$gg \to ZH$ at NLO matched to parton showers with ggxy and POWHEG

This paper implements next-to-leading order QCD corrections for the $gg \to ZH$ process within the ggxy framework, augmented with leptonic decays and an interface to POWHEG to enable parton shower simulations using Pythia.

The Gist

Imagine the Large Hadron Collider (LHC) as the world's most powerful particle smasher. Scientists smash protons together to create rare events, hoping to catch a glimpse of the Higgs boson, the particle that gives other particles mass. One of the most interesting ways the Higgs appears is when it is produced alongside a Z boson (a heavy cousin of the photon).

This paper is essentially a user manual and a quality control report for a new, highly sophisticated "calculator" that physicists use to predict exactly how often this happens and what it looks like.

Here is the breakdown of what the authors did, using everyday analogies:

1. The Problem: A Missing Piece of the Puzzle

For years, physicists had a very accurate calculator for one way the Higgs and Z boson are made (like two cars crashing into each other). However, there is a second, sneaky way they are made: two invisible "gluon" particles (the glue holding protons together) collide to create them.

Until now, the math for this "gluon collision" was too complex to put into a standard calculator. It was like having a recipe for a cake but only knowing how to bake it in a lab oven, not in a home kitchen. This new paper puts that complex recipe into a user-friendly app called ggxy.

2. The Solution: The "ggxy" App

The authors took the complex mathematical formulas (calculated by other scientists) and coded them into a C++ library called ggxy.

• The Analogy: Think of the math as a massive, intricate blueprint for a skyscraper. The authors didn't just draw the blueprint; they built a construction robot (the software) that can read the blueprint and instantly tell you how much concrete and steel you need, even if you change the design slightly.
• Flexibility: The app is smart enough to handle different "mass settings" for the top quark (a heavy particle involved in the process). It's like a chef who can adjust a recipe whether you are using fresh, frozen, or dried ingredients, ensuring the cake still tastes right.

3. Adding the "Special Effects": The Z Boson's Decay

In the real world, the Z boson doesn't just sit there; it immediately falls apart (decays) into other particles like electrons or neutrinos.

• The Old Way: Some calculators treated the Z boson like a solid, stable ball.
• The New Way: This paper adds "special effects" to the simulation. It accounts for the fact that the Z boson is unstable and can be "off-shell" (a bit like a ghost that hasn't fully materialized yet). It also tracks how the "spin" of the particles affects how they fly apart.
• Why it matters: If you ignore these details, your prediction of where the particles will land is like trying to predict the path of a spinning top without knowing which way it's spinning. The new app gets the spin right.

4. The "Traffic Controller": Matching to Parton Showers

This is the most technical but crucial part.

• The Concept: When particles collide, they don't just stop. They spray out a shower of other particles, like sparks flying off a grinding wheel. This is called a "parton shower."
• The Challenge: You have a precise calculation for the initial crash (the "hard" event) and a simulation for the sparks (the "shower"). If you just glue them together, you might double-count the sparks or miss some.
• The Solution: The authors built a bridge (an interface) between their calculator and POWHEG, a famous traffic controller. POWHEG ensures the initial crash and the subsequent sparks fit together perfectly without overlap.
• The Result: They can now simulate the entire event, from the initial collision to the final spray of particles, using a program called Pythia. This is like going from a static photo of a car crash to a full-motion video of the crash, the airbags deploying, and the debris scattering.

5. The Results: Why Should We Care?

The authors tested their new calculator against known results and found it works perfectly.

• The Discovery: They found that if you ignore the "spin" and the "decay" details, your predictions for where the particles end up can be off by 40% to 100%. That's a huge difference!
• The Impact: As the LHC gets more precise, we need our predictions to be just as precise. If the theory is off, we might think we found "new physics" (like a new particle) when it was just a miscalculation. This new tool ensures that when we see something strange at the LHC, it's actually something new, not just a math error.

Summary

In short, this paper is about upgrading the software physicists use to predict Higgs boson production. They took a complex, theoretical calculation, turned it into a flexible, easy-to-use tool, added realistic "special effects" for particle decay, and connected it to a simulation engine that can model the chaotic aftermath of a particle collision. This ensures that when the LHC scientists look at their data, they are comparing it to the most accurate prediction possible.

Technical Summary

1. Problem Statement

The associated production of a Higgs boson with a Z boson ($ZH$) is a critical process for studying Higgs properties at the Large Hadron Collider (LHC). While the Drell-Yan ($q\bar{q} \to ZH$) channel is well understood and implemented in public codes like vh@nnlo, the gluon-induced channel ($gg \to ZH$) is numerically significant, particularly at high energies.

• Gap: Prior to this work, Next-to-Leading Order (NLO) QCD corrections for the $gg \to ZH$ process were not available in flexible, public computer codes.
• Complexity: The calculation requires two-loop amplitudes involving top quark loops. These are analytically complex and depend on the top quark mass renormalization scheme (On-Shell vs. $\overline{\text{MS}}$).
• Phenomenology: Accurate predictions require handling off-shell Z boson effects, spin correlations in Z decays, and matching to parton showers for realistic event simulation.

2. Methodology

The authors implemented the $gg \to ZH$ process at NLO QCD within the C++ library ggxy and created an interface to POWHEG for parton shower matching.

A. Theoretical Framework & Expansions

The implementation relies on analytic expressions for two-loop amplitudes derived from three complementary expansion methods (Ref. [14]):

1. Forward Expansion: Valid for $|t|, q_Z^2, q_H^2 \ll m_t^2, s$.
2. High-Energy Expansion: Valid for $q_Z^2, q_H^2 \ll m_t^2 \ll s, |t|$.
3. Large Top Mass Limit: Used for low center-of-mass energies ($\sqrt{s} < 200$ GeV) to ensure numerical stability.

• Switching Logic: The code automatically switches between expansions based on kinematic variables (e.g., $p_T$). Padé approximations are employed for $p_T > 180$ GeV to improve convergence.
• Form Factors: The amplitude is decomposed into six independent form factors ($F_{12}^\pm, F_2^-, F_3^\pm, F_4$).

B. Real Corrections & Subtraction

• Subtraction Scheme: The Catani-Seymour subtraction scheme is used to handle infrared singularities.
• Matrix Elements: Real emission ($2 \to 3$) matrix elements are calculated using Recola (recursive approach) combined with Collier for loop integrals.
• Diagram Filtering: A critical distinction is made between Drell-Yan-type diagrams (quark-initiated) and gluon-fusion diagrams. The implementation explicitly filters out Drell-Yan contributions to isolate the pure $gg \to ZH$ channel, while retaining contributions where the Z boson is radiated from external quark lines (gauge-invariant subset).
• Off-Shell Decays: The code handles $Z \to \ell^-\ell^+$ and $Z \to \nu\bar{\nu}$ decays by combining helicity amplitudes of $gg \to Z^*H$ with $Z^* \to \ell\ell$, including spin correlations and off-shell effects (limited to $M_{\ell\ell} \lesssim 150$ GeV).

C. POWHEG Interface

• A new POWHEG process (ggxy_ggZH) was developed, interfacing ggxy for virtual corrections and Recola for one-loop matrix elements.
• This allows matching to parton showers (specifically Pythia 8) while preserving NLO accuracy.
• The interface supports both stable Z bosons (decays handled by shower) and unstable Z bosons (decays handled at the matrix element level).

3. Key Contributions

1. First Public Implementation: Provided the first public, flexible C++ implementation of NLO QCD corrections for $gg \to ZH$ via the ggxy library.
2. Flexible Renormalization Schemes: Allows users to choose between On-Shell (OS) and $\overline{\text{MS}}$ schemes for the top quark mass, with arbitrary renormalization scales ($\mu_t$).
3. Spin Correlations & Off-Shell Effects: Implemented full spin correlations and off-shell Z boson effects at the amplitude level, crucial for differential distributions involving leptons.
4. POWHEG Matching: Established a robust interface to POWHEG, enabling NLO+PS simulations.
5. Validation: Cross-checked against independent calculations (Ref. [12], [13], [14], [34]) and the vh@nnlo program.

4. Key Results

Fixed-Order Results (ggxy)

• Total Cross Sections: Reproduced reference results from literature (Ref. [13], [34]) with high precision at $\sqrt{s} = 13, 13.6, 14$ TeV.
  • Example at 13.6 TeV: $\sigma_{NLO} \approx 130.5$ fb (consistent with Ref. [34]).
• Renormalization Scheme Dependence: Switching from OS to $\overline{\text{MS}}$ scheme introduces a downward shift of 10–40% in the high invariant mass region ($M_{ZH}$), comparable to scale variation uncertainties.
• Off-Shell Decays: Integrated cross sections for $gg \to \ell^-\ell^+H$ and $gg \to \nu\bar{\nu}H$ were computed. The NLO K-factor remains $\approx 85\%$ regardless of decay mode.
• Combination with Drell-Yan: Combined with vh@nnlo (Drell-Yan channel). The gluon-induced channel contributes $\sim 15\%$ at LHC energies, rising to 25% at higher energies ($\sqrt{s} \to 100$ TeV).

Parton Shower Results (POWHEG + Pythia)

• Spin Correlations: Comparing "NLO+PS" (matrix element decays with spin correlations) vs. "NLO+PS-stable" (shower decays without spin correlations):
  • Angular Distributions: The naive approach (no spin correlations) fails to describe azimuthal angles ($\Delta\phi_{e^-e^+}$) and lepton $p_T$ distributions, showing deviations of 40–100%.
  • Rapidity: Lepton rapidity ($y_{e^+}$) is insensitive to spin correlations.
• Parton Shower Effects:
  • For $p_{T,ZH}$ and $p_{T,H}$, parton showers significantly alter the shape.
  • Using the default hdamp = ∞ leads to a harder spectrum in the tails (differences up to 100% for $p_{T,H}$).
  • Tuning hdamp = 250 reduces these discrepancies, bringing the shower tails closer to fixed-order predictions (reducing effects to ~40–50%).

5. Significance

• Precision Physics: This work enables precise theoretical predictions for the $gg \to ZH$ channel, which is essential for extracting Higgs couplings and searching for New Physics at the LHC and future colliders (FCC).
• Methodological Benchmark: It demonstrates the necessity of including spin correlations and off-shell effects in NLO+PS simulations for processes involving vector boson decays. Neglecting these leads to large systematic errors in differential observables.
• Community Resource: By providing the code (ggxy and ggxy_ggZH) on GitLab, the authors allow the community to easily integrate these corrections into their own analysis frameworks, moving beyond the limitations of previous tools that lacked this specific channel at NLO.
• Future Outlook: The results motivate the full calculation of NNLO corrections for $gg \to ZH$ in the heavy-top limit, as the NLO scale uncertainties are now dominated by the gluon channel.