Determination of the top-quark mass from top-quark pair events with the matrix element method at next-to-leading order: Potential and prospects

TL;DR

This paper investigates the use of the matrix element method at next-to-leading order accuracy to determine the top-quark mass from pair-production events, demonstrating reduced theoretical uncertainties and improved robustness.

Contribution

It extends the matrix element method to NLO accuracy for top-quark mass measurement, showing potential for more precise results at high-luminosity colliders.

Findings

NLO matrix element method reduces theoretical uncertainties.

Realistic simulations with POWHEG improve robustness.

Potential for more precise top-quark mass measurements.

Abstract

In 2004 the matrix element method was used in a pioneering work by the Tevatron experiment D0 to determine the top-quark mass from a handful of events. Since then the method has been matured into a powerful analysis tool. While the first applications were restricted to leading-order accuracy, in the meantime also the extension to next-to-leading order (NLO) accuracy has been studied. In this article we explore the potential of the matrix element method at NLO to determine the top-quark mass using events with pair-produced top quarks. We simulate a toy experiment by generating unweighted events with a fixed input mass and apply the matrix element method to construct an estimator for the top-quark mass. Two different setups are investigated: unweighted events obtained from the fixed-order cross section at NLO accuracy as well as events obtained using POWHEG matched to a parton shower. The latter lead to a more realistic simulation and allow to study the impact of higher-order corrections as well as the robustness of the approach. We find that the matrix element method in NLO accuracy leads to a significant reduction of the theoretical uncertainties compared to leading order. In view of the high luminosity phase of the LHC, this observation is especially relevant in analyses which are no longer dominated by statistical uncertainties.