Parton distributions with higher twist and jet power corrections

TL;DR

The paper develops a global PDF determination that accounts for higher twist (HT) corrections in DIS and linear power corrections (PCs) in jet/dijet production using a theory covariance framework. HT and PC effects are modeled as multiplicative, linearly parameterized shifts with full correlation across related observables, and their uncertainties are propagated into PDFs via Monte Carlo replicas. The resulting NNPDF4.0HT sets show modest but non-negligible impacts, most notably in the gluon distribution at intermediate $x$ and in Higgs production via gluon fusion, with improvements in perturbative convergence and data description when HT/PCs are included. The work provides public PDFs with HT and PCs (LHAPDF sets) and highlights the necessity of accounting for such power corrections in high-precision LHC phenomenology and $\alpha_s$ extractions.

Abstract

We present a global determination of parton distribution functions (PDFs) that accounts for higher twist corrections in deep-inelastic scattering (DIS) and linear power corrections for single inclusive jet and dijet production data from the LHC. We determine these corrections and their associated correlated uncertainties using a methodology based on the theory covariance formalism, previously used to account for nuclear uncertainties and missing higher order uncertainties (MHOUs) in global PDF determinations. We then study the impact of the power corrections on the extracted PDFs, and demonstrate an improved description of the data due to a reduced sensitivity to DIS data in the low-$x$ region where higher twist uncertainties are relatively large, and a reduced sensitivity to single inclusive jet data at relatively low $p_T$, where linear power corrections can be significant. Finally, we assess the impact of power corrections on observables relevant to LHC phenomenology, including Higgs production via gluon fusion, and the determination of $α_s$. We find that these effects, while small, can be significant, improving perturbative convergence.

Figures (11)

• Figure 3.1: The three HT functions $H_2^p(x)$ (top), $H_2^d(x)$ (middle) and $H_{CC}(x)$ (lower). The prior uncertainty is shown in gray. The posterior distribution is then shown for HT determinations where the leading twist is computed at NNLO (blue), NNLO+MHOU (green) and aN$^3$LO+MHOU (red).
• Figure 3.2: The $x,Q^2$ plane of the DIS data points used in the fits: the NC data (left) and CC data (right). The solid line indicates the position of the standard cut used in NNPDF4.0. The additional data points below this line are the data added when the cuts are lowered to determine the HT. The size of the points indicates the size of the the absolute value of the relative shift due to HT, determined in the NNLO fit. The colour code represents the ratio between the posterior uncertainty due to HT and the experimental uncertainty.
• Figure 3.3: Comparison of our NNLO $F_2$ higher twist to that determined in protons and deuteron by Virchaux and Milsztajn in a NLO PDF fit to SLAC and BCDMS data Virchaux:1991xp. They include TMCs, and use the same multiplicative parametrization of the HT, fitting proton and deuteron data separately, so the results are directly comparable to our own.
• Figure 3.4: Same as Fig. \ref{['fig:posterior_dis']}, but for single-inclusive jets (top), ATLAS dijet (lower left) and CMS dijet (lower right) observables.
• Figure 3.5: The $y,p_T$ plane of the single-inclusive jet data points (left) and the $y, m_{jj}$ plane of the dijet data points (right). For the dijet data, the ATLAS and CMS definitions of rapidity are both used on the same plot. As in Fig. \ref{['fig:kin_jet_nnlo']}, the size of the points indicates the size of the the absolute value of the relative shift due to the PCs, determined in the NNLO fit. The colour code represents the ratio between the posterior uncertainty due to PCs and the experimental uncertainty.