The Path to Proton Structure at One-Percent Accuracy

TL;DR

This paper introduces NNPDF4.0, a new set of parton distribution functions derived from an expanded dataset and advanced machine learning techniques, achieving high-precision proton structure insights relevant for LHC physics.

Contribution

It presents a novel methodology using hyperparameter optimization and stochastic gradient descent for PDF fitting, incorporating theoretical and experimental improvements for enhanced accuracy.

Findings

NNPDF4.0 achieves one-percent accuracy in proton structure determination.

The methodology ensures dataset compatibility and stability across different parametrizations.

Phenomenological studies show improved predictions for LHC processes.

Abstract

We present a new set of parton distribution functions (PDFs) based on a fully global dataset and machine learning techniques: NNPDF4.0. We expand the NNPDF3.1 determination with 44 new datasets, mostly from the LHC. We derive a novel methodology through hyperparameter optimisation, leading to an efficient fitting algorithm built upon stochastic gradient descent. We use NNLO QCD calculations and account for NLO electroweak corrections and nuclear uncertainties. Theoretical improvements in the PDF description include a systematic implementation of positivity constraints and integrability of sum rules. We validate our methodology by means of closure tests and "future tests" (i.e. tests of backward and forward data compatibility), and assess its stability, specifically upon changes of PDF parametrization basis. We study the internal compatibility of our dataset, and investigate the dependence of results both upon the choice of input dataset and of fitting methodology. We perform a first study of the phenomenological implications of NNPDF4.0 on representative LHC processes. The software framework used to produce NNPDF4.0 is made available as an open-source package together with documentation and examples.