Closure testing NNPDF3.0 with LHC observables

Abstract

A thorough understanding of PDFs and their uncertainties is important for the LHC and for future collider experiments. The recently released NNPDF3.0 set was presented alongside results from closure tests, where PDF fits were performed on pseudo-data generated from a chosen input PDF set. The results there demonstrate the validity of the NNPDF methodology and also provide some information about different contributions to the PDF uncertainties. Here I present a number of additional closure test results, specifically an investigation into the effect of using cross-validation in the fits, and an assessment of the successful reproduction of LHC cross-sections in closure tests. The results are consistent with those previously shown in the NNPDF3.0 paper.