An investigation of the αS and heavy quark mass dependence in the MSHT20 global PDF analysis

We investigate the MSHT20 global PDF sets, demonstrating the effects of varying the strong coupling alpha(S)(M-Z(2)) and the masses of the charm and bottom quarks. We determine the preferred value, and accompanying uncertainties, when we allow alpha(S)(M-Z(2)) to be a free parameter in the MSHT20 global analyses of deep-inelastic and related hard scattering data, at both NLO and NNLO in QCD perturbation theory. We also study the constraints on alpha(S)(M-Z(2)) which come from the individual data sets in the global fit by repeating the NNLO and NLO global analyses at various fixed values of alpha(S)(M-Z(2)), spanning the range alpha(S)(M-Z(2)) = 0.108 to 0.130 in units of 0.001. We make all resulting PDFs sets available. We find that the best fit values are alpha(S)(M-Z(2)) = 0.1203 +/- 0.0015 and 0.1174 +/- 0.0013 at NLO and NNLO respectively. We investigate the relationship between the variations in alpha(S)(M-Z(2)) and the uncertainties on the PDF5, and illustrate this by calculating the cross sections for key processes at the LHC. We also perform fits where we allow the heavy quark masses m(c) and m(b) to vary away from their default values and make PDF sets available in steps of Delta m(c) = 0.05 GeV and Delta m(b) = 0.25 GeV, using the pole mass definition of the quark masses. As for varying alpha(S)(M-Z(2)) values, we present the variation in the PDFs and in the predictions. We examine the comparison to data, particularly the HERA data on charm and bottom cross sections and note that our default values are very largely compatible with best fits to data. We provide PDF sets with 3 and 4 active quark flavours, as well as the standard value of 5 flavours.

Automated summary

The study investigated the MSHT20 global PDF sets, demonstrating the effects of varying the strong coupling alpha(S)(M-Z(2)) and the masses of charm and bottom quarks, leading to determination of preferred values and uncertainties. The research also explored constraints on alpha(S)(M-Z(2)) from individual data sets and provided resulting PDF sets. The study found the best-fit values at NLO and NNLO, and examined the relationship between variations in alpha(S)(M-Z(2)) and uncertainties in PDFs, illustrating this by calculating cross sections for key processes at the LHC.