. The fit constrains the total expected yield to the observed yield.
png (22kB)
pdf (33kB)
Table 07
Breakdown of the main contributions to the total uncertainty in tan(α) based on the fit that exploits both shape and rate information. Individual sources of systematic uncertainty are grouped into either the theoretical or the experimental uncertainty. The sum in quadrature of the individual components differs from the total uncertainty due to correlations between the components.
png (25kB)
pdf (35kB)
Table 08
Post-fit event yields in the signal and control regions obtained from a scan over ϵ
VV exploiting both shape and rate information. The quoted uncertainties include those from theoretical and experimental systematic sources and those due to sample statistics. The fit constrains the total expected yields to the observed yields.
png (22kB)
pdf (34kB)
Table 09
Best-fit values and their uncertainties as obtained from the shape-only and shape-plus-rate likelihood fits to the Asimov dataset and to ATLAS data. Results of both the shape-only and shape+rate fits for a
L and a
T are shown. Results of fits to one parameter with the other one fixed or profiled are presented.
png (30kB)
pdf (54kB)
Table 10
Best-fit values and their uncertainties as obtained from the shape-only and shape-plus-rate likelihood fits to the Asimov dataset and to ATLAS data. Results of both shape-only and shape+rate fits for ϵ
VV and κ
VV are shown. Results of fits to one parameter with the other one fixed or profiled are presented.
png (31kB)
pdf (47kB)
Table 11a
The contributions of the leading individual systematic uncertainties together with the data statistical uncertainties, in the one-dimensional fit for the pseudo-observables κ
VV (a) and ϵ
VV (b) for electroweak-boson polarisation in the VBF H→ WW channel. Both the shape and rate information is exploited in the fit. The theoretical and experimental uncertainties are subdivided further into categories.
png (26kB)
pdf (43kB)
Table 11b
The contributions of the leading individual systematic uncertainties together with the data statistical uncertainties, in the one-dimensional fit for the pseudo-observables κ
VV (a) and ϵ
VV (b) for electroweak-boson polarisation in the VBF H→ WW channel. Both the shape and rate information is exploited in the fit. The theoretical and experimental uncertainties are subdivided further into categories.
png (26kB)
pdf (43kB)
Figure 01
Shape comparison of the BDT response w
BDT distribution between H(→ WW
* → eνμν) + 2 jets events and the sum of backgrounds. The distribution for 𝐻 + 2 jets represents the sum of events produced via the ggF +2 jets and VBF production modes. Both distributions are normalised to unit area.
png (37kB)
pdf (15kB)
Figure 02
The ΔΦ
jj distribution in the top control region of the ggF +2 jets analysis, with signal and background yields fixed from the fit to tan(α) using shape and rate information. The shaded areas depict the total uncertainty.
png (44kB)
pdf (16kB)
Figure 03
Weighted w
BDT post-fit distribution in the ggF +2 jets signal region, with signal and background yields fixed from the fit to μ
ggF+2 jets. Data-to-simulation ratios are shown at the bottom of the plot. The shaded areas depict the total uncertainty.
png (101kB)
pdf (17kB)
Figure 04
68% and 95% CL two-dimensional likelihood contours of the coupling strength parameters μ
ggF+2 jets and μ
VBF for a fit to the ATLAS data within the ggF + 2 jets SR. The best-fit point is represented as a black star, while the SM prediction is represented by a red star. The signal strength parameters of the ggF + 2 jets and VBF processes are determined to be μ
ggF + 2 jets=0.7 ± 0.4 (stat.)
+0.8-0.7 (syst.) and μ
VBF= 0.3 ± 0.4 (stat.) ± 0.4 (syst.) respectively, with a linear correlation coefficient of -0.34.
png (159kB)
pdf (51kB)
Figure 05a
The signed ΔΦ
jj distribution in the signal region of the ggF analysis for the CP-even hypothesis and a CP-mixed hypothesis corresponding to tan(α) = 1.9, i.e. the lowest tan(α) value excluded at the 95% CL via the shape + rate fit on an Asimov dataset. Distributions are shown separately for the inclusive |η| range (a) and for |η| > 3.0 (b).
png (48kB)
pdf (15kB)
Figure 05b
The signed ΔΦ
jj distribution in the signal region of the ggF analysis for the CP-even hypothesis and a CP-mixed hypothesis corresponding to tan(α) = 1.9, i.e. the lowest tan(α) value excluded at the 95% CL via the shape + rate fit on an Asimov dataset. Distributions are shown separately for the inclusive |η| range (a) and for |η| > 3.0 (b).
png (53kB)
pdf (15kB)
Figure 06
Expected and observed likelihood curves for a scan over the CP-mixing angle α when both shape and normalisation are used in the fit.
png (102kB)
pdf (15kB)
Figure 07
The signed ΔΦ
jj distribution in the signal region of the VBF analysis, for the SM hypothesis and two BSM hypotheses corresponding to (a
L, a
T) = (1.2, 1.0) and (a
L, a
T) = (1.0, 1.8), i.e. the lowest a
L and a
T values excluded at the 95% CL via the shape + rate fits on a
L and a
T using an Asimov dataset.
png (48kB)
pdf (14kB)
Figure 08
The ΔΦ
jj distribution in the top control region of the VBF analysis, with signal and background yields fixed from the fit for a
L using shape and rate information. The shaded areas depict the total uncertainty.
png (41kB)
pdf (16kB)
Figure 09a
Likelihood scans over κ
VV (a) and ϵ
VV (b). The fits are performed using both shape + rate information. All relevant experimental and theoretical systematics are considered in the fit.
png (40kB)
pdf (14kB)
Figure 09b
Likelihood scans over κ
VV (a) and ϵ
VV (b). The fits are performed using both shape + rate information. All relevant experimental and theoretical systematics are considered in the fit.
png (86kB)
pdf (14kB)
Figure 10a
Generator-level comparison of the functions Δ
L(q
1,q
2) and Δ
T(q
1,q
2) for the incoming (left) and outgoing (right) bosons’ momenta q
1 and q
2.
png (51kB)
pdf (17kB)
Figure 10b
Generator-level comparison of the functions Δ
L(q
1,q
2) and Δ
T(q
1,q
2) for the incoming (left) and outgoing (right) bosons’ momenta q
1 and q
2.
png (47kB)
pdf (16kB)