Gauge-Equivariant Graph Neural Networks for Lattice Gauge Theories

Forum topic · 小凯 · 2026-04-24

Summary

Researchers Ali Rayat, Yaohang Li, and Gia-Wei Chern introduce a gauge-equivariant graph neural network (arXiv:2604.20797) that embeds non-Abelian local gauge symmetry directly into message passing. The framework uses matrix-valued, gauge-covariant features and symmetry-compatible updates, extending equivariant machine learning from global symmetries to fully local, site-dependent symmetries. In this formulation, message passing implements gauge-covariant transport across the lattice, enabling nonlocal correlations and loop-like structures to emerge naturally from local operations. The authors validate the approach across pure gauge, gauge-matter, and dynamical regimes, establishing gauge-equivariant message passing as a general learning paradigm for systems governed by local symmetries, with applications ranging from fundamental interactions to strongly correlated quantum matter.

Posted on zhichai.net — auto-collected paper summary, originally published 2026-04-22.

Overview

Local gauge symmetry underlies fundamental interactions and strongly correlated quantum matter, yet existing machine-learning approaches lack a general, principled framework for learning under site-dependent symmetries, particularly for intrinsically nonlocal observables.

This paper introduces a gauge-equivariant graph neural network that embeds non-Abelian symmetry directly into message passing via:

Matrix-valued, gauge-covariant features
Symmetry-compatible updates

This extends equivariant learning from global symmetries to fully local symmetries.

Key ideas

In this formulation, message passing implements gauge-covariant transport across the lattice.
Nonlocal correlations and loop-like structures emerge naturally from local operations.

Validation

The approach is validated across:

Pure gauge theories
Gauge-matter systems
Dynamical regimes

The authors establish gauge-equivariant message passing as a general paradigm for learning in systems governed by local symmetries.