Unpolarized quasi- and pseudo-distributions at one loop: gluon correlator decomposition, matching, and the region $|x|>1$

Abstract

We compute one-loop unpolarized quark and gluon quasi- and pseudo-distributions in $\overline{\mathrm{MS}}$ and match them to light-cone PDFs. Decomposing the gluon correlator into six covariant form factors, we derive the $6\times6$ renormalization mixing matrix: its eigenvectors classify multiplicatively renormalizable combinations, and any lattice operator choice projects onto our results without a further loop calculation. Evaluating all Fourier transforms in $d$ dimensions, we resolve the structure of the exterior $|x|>1$ in quasi-distributions. The correlator carries a short-distance logarithm with a branch cut at $ν=0$. By the Paley-Wiener theorem, the analytic part transforms into $|x|\le1$, so the exterior is generated solely by this branch cut. At leading power, the exterior carries no independent non-perturbative physics: it is fixed by the light-cone PDF through the renormalization group. In LaMET, lattice data therefore cannot determine the exterior independently of the interior, and unconstrained exterior parameters in reconstructions risk absorbing genuine power corrections into the leading-twist PDF. In the gluon channel, the matching kernel of Balitsky, Morris and Radyushkin differs from our $\overline{\mathrm{MS}}$ result by a finite polynomial, identical across index choices. We trace this to contracting an internal loop index over two instead of $d-2$ transverse directions, omitting an evanescent operator on the collinear pole. Using our basis, we show that the independent kernel of Yao, Ji and Zhang agrees with ours, separated from Balitsky et al. by the same polynomial. While the ratio scheme protects the momentum fraction $\langle x\rangle_g$, this kernel difference suppresses the extracted normalized second and third gluon moments by $11\%$ and $13\%$ at $α_s=0.3$, so the true $\overline{\mathrm{MS}}$ gluon is systematically harder than reported.