Photon-initiated production of a di-lepton final state at the LHC: cross section versus forward-backward asymmetry studies

TL;DR

This work analyzes photon-induced dilepton production at the LHC and its competition with Drell-Yan, emphasizing the large PDF-driven uncertainties in photon-initiated contributions at high dilepton masses. Using QED PDFs from MRST, NNPDF, and CT14, and a replica-based uncertainty framework, the authors show PI can become sizeable relative to DY for $M_{ll}$ in the multi-TeV range, but with substantial theoretical uncertainties that challenge spectrum-based searches. Importantly, the reconstructed Forward-Backward Asymmetry $A_{FB}^*$ remains relatively robust to these uncertainties, making it a powerful observable for both discovery and interpretation of potential $Z′$ signals in resonant and non-resonant scenarios. The study also explores kinematic cuts as a way to suppress PI, while highlighting the trade-offs with signal efficiency, and argues for using a combined strategy involving spectrum and $A_{FB}^*$ alongside improved photon PDFs and higher-order corrections. The findings motivate further theoretical development of photon-initiated processes and their proper inclusion in LHC analyses at high masses.

Abstract

We explore the effects of Photon Induced (PI) production of a dilepton final state in the Large Hadron Collider environment. Using QED Parton Distribution Function (PDF) sets we can treat the photons as real partons inside the protons and compare their yield directly to that of the Drell-Yan (DY) process. In particular, we concentrate on an error analysis of the two mechanisms. In order to do so, we use the NNPDF set, which comes with a set of replicas to estimate the systematic PDF error. On the one hand, we find that the PI contribution becomes dominant over DY above a dilepton invariant mass of 3 TeV. On the other hand, the PI predictions are affected by a large error coming from the QED PDFs, well above the one affecting the DY mode. We assess the impact of these uncertainties in the context of resonant and non-resonant searches for a neutral massive vector boson (Z') through the differential cross section and Forward-Backward Asymmetry (AFB) observables as a function of the dilepton invariant mass. While the former is subject to the aforementioned significant residual errors the latter shows the systematic error cancellation expected (recall that AFB is a ratio of cross sections) even in presence of PI contributions, so that the recently emphasized key role played by AFB as a valid tool for both Z' discovery and interpretation in both resonant and non-resonant mode is further consolidated.

Figures (14)

• Figure 1: Photon Induced process contributing to the dilepton final state.
• Figure 2: (a) PI differential cross section in the dilepton invariant mass for the LHC at 8 TeV (blue line) and 13 TeV (red line). (b) Ratio of the two differential cross sections at 13 TeV and 8 TeV (magenta line) compared with the analogous ratio for the DY case (green line). Standard acceptance cuts are applied: $|\eta_l| < 2.5$ and $p_T^l > 20~$GeV.
• Figure 3: (a) Relative importance of the PI term in the dilepton SM background at the LHC with energies of 8 TeV (blue line) and 13 TeV (red line). Standard acceptance cuts are applied ($|\eta_l| < 2.5$ and $p_T^l > 20~GeV$). (b) Number of events expected in the dielectron channel as a function of the lower cut applied on the dielectron invariant mass. The solid lines represent the pure DY background while the dashed ones include also the PI contribution. Different colors refer to different stages of the LHC as described in the legend. Standard acceptance cuts are applied ($|\eta_l| < 2.5$ and $p_T^l > 20~$GeV) as well as the declared efficiency of the electron channel Khachatryan:2014fba. NNLO QCD corrections are accounted for in the DY term Hamberg:1990np.
• Figure 4: (a) Ratio between the central values of the photon induced dilepton spectrum obtained with the MRST2004QED and the NNPDF2.3QED sets. (b) Same as (a) but now comparing the MRST2004QED and the CT14QED sets. (c) Ratio between the dilepton spectrum at two different factorisation scales, $Q^2=\hat{s}$ and $Q^2=P_T^2$, for MRST2004QED (dashed lines), for CT14QED (dotted lines) and NNPDF2.3QED (solid lines) at the 8 TeV LHC (blue lines) and the 13 TeV LHC (red lines). (d) Differential cross section distribution for the PI process as predicted by the three PDF collaborations specified. Standard acceptance cuts are applied ($|\eta_l| < 2.5$ and $p_T^l > 20~$GeV).
• Figure 5: (a) DY dilepton spectrum for the 100 NNPDF replicas. (b) Central value and PDF error for the DY dilepton spectrum via NNPDF. (c) Relative size of the PDF error for the DY process at the 8 TeV (blue line) and 13 TeV (red line) LHC. Standard acceptance cuts are applied ($|\eta_l| < 2.5$ and $p_T^l > 20~$GeV).