Stability of NLO Global Analysis and Implications for Hadron Collider Physics

TL;DR

This paper assesses the stability of NLO global QCD analyses of PDFs against variations in input data cuts, gluon parametrization (including negative gluons at small x), and the strong coupling α_s. Using the CTEQ framework and the Lagrange Multiplier method, it demonstrates that PDFs and key hadroproduction predictions, especially σ_W at the LHC, remain stable within a few percent under realistic variations. The results contrast with some MRST findings, attributing differences to gluon parameterization and high-x behavior, and reinforce the validity of NLO PDFs for current collider phenomenology while highlighting the potential gains from NNLO analyses in the future.

Abstract

The phenomenology of Standard Model and New Physics at hadron colliders depends critically on results from global QCD analysis for parton distribution functions (PDFs). The accuracy of the standard next-to-leading-order (NLO) global analysis, nominally a few percent, is generally well matched to the expected experimental precision. However, serious questions have been raised recently about the stability of the NLO analysis with respect to certain inputs, including the choice of kinematic cuts on the data sets and the parametrization of the gluon distribution. In this paper, we investigate this stability issue systematically within the CTEQ framework. We find that both the PDFs and their physical predictions are stable, well within the few percent level. Further, we have applied the Lagrange Multiplier method to explore the stability of the predicted cross sections for W production at the Tevatron and the LHC, since W production is often proposed as a standard candle for these colliders. We find the NLO predictions on sigma_W to be stable well within their previously-estimated uncertainty ranges.

Figures (11)

• Figure 1: (a) Dependence of the MRST predictions for the total $W$ production cross section at the LHC on kinematic cuts on input data used in the global analysis; (b) the $W$ rapidity distribution according to the MRST default PDFs, and the "conservative" PDFs (based on relatively high $x$ and $Q$ cuts). Both figures are reproduced from mrst03.
• Figure 2: Comparison of CTEQ and MRST PDFs, for (a) the u-quark and (b) the gluon distributions at $Q^2=10 \, \mathrm{GeV}^2$. The $x$ scale is chosen $\propto x^{1/3}$ to show details from both small- and large-$x$ regions clearly. The vertical axis is the ratio of the specified PDF set to CTEQ6.1M. The dotted curve is CTEQ6; solid curve is MRST2002; dashed curve is MRST2003c.
• Figure 3: Comparison between the D0 jet data and cross sections calculated with CTEQ6.1 (horizontal line), MRST2002, and MSRT2003c PDFs. The vertical axis is the fractional difference (data-CTEQ6.1)/CTEQ6.1.
• Figure 4: Predicted total cross section of $W^+ + W^-$ production at the LHC for the fits obtained in our stability study, compared to the NLO results of Ref. mrst03 (cf. Fig. \ref{['fig:mrst']}). The $Q$-cut values associated with the CTEQ points are given in the two tables. The overall PDF uncertainty of the prediction is $\sim 5\%$.
• Figure 5: (a) The global fit $\chi^2$ as a function of $\alpha_s(m_Z)$ using the standard data cuts. The solid curve assumes $g(x) > 0$; the dotted curve allows $g(x) < 0$. (b) Analogous curves using the strong data cuts.