Closure testing the NNPDF3.0 methodology

Abstract

A thorough understanding of the issues surrounding the determination of parton distributions is crucial due to their importance to calculations of LHC observables. However, it is still not fully understood how much of an impact methodological bias has on PDF fits. Closure tests, where a fit is performed to pseudo-data generated using an existing PDF set, provide a way of directly investigating whether current PDF fitting methodologies are successful. Here, we present a sample of results from closure tests applying the NNPDF methodology to data created using a variety of different PDF sets. The results validate our methodology by showing that the initial PDFs can be reproduced within uncertainties. We also briefly discuss our latest PDF determination, NNPDF3.0, which has been developed making extensive use of the closure test technique.