[Paper Review] Study of Systematic Uncertainties of Single Top Production at ATLAS

G. Khoriauli|arXiv (Cornell University)|Aug 19, 2008

Particle physics theoretical and experimental studies1 references3 citations

TL;DR

This study quantifies systematic uncertainties in single top quark cross-section measurements at ATLAS using Monte Carlo simulations and reweighting techniques for key sources including PDFs, b-quark fragmentation, jet energy calibration, and background normalization. At 1 fb⁻¹ and 10 fb⁻¹ luminosities, multivariate analysis reduces total systematic uncertainties to 9.8% (t-channel), 48.0% (s-channel), and 19.4% (Wt-channel), with background normalization being the dominant contributor.

ABSTRACT

Sytematic uncertainties to the single top production cross section measurement at the ATLAS experiment has been studied. Different sources of systematic uncertainties such as detector luminosity, jet energy calibration, SM background normalization, PDF parameterization and others have been considered. Large scale Monte-Carlo events simulation has been performed to estimate the contribution of each source in the overall uncertainty. The study was done for cut based analysis as well as for multivariate analysis of the single top measurements in ATLAS. The total systematic uncertainties of the single top cross-section measurements in its three production channels have been estimated at 1 $\mathrm{fb}^{-1}$ and 10 $\mathrm{fb}^{-1}$ integrated luminosity of the LHC.

Motivation & Objective

• To assess the impact of multiple systematic uncertainties on single top quark cross-section measurements at ATLAS.
• To evaluate the performance of cut-based versus multivariate analysis techniques in minimizing systematic effects.
• To quantify the contributions of PDF uncertainties, b-quark fragmentation, jet energy calibration, and background normalization to the total uncertainty.
• To apply reweighting methods for PDF and b-fragmentation uncertainties where large-scale simulations would be infeasible.
• To estimate total systematic uncertainties at 1 fb⁻¹ and 10 fb⁻¹ integrated luminosities using a toy Monte Carlo method.

Proposed method

• Used MC@NLO and AcerMC generators with CTEQ6.1 NLO PDF error sets to simulate single top production and top-antitop backgrounds.
• Applied reweighting techniques to estimate PDF uncertainties by computing event weights based on alternative PDF error sets relative to the central value.
• Calculated b-fragmentation uncertainties using binomial probabilities for b-tagging efficiency variations between Peterson and Lund-Bowler models.
• Propagated individual systematic effects through a toy Monte Carlo simulation, randomly varying background and efficiency within Gaussian uncertainties.
• Combined all systematic contributions using a master formula for cross-section uncertainty, with total uncertainty derived from the RMS of the simulated distribution.
• Performed uncertainty estimation for both cut-based and multivariate analyses, with the latter used to optimize signal selection beyond simple kinematic cuts.

Experimental results

Research questions

• RQ1 What is the relative contribution of each systematic uncertainty source—such as PDF, b-fragmentation, jet energy calibration, and background normalization—to the total uncertainty in single top cross-section measurements?
• RQ2 How do systematic uncertainties in the t-, s-, and Wt-channels of single top production scale with integrated luminosity at 1 fb⁻¹ and 10 fb⁻¹?
• RQ3 To what extent can multivariate analysis reduce systematic uncertainties compared to traditional cut-based selection in single top quark measurements?
• RQ4 How effective are reweighting techniques in estimating PDF and b-fragmentation uncertainties without requiring prohibitively large Monte Carlo samples?
• RQ5 Which systematic source dominates the total uncertainty in single top cross-section measurements, and how can it be mitigated?

Key findings

• Background normalization is the dominant source of systematic uncertainty in single top cross-section measurements across all channels.
• At 1 fb⁻¹, the total systematic uncertainty is 21.7% for the t-channel, 95.0% for the s-channel, and 48.0% for the Wt-channel using multivariate analysis.
• At 10 fb⁻¹, systematic uncertainties decrease to 9.8% (t-channel), 48.0% (s-channel), and 19.4% (Wt-channel), indicating significant improvement with increased luminosity.
• The reweighting method successfully estimates PDF and b-fragmentation uncertainties with high precision, avoiding the need for massive additional Monte Carlo simulations.
• The s-channel exhibits the largest systematic uncertainty due to low signal yield and high background contamination, making it the most challenging channel to measure precisely.
• Multivariate analysis reduces systematic uncertainties compared to cut-based analysis, particularly in the t- and Wt-channels, by optimizing signal selection beyond simple kinematic cuts.