Pith. sign in

REVIEW 3 major objections 5 minor 3 cited by

Magnetic correlations in the $SU(3)$ triangular-lattice $t$-$J$ model at finite doping

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The SU(3) triangular-lattice t-J model binds two doped holes more strongly than the SU(2) square-lattice model at the same t/J, with magnetic correlations that closely track the square-lattice case.

desk verdict A solid variational advance on finite-doping SU(3) t-J physics, with the headline binding-energy number resting on a variational bias that needs shoring up. read the letter →

arxiv 2506.01915 v2 pith:I43TPMOL submitted 2025-06-02 cond-mat.quant-gas cond-mat.dis-nncond-mat.str-el

Annika Böhler , Fabian Grusdt , Annabelle Bohrdt This is my paper
classification cond-mat.quant-gascond-mat.dis-nncond-mat.str-el
keywords SU(3)Fermi-Hubbardmodeltriangularlatticet-Jneuralquantumstateshiddenfermiondeterminantmagneticpolaronsholepairinggeometricstrings
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper studies the strong-coupling limit of the SU(3) Fermi-Hubbard model on the triangular lattice at finite hole doping, using a three-flavor extension of Gutzwiller-projected hidden fermion determinant states (G-HFDS). The central claim is that two doped holes bind, with a negative binding energy EB = E2h − 2E1h whose magnitude is larger than in the analogous SU(2) square-lattice t-J model across t/J from 1 to 6. The paper also reports that magnetic correlations, including two- and three-point spin and spin-hole correlations built from the SU(3) Cartan generators, evolve with doping in a way strikingly similar to the SU(2) square lattice, despite the larger local Hilbert space and connectivity. This matters because ultracold alkaline-earth atoms can realize SU(N)-symmetric Hubbard models, and a theoretical handle on their doped regime is currently scarce.

What carries the argument

The central object is the three-flavor G-HFDS ansatz, a neural-network-based variational wavefunction in which the physical amplitude for each Fock configuration is a Slater determinant det M(s) built from hidden fermions, with the configuration-dependent part generated by a feed-forward neural network and symmetrized over S3 spin permutations. This ansatz is used to compute energies and correlation functions on fully periodic 6x6 and 9x9 tori. The binding energy EB = E2h − 2E1h is the key diagnostic, and the geometric-string mapping to a Bethe lattice with coordination number z, leading to the scaling EB = (2α1 − $2^{{1/3}}$α2) $t^{{1/3}}$$J^{{2/3}}$, is the mechanism invoked to explain the enhanced pairing.

What would settle it

Compute EB = E2h − 2E1h on a 9x9 periodic torus using a method with controlled error, such as DMRG with very large bond dimension or exact diagonalization on smaller periodic clusters, for the same t/J values (1, 2, 3, 6). If EB is found to be nonnegative, or its magnitude is smaller than the SU(2) square-lattice values, the central claim of enhanced binding fails.

Watch

Extended reading notes

Core claim

The paper's central discovery is that holes in the SU(3) triangular-lattice t-J model form bound pairs with a negative binding energy, EB < 0, and that the magnitude of this binding is larger than in the SU(2) square-lattice case, contrary to a naive BCS expectation based on the larger coordination number. The binding weakens as t/J increases, and the hole-hole distance distribution shows tightly bound pairs at small t/J that become more delocalized at larger t/J. The enhanced binding is attributed to a geometric string mechanism: in the frozen-spin picture, hole motion creates strings of displaced spins, and the effective spinon-chargon and chargon-chargon interactions on a Bethe lattice of coordination number z lead to a binding energy that grows with z, as derived in the supplementary material. In addition, the paper finds that the magnetic correlations in the SU(3) model track those of the SU(2) square lattice remarkably closely, including a sign change in the next-nearest-neighbor correlations at comparable doping, and it reports non-s-wave pairing symmetry.

Load-bearing premise

The neural-network variational ansatz, with its fixed hidden-fermion architecture, estimates the one-hole and two-hole energies accurately enough on the 9x9 periodic tori that the small difference EB is reliable.

Editorial extensions

If this is right

  • If the central claim holds, two doped holes in the SU(3) triangular-lattice t-J model form pairs for t/J between 1 and 6, with binding energy magnitude larger than in the SU(2) square-lattice model.
  • The close analogy in magnetic correlations suggests the SU(3) triangular lattice can serve as a controlled testbed for SU(2) square-lattice physics without geometric frustration, extending polaronic and pairing concepts to higher symmetry.
  • The enhanced binding at higher coordination number supports the geometric-string mechanism over a simple BCS effective-mass picture for pairing in doped Mott insulators.
  • The computed spin and spin-hole correlations, expressed through the Cartan generators λ3 and λ8, are directly measurable in cold-atom quantum gas microscope experiments with SU(3)-symmetric fermions.
  • The sign change of the next-nearest-neighbor correlations near δ ≈ 0.25 provides a doping-dependent marker that could be tracked experimentally to locate the crossover out of the magnetic polaron regime.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if the enhanced binding is robust, then SU(N) triangular-lattice models with N > 3, which are accessible with ultracold molecules, may show even stronger pairing, and the geometric-string argument suggests the trend with z could be tested by comparing lattices with different connectivity.
  • Editorial inference: the paper's DMRG benchmarks are limited to open-boundary systems, so a controlled periodic-system benchmark with an independent method would directly test whether the variational bias between one-hole and two-hole sectors is the source of the enhanced binding.
  • Editorial inference: the abstract's mention of non-s-wave pairing symmetry is not elaborated in the main text; an explicit angular-momentum analysis of the two-hole wavefunction would clarify the pairing symmetry and connect it to the hole-distance distributions shown in Fig. 4b.
  • Editorial inference: the SU(3) model's correlations show the same qualitative doping dependence as SU(2), suggesting that the universal features of doped Mott insulators may be captured by the string picture, while the reduced correlation range is a quantitative effect of the larger on-site Hilbert space.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript extends the Gutzwiller-projected hidden-fermion determinant state (G-HFDS) variational approach to the three-flavor SU(3) t-J model on the triangular lattice, studying zero-doping magnetism, doping-dependent two- and three-point spin and spin-hole correlations, and two-hole binding. The central quantitative claim is a negative binding energy EB = E2h - 2E1h (Eq. 7, Fig. 4a) whose magnitude decreases with t/J and is larger than the corresponding SU(2) square-lattice value, interpreted through the geometric-string mechanism. The paper also presents benchmarks against DMRG on open 6x6 and 9x9 systems, a combinatorial derivation of the SU(3) correlation range, and an analytic string-theory scaling argument for binding energies.

Significance. If the binding-energy result is correct, the paper is a valuable contribution: it provides a tractable neural-network variational treatment of a doped SU(3) Mott insulator on a frustration-free triangular lattice, with observables that can be probed in cold-atom experiments, and it makes a concrete falsifiable prediction of enhanced pairing relative to the SU(2) square lattice. The authors ship code, benchmark total energies against DMRG at bond dimension 4096 (SM Appendix A), and the combinatorial correlation-range calculation in SM B1 is clean and correct. The main limitation is that the central binding-energy claim is not yet protected against systematic variational bias between particle-number sectors, and the open-boundary DMRG benchmarks do not address that bias.

major comments (3)
  1. [Pair Structure and Binding Energies, Eq. (7), Fig. 4a] The binding energy is an uncontrolled difference of two variational upper bounds: E1h and E2h are independently optimized G-HFDS energies, and EB is their difference. The error bars in Fig. 4a come from only two independent optimizations, which can quantify local-minimum scatter but not the systematic difference in representational bias between the one-hole and two-hole sectors. The DMRG benchmarks in SM Appendix A (Fig. 5) are performed on open-boundary systems and report only total energies, not EB under the fully periodic boundary conditions used in Fig. 4a. Since the enhanced-binding claim is the paper's headline result and the energies involved are small (order 0.1t), the manuscript should provide a controlled estimate of EB, for example by computing EB on the same periodic torus with an independent method for at least one value of t/J, or by demonstrating convergence of EB across G-HFDS architectures, initialization ensembles, and system sizes.
  2. [SM B2, Eq. (B5)] The analytic string argument is not quantitative for the sign or magnitude of EB. Equation (B5) gives EB = (2 alpha_1 - 2^{1/3} alpha_2) t^{1/3} J^{2/3}, with alpha_1 and alpha_2 non-universal constants; the subsequent discussion argues that a geometric spinon-chargon repulsion increases alpha_1 for larger coordination number z, but this is not derived or computed. As written, the argument is a plausible qualitative mechanism, not a derivation of enhanced binding. I recommend clearly separating the numerical result from this explanatory mechanism and stating explicitly that the string-theory comparison is qualitative.
  3. [Fig. 4a and 'Pair Structure and Binding Energies' text] The comparison of binding energies between the SU(3) triangular lattice and the SU(2) square lattice does not specify the system sizes, boundary conditions, lattice sizes, and t/J values used for the SU(2) data in Fig. 4a. The cited literature values (Refs [57-59]) are obtained on different geometries and sizes, and the SU(2) NQS data presumably come from Ref. [45] with its own finite-size setup. A direct comparison of absolute binding energies across different geometries requires identical, or explicitly extrapolated, finite-size treatments; otherwise the enhanced-binding conclusion may include finite-size contamination. Please state precisely the SU(2) parameter sets and any finite-size checks.
minor comments (5)
  1. [Abstract] The abstract claims 'non-s-wave pairing symmetry,' but the manuscript does not present an analysis of the pairing symmetry or pair wavefunction form factor; the only two-hole information is EB and the distance distribution in Fig. 4b. Please either remove this claim or substantiate it with a symmetry-resolved analysis of the pair state.
  2. [SM Appendix A] Please state explicitly whether the same G-HFDS architecture, number of hidden fermions, features, and optimization settings are used for the one-hole and two-hole states in Fig. 4a, since the two-hole energy difference is the quantity of interest.
  3. [Eq. (6)] The normalization tilde_eta in Eq. (6) is written using sigma(lambda_3) even though the correlator is defined for a general Cartan generator lambda_alpha; please clarify how the normalization is defined for lambda_8 and whether lambda_3 and lambda_8 results are identical by symmetry.
  4. [Figs. 2c and 3] For the SU(2) comparison data, please state explicitly whether the points are taken from Refs. [45] and [42] at the same system sizes and parameters or recomputed for this manuscript, so that the reader can judge whether differences are physical or numerical.
  5. [Introduction] The sentence 'the SU(3) t-J model on the triangular lattice does not exhibit geometric frustration' is too categorical: the triangular lattice is geometrically frustrated for SU(2) spin order, and the absence of frustration here refers to the three-sublattice SU(3) Neel state. Please rephrase to avoid confusion.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity; self-citations are limited to comparison data and post-hoc interpretation, while the central binding-energy result is an independent variational computation.

full rationale

The paper's central quantitative claim, enhanced binding energy EB = E2h - 2E1h in the SU(3) triangular-lattice t-J model, is computed directly from independently optimized G-HFDS variational energies for the one-hole and two-hole sectors (Eq. 7, Fig. 4a). No parameter is fitted to the binding energy, and the result is not defined in terms of the geometric-string interpretation that is later invoked to explain it. The SU(2) square-lattice comparison data do come partly from the same group's prior NQS work (Ref. [45]), and the geometric-string explanation cites the authors' own earlier papers (Refs. [50,51,53]); however, these are used for comparison and interpretation, not as inputs that force the SU(3) result. Critically, the SU(2) binding energies are anchored by independent DMRG and exact-diagonalization studies (Refs. [57-59]), so the enhanced-binding claim does not reduce to a self-citation chain. The variational-bias concern raised in the reader's take is a numerical accuracy risk, not a circularity: it questions whether the fixed neural-network ansatz represents one particle-number sector better than another, but that is a systematic error issue external to the derivation logic. The ansatz itself is openly presented as a variational method, and its accuracy is benchmarked against DMRG in the SM, albeit on open-boundary systems. No equation in the paper is equivalent by construction to the binding-energy result, and no fitted parameter is renamed as a prediction. Accordingly, the derivation is self-contained with respect to circularity, with only minor self-citations that are not load-bearing.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central claims rest on the variational suitability of the G-HFDS ansatz (with DMRG benchmarks only on open systems), the standard t-J reduction of the SU(3) Hubbard model, the accepted three-sublattice order of the undoped model, and the geometric string framework inherited from the authors' prior work. The binding-energy explanation additionally depends on non-universal constants alpha_1, alpha_2. No new physical entities are postulated.

free parameters (2)
  • NQS hidden fermion and feature counts = 20 hidden fermions; 96 features (6x6), 128 features (9x9)
    Architecture hyperparameters chosen by hand (SM Appendix A). The variational accuracy of the ansatz, and hence the binding-energy estimate, depends on these choices; they are not fitted to observables but are also not shown to be converged.
  • alpha_1, alpha_2 (string-theory constants) = not computed
    Non-universal constants in Eq. (B4)-(B5). The claimed explanation of the enhanced binding energy depends on these undetermined constants; without them the string argument predicts only a trend with coordination number z, not the observed magnitude.
assumptions (5)
  • domain assumption The t-J Hamiltonian (Eq. 1) is the strong-coupling limit of the SU(3) Fermi-Hubbard model on the triangular lattice.
    Invoked in the introduction. Standard strong-coupling reduction, valid for large U (small J/t); the paper studies t/J = 1 to 6, a range in which the strict t-J reduction is not controlled, although the t-J model is treated as a standalone model as is conventional.
  • domain assumption The SU(3) Heisenberg ground state on the triangular lattice has three-sublattice (3-SL) order.
    Used as the zero-doping reference for the polaron and correlation analysis (Fig. 2). Cited to Refs [25,27]; the paper's own undoped NQS reproduces the pattern, so this assumption is corroborated in-text.
  • domain assumption The G-HFDS ansatz with S3 symmetrization (Eq. 3) spans the relevant ground-state manifold at all studied dopings.
    The central numerical results depend on this variational ansatz. Benchmarks against DMRG exist only on open-boundary 6x6 and 9x9 systems (SM Fig. 5); the fully periodic systems used for the main claims are not independently verified.
  • domain assumption The frozen-spin / geometric string picture (Fig. 1b) describes hole motion in the 3-SL background.
    Used to interpret the doping dependence of the correlations and the binding-energy enhancement. Inherited from Refs [47,48,50,51,53], several of which share authors with this paper.
  • domain assumption The two-hole problem maps to a Bethe lattice with a linear confining potential, giving Eq. (B4).
    Appendix B. The mapping and the energy expression come from Ref [50]; the constants alpha_n are non-universal and the resulting binding-energy expression (Eq. B5) cannot be evaluated without them.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Magnetic correlations in the $SU(3)$ triangular-lattice $t$-$J$ model at finite doping." pith.science (2026). https://pith.science/paper/I43TPMOL

@misc{pith2026250601915,
  author       = {Pith},
  title        = {Pith review of: Magnetic correlations in the $SU(3)$ triangular-lattice $t$-$J$ model at finite doping},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/I43TPMOL}},
  note         = {Machine review of arXiv:2506.01915}
}
abstract

Ultracold alkaline-earth atoms and molecules now enable experimental realizations of SU(N)-symmetric Fermi-Hubbard models, yet theoretical understanding of these systems, particularly at finite doping remains limited. Here we investigate the strong-coupling limit of the $SU(3)$ symmetric Fermi-Hubbard model on the triangular lattice across the full doping range. Using a three-flavor extension of Gutzwiller-projected hidden fermion determinant states (G-HFDS), a neural network based variational ansatz, we analyze two- and three-point spin-spin and spin-spin-hole correlations of the $SU(3)$ Cartan generators. We further study the structure of a pair of doped holes for large periodic systems, and compare our results to the paradigmatic $SU(2)$ square lattice equivalent, finding strikingly similar magnetic correlations, non-s-wave pairing symmetry, and enhanced binding energies. Our results provide a foundation for future exploration of doped SU(N) Mott insulators, providing valuable insights for both theoretical developments and quantum simulation experiments.

Figures

Figures reproduced from arXiv: 2506.01915 by the authors.

Figure 1
Figure 1. a. SU(3) t-J model, as given in Eq. (1). The spin￾exchange allows spin flips between all three flavors via vir￾tual double occupancies, with the SU(3) symmetry requiring Jrg = Jbr = Jgb. b. String patterns in the SU(3) triangular and SU(2) square lattice. In the frozen spin approximation, the motion of a hole through the classical Néel state perturbs the spin background, creating a string of displaced spins. Model a… view at source ↗
Figure 3
Figure 3. Connected nearest and next-nearest neighbor corre [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. a. Binding energies EB for different ratios of t/J. Error bars are estimated by comparing two different optimiza￾tions of the NQS wavefunctions. For the SU(2) results error bars are smaller than the markers. A negative binding energy indicates a lower energy for the two-hole state. b. Distribu￾tion of hole distances d from snapshots of the optimized NQS with two doped holes on a 9 × 9 torus. Due to the triangular ge… view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Energies vs doping for 6×6 and 9×9 open boundary systems as compared to DMRG simulations. Dotted gray lines indicate the DMRG result for a bond dimension of χ = 4096. (⟨↑↓| + ⟨↓↑|)S z i S z j (|↑↓⟩ + |↓↑⟩) = − 1 2 , (B1) which upon normalizing by the variance σ(S z i )…

Discussion (0). Sign in to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Charge-6e superconductivity from doping SU(3) spin liquids

    cond-mat.str-el 2026-07 conditional novelty 7.0 of 10

    Doping SU(3) spin liquids can yield charge-6e superconductors, including a non-Abelian chiral version with h/(6e) vortices.

  2. Engineering SU($N$)-Symmetric Hubbard Models with Microwave-Shielded Dipolar Molecules

    cond-mat.quant-gas 2026-07 conditional novelty 6.0 of 10

    Tuning microwave Rabi frequency resonantly couples monomer pairs to on-site field-linked dimers, yielding independently controllable on-site and off-site interactions in an SU(N)-symmetric extended Hubbard model.

  3. Dominant Excitonic Superconductivity in a Three-component Hubbard Chain

    cond-mat.supr-con 2025-12 conditional novelty 6.0 of 10

    DMRG and analytic theory show dominant quasi-long-range pairing among flavors 1 and 3 in a one-dimensional three-component repulsive Hubbard model when the 1-3 repulsion is weaker than the other two.

Reference graph

Works this paper leans on

67 extracted references · 41 canonical work pages · cited by 3 Pith papers

  1. [45]

    Simu- lating the two-dimensionalt−jmodel at finite doping with neural quantum states,

    H. Lange, A. Böhler, C. Roth, and A. Bohrdt, “Simu- lating the two-dimensionalt−jmodel at finite doping with neural quantum states,” (2024), arXiv:2411.10430 [cond-mat.str-el]

    arXiv 2024

  2. [1]

    D. P. Arovas, E. Berg, S. A. Kivelson, and S. Raghu, An- nualReviewofCondensedMatterPhysics13,239(2022)

    2022

  3. [2]

    J. P. F. LeBlanc, A. E. Antipov, F. Becca, I. W. Bulik, G. K.-L. Chan, C.-M. Chung, Y. Deng, M. Ferrero, T. M. Henderson, C. A. Jiménez-Hoyos, E. Kozik, X.-W. Liu, A. J. Millis, N. V. Prokof’ev, M. Qin, G. E. Scuseria, H. Shi, B. V. Svistunov, L. F. Tocchio, I. S. Tupitsyn, S. R. White, S. Zhang, B.-X. Zheng, Z. Zhu, and E. Gull (Simons Collaboration on t...

    work page 2015

  4. [3]

    Jiang and S

    H.-C. Jiang and S. A. Kivelson, Proceedings of the National Academy of Sciences119, e2109406119 (2022), https://www.pnas.org/doi/pdf/10.1073/pnas.2109406119

    doi:10.1073/pnas.2109406119 2022

  5. [4]

    Qin, C.-M

    M. Qin, C.-M. Chung, H. Shi, E. Vitali, C. Hubig, U. Schollwöck, S. R. White, and S. Zhang (Simons Col- laboration on the Many-Electron Problem), Phys. Rev. X10, 031016 (2020)

    work page 2020

  6. [5]

    Zheng, C.-M

    B.-X. Zheng, C.-M. Chung, P. Corboz, G. Ehlers, M.-P. Qin, R. M. Noack, H. Shi, S. R. White, S. Zhang, and G. K.-L. Chan, Science358, 1155 (2017), https://www.science.org/doi/pdf/10.1126/science.aam7127

    doi:10.1126/science.aam7127 2017

  7. [6]

    Bohrdt, L

    A. Bohrdt, L. Homeier, C. Reinmoser, E. Demler, and F. Grusdt, Annals of Physics435, 168651 (2021), special issue on Philip W. Anderson

    2021

  8. [7]

    Raghu, S

    S. Raghu, S. A. Kivelson, and D. J. Scalapino, Phys. Rev. B81, 224505 (2010)

    work page 2010

Show all 67 references
  1. [8]

    Keimer, S

    B. Keimer, S. A. Kivelson, M. R. Norman, S. Uchida, and J. Zaanen, Nature518, 179 (2015)

    2015

  2. [9]

    M. A. Cazalilla and A. M. Rey, Reports on Progress in Physics77, 124401 (2014)

    2014

  3. [10]

    Ibarra-García-Padilla and S

    E. Ibarra-García-Padilla and S. Choudhury, Journal of Physics: Condensed Matter37, 083003 (2024)

    2024

  4. [11]

    Tokura and N

    Y. Tokura and N. Nagaosa, Science288, 462 (2000), https://www.science.org/doi/pdf/10.1126/science.288.5465.462

    2000 doi

  5. [12]

    Y. Q. Li, M. Ma, D. N. Shi, and F. C. Zhang, Phys. Rev. Lett.81, 3527 (1998)

    1998

  6. [13]

    F. F. Assaad, Phys. Rev. B71, 075103 (2005)

    2005

  7. [14]

    Honerkamp and W

    C. Honerkamp and W. Hofstetter, Phys. Rev. Lett.92, 170403 (2004)

    2004

  8. [15]

    Ibarra-García-Padilla, S

    E. Ibarra-García-Padilla, S. Dasgupta, H.-T. Wei, S. Taie, Y. Takahashi, R. T. Scalettar, and K. R. A. Hazzard, Phys. Rev. A104, 043316 (2021). 6

    2021

  9. [16]

    Sotnikov and W

    A. Sotnikov and W. Hofstetter, Phys. Rev. A89, 063601 (2014)

    2014

  10. [17]

    D. Tusi, L. Franchi, L. F. Livi, K. Baumann, D. Bene- dicto Orenes, L. Del Re, R. E. Barfknecht, T.-W. Zhou, M. Inguscio, G. Cappellini, M. Capone, J. Catani, and L. Fallani, Nature Physics18, 1201–1205 (2022)

    2022

  11. [18]

    S. Taie, R. Yamazaki, S. Sugawa, and Y. Takahashi, Nature Physics8, 825–830 (2012)

    2012

  12. [19]

    Hofrichter, L

    C. Hofrichter, L. Riegger, F. Scazza, M. Höfer, D. R. Fernandes, I. Bloch, and S. Fölling, Phys. Rev. X6, 021030 (2016)

    2016

  13. [20]

    Ozawa, S

    H. Ozawa, S. Taie, Y. Takasu, and Y. Takahashi, Phys. Rev. Lett.121, 225303 (2018)

    2018

  14. [21]

    S. Taie, E. Ibarra-García-Padilla, N. Nishizawa, Y. Takasu, Y. Kuno, H.-T. Wei, R. T. Scalettar, K. R. A. Hazzard, and Y. Takahashi, Nature Physics 18, 1356–1361 (2022)

    2022

  15. [22]

    S. Buob, J. Höschele, V. Makhalov, A. Rubio- Abadal, and L. Tarruell, PRX Quantum5(2024), 10.1103/prxquantum.5.020316

    2024 doi

  16. [23]

    Quantum gas microscopy of three-flavor hubbard systems,

    J. Mongkolkiattichai, L. Liu, S. Dasgupta, K. R. A. Hazzard, and P. Schauss, “Quantum gas microscopy of three-flavor hubbard systems,” (2025), arXiv:2503.05687 [cond-mat.quant-gas]

    2025 arXiv

  17. [24]

    Mukherjee, J

    B. Mukherjee, J. M. Hutson, and K. R. A. Hazzard, New Journal of Physics27, 013013 (2025)

    2025

  18. [25]

    T. A. Tóth, A. M. Läuchli, F. Mila, and K. Penc, Phys. Rev. Lett.105, 265301 (2010)

    2010

  19. [26]

    Nataf and F

    P. Nataf and F. Mila, Phys. Rev. Lett.113, 127204 (2014)

    2014

  20. [27]

    Bauer, P

    B. Bauer, P. Corboz, A. M. Läuchli, L. Messio, K. Penc, M. Troyer, and F. Mila, Phys. Rev. B85, 125116 (2012)

    2012

  21. [28]

    Harada, N

    K. Harada, N. Kawashima, and M. Troyer, Phys. Rev. Lett.90, 117203 (2003)

    2003

  22. [29]

    Nataf, M

    P. Nataf, M. Lajkó, P. Corboz, A. M. Läuchli, K. Penc, and F. Mila, Phys. Rev. B93, 201113 (2016)

    2016

  23. [30]

    Corboz, A

    P. Corboz, A. M. Läuchli, K. Penc, M. Troyer, and F. Mila, Phys. Rev. Lett.107, 215301 (2011)

    2011

  24. [31]

    Schlömer, F

    H. Schlömer, F. Grusdt, U. Schollwöck, K. R. A. Haz- zard, and A. Bohrdt, Phys. Rev. B110, 125134 (2024)

    2024

  25. [32]

    C. Feng, E. Ibarra-García-Padilla, K. R. A. Hazzard, R. Scalettar, S. Zhang, and E. Vitali, Phys. Rev. Res. 5, 043267 (2023)

    2023

  26. [33]

    Ibarra-García-Padilla, C

    E. Ibarra-García-Padilla, C. Feng, G. Pasqualetti, S. Fölling, R. T. Scalettar, E. Khatami, and K. R. A. Hazzard, Phys. Rev. A108, 053312 (2023)

    2023

  27. [34]

    Carleo and M

    G. Carleo and M. Troyer, Science355, 602–606 (2017)

    2017

  28. [35]

    Hornik, Neural Networks4, 251 (1991)

    K. Hornik, Neural Networks4, 251 (1991)

    1991

  29. [36]

    Rende, L

    R. Rende, L. L. Viteritti, L. Bardone, F. Becca, and S. Goldt, Communications Physics7, 260 (2024)

    2024

  30. [37]

    Convolutional trans- former wave functions,

    A. Chen, V. D. Naik, and M. Heyl, “Convolutional trans- former wave functions,” (2025), arXiv:2503.10462 [cond- mat.dis-nn]

    2025 arXiv

  31. [38]

    Group convolu- tional neural networks improve quantum state accuracy,

    C. Roth and A. H. MacDonald, “Group convolu- tional neural networks improve quantum state accuracy,” (2021), arXiv:https://arxiv.org/abs/2104.05085 [quant- ph]

    2021 arXiv

  32. [39]

    Sharir, Y

    O. Sharir, Y. Levine, N. Wies, G. Carleo, and A. Shashua, Phys. Rev. Lett.124, 020503 (2020)

    2020

  33. [40]

    D.-L. Deng, X. Li, and S. Das Sarma, Phys. Rev. X7, 021021 (2017)

    2017

  34. [41]

    Levine, O

    Y. Levine, O. Sharir, N. Cohen, and A. Shashua, Phys. Rev. Lett.122, 065301 (2019)

    2019

  35. [42]

    From architectures to applications: a review of neural quantum states,

    H. Lange, A. V. de Walle, A. Abedinnia, and A. Bohrdt, “From architectures to applications: a review of neural quantum states,” (2024)

    2024

  36. [43]

    Quantum many-body solver using artificial neural networks and its applica- tions to strongly correlated electron systems,

    Y. Nomura and M. Imada, “Quantum many-body solver using artificial neural networks and its applica- tions to strongly correlated electron systems,” (2024), arXiv:2410.02633 [cond-mat.str-el]

    2024 arXiv

  37. [44]

    Romero, J

    I. Romero, J. Nys, and G. Carleo, Communications Physics8(2025), 10.1038/s42005-025-01955-z

    2025 doi

  38. [46]

    Robledo Moreno, G

    J. Robledo Moreno, G. Carleo, A. Georges, and J. Stokes, Proceedings of the National Academy of Sci- ences119(2022), 10.1073/pnas.2122059119

    2022 doi

  39. [47]

    C. S. Chiu, G. Ji, A. Bohrdt, M. Xu, M. Knap, E. Dem- ler, F. Grusdt, M. Greiner, and D. Greif, Science365, 251–256 (2019)

    2019

  40. [48]

    Grusdt, A

    F. Grusdt, A. Bohrdt, and E. Demler, Phys. Rev. B99, 224422 (2019)

    2019

  41. [49]

    Brauner, Symmetry2, 609–657 (2010)

    T. Brauner, Symmetry2, 609–657 (2010)

    2010

  42. [50]

    Grusdt, E

    F. Grusdt, E. Demler, and A. Bohrdt, SciPost Phys.14, 090 (2023)

    2023

  43. [51]

    Bohrdt, L

    A. Bohrdt, L. Homeier, I. Bloch, E. Demler, and F. Grusdt, Nature Physics18, 651–656 (2022)

    2022

  44. [52]

    Homeier, H

    L. Homeier, H. Lange, E. Demler, A. Bohrdt, and F. Grusdt, Nature Communications16(2025), 10.1038/s41467-024-55549-4

    2025 doi

  45. [53]

    Bohrdt, E

    A. Bohrdt, E. Demler, F. Pollmann, M. Knap, and F. Grusdt, Phys. Rev. B102, 035139 (2020)

    2020

  46. [54]

    Chen and M

    A. Chen and M. Heyl, Nature Physics20, 1476–1481 (2024)

    2024

  47. [55]

    Kawakami, Phys

    N. Kawakami, Phys. Rev. B46, 3191 (1992)

    1992

  48. [56]

    Koepsell, D

    J. Koepsell, D. Bourgund, P. Sompet, S. Hirthe, A. Bohrdt, Y. Wang, F. Grusdt, E. Demler, G. Sa- lomon, C. Gross, and I. Bloch, Science374, 82 (2021), https://www.science.org/doi/pdf/10.1126/science.abe7165

    2021 doi

  49. [57]

    Two-dopant origin of competing stripe and pair formation in hubbard andt-jmodels,

    T. Blatz, U. Schollwöck, F. Grusdt, and A. Bohrdt, “Two-dopant origin of competing stripe and pair formation in hubbard andt-jmodels,” (2024), arXiv:2409.18131 [cond-mat.str-el]

    2024

  50. [58]

    A. L. Chernyshev, P. W. Leung, and R. J. Gooding, Phys. Rev. B58, 13594 (1998)

    1998

  51. [59]

    P. W. Leung, Phys. Rev. B65, 205101 (2002)

    2002

  52. [60]

    A. Rapp, G. Zaránd, C. Honerkamp, and W. Hofstet- ter, Physical Review Letters98(2007), 10.1103/phys- revlett.98.160405

    2007 doi

  53. [61]

    Vicentini, D

    F. Vicentini, D. Hofmann, A. Szabó, D. Wu, C. Roth, C. Giuliani, G. Pescia, J. Nys, V. Vargas-Calderón, N. Astrakhantsev, and G. Carleo, SciPost Phys. Code- bases , 7 (2022)

    2022

  54. [62]

    TheSyTentoolkit,

    C. Hubig, F. Lachenmaier, N.-O. Linden, T. Reinhard, L. Stenzel, A. Swoboda, M. Grundner, S. Mardazad, and S. Paeckel, “TheSyTentoolkit,”

  55. [63]

    Hubig,Symmetry-Protected Tensor Networks, Ph.D

    C. Hubig,Symmetry-Protected Tensor Networks, Ph.D. thesis, LMU München (2017). 7 SUPPLEMENT AL MA TERIAL Appendix A: Architecture and Benchmarks

    2017

  56. [64]

    Initialization and T raining Details We initialize each state as a three-flavor Gutzwiller projected Fermi sea. We diagonalize a single particle HamiltonianandfillthecolumnsofeachmeanfieldSlater determinantΦ α according to the lowest energy eigen- states such that in the matri...

  57. [65]

    Due to the limitations of the MPS method, we choose an open boundary system and com- pare the energies for both6×6and9×9lattice sites

    DMRG Benchmarks InordertobenchmarktheperformanceoftheG-HFDS ansatz, we compare the obtained energies to those of a DMRG simulation. Due to the limitations of the MPS method, we choose an open boundary system and com- pare the energies for both6×6and9×9lattice sites. Results ar...

  58. [66]

    Correlation range ofSU(3)Cartan generators As discussed in the main text, the correlation range of theSU(3)Cartan generators⟨ ˆλα,iˆλα,j⟩is reduced com- pared to the⟨ˆSz i ˆSz j ⟩correlations. This can be understood from a combinatorial argument: In the maximally anti- symmetr...

  59. [67]

    From a simple BCS per- spective, one would expect a lower binding energy

    Binding energy from BCS and geometric strings We present here an argument to support the higher binding energies in theSU(3)triangular lattice due to 8 the higher lattice connectivity. From a simple BCS per- spective, one would expect a lower binding energy. The single particl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.