Higgs Production: A Comparison of Parton Showers and Resummation

TL;DR

The paper systematically compares parton-shower Monte Carlo methods with analytic resummation for predicting Higgs boson transverse momentum distributions, using Z0 and diphoton processes as experimental stand-ins to test soft-gluon emission modeling at the Tevatron and LHC. It analyzes the CSS low-$p_T$ factorization framework, the mechanics of backward-evolution parton showers, and the role of non-perturbative $k_T$ in shaping pT spectra. The study finds broad agreement in some regions but notable discrepancies in Higgs pT distributions across MC versions, underscoring the need for matrix-element corrections and careful non-perturbative tuning. The results highlight how data from Run 2 and the LHC will be crucial to validate and refine these approaches for robust Higgs searches.

Abstract

The search for the Higgs boson(s) is one of the major priorities at the upgraded Fermilab Tevatron and at the CERN Large Hadron Collider (LHC). Monte Carlo event generators (MCs) are heavily utilized to extract and interpret the Higgs signal, which depends on the details of the multiple soft-gluon emission from the initial state partons in hadronic collisions. Thus, it is crucial to establish the reliability of the MCs used by the experimentalists. In this paper, the MC based parton shower formalism is compared to that of an analytic resummation calculation. Theoretical input, predictions and, where they exist, data for the transverse momentum distribution of Higgs bosons, $Z^0$ bosons, and photon pairs are compared for the Tevatron and the LHC. This comparison is useful in understanding the strengths and the weaknesses of the different theoretical approaches, and in testing their reliability.

Figures (16)

• Figure 1: The $Z^0$$p_T$ distribution (at low $p_T$) from CDF for Run 1 compared to predictions from ResBos (curve) and from PYTHIA (histograms). The two PYTHIA predictions use the default (rms) value for the non-perturbative $k_T$ (0.44 GeV) and the value that gives the best agreement with the shape of the data (2.15 GeV). The normalization of the resummed prediction was rescaled upwards by 8.4%. The PYTHIA prediction was rescaled by a factor of 1.4 for the shape comparison. (Including only soft--gluon QCD corrections, PYTHIA does not contain the QCD $K$-factor.)
• Figure 2: The $Z^0$$p_T$ distribution (for the full range of $p_T$) from CDF for Run 1 compared to predictions from ResBos (curve) and from PYTHIA (histogram). The normalization of the resummed prediction was rescaled upwards by 8.4%. The PYTHIA prediction was rescaled by a factor of 1.4 for the shape comparison.
• Figure 3: A comparison of the PYTHIA predictions for diphoton production at the Tevatron for the two different subprocesses, $q\overline{q}$ and $gg$. The same cuts are applied to PYTHIA as in the CDF diphoton analysis.
• Figure 4: A comparison of the PYTHIA predictions for diphoton production at the Tevatron for the two different subprocesses, $gg$ (top) and $q\overline{q}$ (bottom), for two recent versions of PYTHIA. The same cuts are applied to PYTHIA as in the CDF diphoton analysis.
• Figure 5: A comparison of the PYTHIA and ResBos predictions for diphoton production at the Tevatron for the two different subprocesses, $gg$ (left) and $q\overline{q}$ (right). The same cuts are applied to PYTHIA and ResBos as in the CDF diphoton analysis. The bottom figures show the same in logarithmic scale.