Constraining mathcal{CP} -violation in the Higgs-top-quark interaction using machine-learning-based inference

Henning Bahl,Simon Brass

doi:10.1007/jhep03(2022)017

Abstract

While mathcal{CP} violation in the Higgs interactions with massive vector boson is already tightly constrained, the mathcal{CP} nature of the Higgs interactions with fermions is far less constrained. In this work, we assess the potential of machine-learning-based inference methods to constrain mathcal{CP} violation in the Higgs top-Yukawa coupling. This approach enables the use of the full available kinematic information. Concentrating on top-associated Higgs production with the Higgs decaying to two photons, we derive expected exclusion bounds for the LHC and the high-luminosity phase of the LHC. We also study the dependence of these bounds on the Higgs interaction with massive vector bosons and their robustness against theoretical uncertainties. In addition to deriving expected exclusion bounds, we discuss at which level a non-zero mathcal{CP} -violating top-Yukawa coupling can be distinguished from the SM. Moreover, we analyze which kinematic distributions are most sensitive to a mathcal{CP} -violating top-Yukawa coupling.

Highlights

Electric dipole moments (EDMs) are prominent examples for CP-odd observables
Concentrating on top-associated Higgs production with the Higgs decaying to two photons, we derive expected exclusion bounds for the LHC and the high-luminosity phase of the LHC
A detailed investigation of the top-Yukawa coupling is crucial to probe the CP nature of the Higgs boson discovered at the LHC

Summary

Introduction

Electric dipole moments (EDMs) are prominent examples for CP-odd observables. While recent EDM measurements (see e.g. [7, 8]) severely constrain CP violation in the Higgs sector [9–12], different contributions to the electric dipole moments — originating e.g. from CP violation in more than one Higgs coupling — can cancel each other [13–15]. Searching for CP violation in the Higgs sector at colliders is an important complementary approach, especially since collider studies allow to disentangle the different couplings. Even though CP-odd variables for top-associated Higgs production have been proposed [48– 51], the current experimental studies [23–26] — targeting the Higgs decay to two photons — exploit the effects of the CP-odd Yukawa coupling on the kinematic distributions — effectively mixing CP-even and CP-odd observables. The main goal of the present paper to evaluate the potential of machine-learningbased inference methods to probe a CP-odd top-Yukawa coupling. It uses machine learning to approximate the full likelihood fully taking into account parton shower as well as detector effects, which are only approximated in similar approaches like the matrix element method (see e.g. [36, 57–69]) or the optimal observable approach [70–72]

Methods

Results

Conclusion