arxiv: 2604.20797 · v1 · submitted 2026-04-22 · ❄️ cond-mat.str-el · cs.LG· hep-lat

Recognition: unknown

Gauge-Equivariant Graph Neural Networks for Lattice Gauge Theories

Ali Rayat , Yaohang Li , Gia-Wei Chern

Authors on Pith no claims yet

Pith reviewed 2026-05-09 23:20 UTC · model grok-4.3

classification ❄️ cond-mat.str-el cs.LGhep-lat

keywords gauge-equivariant neural networkslattice gauge theoriesgraph neural networkslocal gauge symmetrynon-Abelian gauge fieldsequivariant machine learningquantum matter simulation

0 comments

The pith

Gauge-equivariant graph neural networks embed non-Abelian local symmetry into message passing on lattices via matrix-valued covariant features.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a graph neural network architecture designed to respect site-dependent gauge symmetries that appear in lattice gauge theories and strongly correlated quantum matter. It achieves this by replacing scalar features with matrix-valued ones that transform covariantly under local gauge transformations and by defining update rules that preserve this covariance during message passing. This setup turns the network's local operations into gauge-covariant transport, so that loop structures and nonlocal correlations arise automatically rather than being imposed by hand. The authors demonstrate the method on pure gauge fields, gauge-matter systems, and dynamical regimes, positioning it as a general framework that extends equivariant machine learning from global to fully local symmetries.

Core claim

We introduce a gauge-equivariant graph neural network that embeds non-Abelian symmetry directly into message passing via matrix-valued, gauge-covariant features and symmetry-compatible updates, extending equivariant learning from global to fully local symmetries. In this formulation, message passing implements gauge-covariant transport across the lattice, allowing nonlocal correlations and loop-like structures to emerge naturally from local operations. We validate the approach across pure gauge, gauge-matter, and dynamical regimes, establishing gauge-equivariant message passing as a general paradigm for learning in systems governed by local symmetry.

What carries the argument

Matrix-valued gauge-covariant features together with symmetry-compatible update rules that perform gauge-covariant transport during message passing on the lattice graph.

If this is right

Nonlocal observables such as Wilson loops arise automatically from repeated local message-passing steps.
The same architecture applies without modification to pure gauge theories, systems with matter fields, and dynamical gauge theories.
Training can proceed with standard loss functions because gauge invariance is built into the feature representation rather than enforced through penalties.
The method extends the reach of equivariant neural networks from global symmetries to the local symmetries required by fundamental interactions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The approach may reduce the need for post-processing or projection steps that are common when symmetry is only approximate in learned models.
It opens a route to learning effective theories directly on the lattice while preserving the underlying gauge structure exactly.
Similar matrix-valued covariant features could be adapted to other locally symmetric systems such as spin liquids or topological phases.

Load-bearing premise

The proposed matrix-valued features and update rules correctly implement gauge covariance for arbitrary non-Abelian groups and allow stable training on realistic lattice sizes without extra regularization or symmetry-breaking effects.

What would settle it

A concrete numerical check on a small lattice where a trained network's output for a Wilson loop or plaquette observable changes under a local gauge transformation that should leave it invariant.

Figures

Figures reproduced from arXiv: 2604.20797 by Ali Rayat, Gia-Wei Chern, Yaohang Li.

**Figure 2.** Figure 2: FIG. 2. Benchmark of the gauge-equivariant GNN for the [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: FIG. 3. Benchmark of the gauge-equivariant GNN for non-Abelian gauge–matter systems on a 20 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

**Figure 4.** Figure 4: FIG. 4. Benchmark of gauge-covariant force prediction [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗

**Figure 5.** Figure 5: FIG. 5. Dynamical benchmark of the GNN force-field for [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

read the original abstract

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. Here we introduce a gauge-equivariant graph neural network that embeds non-Abelian symmetry directly into message passing via matrix-valued, gauge-covariant features and symmetry-compatible updates, extending equivariant learning from global to fully local symmetries. In this formulation, message passing implements gauge-covariant transport across the lattice, allowing nonlocal correlations and loop-like structures to emerge naturally from local operations. We validate the approach across pure gauge, gauge-matter, and dynamical regimes, establishing gauge-equivariant message passing as a general paradigm for learning in systems governed by local symmetry.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper proposes matrix-valued gauge-covariant features in GNN message passing to handle local non-Abelian symmetries on lattices, but the abstract leaves the exact rules and validation thin.

read the letter

The main thing to know is that this work tries to extend equivariant graph networks from global symmetries to local gauge symmetries by introducing matrix-valued features that stay covariant under site-dependent transformations during message passing. That lets the model do gauge-covariant transport across links so nonlocal correlations can build up from local steps without explicit symmetry breaking. The abstract frames this as a general paradigm for pure gauge, gauge-matter, and dynamical lattice systems, which is a reasonable target given how central local gauge invariance is in condensed matter and particle physics models. What they do well is identify the gap clearly: standard lattice ML either ignores gauge symmetry or sticks to global equivariance, and this direction aims to fix that by baking the link variables and representation transport into the updates. The claim that loop-like structures emerge naturally from the local operations is a nice conceptual point if it holds. The soft spots are exactly where the stress-test note lands. The abstract asserts the construction and says they validated it, but supplies no update equations, no derivation showing how the matrix multiplications commute with arbitrary local group actions for non-Abelian cases like SU(2), and no numerical checks that a transformed input produces the correspondingly transformed output. Without those, it is hard to tell whether covariance is preserved at every layer or whether training stays stable on realistic lattice sizes. If the full manuscript has the explicit rules, an equivariance proof, and concrete metrics, that would address the concern; based on what is here the evidence is still preliminary. This paper is for people working at the ML-physics boundary who need architectures that respect local symmetries. A reader who wants to experiment with new symmetry-aware layers would get value from the high-level idea and could try to implement the missing pieces. It deserves a serious referee because the problem matters and the architectural direction is coherent, even though the current write-up needs more detail on the mechanics and results before it can be fully assessed.

Referee Report

1 major / 2 minor

Summary. The manuscript introduces a gauge-equivariant graph neural network architecture for lattice gauge theories. It embeds non-Abelian local gauge symmetries into the message-passing framework by using matrix-valued, gauge-covariant features together with symmetry-compatible update rules, thereby extending global-equivariant networks to fully local symmetries. The approach is claimed to allow natural emergence of nonlocal correlations and loop structures from local operations, and is validated on pure-gauge, gauge-matter, and dynamical regimes.

Significance. If the construction preserves exact gauge covariance under arbitrary local non-Abelian transformations, the work would constitute a meaningful extension of equivariant learning to systems with site-dependent symmetries. This could improve modeling of intrinsically nonlocal observables in lattice gauge theories without auxiliary regularization or explicit symmetry breaking, building on existing global-equivariant GNNs while addressing a gap in handling local gauge structure.

major comments (1)

[§3] §3 (Gauge-Equivariant Message Passing): the update rules for matrix-valued features are presented via matrix multiplications and parallel transport by link variables, but an explicit verification that these rules commute with arbitrary local SU(2) or SU(3) transformations (i.e., that transformed inputs produce correspondingly transformed outputs after one full message-passing step) is not supplied. For non-Abelian groups this is load-bearing, as non-commutativity of neighboring transformations can break covariance even when it holds for U(1).

minor comments (2)

[Figure 2] Figure 2 and the associated caption: the visualization of feature transport would be clearer if the gauge transformation applied to the input were shown side-by-side with the output transformation.
[§4] The abstract and §4 would benefit from a short statement of the largest lattice volume and group (e.g., SU(3)) on which stable training was demonstrated.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thorough review and positive assessment of our manuscript. We address the single major comment below and agree to incorporate the requested verification in the revised version.

read point-by-point responses

Referee: [§3] §3 (Gauge-Equivariant Message Passing): the update rules for matrix-valued features are presented via matrix multiplications and parallel transport by link variables, but an explicit verification that these rules commute with arbitrary local SU(2) or SU(3) transformations (i.e., that transformed inputs produce correspondingly transformed outputs after one full message-passing step) is not supplied. For non-Abelian groups this is load-bearing, as non-commutativity of neighboring transformations can break covariance even when it holds for U(1).

Authors: We thank the referee for this observation. The message-passing rules are constructed so that parallel transport by the link variables U_{ij} (which transform as U'_{ij} = g_i U_{ij} g_j^†) compensates for local gauge transformations, ensuring covariance of the matrix-valued features by design. Nevertheless, we agree that an explicit, step-by-step verification for non-Abelian groups—accounting for the non-commutativity of neighboring transformations—was omitted and is necessary to confirm that a full message-passing step maps transformed inputs to correspondingly transformed outputs. In the revised manuscript we will add this verification directly in §3 (or as a short appendix), explicitly showing the transformation properties of the message function and the update rule under arbitrary local SU(2) and SU(3) transformations. revision: yes

Circularity Check

0 steps flagged

New architectural construction for gauge-equivariant GNNs is self-contained with no reduction to inputs

full rationale

The paper proposes a novel gauge-equivariant graph neural network that incorporates matrix-valued gauge-covariant features and symmetry-compatible updates to handle local non-Abelian symmetries on lattices. This is presented as an original extension of equivariant message passing, with the central claim being the definition and implementation of the architecture itself rather than any fitted prediction or derived quantity. No equations or steps in the provided abstract or description reduce a result to a parameter chosen from the same data, a self-citation chain, or an ansatz smuggled in by prior work. Validation across pure gauge, gauge-matter, and dynamical regimes is described as empirical testing of the new model, not a self-referential fit. The derivation chain consists of the architectural definitions, which stand independently of their own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The architecture rests on the assumption that a matrix-valued representation can be made to transform covariantly under arbitrary local gauge transformations while remaining compatible with standard graph-neural-network aggregation; no new particles or forces are postulated.

axioms (1)

domain assumption Gauge transformations act independently at each lattice site and the network features must transform in the appropriate representation of the gauge group.
Invoked in the description of embedding non-Abelian symmetry directly into message passing.

invented entities (1)

Matrix-valued gauge-covariant features no independent evidence
purpose: Carry local symmetry information through message passing while transforming correctly under gauge changes.
New representational primitive introduced to extend global equivariance to local gauge symmetry.

pith-pipeline@v0.9.0 · 5422 in / 1335 out tokens · 37114 ms · 2026-05-09T23:20:04.129230+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

68 extracted references · 8 canonical work pages · 1 internal anchor

[1]

Carleo, I

G. Carleo, I. Cirac, K. Cranmer, L. Daudet, M. Schuld, N. Tishby, L. Vogt-Maranto, and L. Zdeborov´ a, Machine learning and the physical sciences, Rev. Mod. Phys.91, 045002 (2019)

2019
[2]

Das Sarma, D.-L

S. Das Sarma, D.-L. Deng, and L.-M. Duan, Machine learning meets quantum physics, Physics Today72, 48 (2019)

2019
[3]

Bedolla, L

E. Bedolla, L. C. Padierna, and R. Casta˜ neda-Priego, Machine learning for condensed matter physics, Journal of Physics: Condensed Matter33, 053001 (2020)

2020
[4]

M. D. Schwartz, Modern machine learning and particle physics, Harvard Data Science Review3, 2 (2021). 10

2021
[5]

G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang, Physics-informed machine learn- ing, Nature Reviews Physics3, 422 (2021)

2021
[6]

Boehnlein, M

A. Boehnlein, M. Diefenthaler, N. Sato, M. Schram, V. Ziegler, C. Fanelli, M. Hjorth-Jensen, T. Horn, M. P. Kuchera, D. Lee, W. Nazarewicz, P. Ostroumov, K. Orginos, A. Poon, X.-N. Wang, A. Scheinker, M. S. Smith, and L.-G. Pang, Colloquium: Machine learning in nuclear physics, Rev. Mod. Phys.94, 031003 (2022)

2022
[7]

Karagiorgi, G

G. Karagiorgi, G. Kasieczka, S. Kravitz, B. Nachman, and D. Shih, Machine learning in the search for new fun- damental physics, Nature Reviews Physics4, 399 (2022)

2022
[8]

Ramprasad, R

R. Ramprasad, R. Batra, G. Pilania, A. Mannodi- Kanakkithodi, and C. Kim, Machine learning in mate- rials informatics: recent applications and prospects, npj Computational Materials3, 54 (2017)

2017
[9]

K. T. Butler, D. W. Davies, H. Cartwright, O. Isayev, and A. Walsh, Machine learning for molecular and ma- terials science, Nature559, 547 (2018)

2018
[10]

Schmidt, M

J. Schmidt, M. R. G. Marques, S. Botti, and M. A. L. Marques, Recent advances and applications of machine learning in solid-state materials science, npj Computa- tional Materials5, 83 (2019)

2019
[11]

Bapst, T

V. Bapst, T. Keck, A. Grabska-Barwi´ nska, C. Donner, E. D. Cubuk, S. S. Schoenholz, A. Obika, A. W. R. Nel- son, T. Back, D. Hassabis, and P. Kohli, Unveiling the predictive power of static structure in glassy systems, Na- ture Physics16, 448 (2020)

2020
[12]

T. S. Cohen and M. Welling, Group equivariant con- volutional networks, inProceedings of the 33rd Interna- tional Conference on Machine Learning, Proceedings of Machine Learning Research, Vol. 48 (PMLR, 2016) pp. 2990–2999, arXiv:1602.07576 [cs.LG]

work page Pith review arXiv 2016
[13]

T. S. Cohen, M. Geiger, M. Weiler, and M. Welling, A general theory of equivariant cnns on homogeneous spaces, inAdvances in Neural Information Processing Systems, Vol. 31 (2018) arXiv:1811.02017 [cs.LG]

work page arXiv 2018
[14]

3D Steerable CNNs: Learning Rotationally Equivariant Features in Volumetric Data

M. Weiler, M. Geiger, M. Welling, W. Boomsma, and T. S. Cohen, 3d steerable cnns: Learning rotationally equivariant features in volumetric data, inAdvances in Neural Information Processing Systems, Vol. 31 (Curran Associates, Inc., 2018) arXiv:1807.02547 [cs.LG]

work page Pith review arXiv 2018
[15]

Kondor, The principles behind equivariant neural net- works for physics and chemistry, Proceedings of the Na- tional Academy of Sciences122, e2415656122 (2025)

R. Kondor, The principles behind equivariant neural net- works for physics and chemistry, Proceedings of the Na- tional Academy of Sciences122, e2415656122 (2025)

2025
[16]

Batzner, A

S. Batzner, A. Musaelian, L. Sun, M. Geiger, J. P. Mailoa, M. Kornbluth, N. Molinari, T. E. Smidt, and B. Kozinsky, E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials, Nature Communications13, 2453 (2022)

2022
[17]

Kaba and S

S.-O. Kaba and S. Ravanbakhsh, Equivariant networks for crystal structures, inAdvances in Neural Informa- tion Processing Systems, Vol. 35 (2022) pp. 300–314, arXiv:2211.15420 [cs.LG]

work page arXiv 2022
[18]

Musaelian, S

A. Musaelian, S. Batzner, A. Johansson, L. Sun, C. J. Owen, M. Kornbluth, and B. Kozinsky, Learning local equivariant representations for large-scale atomistic dy- namics, Nature Communications14, 579 (2023)

2023
[19]

X. Gong, H. Li, N. Zou, R. Xu, W. Duan, and Y. Xu, General framework for e(3)-equivariant neural network representation of density functional theory hamiltonian, Nature Communications14, 2848 (2023)

2023
[20]

Batatia, S

I. Batatia, S. Batzner, D. P. Kov´ acs, A. Musaelian, G. N. C. Simm, R. Drautz, C. Ortner, B. Kozinsky, and G. Cs´ anyi, The design space of e(3)-equivariant atom- centred interatomic potentials, Nature Machine Intelli- gence7, 56 (2025)

2025
[21]

Z. Yang, X. Wang, Y. Li, Q. Lv, C. Y.-C. Chen, and L. Shen, Efficient equivariant model for machine learning interatomic potentials, npj Computational Materials11, 49 (2025)

2025
[22]

Scarselli, M

F. Scarselli, M. Gori, A. C. Tsoi, M. Hagenbuchner, and G. Monfardini, The graph neural network model, IEEE Transactions on Neural Networks20, 61 (2009)

2009
[23]

Neural message passing for quantum chemistry

J. Gilmer, S. S. Schoenholz, P. F. Riley, O. Vinyals, and G. E. Dahl, Neural message passing for quantum chem- istry, inProceedings of the 34th International Conference on Machine Learning, Vol. 70 (PMLR, 2017) pp. 1263– 1272, arXiv:1704.01212

work page arXiv 2017
[24]

W. L. Hamilton, R. Ying, and J. Leskovec, Inductive representation learning on large graphs, inAdvances in Neural Information Processing Systems, Vol. 30 (2017) arXiv:1706.02216

work page Pith review arXiv 2017
[25]

K. Xu, W. Hu, J. Leskovec, and S. Jegelka, How powerful are graph neural networks?, inInternational Conference on Learning Representations (ICLR)(2019) arXiv:1810.00826

work page internal anchor Pith review arXiv 2019
[26]

Invariant and equivariant graph networks.arXiv:1812.09902,

H. Maron, H. Ben-Hamu, N. Shamir, and Y. Lipman, In- variant and equivariant graph networks, inInternational Conference on Learning Representations (ICLR)(2019) arXiv:1812.09902

work page arXiv 2019
[27]

K. G. Wilson, Confinement of quarks, Physical Review D10, 2445 (1974)

1974
[28]

J. B. Kogut and L. Susskind, Hamiltonian formulation of wilson’s lattice gauge theories, Physical Review D11, 395 (1975)

1975
[29]

Fradkin and S

E. Fradkin and S. H. Shenker, Phase diagrams of lattice gauge theories with higgs fields, Physical Review D19, 3682 (1979)

1979
[30]

J. B. Kogut, An introduction to lattice gauge theory and spin systems, Reviews of Modern Physics51, 659 (1979)

1979
[31]

Creutz,Quarks, Gluons and Lattices(Cambridge Uni- versity Press, 1983)

M. Creutz,Quarks, Gluons and Lattices(Cambridge Uni- versity Press, 1983)

1983
[32]

H. J. Rothe,Lattice Gauge Theories: An Introduction, 4th ed. (World Scientific, 2012)

2012
[33]

Wen,Quantum Field Theory of Many-Body Sys- tems(Oxford University Press, 2004)

X.-G. Wen,Quantum Field Theory of Many-Body Sys- tems(Oxford University Press, 2004)

2004
[34]

P. A. Lee, N. Nagaosa, and X.-G. Wen, Doping a mott in- sulator: Physics of high-temperature superconductivity, Reviews of Modern Physics78, 17 (2006)

2006
[35]

Baskaran, Z

G. Baskaran, Z. Zou, and P. W. Anderson, The resonat- ing valence bond state and high-t c superconductivity: A mean field theory, Solid State Communications63, 973 (1987)

1987
[36]

Affleck and J

I. Affleck and J. B. Marston, Large-nlimit of the heisenberg-hubbard model: Implications for high-t c su- perconductors, Physical Review B37, 3774 (1988)

1988
[37]

Wen, Mean-field theory of spin-liquid states with finite energy gap and topological orders, Physical Review B44, 2664 (1991)

X.-G. Wen, Mean-field theory of spin-liquid states with finite energy gap and topological orders, Physical Review B44, 2664 (1991)

1991
[38]

Wen, Quantum orders and symmetric spin liquids, Physical Review B65, 165113 (2002)

X.-G. Wen, Quantum orders and symmetric spin liquids, Physical Review B65, 165113 (2002)

2002
[39]

Savary and L

L. Savary and L. Balents, Quantum spin liquids: A re- view, Reports on Progress in Physics80, 016502 (2017)

2017
[40]

Y. Zhou, K. Kanoda, and T.-K. Ng, Quantum spin liquid states, Reviews of Modern Physics89, 025003 (2017)

2017
[41]

Chandrasekharan and U.-J

S. Chandrasekharan and U.-J. Wiese, Quantum link models: A discrete approach to gauge theories, Nuclear 11 Physics B492, 455 (1997)

1997
[42]

Brower, S

R. Brower, S. Chandrasekharan, and U.-J. Wiese, Qcd as a quantum link model, Phys. Rev. D60, 094502 (1999)

1999
[43]

Wiese, Ultracold quantum gases and lattice sys- tems: quantum simulation of lattice gauge theories, An- nalen der Physik525, 777 (2013)

U.-J. Wiese, Ultracold quantum gases and lattice sys- tems: quantum simulation of lattice gauge theories, An- nalen der Physik525, 777 (2013)

2013
[44]

Banerjee, M

D. Banerjee, M. Dalmonte, M. M¨ uller, E. Rico, P. Ste- bler, U.-J. Wiese, and P. Zoller, Atomic quantum simula- tion of dynamical gauge fields coupled to fermionic mat- ter: From string breaking to evolution after a quench, Phys. Rev. Lett.109, 175302 (2012)

2012
[45]

Zohar, J

E. Zohar, J. I. Cirac, and B. Reznik, Quantum simula- tions of gauge theories with ultracold atoms: Local gauge invariance from angular-momentum conservation, Phys. Rev. A88, 023617 (2013)

2013
[46]

Tagliacozzo, A

L. Tagliacozzo, A. Celi, P. Orland, M. W. Mitchell, and M. Lewenstein, Simulation of non-abelian gauge theo- ries with optical lattices, Nature Communications4, 2615 (2013)

2013
[47]

Zohar, J

E. Zohar, J. I. Cirac, and B. Reznik, Quantum simula- tions of lattice gauge theories using ultracold atoms in op- tical lattices, Reports on Progress in Physics79, 014401 (2015)

2015
[48]

Cheng and H

Y. Cheng and H. Zhai, Emergent u(1) lattice gauge the- ory in rydberg atom arrays, Nature Reviews Physics6, 566 (2024)

2024
[49]

F. M. Surace, P. P. Mazza, G. Giudici, A. Lerose, A. Gambassi, and M. Dalmonte, Lattice gauge theories and string dynamics in rydberg atom quantum simula- tors, Phys. Rev. X10, 021041 (2020)

2020
[50]

Zhou, G.-X

Z.-Y. Zhou, G.-X. Su, J. C. Halimeh, R. Ott, H. Sun, P. Hauke, B. Yang, Z.-S. Yuan, J. Berges, and J.-W. Pan, Thermalization dynamics of a gauge theory on a quantum simulator, Science377, 311 (2022)

2022
[51]

Cranmer, G

K. Cranmer, G. Kanwar, S. Racani` ere, D. J. Rezende, and P. E. Shanahan, Advances in machine-learning-based sampling motivated by lattice quantum chromodynam- ics, Nature Reviews Physics5, 526 (2023)

2023
[52]

Tomiya, Machine learning for lattice qcd, Journal of the Physical Society of Japan94, 031006 (2025)

A. Tomiya, Machine learning for lattice qcd, Journal of the Physical Society of Japan94, 031006 (2025)

2025
[53]

Favoni, A

M. Favoni, A. Ipp, D. I. M¨ uller, and D. Schuh, Lattice gauge equivariant convolutional neural networks, Phys. Rev. Lett.128, 032003 (2022)

2022
[54]

Bulusu, M

S. Bulusu, M. Favoni, A. Ipp, D. I. M¨ uller, and D. Schuh, Generalization capabilities of translationally equivariant neural networks, Phys. Rev. D104, 074504 (2021)

2021
[55]

Lehner and T

C. Lehner and T. Wettig, Gauge-equivariant neural net- works as preconditioners in lattice qcd, Phys. Rev. D108, 034503 (2023)

2023
[56]

Lehner and T

C. Lehner and T. Wettig, Gauge-equivariant pooling lay- ers for preconditioners in lattice qcd, Phys. Rev. D110, 034517 (2024)

2024
[57]

A. Apte, C. C´ ordova, T.-C. Huang, and A. Ashmore, Deep learning lattice gauge theories, Phys. Rev. B110, 165133 (2024)

2024
[58]

T. S. Cohen, M. Weiler, B. Kicanaoglu, and M. Welling, Gauge equivariant convolutional networks and the icosa- hedral cnn, inProceedings of the 36th International Con- ference on Machine Learning (ICML), Proceedings of Machine Learning Research, Vol. 97 (2019) pp. 1321– 1330

2019
[59]

Boyda, G

D. Boyda, G. Kanwar, S. Racani` ere, D. J. Rezende, M. S. Albergo, K. Cranmer, D. C. Hackett, and P. E. Shana- han, Sampling using SU(n) gauge equivariant flows, Phys. Rev. D103, 074504 (2021)

2021
[60]

Nagai and A

Y. Nagai and A. Tomiya, Gauge covariant neural network for quarks and gluons, Phys. Rev. D111, 074501 (2025)

2025
[61]

M. S. Albergo, G. Kanwar, and P. E. Shanahan, Flow- based generative models for markov chain monte carlo in lattice field theory, Phys. Rev. D100, 034515 (2019)

2019
[62]

Kanwar, M

G. Kanwar, M. S. Albergo, D. Boyda, K. Cranmer, D. C. Hackett, S. Racani` ere, D. J. Rezende, and P. E. Shana- han, Equivariant flow-based sampling for lattice gauge theory, Phys. Rev. Lett.125, 121601 (2020)

2020
[63]

Abbott, M

R. Abbott, M. S. Albergo, D. Boyda, K. Cranmer, D. C. Hackett, G. Kanwar, S. Racani` ere, D. J. Rezende, F. Romero-L´ opez, P. E. Shanahan, B. Tian, and J. M. Urban, Gauge-equivariant flow models for sampling in lattice field theories with pseudofermions, Phys. Rev. D 106, 074506 (2022)

2022
[64]

Nagai, A

Y. Nagai, A. Tanaka, and A. Tomiya, Self-learning monte carlo for non-abelian gauge theory with dynam- ical fermions, Phys. Rev. D107, 054501 (2023)

2023
[65]

Nagai and A

Y. Nagai and A. Tomiya, Self-learning monte carlo with equivariant transformer, Journal of the Physical Society of Japan93, 114007 (2024)

2024
[66]

D. Luo, G. Carleo, B. K. Clark, and J. Stokes, Gauge equivariant neural networks for quantum lattice gauge theories, Phys. Rev. Lett.127, 276402 (2021)

2021
[67]

D. Luo, Z. Chen, K. Hu, Z. Zhao, V. M. Hur, and B. K. Clark, Gauge-invariant and anyonic-symmetric au- toregressive neural network for quantum lattice models, Phys. Rev. Res.5, 013216 (2023)

2023
[68]

Cheng and D.-B

L.-X. Cheng and D.-B. Zhang, Variational quantum sim- ulation of ground states and thermal states for lattice gauge theory with multi-objective optimization, Physics Letters A546, 130516 (2025)

2025