Neural Network Representation of Generalized Parton Distributions (NNGPD)
Abstract
Extracting generalized parton distributions (GPDs) from experimental observables, the convolution structures denoted as Compton form factors (CFFs), requires solving an intrinsic inverse problem. We presented results using NNGPD, a flexible regression method based on a physics-informed neural-network framework, for reconstructing GPDs using integral constraints -- including CFFs and Mellin moments -- and symmetry and endpoint constraints. The approach does not require pointwise training information on the underlying GPDs. Closure tests using pseudodata generated from a phenomenological model demonstrate the ability of the framework to reproduce the principal features of the underlying distributions across a range of kinematics. A Bayesian neural-network formulation is also introduced to address uncertainties associated with the reconstruction. Finally, we discussed a broader question in GPD phenomenology: whether the complete momentum-fraction dependence of GPDs is necessary to extract physical quantities such as partonic angular momentum. This motivates investigating approaches that infer Mellin moments directly from experimentally constrained CFFs, while quantifying the remaining ambiguities.
Related
- Deep-neural-network extraction of unpolarized transverse-momentum-dependent parton distributions in bT space from Drell-Yan data
- Generalized parton distributions of the nucleon in the chiral quark soliton model: Sum over all quark states at nonzero skewness and momentum transfer
- Generalized parton distributions of the nucleon in the chiral quark soliton model II: Chiral anomaly mediation in the polarized GPDs
- Leading-twist to higher-twist generalized parton distributions of the pseudoscalar mesons at non-zero skewness
- Generalized parton distributions: Theory meets experiment
BibTeX
@misc{liuti2026neural,
title = {{Neural Network Representation of Generalized Parton Distributions (NNGPD)}},
author = {Simonetta Liuti and Jitao Xu and Ho Jang and Zaki Panjsheeri and Gia-Wei Chern and Yaohang Li and Douglas Q. Adams and Michael Engelhardt and Gary R. Goldstein and Adil Khawaja and Huey-Wen Lin and Saraswati Pandey and Kemal Tezgin},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.23390},
eprint = {2610.23390},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.23390},
}
RIS
TY - UNPB
TI - Neural Network Representation of Generalized Parton Distributions (NNGPD)
AU - Simonetta Liuti
AU - Jitao Xu
AU - Ho Jang
AU - Zaki Panjsheeri
AU - Gia-Wei Chern
AU - Yaohang Li
AU - Douglas Q. Adams
AU - Michael Engelhardt
AU - Gary R. Goldstein
AU - Adil Khawaja
AU - Huey-Wen Lin
AU - Saraswati Pandey
AU - Kemal Tezgin
PY - 2026
DA - 2026/10/16
PB - arXiv
DO - 10.48550/arxiv.2610.23390
UR - https://arxiv.org/abs/2610.23390
AB - Extracting generalized parton distributions (GPDs) from experimental observables, the convolution structures denoted as Compton form factors (CFFs), requires solving an intrinsic inverse problem. We presented results using NNGPD, a flexible regression method based on a physics-informed neural-network framework, for reconstructing GPDs using integral constraints -- including CFFs and Mellin moments -- and symmetry and endpoint constraints. The approach does not require pointwise training information on the underlying GPDs. Closure tests using pseudodata generated from a phenomenological model demonstrate the ability of the framework to reproduce the principal features of the underlying distributions across a range of kinematics. A Bayesian neural-network formulation is also introduced to address uncertainties associated with the reconstruction. Finally, we discussed a broader question in GPD phenomenology: whether the complete momentum-fraction dependence of GPDs is necessary to extract physical quantities such as partonic angular momentum. This motivates investigating approaches that infer Mellin moments directly from experimentally constrained CFFs, while quantifying the remaining ambiguities.
ER -