Resurrecting $$ b\overline{b}h $$ with kinematic shapes

Christophe Grojean, Ayan Paul, Z. QianJournal of High Energy Physics 2021 · DOI

Abstract

The associated production of a $b\bar{b}$ pair with a Higgs boson could provide an important probe to both the size and the phase of the bottom-quark Yukawa coupling, $y_b$. However, the signal is shrouded by several background processes including the irreducible $Zh, Z \to b\bar{b}$ background. We show that the analysis of kinematic shapes provides us with a concrete prescription for separating the $y_b$-sensitive production modes from both the irreducible and the QCD-QED backgrounds using the $b\bar{b}\gamma\gamma$ final state. We draw a page from game theory and use Shapley values to make Boosted Decision Trees interpretable in terms of kinematic measurables and provide physics insights into the variances in the kinematic shapes of the different channels that help us complete this feat. Adding interpretability to the machine learning algorithm opens up the black-box and allows us to cherry-pick only those kinematic variables that matter most in the analysis. We resurrect the hope of constraining the size and, possibly, the phase of $y_b$ using kinematic shape studies of $b\bar{b}h$ production with the full HL-LHC data and at FCC-hh.

Introduction

In the Standard Model of particle physics, the Higgs boson's interaction with fermions is governed by Yukawa couplings. While the coupling to the bottom quark ($y_b$) has been observed through Higgs decays ($h \to b\bar{b}$), measuring it via associated production ($b\bar{b}h$) offers a unique and complementary probe. Specifically, $b\bar{b}h$ production is sensitive not just to the magnitude of the coupling, but also to its CP-violating phase—a key ingredient in understanding the matter-antimatter asymmetry of the universe.

Despite its potential, the $b\bar{b}h$ channel has long been considered nearly impossible to isolate at the Large Hadron Collider (LHC). The signal is extremely faint compared to the massive backgrounds that mimic its signature, leading many to label it a "lost cause."

The Challenge of Backgrounds

The primary obstacle in identifying $b\bar{b}h$ production is the presence of several high-rate background processes that share the same final state. These include:

  1. Irreducible $Zh$ Background: Where a $Z$ boson is produced in association with a Higgs, and the $Z$ subsequently decays into a $b\bar{b}$ pair.
  2. Top-quark Induced Processes: Contributions where the Higgs is produced via top-quark loops or boxes ($y_t$), which do not depend on the $y_b$ coupling.
  3. QCD-QED Backgrounds: Large-scale production of $b\bar{b}\gamma\gamma$ through standard electromagnetic and strong force interactions.
Feynman diagrams for signal ($y_b$), top-quark induced backgrounds ($y_t$), and the irreducible $Zh$ background.

As shown in the diagrams above, the final states for these processes are often identical. Furthermore, simple one-dimensional kinematic distributions, such as the transverse momentum ($p_T$) of the jets, show significant overlap between the signal and the $Zh$ background, making traditional cut-based analyses ineffective.

Methodology: Beyond Black-Box Machine Learning

To overcome these challenges, this study utilizes advanced multivariate analysis (MVA) techniques. The researchers simulated events for the $b\bar{b}\gamma\gamma$ final state (where the Higgs decays into two photons) for both the High-Luminosity LHC (HL-LHC) and the Future Circular Collider (FCC-hh).

They employed Boosted Decision Trees (BDT) and Deep Neural Networks (DNN) to classify events. However, rather than treating these models as "black boxes," the authors integrated Shapley values (SHAP) from game theory. SHAP values allow for the decomposition of a model's prediction into contributions from each input variable. This interpretability revealed that the model was exploiting subtle multidimensional correlations—"kinematic shapes"—rather than just simple magnitudes. Key variables identified include the invariant mass of the leading $b$-jet and the Higgs ($m_{b_1h}$) and the transverse momentum of the photon pair ($p_T^{\gamma\gamma}$).

Results and Future Outlook

The results "resurrect" the $b\bar{b}h$ channel as a viable experimental target.

  • HL-LHC: With $6$ ab$^{-1}$ of data, a statistical significance of $3.33\sigma$ is achievable for the $y_b^2$-driven signal, even when accounting for systematic uncertainties.
  • FCC-hh: At the $100$ TeV collider, the significance reaches a staggering $63.7\sigma$, transforming this into a high-precision measurement channel.

Furthermore, at the FCC-hh, the interference between the top and bottom quark contributions becomes a powerful tool for constraining the CP-phase of the Yukawa coupling. The study indicates that including $b\bar{b}h$ data can improve phase constraints by approximately 15% over standard decay-only measurements. By combining sophisticated machine learning with physical interpretability, this research provides a roadmap for future collider experiments to pin down one of the most fundamental parameters of the Higgs sector.

Summary updated

Abstract

A bstract The associated production of a $$ b\overline{b} $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>b</mml:mi> <mml:mover> <mml:mi>b</mml:mi> <mml:mo>¯</mml:mo> </mml:mover> </mml:math> pair with a Higgs boson could provide an important probe to both the size and the phase of the bottom-quark Yukawa coupling, y b . However, the signal is shrouded by several background processes including the irreducible Zh, Z → $$ b\overline{b} $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>b</mml:mi> <mml:mover> <mml:mi>b</mml:mi> <mml:mo>¯</mml:mo> </mml:mover> </mml:math> background. We show that the analysis of kinematic shapes provides us with a concrete prescription for separating the y b -sensitive production modes from both the irreducible and the QCD-QED backgrounds using the $$ b\overline{b}\gamma \gamma $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>b</mml:mi> <mml:mover> <mml:mi>b</mml:mi> <mml:mo>¯</mml:mo> </mml:mover> <mml:mi>γγ</mml:mi> </mml:math> final state. We draw a page from game theory and use Shapley values to make Boosted Decision Trees interpretable in terms of kinematic measurables and provide physics insights into the variances in the kinematic shapes of the different channels that help us complete this feat. Adding interpretability to the machine learning algorithm opens up the black-box and allows us to cherry-pick only those kinematic variables that matter most in the analysis. We resurrect the hope of constraining the size and, possibly, the phase of y b using kinematic shape studies of $$ b\overline{b}h $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>b</mml:mi> <mml:mover> <mml:mi>b</mml:mi> <mml:mo>¯</mml:mo> </mml:mover> <mml:mi>h</mml:mi> </mml:math> production with the full HL-LHC data and at FCC-hh.
BibTeX
@article{grojean2021resurrecting,
  author = {Christophe Grojean and Ayan Paul and Z. Qian},
  title = {{Resurrecting $$ b\textbackslash\{\}overline\{b\}h $$ with kinematic shapes}},
  journal = {Journal of High Energy Physics},
  year = {2021},
  doi = {10.1007/jhep04(2021)139},
}
Figure