The negligible impact of experimental inconsistencies in the NNPDF4.0 global dataset

TL;DR

This paper investigates the impact of data inconsistencies on the NNPDF4.0 global dataset and finds that, after accounting for higher order uncertainties, their effect is negligible compared to statistical fluctuations.

Contribution

The study demonstrates that data inconsistencies have minimal impact on PDF uncertainties when higher order uncertainties are included.

Findings

Data inconsistencies contribute negligibly to PDF uncertainties.

Higher order uncertainties dominate the error budget.

Impact of inconsistencies is comparable to statistical fluctuations.

Abstract

As both predictions and measurements of high-energy physics observables become more precise, controlling all sources of uncertainties in determinations of parton distribution functions (PDFs) becomes increasingly important. One source of PDF uncertainty is the result of data not being consistent under a chosen theoretical framework. In these proceedings we investigate the impact these inconsistencies present in the global NNPDF4.0 dataset. We show that, when accounting for missing higher order uncertainties, the missing contribution to the PDF uncertainty due to data inconsistencies are at the level of statistical fluctuations.