Modelling Wt and tWZ production at NLO for ATLAS analyses

TL;DR

This paper compares different theoretical methods for modeling Wt and tWZ production at NLO, focusing on Diagram Removal 2, and discusses its implications for ATLAS analyses and background estimation.

Contribution

It introduces and evaluates Diagram Removal 2 as a method to model Wt and tWZ production without reshuffling, preserving interference effects, and compares it with other approaches.

Findings

Diagram Removal 2 preserves interference without reshuffling.

Predictions differ significantly in certain kinematic regions.

Implications for background estimation in ATLAS analyses.

Abstract

The generation of single top production in the $Wt$ channel at NLO gives rise to doubly-resonant diagrams that overlap with $t \bar t$, and similarly for associated production with a $Z$ or Higgs boson. Several methods exist to remove this overlap, thus avoiding double counting and separating the two processes. In Diagram Removal 1, the amplitudes of doubly-resonant diagrams are set to zero, which also removes the interference of $Wt$ with $t \bar t$. If a measurement is performed in a region where this interference is not negligible, it would be of interest to take it into account. Another method, Diagram Subtraction, preserves the interference. However, it relies upon momentum reshuffling, which introduces uncertainties to the prediction. An alternative method, Diagram Removal 2, does not involve reshuffling and preserves the interference by subtracting the amplitude squared of the doubly-resonant diagrams from the total. The predictions of Diagram Removal 2 from MG5_aMC@NLO are compared with other predictions for $Wt$ and $tWZ$ production in the context of analyses at the ATLAS experiment. Implications of using Diagram Removal 2 are discussed, both for measurements of single top processes and when $Wt$ or $tWZ$ are treated as a background. Differential distributions are shown, investigating in what regions the predictions differ the most.