[Paper Review] A Comparison of Predictions for SM Higgs Boson Production at the LHC

Csaba Balázs, Massimiliano Grazzini+3 more|ArXiv.org|Mar 4, 2004

Particle physics theoretical and experimental studies5 citations

TL;DR

This paper compares theoretical predictions for the transverse momentum (pT) distribution of the Standard Model Higgs boson in gluon-gluon fusion at the LHC, evaluating methods ranging from parton shower Monte Carlo (HERWIG, PYTHIA) to analytical resummation (ResBos) and NLO+PS matching (MC@NLO). It finds that while most predictions agree on the peak position and low/moderate pT behavior, significant discrepancies emerge at high pT due to differing treatments of higher-order corrections, with PYTHIA yielding a notably softer spectrum and NNLO-resummed results (Grazzini et al.) being the hardest, highlighting the importance of accurate higher-order effects for experimental analysis strategies.

ABSTRACT

This paper describes a comparison of most of the available predictions for the cross section and transverse momentum distribution for a 125 GeV mass Higgs at the LHC, including those from the PYTHIA and HERWIG parton shower Monte Carlos and from four resummation calculations.

Motivation & Objective

• To compare the reliability and shape of Higgs boson pT distributions predicted by various theoretical methods at the LHC.
• To assess the impact of higher-order QCD corrections—especially NNLO and resummation—on the pT spectrum and total cross section.
• To evaluate discrepancies between Monte Carlo generators (HERWIG, PYTHIA) and analytical resummation (ResBos) and NLO+PS matching (MC@NLO) in the high-pT region.
• To determine whether differences in pT spectrum predictions are experimentally resolvable and could affect search strategies.

Proposed method

• Uses analytical resummation in impact parameter (b) space via the CSS formalism to exponentiate large logarithmic corrections of the form αs^n ln^m(mH²/pT²).
• Applies the Sudakov form factor S_c with coefficients A_c(αs) and B_c(αs), expanded in αs, to resum soft and collinear radiation up to NNLL accuracy.
• Compares predictions from HERWIG (angular-ordered shower), PYTHIA (with matrix element corrections), MC@NLO (NLO+PS matching), and ResBos (analytical resummation with matching).
• Incorporates NNLO total cross sections (e.g., Grazzini et al.) and uses MRST2001/2002 and CTEQ5M PDFs to assess PDF dependence.
• Normalizes all predictions to the same total cross section (39.4 pb) to isolate differences in pT shape for direct comparison.
• Evaluates the role of scale choices (μ = mH vs. μ = √(mH² + pT²)) and hard matrix element corrections in shaping the high-pT tail.

Experimental results

Research questions

• RQ1 How do different theoretical approaches—parton shower MC, analytical resummation, and NLO+PS matching—predict the Higgs boson transverse momentum distribution at the LHC?
• RQ2 To what extent do higher-order QCD corrections (NNLO, NNLL) alter the shape and normalization of the pT spectrum compared to fixed-order or leading-log approximations?
• RQ3 Why does PYTHIA predict a significantly softer pT spectrum than other methods, and what does this imply for experimental analysis strategies?
• RQ4 How do scale choices and matrix element corrections influence the high-pT behavior of the Higgs pT distribution across different models?
• RQ5 Are the observed differences in pT spectrum predictions experimentally resolvable, and could they affect Higgs boson search strategies at the LHC?

Key findings

• The PYTHIA prediction is significantly softer than all other models, especially in the high-pT region, due to the lack of hard matrix element corrections in its default implementation.
• HERWIG, which uses angular ordering, lacks reliable hard radiation effects and becomes unphysical at high pT, though it correctly captures low-pT behavior.
• The MC@NLO prediction, fixed to the NLO total cross section (~32.4 pb), is softer than the NNLO result (39.4 pb) and lies below the ResBos and Grazzini et al. curves at high pT.
• ResBos and Grazzini et al. predict harder pT spectra than MC@NLO, primarily due to inclusion of NNLO matrix elements and K-factor matching, respectively.
• The Grazzini et al. prediction, based on NNLO matrix elements and three-loop αs, yields a total cross section of 39.4 pb, which is ~10% higher than the pure threshold result (35 pb) from Kulesza et al., who include subleading low-x contributions.
• Normalization to the same total cross section (39.4 pb) reveals that only PYTHIA and Kulesza et al. show significant deviations in shape, with PYTHIA being the softest and Kulesza et al. softer than the NNLO baseline, indicating sensitivity to non-peak region dynamics.