[Paper Review] The Top Quark Mass at the LHC

Paolo Nason|arXiv (Cornell University)|Jan 15, 2018

Particle physics theoretical and experimental studies17 references3 citations

TL;DR

This paper investigates theoretical uncertainties in top quark mass measurements at the LHC by comparing NLO+PS simulations using Pythia8.2 and Herwig7.1, revealing a 1 GeV discrepancy in reconstructed top mass peak positions between the two generators—especially under detector smearing—highlighting significant theoretical uncertainty from parton shower models, despite agreement in bare distributions.

ABSTRACT

I briefly discuss some theoretical aspects of top mass measurements at the LHC. In particular, I illustrate a recent theoretical study performed using next-to-leading order (NLO) calculations interfaced to shower generators (NLO+PS) of increasing accuracy, interfaced to both Pythia8 and Herwig7 Monte Carlo generators.

Motivation & Objective

• To assess theoretical uncertainties in top quark mass measurements at the LHC arising from different Monte Carlo generators and their parton shower implementations.
• To evaluate whether NLO+PS generators (hvq, t\bar{t}dec, b\bar{b}4\ell) yield consistent top mass peak positions when interfaced with Pythia8.2 and Herwig7.1.
• To investigate the impact of detector resolution (via Gaussian smearing) on the consistency of reconstructed top mass distributions across different generators.
• To determine whether differences in parton shower models (dipole vs. angular ordering) introduce unaccounted theoretical errors in direct top mass measurements.
• To examine whether leptonic and jet-based observables are robust against generator-dependent effects, challenging the assumption of their insensitivity to hadronization and shower models.

Proposed method

• Uses three NLO+PS generators: hvq (NLO in production only), t\bar{t}dec (NLO in production and decay with spin correlations), and b\bar{b}4\ell (includes finite width, non-resonant, and interference effects).
• Interfaces each generator with both Pythia8.2 and Herwig7.1 to compare parton shower effects on the reconstructed top mass peak.
• Defines a simplified 'particle-level top' mass observable, m_{Wb_j}^{max}, from the hardest lepton, neutrino, and b-jet to isolate mass sensitivity.
• Applies Gaussian smearing with 15 GeV width to mimic experimental resolution and assess its impact on peak position differences.
• Compares peak positions across generators and shower models, quantifying shifts in m_{Wb_j}^{max} for a fixed input pole mass of 172.5 GeV.
• Analyzes dependence on jet radius and other kinematic parameters to test robustness of the observed discrepancies.

Experimental results

Research questions

• RQ1 How do different Monte Carlo generators (hvq, t\bar{t}dec, b\bar{b}4\ell) with increasing NLO+PS accuracy affect the reconstructed top mass peak position?
• RQ2 To what extent do Pythia8.2 and Herwig7.1 parton shower models alter the m_{Wb_j}^{max} peak position, especially after detector smearing?
• RQ3 Are leptonic observables truly insensitive to parton shower and hadronization effects, as commonly assumed in top mass measurements?
• RQ4 What is the magnitude of theoretical uncertainty introduced by the choice of parton shower model, independent of NLO accuracy?
• RQ5 Can the observed 1 GeV discrepancy between Pythia and Herwig in the smeared distribution be attributed to systematic differences in shower algorithms rather than statistical fluctuations?

Key findings

• The peak position of m_{Wb_j}^{max} agrees within ~10 MeV across all three NLO+PS generators when interfaced with Pythia8.2, even for the less accurate hvq generator.
• With Herwig7.1, the m_{Wb_j}^{max} peak for hvq exceeds that of b\bar{b}4\ell by 300 MeV in the bare distribution and by nearly 700 MeV after Gaussian smearing (15 GeV width).
• The difference in peak position between Pythia8.2 and Herwig7.1 exceeds 1 GeV for the b\bar{b}4\ell and t\bar{t}dec generators after smearing, indicating a major theoretical uncertainty source.
• The discrepancy is not due to NLO accuracy but arises from fundamental differences in shower models (dipole vs. angular ordering), with Pythia and Herwig showing opposite sign shifts in the smeared case.
• The 1 GeV shift in the smeared distribution is not accidental but results from cancellation effects between different shower descriptions, suggesting a real theoretical uncertainty.
• Leptonic observables and b-jet energy peak positions also show large discrepancies between Pythia and Herwig, challenging the assumption of their insensitivity to shower and hadronization models.