Testing the Higgs Boson Coupling to Gluons

TL;DR

The paper investigates whether the Higgs coupling to gluons is loop-induced or point-like by analyzing high-pT Higgs production in the H→γγ channel. It uses a detailed simulation and a two-dimensional fit to the diphoton invariant mass and pT to separate the loop-induced and point-like hypotheses, incorporating detector effects and background systematics. The analysis finds that a ~2σ separation is attainable with about 500 fb^-1, improving to ~3σ at 1000 fb^-1, but both experimental background and theoretical top-mass uncertainties significantly limit sensitivity. Achieving stronger discrimination will require higher-order calculations that incorporate full top-mass effects in the pT spectrum and better pile-up mitigation strategies for the HL-LHC.

Abstract

We study the possibility to separate in gluon fusion loop-induced Higgs boson production from point-like production. The Higgs boson is reconstructed in the Hgg final state at very large transverse momentum. Using the Higgs boson yields (normalized to the overall rate) and the shape of the Higgs boson pt distribution the two hypotheses can be separated with 2 standard deviations with an integrated luminosity of about 500 fb^-1. The largest experimental uncertainty affecting this estimate is the background event yield. The theoretical uncertainties from missing top mass effects are large, but can be decreased with dedicated calculations.

Figures (6)

• Figure 1: Gluon fusion $gg\rightarrow\xspace H$ at leading order mediated by top and bottom triangle loops.
• Figure 2: Generic diagrams for Higgs production in association with a jet via gluon fusion at leading order mediated by top and bottom triangle loops generated by $gg, gq, q\bar{q}$ initial states.
• Figure 3: Differential production cross section as generated (left) and reconstructed after all analysis requirements (right). The effective cross section includes ${\cal B}(H \rightarrow\xspace \gamma\gamma)$ for the Higgs boson cross sections (left and right), and the analysis efficiency for all cross sections (right). The error bars show the statistical uncertainty only.
• Figure 4: Average expected event yields vs diphoton invariant mass (left) and diphoton $p_\perp$ (right), normalized to 1000$\hbox{,fb}^{-1}$. The histograms contain the background plus the point-like Higgs production (open triangles) and the background plus the loop-induced Higgs production (open circles). The background component is identical for both hypotheses. The error bars show the statistical uncertainty only.
• Figure 5: Example toy data set with overlayed projections of the extended unbinned maximum likelihood fits for the diphoton invariant mass (left) and the diphoton $p_\perp$ (right). The $H_0$ hypothesis (loop-induced Higgs boson production) is shown in the top row, the $H_1$ hypothesis (point-like Higgs boson production) is shown in the bottom row. The background component is identical for both hypotheses. The (black) solid curves shows the signal+background fit, the (red) dashed line shows the background component. The error bars show the statistical uncertainty only.