Pith. sign in

REVIEW 3 major objections 6 minor 57 references

Large scale neural quantum states reveal the interplay between superconductivity and quantum criticality in the Hofstadter-Hubbard model

T0 review · 3 major / 6 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 Doping near a topological quantum critical point produces a d+id superconductor with stiffness peaked at the transition.

desk verdict Large-scale NQS results worth a serious look, but the d+id superconductivity claim rests on a seed that is never shown to be unbiased. read the letter →

arxiv 2608.02753 v1 pith:LD2AX7L5 submitted 2026-08-03 cond-mat.str-el cond-mat.dis-nncond-mat.supr-con

Christopher Roth , Andrew Millis , Tomohiro Soejima This is my paper · ORCID
classification cond-mat.str-elcond-mat.dis-nncond-mat.supr-con
keywords Hofstadter-HubbardmodelchiralspinliquidintegerquantumHallinsulatorcriticalpointtopologicalsuperconductord+idpairingsuperfluidstiffnessneuralstates
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

The paper studies the triangular-lattice Hofstadter-Hubbard model at $\pi/2$ flux per plaquette, a setting where half filling offers two distinct parent insulators: an integer quantum Hall state at weak coupling and a chiral spin liquid at intermediate coupling. Using a hidden-fermion Pfaffian neural quantum state on tori of up to 432 sites, it argues that the transition between these parents is continuous, with the $2e$ charge gap vanishing and charge fluctuations becoming critical near $U \approx 13$. Doping the model yields a topological superconductor with $d+id$ pairing order on both sides of the transition. The central claim is that the two ingredients of superconductivity respond differently: the pairing amplitude stays almost unchanged across the transition, while the superfluid stiffness is strongly enhanced near the critical point. The paper concludes that the superconducting energy scale is set by proximity to the quantum critical point rather than by which insulator is doped, and that critical charge fluctuations promote pair mobility.

What carries the argument

The central computational object is the Hidden Fermion Pfaffian State (HFPS): a variational wavefunction that combines a Pfaffian of visible fermions with hidden-fermion correlations parameterized by a residual convolutional neural network, using block-circulant pairing orbitals so that a single Pfaffian evaluation respects the lattice translation symmetry. This ansatz allows ground states on tori of 192, 300, and 432 sites to be optimized and then probed by flux threading, which yields the Hall conductances, the $2e$ charge gap (from energy differences at fixed particle number and with two holes), and the superfluid stiffness (from the second derivative of energy with respect to threaded flux). The charge structure factor is used as a diagnostic of critical fluctuations because a power-law tail in real space shows up as a non-analytic small-momentum behavior in Fourier space.

What would settle it

Train the same neural quantum state starting from a mean-field state with no pairing pinning field (or with a pinning field of opposite chirality) and continue to full variational convergence; if the final wavefunction shows no off-diagonal long-range order with the $d+id$ winding pattern, the superconductivity claim is an artifact of the initialization.

Watch

Extended reading notes

Core claim

The study establishes a continuous quantum phase transition between an integer quantum Hall insulator and a chiral spin liquid at half filling, and shows that doping either side produces a topological superconductor whose stiffness is largest right at the critical point. The evidence for the transition is a $2e$ charge gap that decreases with system size and approaches zero near $U = 13.25$ on a 432-site torus, together with a charge structure factor that acquires a power-law tail, while the spin structure factor remains analytic. Upon doping, the pair correlation function plateaus to a constant (off-diagonal long-range order) with a pairing pattern that winds twice around each plaquette, consistent with $d+id$ pairing, and the spin Hall conductance takes the value $2$ in units of $\hbar/(8\pi)$, identifying the superconductor as topological. The superfluid stiffness, extracted from the flux curvature of the ground-state energy, shows a peak near the critical point that sharpens with system size and low doping; the pairing order parameter, by contrast, is featureless around the transition. The paper interprets this dichotomy as evidence that quantum critical charge fluctuations enhance phase coherence and pair mobility without enhancing pair formation.

Load-bearing premise

The superconductivity claim rests on the assumption that the small $d+id$ pairing field (amplitude $0.2$) used to initialize the mean-field starting point of the variational optimization does not bias the final wavefunction; the paper checks this only by comparing pinned and unpinned training after 1000 iterations, not at full convergence.

Editorial extensions

If this is right

  • If the transition is continuous with a vanishing $2e$ gap, the half-filled system near $U \approx 13$ hosts gapless charge fluctuations, so the doped superconductor inherits a divergent correlation length that can enhance phase coherence.
  • The $d+id$ topological superconductor appears on both sides of the transition, so the same superconducting phase emerges from two topologically distinct parent insulators, carrying a spin Hall conductance of $2\hbar/(8\pi)$ with chiral edge modes.
  • Because the superfluid stiffness peaks near the critical point while the pairing amplitude stays flat, the mean-field transition temperature estimated from the Nelson-Kosterlitz criterion should be highest near $U_c$ at low doping.
  • The dimensional scaling form $D_s = \sqrt{\delta} F(\sqrt{\delta}/\xi)$ implies the stiffness peak is controlled by the ratio of the doping length scale to the correlation length, so lower doping should yield a sharper, taller peak.

Reading between the lines

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

  • Beyond the paper: the same separation of pairing amplitude and stiffness could be tested in other topological transitions, such as doping a Chern insulator near its transition to a fractional Chern insulator; if the pattern holds, critical charge fluctuations may be a generic route to enhanced superconductivity.
  • Beyond the paper: the paper does not determine critical exponents or match operator content to a field theory; computing the charge structure factor exponent $\eta$ at larger momenta or on larger systems could identify whether the critical point is the proposed conformal field theory.
  • Beyond the paper: because the stiffness peak sharpens with system size at fixed doping, finite-size scaling of $D_s$ across $U$ and $N$ could locate $U_c$ more precisely than the gap minimum alone, and might be combined with the kinetic-energy structure factor to define a doping-dependent crossover scale.
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 / 6 minor

Summary. This paper uses hidden fermion Pfaffian states (HFPS) with residual convolutional neural networks to simulate the triangular-lattice Hofstadter-Hubbard model at pi/2 flux on tori of up to 432 sites. At half filling, the authors report a continuous IQH to chiral spin liquid transition, supported by a decreasing 2e charge gap and power-law charge structure factors near U~13.25. Upon doping, they report a topological superconductor with d+id pairing ODLRO and spin Hall conductance, and they argue that the superfluid stiffness is enhanced near the quantum critical point while the pairing order parameter is not. The paper includes detailed appendices on flux threading, structure factor fits, stiffness extraction, and variational optimization.

Significance. If the central claims hold, this is a substantial computational advance: the system sizes (300--432 sites) are far beyond typical DMRG limits for this frustrated model, and the clean separation of pairing scale from phase coherence would be a new qualitative result. The paper gives credit where it is due; it is transparent about the variational energy errors, provides data availability, and honestly states in the Discussion that critical exponents and z=1 scaling are not determined. These strengths make the manuscript a candidate for publication, but the superconductivity claim and the 'vanishing gap' claim are not yet fully supported by the evidence as presented.

major comments (3)
  1. [Appendix E1, Eq. (E1), Table II, Fig. 3] The superconducting ODLRO and spin-Hall winding are computed from variational states initialized from a mean-field Hamiltonian with a pinning field of amplitude Delta=0.2 whose phase pattern is exactly the d+id order parameter used in Eq. (2). The only comparison to an unpinned run (Table II) is after 1000 iterations, and the text explicitly states that the unpinned state 'has still not found the correct superconducting order.' A higher-energy, unconverged state is not evidence that the converged unbiased ground state lacks ODLRO. Since the HFPS ansatz contains a Pfaffian with explicitly paired visible fermions, the variational bias could produce the observed C_p(r) plateau and W^s=2 without the true ground state being superconducting. Please provide a converged unpinned calculation (or a controlled extrapolation in training time) and demonstrate that the ODLRO and spin-Hall response are independent of the initial pairing seed.
  2. [Fig. 2(a), section 'Vanishing charge gap'] The claim that the transition is continuous rests substantially on the 2e gap decreasing with system size, but the minimum gap at N=432 is still 0.22t and no thermodynamic extrapolation is presented. The text says the gap 'trends toward zero' but does not quantify the trend or rule out a small but nonzero value in the thermodynamic limit. Please provide a finite-size scaling analysis (e.g., gap versus 1/N with an estimate of the extrapolated value at U_c) or explicitly state the precision with which the vanishing gap can be claimed at the accessible sizes.
  3. [Appendix D2, Fig. 10, Table I] The superfluid stiffness is computed from a finite difference with flux step phi_s = 0.4 x 2pi, which is a large excursion from phi=0. The quadratic form of E(phi_y) is verified at only one parameter set (U=8, delta=1/8, N=192). Since the central 'interplay' narrative and the BKT temperature estimate rely on the stiffness peak near U_c, please verify that E(phi_y) is quadratic over the range [0, 0.4 x 2pi] at representative points near the critical point and at the largest system size, or use a smaller flux step with a convergence check.
minor comments (6)
  1. [Fig. 2 caption] The caption 'Power law of Cn/s(k)' is ambiguous; please state explicitly that the plotted quantity is the fitted low-|k| exponent of the charge and spin structure factors, and include uncertainty estimates for the fits.
  2. [Table II] The table headings 'Delta = 0' and 'Delta = 0.2' compare energies after an equal number of iterations, but the text uses this to argue for the benefit of the pinning field; please also state the converged energy or the asymptotic energy difference, or add a caveat that the comparison is a training-progress measure rather than a converged-energy comparison.
  3. [Introduction] The sentence referring to 'the theoretically predicted vanishing of the 2e charge gap' would benefit from a specific citation or a one-sentence derivation, since the prediction follows from the difference in Hall conductivity but is not obvious to all readers.
  4. [Dimensional analysis paragraph] The scaling form D_s = sqrt(delta) F(sqrt(delta)/xi) is introduced without derivation; please add a sentence explaining the origin of this form (e.g., the degeneracy temperature of a 2D boson gas) and the meaning of the F(0)=1 normalization.
  5. [Abstract and Discussion] The abstract says 'strong evidence' for a continuous transition, while the Discussion notes that critical exponents and z=1 scaling are not determined; consider softening the abstract to 'evidence consistent with' a continuous transition, or explicitly list the missing diagnostics in the abstract.
  6. [Eq. (D4)] The finite-difference expression for D_s uses phi_s = 0.4 x 2pi; please state the uncertainty in D_s arising from the energy difference, since the stiffness peak near U_c is a central observable.

Circularity Check

1 steps flagged · score 6.0 of 10

Doped-superconductor claim is seeded by the same d+id pairing operator used to measure ODLRO; the unpinned control is explicitly unconverged.

  1. self definitional [Main text 'Superconductivity upon doping', Eq. (2)-(3), Fig. 3; Appendix E 1, Eq. (E1), Table II]
    "We consider nearest neighbor pairing of the form ∆x = 1/(12√2) Σ_{⟨α,β⟩∈Ux} phase(α, β)(cα,↑cβ,↓ −cα,↓cβ,↑) (2) ... First, we optimized the mean field on the Hubbard model at U=3 with a pinning field consistent with the order parameter shown in the inset of Fig. 3(a)."

    The variational search is initialized from H_MF (Eq. E1), which contains a pairing pinning field Δ=0.2 whose phase pattern phase(α,β) is exactly the d+id order parameter of Eq. (2) that is later used to compute Cp(r) and declared to show ODLRO. All doped runs start from this seed and are transferred adiabatically in U, so the final wavefunction inherits the seed's pairing symmetry unless optimization actively removes it. The only unbiased control (Table II) runs for 1,000 iterations, and the text admits the unpinned state 'has still not found the correct superconducting order'—i.e., it is unconverged, so it cannot demonstrate that the converged unbiased ground state lacks ODLRO.

full rationale

The half-filling results—the IQH-CSL transition, the vanishing 2e charge gap, and the critical charge fluctuations in Fig. 2—are computed from variational energies and structure factors and are not constructed from the d+id pairing operator; those claims stand as independent variational evidence. The superconductivity claim, however, is the load-bearing result for the announced 'interplay' narrative, and it is seeded by the same d+id pairing operator that is later measured: the mean-field initialization in Eq. (E1) contains Δ=0.2 with the phase pattern of the order parameter Eq. (2), and the paper's only unpinned control is explicitly run for only 1000 iterations and conceded to be unconverged. Thus the ODLRO plateau and W^s=2 are not shown to be independent of the variational input. The self-citations to Refs. [24,25] are not themselves the circularity; the circularity is that the cited protocol places the claimed order parameter into the initialization and then reports the same order as a finding. Score 6 reflects partial circularity: the central SC prediction reduces, in practice, to the seed, while the criticality results remain self-contained.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claims rest on the representational power and unbiased optimization of the HFPS ansatz, on the pinning-field initialization not biasing the superconducting order, on finite-size scaling of the charge gap, and on a heuristically assumed scaling form for the stiffness.

free parameters (3)
  • Pinning field amplitude Delta = 0.2
    Chosen by hand in the mean-field initialization (Eq. E1) with the d+id phase pattern; used for all calculations and biases the variational search toward the claimed superconducting order.
  • Flux step phi_s for stiffness finite difference = 0.4*2pi
    Numerical parameter for Eq. D4; validated against a quadratic fit at U=8, delta=1/8 but not at all parameter points.
  • Structure factor power-law exponent = not reported
    Fitted to low-|k| charge structure factor data (|k|<pi/3); the critical-fluctuation claim depends on this fit, and no error bars are given.
assumptions (6)
  • domain assumption The HFPS ansatz with R-CNN parameterization can represent the ground state of the Hofstadter-Hubbard model at these system sizes.
    Invoked throughout; the variational results assume the ansatz is sufficiently expressive and the optimization finds the global minimum (App. E).
  • domain assumption Periodic boundary conditions on sqrt(3)L x sqrt(3)L tori with L=8,10,12 capture the thermodynamic limit behavior.
    Finite-size effects are discussed, but no systematic extrapolation to infinite size is performed for the gap.
  • domain assumption Zero-variance extrapolation (Eq. E3) yields an unbiased estimate of the exact ground-state energy.
    The error estimates in Fig. 11 rely on this; the method is from prior literature (Refs. [27,47]).
  • standard math The small-|k| structure factor can be decomposed as analytic plus power-law with C(k)=alpha|k|^2 + beta|k|^{eta-2}.
    App. C1; standard for power-law correlations.
  • ad hoc to paper Dimensional analysis scaling D_s = sqrt(delta) F(sqrt(delta)/xi) with F(0)=1 holds.
    Main text 'A simple dimensional analysis...' asserts this form without derivation; used to explain the stiffness peak.
  • ad hoc to paper The mean-field starting point with pinning field does not prevent the variational wavefunction from reaching the true ground state.
    Unpinned calculation is not run to convergence; the comparison in Table II is only after 1000 iterations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Large scale neural quantum states reveal the interplay between superconductivity and quantum criticality in the Hofstadter-Hubbard model." pith.science (2026). https://pith.science/paper/LD2AX7L5

@misc{pith2026260802753,
  author       = {Pith},
  title        = {Pith review of: Large scale neural quantum states reveal the interplay between superconductivity and quantum criticality in the Hofstadter-Hubbard model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LD2AX7L5}},
  note         = {Machine review of arXiv:2608.02753}
}
abstract

Understanding how a parent insulating state shapes the superconductivity that emerges upon doping is a long-standing problem dating back to Anderson's resonating-valence-bond proposal. The triangular-lattice Hofstadter-Hubbard model with $\pi/2$ flux per plaquette offers an ideal setting for this question: at half filling it hosts two distinct parent states---an integer quantum Hall insulator at weak coupling and a chiral spin liquid at intermediate coupling---separated by a topological phase transition. Using neural quantum states on tori of up to $432$ sites, we present strong evidence that the transition is continuous, with a vanishing $2e$ charge gap and critical charge fluctuations. Upon doping, we find a topological superconductor with off-diagonal long-range order consistent with $d+id$ pairing on either side of the transition. The two ingredients of superconductivity, pair formation and phase coherence, respond to the parent state in sharply different ways: while the pairing order parameter remains nearly unchanged across the transition, the superfluid stiffness is strongly enhanced near the critical point. The energy scale of the superconductor is therefore set not by which parent state is doped, but by proximity to the transition between them. Our results establish neural quantum states as a powerful tool for understanding the delicate interplay between long-ranged electronic correlations and superconductivity.

Figures

Figures reproduced from arXiv: 2608.02753 by the authors.

Figure 1
Figure 1. FIG. 1. Overview of our numerical simulations. We use a [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Characterizing topological superconductivity. (a) [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. FIG. 4. Doping the Hofstadter-Hubbard model around the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: FIG. 5. Diagram of the tori simulated in this work. Here we show a [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Diagram of the Hofstadter-Hubbard Hamiltonian at [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Identifying phases of the [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8. Behavior of the kinetic energy structure factors at small momenta. The linear component of the kinetic energy structure [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: FIG. 9. Fitting the small- [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: FIG. 10. (a) Energy as a function of flux threaded on the 192 site cluster at [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: FIG. 11. Estimates of variational accuracy. The error is estimated from the difference between the variational energy after [PITH_FULL_IMAGE:figures/full_fig_p014_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 49 canonical work pages

  1. [1]

    P. W. Anderson. The resonating valence bond state in la¡sub¿2¡/sub¿cuo¡sub¿4¡/sub¿ and superconductivity. Science, 235(4793):1196–1198, 1987

    work page 1987

  2. [2]

    R. B. Laughlin. The Relationship Between High- Temperature Superconductivity and the Fractional Quantum Hall Effect.Science, 242(4878):525–533, Oc- tober 1988. Publisher: American Association for the Ad- vancement of Science

    work page 1988

  3. [3]

    Quantum spin liquids: a review.Reports on Progress in Physics, 80(1):016502, 2017

    Lucile Savary and Leon Balents. Quantum spin liquids: a review.Reports on Progress in Physics, 80(1):016502, 2017

    work page 2017

  4. [4]

    Quan- tum spin liquid states.Reviews of Modern Physics, 89(2):025003, 2017

    Yi Zhou, Kazushi Kanoda, and Tai-Kai Ng. Quan- tum spin liquid states.Reviews of Modern Physics, 89(2):025003, 2017

    work page 2017

  5. [5]

    Colloquium: Zoo of quantum- topological phases of matter.Reviews of Modern Physics, 89(4):041004, 2017

    Xiao-Gang Wen. Colloquium: Zoo of quantum- topological phases of matter.Reviews of Modern Physics, 89(4):041004, 2017

    work page 2017

  6. [6]

    Doping a mott insulator: Physics of high-temperature superconductivity.Reviews of modern physics, 78(1):17– 85, 2006

    Patrick A Lee, Naoto Nagaosa, and Xiao-Gang Wen. Doping a mott insulator: Physics of high-temperature superconductivity.Reviews of modern physics, 78(1):17– 85, 2006

    work page 2006

  7. [7]

    Doping the chiral spin liquid: Topological superconduc- tor or chiral metal.Physical Review B, 103(16):165138, 2021

    Xue-Yang Song, Ashvin Vishwanath, and Ya-Hui Zhang. Doping the chiral spin liquid: Topological superconduc- tor or chiral metal.Physical Review B, 103(16):165138, 2021

    work page 2021

  8. [8]

    Chiral pseudospin liquids in moir´ e heterostructures.Physical Review X, 14(2):021013, 2024

    Clemens Kuhlenkamp, Wilhelm Kadow, Ata¸ c Imamo˘ glu, and Michael Knap. Chiral pseudospin liquids in moir´ e heterostructures.Physical Review X, 14(2):021013, 2024

    work page 2024

Show all 57 references
  1. [9]

    Chiral spin liq- uid and quantum phase transition in the triangular- lattice hofstadter-hubbard model.Physical Review B, 113(12):L121107, 2026

    Stefan Divic, Tomohiro Soejima, Valentin Cr´ epel, Michael P Zaletel, and Andrew Millis. Chiral spin liq- uid and quantum phase transition in the triangular- lattice hofstadter-hubbard model.Physical Review B, 113(12):L121107, 2026

    2026

  2. [10]

    Gallegos, Rafael M

    Cesar A. Gallegos, Rafael M. Magaldi, Andrew Millis, and Steven R. White. Quantum hall to chiral spin liq- uid transition in a triangular lattice hofstadter-hubbard model.Phys. Rev. B, 113:064412, Feb 2026

    2026

  3. [11]

    A. H. MacDonald, S. M. Girvin, and D. Yoshioka. t U expansion for the hubbard model.Phys. Rev. B, 37:9753– 9756, Jun 1988

    1988

  4. [12]

    Approaching the thermodynamic limit with neural-network quantum states.arXiv preprint arXiv:2602.02665, 2026

    Luciano Loris Viteritti, Riccardo Rende, Subir Sachdev, and Giuseppe Carleo. Approaching the thermodynamic limit with neural-network quantum states.arXiv preprint arXiv:2602.02665, 2026

    2026

  5. [13]

    This is in contrast to an earlier DMRG study that only found a∼25% decrease relative to the band gap [13]

    On the 432 site cluster, the smallest gap is atU= 13.25, and has a value of 0.22t, roughly a factor of 20 smaller than the band gap. This is in contrast to an earlier DMRG study that only found a∼25% decrease relative to the band gap [13]. Structure factor—The small-|k|behavio...

  6. [14]

    Anyon superconductiv- ity from topological criticality in a hofstadter–hubbard model.Proceedings of the National Academy of Sciences, 122(33):e2426680122, 2025

    Stefan Divic, Valentin Cr´ epel, Tomohiro Soejima, Xue-Yang Song, Andrew J Millis, Michael P Zale- tel, and Ashvin Vishwanath. Anyon superconductiv- ity from topological criticality in a hofstadter–hubbard model.Proceedings of the National Academy of Sciences, 122(33):e2426680...

    2025

  7. [15]

    Robust su- perconductivity upon doping chiral spin liquid and chern insulators in a hubbard-hofstadter model.arXiv preprint arXiv:2509.02675, 2025

    Clemens Kuhlenkamp, Stefan Divic, Michael P Zaletel, Tomohiro Soejima, and Ashvin Vishwanath. Robust su- perconductivity upon doping chiral spin liquid and chern insulators in a hubbard-hofstadter model.arXiv preprint arXiv:2509.02675, 2025

    2025 arXiv

  8. [16]

    Topological chiral superconductivity in the triangular-lattice hofstadter-hubbard model.Physical Review Letters, 136(8):086503, 2026

    Feng Chen, Wen O Wang, Jia-Xin Zhang, Leon Balents, and DN Sheng. Topological chiral superconductivity in the triangular-lattice hofstadter-hubbard model.Physical Review Letters, 136(8):086503, 2026

    2026

  9. [17]

    Solving the quan- tum many-body problem with artificial neural networks

    Giuseppe Carleo and Matthias Troyer. Solving the quan- tum many-body problem with artificial neural networks. Science, 355(6325):602–606, 2017

    2017

  10. [18]

    Better, faster fermionic neural networks.arXiv preprint arXiv:2011.07125, 2020

    James S Spencer, David Pfau, Aleksandar Botev, and W Matthew C Foulkes. Better, faster fermionic neural networks.arXiv preprint arXiv:2011.07125, 2020

    2011 arXiv

  11. [19]

    Yusuke Nomura and Masatoshi Imada. Dirac-type nodal spin liquid revealed by refined quantum many- body solver using neural-network wave function, corre- lation ratio, and level spectroscopy.Physical Review X, 11(3):031034, 2021

    2021

  12. [20]

    From architectures to applica- tions: a review of neural quantum states.Quantum Sci- ence and Technology, 9(4):040501, 2024

    Hannah Lange, Anka Van de Walle, Atiye Abedinnia, and Annabelle Bohrdt. From architectures to applica- tions: a review of neural quantum states.Quantum Sci- ence and Technology, 9(4):040501, 2024

    2024

  13. [21]

    Chen Ning Yang.ηpairing and off-diagonal long-range order in a hubbard model.Phys. Rev. Lett., 63:2144– 2147, Nov 1989

    1989

  14. [22]

    Chen Ning Yang and S. C. Zhang. SO 4 Symmetry in a Hubbard Model.Modern Physics Letters B, 4(11):759– 766, January 1990

    1990

  15. [23]

    Remarks and generalizations about SU 2×SU 2 symmetry of Hubbard models.Physics Letters A, 161(3):292–294, December 1991

    Chen Ning Yang. Remarks and generalizations about SU 2×SU 2 symmetry of Hubbard models.Physics Letters A, 161(3):292–294, December 1991

    1991

  16. [24]

    FIELD THEORY METHODS AND QUAN- TUM CRITICAL PHENOMENA

    Ian Affleck. FIELD THEORY METHODS AND QUAN- TUM CRITICAL PHENOMENA. InLes Houches Sum- mer School in Theoretical Physics: Fields, Strings, Crit- ical Phenomena, 6 1988

    1988

  17. [25]

    Neural network- augmented pfaffian wave-functions for scalable sim- ulations of interacting fermions.arXiv preprint arXiv:2507.10705, 2025

    Ao Chen, Zhou-Quan Wan, Anirvan Sengupta, An- toine Georges, and Christopher Roth. Neural network- augmented pfaffian wave-functions for scalable sim- ulations of interacting fermions.arXiv preprint arXiv:2507.10705, 2025

    2025 arXiv

  18. [26]

    Superconductivity in the two-dimensional hubbard model revealed by neural quantum states.arXiv preprint arXiv:2511.07566, 2025

    Christopher Roth, Ao Chen, Anirvan Sengupta, and An- toine Georges. Superconductivity in the two-dimensional hubbard model revealed by neural quantum states.arXiv preprint arXiv:2511.07566, 2025

    2025

  19. [27]

    Simu- lating superconductivity in mixed-dimensionalt ∥−j∥−j⊥ bilayers with neural quantum states.arXiv preprint arXiv:2602.10091, 2026

    Hannah Lange, Ao Chen, Antoine Georges, Fabian Grusdt, Annabelle Bohrdt, and Christopher Roth. Simu- lating superconductivity in mixed-dimensionalt ∥−j∥−j⊥ bilayers with neural quantum states.arXiv preprint arXiv:2602.10091, 2026

    2026

  20. [28]

    Empowering deep neural quantum states through efficient optimization.Nature Physics, 20(9):1476–1481, 2024

    Ao Chen and Markus Heyl. Empowering deep neural quantum states through efficient optimization.Nature Physics, 20(9):1476–1481, 2024

    2024

  21. [29]

    Universal jump in the superfluid density of two-dimensional super- fluids.Physical Review Letters, 39(19):1201, 1977

    David R Nelson and John Michael Kosterlitz. Universal jump in the superfluid density of two-dimensional super- fluids.Physical Review Letters, 39(19):1201, 1977

    1977

  22. [30]

    R. B. Laughlin. Superconducting Ground State of Non- interacting Particles Obeying Fractional Statistics.Phys- ical Review Letters, 60(25):2677–2680, June 1988. Pub- lisher: American Physical Society

    1988

  23. [31]

    A. L. Fetter, C. B. Hanna, and R. B. Laughlin. Random- phase approximation in the fractional-statistics gas. Physical Review B, 39(13):9679–9681, May 1989. Pub- lisher: American Physical Society

    1989

  24. [32]

    Halperin

    Yi-Hong Chen, Frank Wilczek, Edward Witten, and 6 Bertrand I. Halperin. On Anyon Superconductivity. International Journal of Modern Physics B, 3(7):1001– 1067, January 1989

    1989

  25. [33]

    X. G. Wen, Frank Wilczek, and A. Zee. Chiral spin states and superconductivity.Phys. Rev. B, 39:11413–11423, Jun 1989

    1989

  26. [34]

    X. G. Wen and A. Zee. Effective theory of the t- and p-breaking superconducting state.Phys. Rev. Lett., 62:2873–2876, Jun 1989

    1989

  27. [35]

    X. G. Wen and A. Zee. Compressibility and superfluidity in the fractional-statistics liquid.Phys. Rev. B, 41:240– 253, Jan 1990

    1990

  28. [36]

    Dung-Hai Lee and Matthew P. A. Fisher. Anyon su- perconductivity and the fractional quantum Hall effect. Physical Review Letters, 63(8):903–906, August 1989

    1989

  29. [37]

    Supercon- ductivity in the anyon model.Phys

    Yutaka Hosotani and Sumantra Chakravarty. Supercon- ductivity in the anyon model.Phys. Rev. B, 42:342–346, Jul 1990

    1990

  30. [38]

    Continuous transition between fractional quantum hall and super- fluid states.Phys

    Maissam Barkeshli and John McGreevy. Continuous transition between fractional quantum hall and super- fluid states.Phys. Rev. B, 89:235116, Jun 2014

    2014

  31. [39]

    Zaletel, Ashvin Vishwanath, and Yin-Chen He

    Jong Yeon Lee, Chong Wang, Michael P. Zaletel, Ashvin Vishwanath, and Yin-Chen He. Emergent multi-flavor qed3 at the plateau transition between fractional chern insulators: Applications to graphene heterostructures. Phys. Rev. X, 8:031015, Jul 2018

    2018

  32. [40]

    Chern- simons-matter conformal field theory on fuzzy sphere: Confinement transition of kalmeyer-laughlin chiral spin liquid.arXiv preprint arXiv:2507.19580, 2025

    Zheng Zhou, Chong Wang, and Yin-Chen He. Chern- simons-matter conformal field theory on fuzzy sphere: Confinement transition of kalmeyer-laughlin chiral spin liquid.arXiv preprint arXiv:2507.19580, 2025

    2025 arXiv

  33. [41]

    Microscopic mechanism of anyon superconductivity emerging from fractional chern insulators.Newton, 2(3), 2026

    Fabian Pichler, Clemens Kuhlenkamp, Michael Knap, and Ashvin Vishwanath. Microscopic mechanism of anyon superconductivity emerging from fractional chern insulators.Newton, 2(3), 2026

    2026

  34. [42]

    Re- alization of the hofstadter hamiltonian with ultracold atoms in optical lattices.arXiv preprint arXiv:1308.0321, 2013

    Monika Aidelsburger, Marcos Atala, Michael Lohse, Julio T Barreiro, B Paredes, and Immanuel Bloch. Re- alization of the hofstadter hamiltonian with ultracold atoms in optical lattices.arXiv preprint arXiv:1308.0321, 2013

    2013 arXiv

  35. [43]

    Hubbard model physics in transi- tion metal dichalcogenide moir´ e bands.Physical review letters, 121(2):026402, 2018

    Fengcheng Wu, Timothy Lovorn, Emanuel Tutuc, and Allan H MacDonald. Hubbard model physics in transi- tion metal dichalcogenide moir´ e bands.Physical review letters, 121(2):026402, 2018

    2018

  36. [44]

    Quantum electrodynamics in 2+ 1 dimensions as the organizing principle of a triangular lattice antifer- romagnet.Physical Review X, 14(2):021010, 2024

    Alexander Wietek, Sylvain Capponi, and Andreas M L¨ auchli. Quantum electrodynamics in 2+ 1 dimensions as the organizing principle of a triangular lattice antifer- romagnet.Physical Review X, 14(2):021010, 2024

    2024

  37. [45]

    High-accuracy variational monte carlo for frustrated magnets with deep neural networks.Physical Review B, 108(5):054410, 2023

    Christopher Roth, Attila Szab´ o, and Allan H MacDon- ald. High-accuracy variational monte carlo for frustrated magnets with deep neural networks.Physical Review B, 108(5):054410, 2023

    2023

  38. [46]

    To avoid confusion, we note that the variance of thetotal energy is zero for an eigenstate but the variance of the kineticenergy can still be finite

  39. [47]

    Insulator, metal, or superconductor: The criteria

    Douglas J Scalapino, Steven R White, and Shoucheng Zhang. Insulator, metal, or superconductor: The criteria. Physical Review B, 47(13):7995, 1993

    1993

  40. [48]

    Variance ex- trapolation method for neural-network variational monte carlo.arXiv preprint arXiv:2308.02471, 2023

    Weizhong Fu, Weiluo Ren, and Ji Chen. Variance ex- trapolation method for neural-network variational monte carlo.arXiv preprint arXiv:2308.02471, 2023. 7 flux threading polarization response FIG. 5. Diagram of the tori simulated in this work. Here we show a √ 3L× √ 3Lcluster wi...

    2023 arXiv

  41. [49]

    The exception is quantum critical phases whereC ˆO(r)∼1/|r| η at large|r|

    Structure factor and quantum criticality For an operator ˆO, we define the connected spatial correlation function as follows: C ˆO(r) = 1 N X x h ⟨ ˆO† x+r ˆOx⟩ − ⟨ˆOx+r⟩⟨ ˆOx⟩ i .(C1) For phases with long range order,C ˆO(r) should be non-zero as|r| → ∞, while in most other c...

  42. [50]

    2 we present evidence that there are critical charge fluctuations in the vicinity of the QCP

    Kinetic energy fluctuations In the main text Fig. 2 we present evidence that there are critical charge fluctuations in the vicinity of the QCP. Here we show that there are critical kinetic energy fluctuations as well, which persist to finite doping. First, we define 10 5.0 7.5...

  43. [51]

    The charge structure factor is 0 atk=0and positive definite, therefore we can fit the low-|k|behavior to a power law

    Fits to structure factors In this section we show details of the fits to the small|k|behavior of the charge and energy structure factors. The charge structure factor is 0 atk=0and positive definite, therefore we can fit the low-|k|behavior to a power law. The results forU= 4, ...

  44. [52]

    As defined in App

    Extraction of superfluid stiffness and BKT transition temperature We compute the superconducting stiffness from flux response. As defined in App. B, the fluxϕ y enters as a uniform vector potentialeA= 0, ϕy√ 3L . The superfluid weight in theydirection is the curvature of the g...

  45. [53]

    (D2) may be approximated using a finite difference method on the larger systems, Ds = 4√ 3 E(ϕy =ϕ s)−E(ϕ y = 0) ϕ2s ,(D4) where we setϕ s = 0.4×2π

    Extracting stiffness from finite difference Assuming that the ground-state energy is quadratic in flux nearϕ y = 0, Eq. (D2) may be approximated using a finite difference method on the larger systems, Ds = 4√ 3 E(ϕy =ϕ s)−E(ϕ y = 0) ϕ2s ,(D4) where we setϕ s = 0.4×2π. This gre...

  46. [54]

    10(b) the superfluid stiffness is shown on the 192 and 432 site clusters atδ= 1/24 as a function ofU

    Finite size effects of superfluid stiffness In Fig. 10(b) the superfluid stiffness is shown on the 192 and 432 site clusters atδ= 1/24 as a function ofU. Away fromU c the superfluid stiffness is approximately size invariant, but around the critical point the stiffness has a la...

  47. [55]

    The visible-visible couplings have a 2×2 sublattice structure such that they have the same unit cell as the Hamiltonian couplings

    Hyperparameters and training procedure For all of the calculations a hidden fermion Pfaffian state wavefunction [24] was used in conjunction with a residual convolutional neural network (R-CNN) with 36 features, 8 layers and 8 hidden fermions. The visible-visible couplings hav...

    2000

  48. [56]

    Further- more, the superfluid stiffness is computed from the dependence of the ground-state energy on the threaded flux

    Zero-variance extrapolation Throughout this work we compute gaps by considering the difference in energy between two eigenstates. Further- more, the superfluid stiffness is computed from the dependence of the ground-state energy on the threaded flux. Thus, the error in these r...

  49. [57]

    In Table III the approx- imate time needed for 2000 training iterations at half filling is shown as a function of system size

    Numerical Resources The data shown in this paper required approximately 10 5 hours on NVIDIA H200 GPUs. In Table III the approx- imate time needed for 2000 training iterations at half filling is shown as a function of system size. Upon hole doping, the computation becomes slig...

    2000

Pith tools

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