Artificial neural network modelling of generalised parton distributions

TL;DR

This paper introduces a machine learning approach for nonparametric modeling of generalised parton distributions, reducing model dependency and improving the accuracy of phenomenological analyses in high-energy physics.

Contribution

It presents a novel machine learning strategy that models GPDs while satisfying theoretical constraints, minimizing model dependency compared to traditional parameterisations.

Findings

Reduces systematic uncertainties in GPD modeling

Enhances the precision of GPD extraction from experimental data

Supports the advancement of GPD phenomenology in the era of new experiments

Abstract

We discuss the use of machine learning techniques in effectively nonparametric modelling of generalised parton distributions (GPDs) in view of their future extraction from experimental data. Current parameterisations of GPDs suffer from model dependency that lessens their impact on phenomenology and brings unknown systematics to the estimation of quantities like Mellin moments. The new strategy presented in this study allows to describe GPDs in a way fulfilling theory-driven constraints, keeping model dependency to a minimum. Getting a better grip on the control of systematic effects, our work will help the GPD phenomenology to achieve its maturity in the precision era commenced by the new generation of experiments.