Physics-Informed Neural Network Approach to Quark–Antiquark Color Flux Tube

Wei Kou, Xiaoxuan Lin, Bing’ang Guo, Xurong Chen · 2025

Read Physics-Informed Neural Network Approach to Quark–Antiquark Color Flux Tube by Wei Kou, Xiaoxuan Lin, Bing’ang Guo, Xurong Chen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-03513.txt.

Introduction

Wei Kou, Xiaoxuan Lin, Bing'ang Guo, and Xurong Chen map the glue tube between a quark and an antiquark with a smart network. Their paper, from twenty twenty-five, uses physics-informed neural networks. Such networks bake the governing equations into their training. The equations here come from type-two superconductivity, an analogy for the strong force field.

Lattice simulations supply the raw field data. The network works backward to extract the equation's parameters. Reformulated math lets it respect boundary rules, with a fractional extension. Rebuilt field maps are scored against the data. From them come the string tension and the tube's width. Both numbers speak to how quarks stay confined.

Catalog record

CREATORWei Kou, Xiaoxuan Lin, Bing’ang Guo, Xurong Chen
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-03513

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Physics-Informed Neural Network Approach to Quark–Antiquark Color Flux Tube Wei Kou, Xiaoxuan Lin, Bing’ang Guo, Xurong Chen Abstract We introduce a physics-informed neural network (PINNs) framework for modelling the spatial distribution of chromodynamic fields induced by quark–antiquark pairs, based on lattice Monte Carlo simulations. In contrast to conventional neural networks, PINNs incorporate physical laws—expressed here as differential equations governing type-II superconductivity—directly into the training objective. By embedding these equations into the loss function, we guide the network to learn physically consistent solutions. Adopting an inverse problem approach, we extract the parameters of the superconducting equations from lattice QCD data and subsequently solve them. To accommodate physical boundary conditions, we recast the system into an integro-differential form and extend the analysis within the fractional PINNs framework. The accuracy of the reconstructed field distribution is assessed via relative L 2 L_{2} -error norms. We further extract physical observables such as the string tension and the mean width of the flux tube, offering quantitative insight into the confinement mechanism. This method enables the reconstruction of colour field profiles as functions of quark–antiquark separation without recourse to predefined parametric models. Our results illuminate aspects of the dual Meissner effect and highlight the promise of data-driven strategies in addressing non-perturbative challenges in quantum chromodynamics. I Introduction Quantum Chromodynamics (QCD), as the fundamental theory describing the strong interaction between quarks and gluons, has given rise to a comprehensive framework for hadronic physics through the study of its essential features—namely, asymptotic freedom and spontaneous chiral symmetry breaking. Due to confinement, the constituent degrees of freedom of hadrons—quarks and gluons—cannot be isolated, and confinement itself is an intrinsically non-perturbative phenomenon of QCD at low energies. Consequently, the theoretical treatment of QCD in this regime often relies heavily on phenomenological inputs. Fortunately, QCD is an experimentally grounded theory: extensive data have been accumulated over decades from large-scale collider and fixed-target experiments involving strong interactions. Moreover, advancements in computational hardware and algorithms have enabled lattice Monte Carlo simulations to provide initial yet valuable insights and measurements concerning non-perturbative phenomena in QCD. In high-energy physics, the problem of confinement has remained unresolved for a long time, particularly within the framework of quantum field theory, where multiple theoretical approaches have been proposed. The dual superconductivity picture is one such hypothesis for the confinement mechanism Nambu 1974 . In this scenario, the vacuum may undergo magnetic monopole condensation, resulting in the formation of confining Abelian electric flux tubes that connect quarks and antiquarks. These flux tubes are the dual analogs of Abrikosov-Nielsen-Olesen (ANO) vortices in type-II superconductors Abrikosov 1957a ; Abrikosov 1957b ; Nielsen and Olesen 1973 . Detailed discussions on the dual superconductivity mechanism can be found in the review Kondo et al. 2015 . Lattice QCD simulations have long been employed as one of the primary approaches for obtaining the chromoelectric field or the profile of the flux tube Singh et al. 1993 ; Schilling et al. 1999 ; Chernodub et al. 2001 ; Chernodub et al. 2005 ; Cea et al. 2012 . It is worth noting that these studies largely rely on the choice of a specific gauge condition Suzuki et al. 2009 , namely the so-called “maximal Abelian projection”. Among numerous lattice simulation studies, the Clem ansatz Clem 1975 is frequently employed to describe the short-distance behavior of flux tubes. The Clem ansatz is intended to hypothesize or provide an approximate analytical expression for the transverse profile of the flux tube. More specifically, it introduces a simplified ordinary differential equation (ODE) that captures the spatial variation of the magnetic field or magnetic potential within the flux tube. This ODE is governed by the Ginzburg–Landau parameter, and under specific boundary conditions, its solutions are expressed as combinations of Bessel functions. Such solutions are often used to fit the chromoelectric field distributions obtained from lattice simulations. Many researchers aspire to address such problems using a forward-thinking approach—for instance, by directly solving the second-type superconducting ODE associated with the Clem ansatz, and subsequently analyzing the parameterization of the analytical solution based on lattice simulation data. While studying the solutions of such equations is inherently meaningful, the high complexity of many problems often precludes the availability of explicit analytical expressions, thus hindering forward approaches. Conversely, in QCD research, one frequently encounters inverse problems—namely, deducing physical laws from experimental observations. These problems are inherently challenging and demand sophisticated solution strategies. Various methods exist for inferring model parameters from experimental or simulated data, and recent advances in data-driven approaches have introduced increasingly modern techniques. Machine learning (ML), a modern branch of artificial intelligence (AI), is gaining traction as a powerful tool in QCD studies and applications Larkoski et al. 2020 ; Guest et al. 2018 ; Radovic et al. 2018 ; Albertsson et al. 2018 ; Carleo et al. 2019 ; Bourilkov 2020 ; Schwartz 2021 ; Karagiorgi et al. 2021 ; Boehnlein et al. 2022 ; Shanahan et al. 2022 ; Yang et al. 2023 ; Li et al. 2023 ; He et al. 2023 ; Zhou et al. 2024 ; Zhou et al. 2023 ; Pang 2024 ; Ma et al. 2023 ; Luo et al. 2024 ; Chen et al. 2024 ; Omana Kuttan et al. 2023 ; Shi et al. 2022a ; Shi et al. 2022b ; Shi et al. 2022c ; Mansouri et al. 2024 ; Chen and Huang 2024a ; Chen and Huang 2024b ; Wang et al. 2023 (see also the review Aarts et al. 2025 ). ML provides a data-driven framework for uncovering the behavior of complex systems. In particular, by training deep neural networks (DNNs) on empirical data, one can infer model parameters and associated physical properties. Furthermore, embedding physical knowledge—such as PDEs or ODEs and their symmetries—into DNNs can enhance the fidelity of the learning outcomes. This class of methods is known as physics-informed neural networks (PINNs) Raissi et al. 2019 . Physical equations are often employed as prior conditions for describing a system, with their associated laws and symmetries providing constraints that ensure the solutions obtained during neural network training remain physically consistent. PINNs offer a framework that leverages automatic differentiation (AD) Baydin et al. 2015 to handle PDEs or ODEs. By incorporating auxiliary loss terms corresponding to the governing equations, PINNs enable supervised learning constrained by physical laws, allowing the solution of both linear and nonlinear problems, such as the Navier–Stokes equation in fluid dynamics Amalinadhi et al. 2022 ; Eivazi et al. 2022 . In this work, we introduce the PINNs framework into the study of chromoelectric field distributions. By analyzing color flux tube data generated from lattice simulations of quark–antiquark excitations at varying distances Baker et al. 2020 , we aim to solve the simplified Ginzburg–Landau equation embedded in the Clem ansatz Clem 1975 . Without imposing explicit boundary conditions, and following a purely data-driven approach, we successfully infer both the field profiles and the governing parameters corresponding to different quark separations. The PINNs functionality implemented in the DeepXDE Lu et al. 2021 software package is used to perform the analysis. The structure of this paper is organized as follows: in Sec. II , we introduce the chromoelectric field distributions and outline relevant details of the lattice simulations. Sec. III

Read the complete text