Non-perturbative QCD Effects and the Top Mass at the Tevatron

TL;DR

The paper introduces a universal toy model for colour reconnection (CR) in hadronic final states and implements it in Pythia, constraining the CR strength via Tevatron minimum-bias data. It then assesses how CR/UE and shower variations propagate into semileptonic ttbar top-mass observables using a generator-level toy mass estimator, finding that non-perturbative CR effects contribute about 0.5 GeV and shower differences about 1 GeV to the top mass, with an overall spread around 1.5 GeV. The work highlights the significance of non-perturbative uncertainties for high-precision top-mass measurements and urges further data-driven constraints and cross-checks, including implications for LHC studies. Overall, CR is shown to be a measurable, model-dependent effect that should be accounted for in precision top physics and hadron-collision analyses.

Abstract

We present a new, universally applicable toy model of colour reconnections in hadronic final states. The model is based on hadronising strings and has one free parameter. We next present an implementation of this model in the Pythia event generator and provide several parameter sets ("tunes"), constrained by fits to Tevatron minimum-bias data. Finally, we consider the sensitivity of a simplified top mass analysis to these effects, in exclusive semi-leptonic top events at the Tevatron. A first attempt at isolating the genuine non-perturbative effects gives an estimate of order delta(m_top) ~ 0.5 GeV from non-perturbative uncertainties, and a further delta(m_top) ~ 1 GeV from shower effects.

Figures (3)

• Figure 1: Comparison of the models/tunes discussed in the text. Inelastic non-diffractive (min-bias) events in $p\bar{p}$ collisions at $\sqrt{s} = 1960\,\hbox{GeexV}$. Top: Charged multiplicity distribution. Bottom: mean $p_\perp$ in GeV, as a function of charged multiplicity. The main point is not the precise predictions for each tune, but rather that they all roughly agree, with the notable exception of the NOCR one.
• Figure 2: Calibration curve obtained for Tune A, before JES rescaling. A similar plot was made for each model in Tab. \ref{['tab:tunes']} and their relative offsets compared, both before and after JES rescaling. The inset shows the Gaussian fit to the distribution reconstructed top masses from the hadronic event side for the specific point $m_\mathrm{top}^{\mathrm{gen}}=175\,\hbox{GeexV}$.
• Figure 3: Comparison of calibration offsets obtained for each model, in GeV, here including an additional parameter set, 'BW', from Rick Field. On the left are the results obtained before $\mathrm{JES}$ rescaling (dots) and on the right after rescaling (squares). The coloured bands group models with the same final-state shower (green: virtuality-ordered, blue: $p_\perp$-ordered). The statistical precision due to the finite number of generated events is at the $0.1\,\hbox{GeexV}$ level.