pith. sign in

arxiv: 2506.05050 · v2 · submitted 2025-06-05 · ❄️ cond-mat.str-el · cond-mat.dis-nn· cond-mat.quant-gas· quant-ph

Hybrid between biologically and quantum-inspired many-body states

Miha Srdin\v{s}ek , Xavier Waintal This is my paper

Pith reviewed 2026-05-19 11:05 UTC · model grok-4.3

classification ❄️ cond-mat.str-el cond-mat.dis-nncond-mat.quant-gasquant-ph
keywords perceptrainvariational Monte CarloGreen function Monte Carlotransverse-field Ising modelRydberg atomstensor networkneural network ansatzquantum phase transition
0
0 comments X

The pith

A network of perceptrains delivers ground-state energies accurate to 10^{-5}–10^{-6} for a 10×10 long-range Ising model using ranks of only 2–5.

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

The paper introduces perceptrains as a hybrid between perceptrons and tensor networks to build variational many-body wave functions. A simple network of these objects is optimized with variational Monte Carlo and Green function Monte Carlo on the transverse-field Ising model with 1/r^6 antiferromagnetic couplings on a 10×10 lattice. The approach achieves relative energy errors of order 10^{-5} in VMC and 10^{-6} in GFMC throughout the phase diagram, including near the quantum critical point, while employing very small ranks and a single initialization with fixed hyperparameters. This structure retains local optimization, dynamic rank growth, and compression from tensor networks inside a neural-network-like layout. The result suggests that two-dimensional quantum states can be represented accurately without the parameter explosion typical of matrix-product states.

Core claim

A network of perceptrains forms a variational ansatz whose ground-state energy for the 10×10 transverse-field Ising model with long-range 1/r^6 interactions is accurate to relative precision ~10^{-5} under VMC and ~10^{-6} under GFMC in every regime of the phase diagram, including the vicinity of the quantum phase transition, when only ranks 2–5 are used and optimization proceeds from one initial condition with fixed hyperparameters.

What carries the argument

The perceptrain, a perceptron generalized to inherit local optimization, dynamic parameter increase, state compression, and quantum-inspired structure from tensor networks.

If this is right

  • The entire phase diagram is accessible from one set of hyperparameters and a single initial condition.
  • Ranks of only 2–5 suffice where matrix product states typically require thousands of parameters.
  • Local optimization akin to DMRG can be performed inside the neural-network layout.
  • States can be compressed on the fly during the variational procedure.
  • The method applies directly to Rydberg-atom platforms proposed for quantum annealing.

Where Pith is reading between the lines

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

  • Dynamic rank adjustment during optimization may allow the ansatz to adapt to critical fluctuations without prior knowledge of the correlation length.
  • The same hybrid construction could be tested on other two-dimensional models with power-law interactions to check whether small ranks remain sufficient.
  • Combining perceptrains with quantum-computing simulators might accelerate preparation of low-energy states for annealing protocols.
  • The biological–quantum hybridization points to a route for representing higher-dimensional states that remain intractable for pure tensor networks.

Load-bearing premise

A network of perceptrains limited to ranks 2–5 can represent the ground state faithfully across the full phase diagram without requiring multiple random starts or later adjustments.

What would settle it

A calculation on the same 10×10 lattice or a modestly larger one that yields relative energy error larger than 10^{-4} in any parameter regime when ranks remain capped at 5 and the single-initialization protocol is followed.

Figures

Figures reproduced from arXiv: 2506.05050 by Miha Srdin\v{s}ek, Xavier Waintal.

Figure 1
Figure 1. Figure 1: FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. A standard perceptron defines a hyperplane in the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. A representation of the PN ansatz Eq. ( [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Phase diagram of the 2D model on a 10 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Comparison between the perceptrain based ansatz [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Best obtained energy as a function of the V-score [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Relative error in energy along the optimization of the [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. The V-score along the dynamical- [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9. Evolution of the variational wave function [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10. V-score (right axis) and staggered magnetization [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11. Linear zero variance extrapolation of the ground [PITH_FULL_IMAGE:figures/full_fig_p010_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: a) under the curve E(τ ), κ = Z [E(τ ) − E 0 ]dτ, (24) which is a very good metric for the quality of the vari￾10−6 10−5 10−4 10−3 10−2 δ E / E 0 (a) y = x MPS χ = 2 MPS χ = 8 MPS χ = 16 PN χ = 2 PN χ = 3 PN χ = 6 10−6 10−5 10−4 10−3 10−2 10−1 Nvar[H]/E 2 10−5 10−4 10−3 10−2 10−1 100 ˜κ (b) y = x hx = 10.0 hx = 7.0 hx = 5.0 hx = 3.0 hx = 2.5 FIG. 13. The best relative energy error (a) and ˜κ from Eq. (24)… view at source ↗
Figure 14
Figure 14. Figure 14: FIG. 14. Role of the guiding wave function for the per [PITH_FULL_IMAGE:figures/full_fig_p012_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: FIG. 15. Relative energy error versus rank for the 1D model [PITH_FULL_IMAGE:figures/full_fig_p013_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: FIG. 16. Relative energy error along the epoch for the 2D model at [PITH_FULL_IMAGE:figures/full_fig_p014_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: FIG. 17. The V-score along the VMC optimization of the [PITH_FULL_IMAGE:figures/full_fig_p014_17.png] view at source ↗
read the original abstract

Deep neural networks can represent very different sorts of functions, including complex quantum many-body states. Tensor networks can also represent these states, have more structure and are easier to optimize. However, they can be prohibitively costly computationally in two or higher dimensions. Here, we propose a generalization of the perceptron -- the perceptrain -- which borrows features from the two different formalisms. We construct variational many-body ansatz from a simple network of perceptrains. The network can be thought of as a neural network with a few distinct features inherited from tensor networks. These include efficient local optimization akin to the density matrix renormalization algorithm, instead of optimizing all the parameters at once; the possibility to dynamically increase the number of parameters during the optimization; the possibility to compress the state; and a structure that remains quantum-inspired. We showcase the ansatz using a combination of variational Monte Carlo (VMC) and Green function Monte Carlo (GFMC) on a $10\times 10$ transverse field quantum Ising model with a long-range $1/r^6$ antiferromagnetic interaction. The model corresponds to the Rydberg (cold) atoms platform proposed for quantum annealing. We consistently find a very high relative accuracy for the ground state energy, around $10^{-5}$ for VMC and $10^{-6}$ for GFMC in all regimes of parameters, including in the vicinity of the quantum phase transition. We use very small ranks ($\sim 2$-$5$) of perceptrains, as opposed to multiples of thousand used in matrix product states. The optimization of the energy was very robust. The entire phase diagram was found with a single initial condition and a fixed set of hyperparameters.

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.

Referee Report

2 major / 0 minor

Summary. The manuscript introduces the 'perceptrain' as a hybrid variational ansatz that combines perceptron-style neural networks with tensor-network-inspired features such as local optimization, dynamic rank adjustment, compression, and quantum-inspired structure. The authors construct a network of perceptrains and apply it to the 10×10 transverse-field Ising model with long-range 1/r^6 antiferromagnetic interactions using both variational Monte Carlo (VMC) and Green-function Monte Carlo (GFMC). They report relative ground-state energy accuracies of order 10^{-5} (VMC) and 10^{-6} (GFMC) across the full phase diagram, including near the quantum phase transition, achieved with small perceptrain ranks (2–5) and a single initialization plus fixed hyperparameters.

Significance. If the numerical claims hold, the perceptrain network offers a promising, computationally tractable route to high-accuracy variational states in two dimensions that retains some tensor-network optimization advantages while avoiding their prohibitive cost. The combination of VMC and GFMC, the use of very small ranks relative to matrix-product states, and the reported performance on a long-range model relevant to Rydberg platforms are notable strengths. The work provides concrete evidence that a modest hybrid architecture can reach high accuracy without extensive hyperparameter tuning.

major comments (2)
  1. [Numerical results] Numerical results section: the reported relative accuracies of ~10^{-5} for VMC and ~10^{-6} for GFMC are stated without explicit statistical error bars on the Monte Carlo estimates or details on the convergence criteria and data-exclusion rules employed in the GFMC runs. This information is necessary to evaluate whether the quoted precision is statistically reliable, especially near the quantum phase transition.
  2. [Optimization procedure] Optimization and robustness discussion: the claim that the entire phase diagram was obtained with a single initial condition and a fixed set of hyperparameters, and that the optimization is 'very robust,' is not accompanied by statistics from multiple independent runs or variance estimates across random seeds. In VMC and GFMC studies of models near a QPT, such checks are standard to confirm that the reported energy is not an artifact of a favorable basin.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their positive evaluation of the manuscript and for the constructive major comments. We address each point below and indicate the revisions that will be incorporated in the next version of the manuscript.

read point-by-point responses

  1. Referee: [Numerical results] Numerical results section: the reported relative accuracies of ~10^{-5} for VMC and ~10^{-6} for GFMC are stated without explicit statistical error bars on the Monte Carlo estimates or details on the convergence criteria and data-exclusion rules employed in the GFMC runs. This information is necessary to evaluate whether the quoted precision is statistically reliable, especially near the quantum phase transition.

    Authors: We agree that explicit statistical error bars and procedural details are required to substantiate the quoted accuracies. In the revised manuscript we will add the Monte Carlo statistical uncertainties to all reported energies; these uncertainties are typically 10^{-7} or smaller and thus do not affect the claimed relative accuracies. We will also include a concise description of the GFMC convergence protocol, specifying the number of iterations, the equilibration criterion, and the rule used to discard initial transient data. revision: yes

  2. Referee: [Optimization procedure] Optimization and robustness discussion: the claim that the entire phase diagram was obtained with a single initial condition and a fixed set of hyperparameters, and that the optimization is 'very robust,' is not accompanied by statistics from multiple independent runs or variance estimates across random seeds. In VMC and GFMC studies of models near a QPT, such checks are standard to confirm that the reported energy is not an artifact of a favorable basin.

    Authors: We acknowledge that variance estimates from multiple independent optimizations constitute standard practice near a quantum phase transition. While the original results were obtained from a single initialization with fixed hyperparameters, the consistent accuracy across the full phase diagram already provides supporting evidence. To meet the referee's request, the revised manuscript will contain a brief discussion of the optimization protocol together with results from a small set of additional independent runs (different random seeds) that quantify the observed variance; these checks confirm that the reported energies are not basin-specific artifacts. revision: yes

Circularity Check

0 steps flagged

No significant circularity; variational results are self-contained against external Hamiltonian benchmarks

full rationale

The manuscript proposes a hybrid perceptrain network ansatz by merging perceptron and tensor-network structural features, then evaluates it through direct variational optimization and Monte Carlo sampling on the long-range transverse-field Ising Hamiltonian. Reported relative accuracies (10^{-5} VMC, 10^{-6} GFMC) are obtained by comparing the minimized variational energies to independent reference values for the same model, without any redefinition of fitted parameters as predictions, self-definitional loops in the ansatz construction, or load-bearing reliance on self-citations whose content reduces to the present results. The single-initialization robustness statement is a numerical observation about optimization behavior rather than a derivation that collapses to its own inputs by construction. The entire chain remains externally falsifiable via the Hamiltonian and sampling procedure.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 1 invented entities

The approach rests on the variational principle for ground-state energy, the ability of the perceptrain network to represent relevant states with low rank, and standard Monte Carlo sampling assumptions. No new particles or forces are postulated.

free parameters (2)
  • perceptrain rank r
    Chosen in the range 2-5; controls the number of parameters per perceptrain and is increased dynamically during optimization.
  • network hyperparameters
    Fixed set used for the entire phase diagram; includes learning rates and update schedules for the perceptrain weights.
axioms (2)
  • standard math The variational energy obtained by Monte Carlo sampling is an upper bound to the true ground-state energy.
    Invoked when reporting relative accuracy of the optimized ansatz.
  • domain assumption The perceptrain network can be contracted and updated locally in a manner analogous to DMRG sweeps.
    Central to the claimed computational advantage over generic neural networks.
invented entities (1)
  • perceptrain no independent evidence
    purpose: Hybrid unit that generalizes the perceptron with tensor-network-inspired local optimization and compression.
    New building block introduced to combine features of neural networks and tensor networks.

pith-pipeline@v0.9.0 · 5852 in / 1657 out tokens · 30722 ms · 2026-05-19T11:05:50.233690+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

51 extracted references · 51 canonical work pages · 2 internal anchors

  1. [1]

    Due to the presence of the lattice, GFMC can be formulated directly in the imaginary time continuum, without resorting to Trotterization or other discretization scheme

    GFMC in a nutshell GFMC is a lattice counterpart to the Diffusion Monte Carlo approach [41]. Due to the presence of the lattice, GFMC can be formulated directly in the imaginary time continuum, without resorting to Trotterization or other discretization scheme. For completeness, we describe the method below and refer to the literature for the associ- ated...

    work page

  2. [2]

    12a shows an example of VMC opti- mization for the 1D model (black line)

    Benchmark for the 1D model The inset of Fig. 12a shows an example of VMC opti- mization for the 1D model (black line). During the op- timization, we perform many GFMC calculations (green points). We find consistently that GFMC improves the accuracy by one or two orders of magnitude until the VMC ansatz is so close to the ground state that one is dominated...

    work page

  3. [3]

    EPIQ”, the PEPR “EQUBITFLY

    Benchmark for the 2D model In this section, we present our reference calculations for the 2D model. We have performed various GFMC runs for different ans¨ atze and values of hx and checked the consistency of the results. We have also, for com- pleteness, compared the results with what could be ob- tained from a zero variance extrapolation [15, 40, 46] (se...

    work page 2030

  4. [4]

    Attention Is All You Need

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, Attention is all you need (2023), arXiv:1706.03762

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  5. [5]

    Nunez Fernandez, M

    Y. Nunez Fernandez, M. K. Ritter, M. Jeannin, J.-W. Li, T. Kloss, T. Louvet, S. Terasaki, O. Parcollet, J. von Delft, H. Shinaoka, and X. Waintal, SciPost Physics 18, 104 (2025)

    work page 2025

  6. [6]

    S. R. White, Physical Review Letters 69, 2863 (1992)

    work page 1992

  7. [7]

    S. R. White and R. M. Noack, Physical Review Letters 68, 3487 (1992)

    work page 1992

  8. [8]

    Corboz, R

    P. Corboz, R. Or´ us, B. Bauer, and G. Vidal, Phys. Rev. B 81, 165104 (2010)

    work page 2010

  9. [9]

    Xu, C.-M

    H. Xu, C.-M. Chung, M. Qin, U. Schollw¨ ock, S. R. White, and S. Zhang, Science 384, adh7691 (2024)

    work page 2024

  10. [10]

    Carleo and M

    G. Carleo and M. Troyer, Science 355, 602 (2017)

    work page 2017

  11. [11]

    J. Kim, G. Pescia, B. Fore, J. Nys, G. Carleo, S. Gandolfi, M. Hjorth-Jensen, and A. Lovato, Neural- network quantum states for ultra-cold Fermi gases (2023), arXiv:2305.08831

    work page arXiv 2023

  12. [12]

    Lovato, C

    A. Lovato, C. Adams, G. Carleo, and N. Rocco, Physical Review Research 4, 043178 (2022)

    work page 2022

  13. [13]

    Robledo Moreno, G

    J. Robledo Moreno, G. Carleo, A. Georges, and J. Stokes, Proceedings of the National Academy of Sciences 119, e2122059119 (2022)

    work page 2022

  14. [14]

    Vicentini, A

    F. Vicentini, A. Biella, N. Regnault, and C. Ciuti, Phys- ical Review Letters 122, 250503 (2019)

    work page 2019

  15. [15]

    Hibat-Allah, M

    M. Hibat-Allah, M. Ganahl, L. E. Hayward, R. G. Melko, and J. Carrasquilla, Physical Review Research 2, 023358 (2020)

    work page 2020

  16. [16]

    K. Choo, T. Neupert, and G. Carleo, Physical Review B 100, 125124 (2019)

    work page 2019

  17. [17]

    Nomura, A

    Y. Nomura, A. S. Darmawan, Y. Yamaji, and M. Imada, Physical Review B 96, 205152 (2017)

    work page 2017

  18. [18]

    Chen and M

    A. Chen and M. Heyl, Nature Physics 20, 1476 (2024)

    work page 2024

  19. [19]

    S. Dash, L. Gravina, F. Vicentini, M. Ferrero, and A. Georges, Commun Phys 8, 1 (2025)

    work page 2025

  20. [20]

    A. W. Sandvik and G. Vidal, Physical Review Letters 99, 220602 (2007)

    work page 2007

  21. [21]

    Liu, Y.-Z

    W.-Y. Liu, Y.-Z. Huang, S.-S. Gong, and Z.-C. Gu, Phys- ical Review B 103, 235155 (2021)

    work page 2021

  22. [22]

    Naumann, E

    J. Naumann, E. L. Weerda, M. Rizzi, J. Eisert, and P. Schmoll, SciPost Physics Lecture Notes , 086 (2024)

    work page 2024

  23. [23]

    M. Wang, Y. Pan, Z. Xu, X. Yang, G. Li, and A. Ci- chocki, Tensor Networks Meet Neural Networks: A Sur- 16 vey and Future Perspectives (2023), arXiv:2302.09019

    work page arXiv 2023

  24. [24]

    Sfondrini, J

    A. Sfondrini, J. Cerrillo, N. Schuch, and J. I. Cirac, Phys- ical Review B 81, 214426 (2010)

    work page 2010

  25. [25]

    Schuch, M

    N. Schuch, M. M. Wolf, F. Verstraete, and J. I. Cirac, Physical Review Letters 100, 040501 (2008)

    work page 2008

  26. [26]

    Glasser, N

    I. Glasser, N. Pancotti, M. August, I. D. Rodriguez, and J. I. Cirac, Physical Review X 8, 011006 (2018)

    work page 2018

  27. [27]

    Schollwoeck, Reviews of Modern Physics 77, 259 (2005)

    U. Schollwoeck, Reviews of Modern Physics 77, 259 (2005)

    work page 2005

  28. [28]

    Nunez Fernandez, M

    Y. Nunez Fernandez, M. Jeannin, P. T. Dumitrescu, T. Kloss, J. Kaye, O. Parcollet, and X. Waintal, Physical Review X 12, 041018 (2022)

    work page 2022

  29. [29]

    Dolgov and D

    S. Dolgov and D. Savostyanov, Computer Physics Com- munications 246, 106869 (2020)

    work page 2020

  30. [30]

    Becca and S

    F. Becca and S. Sorella, Quantum Monte Carlo Ap- proaches for Correlated Systems, 1st ed. (Cambridge Uni- versity Press, 2017)

    work page 2017

  31. [31]

    Feasibility of performing quantum chemistry calculations on quantum computers

    T. Louvet, T. Ayral, and X. Waintal, On the feasibility of performing quantum chemistry calculations on quantum computers (2024), arXiv:2306.02620

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  32. [32]

    Schollw¨ ock, Annals of Physics January 2011 Special Issue, 326, 96 (2011)

    U. Schollw¨ ock, Annals of Physics January 2011 Special Issue, 326, 96 (2011)

    work page 2011

  33. [33]

    Sorella, Physical Review Letters 80, 4558 (1998)

    S. Sorella, Physical Review Letters 80, 4558 (1998)

    work page 1998

  34. [34]

    Sorella and L

    S. Sorella and L. Capriotti, Physical Review B 61, 2599 (2000)

    work page 2000

  35. [35]

    Casula and S

    M. Casula and S. Sorella, The Journal of Chemical Physics 119, 6500 (2003)

    work page 2003

  36. [36]

    Filippi and S

    C. Filippi and S. Fahy, The Journal of Chemical Physics 112, 3523 (2000)

    work page 2000

  37. [37]

    C. Roth, A. Szabo, and A. MacDonald, Physical Review B 108, 054410 (2023)

    work page 2023

  38. [38]

    D. Wu, R. Rossi, F. Vicentini, N. Astrakhantsev, F. Becca, X. Cao, J. Carrasquilla, F. Ferrari, A. Georges, M. Hibat-Allah, M. Imada, A. M. L¨ auchli, G. Mazzola, A. Mezzacapo, A. Millis, J. Robledo Moreno, T. Neupert, Y. Nomura, J. Nys, O. Parcollet, R. Pohle, I. Romero, M. Schmid, J. M. Silvester, S. Sorella, L. F. Tocchio, L. Wang, S. R. White, A. Wiet...

    work page 2024

  39. [39]

    Kalinowski, R

    M. Kalinowski, R. Samajdar, R. G. Melko, M. D. Lukin, S. Sachdev, and S. Choi, Physical Review B 105, 174417 (2022)

    work page 2022

  40. [40]

    Ebadi, T

    S. Ebadi, T. T. Wang, H. Levine, A. Keesling, G. Se- meghini, A. Omran, D. Bluvstein, R. Samajdar, H. Pich- ler, W. W. Ho, S. Choi, S. Sachdev, M. Greiner, V. Vuletic, and M. D. Lukin, Nature 595, 227 (2021)

    work page 2021

  41. [41]

    Samajdar, W

    R. Samajdar, W. W. Ho, H. Pichler, M. D. Lukin, and S. Sachdev, Physical Review Letters 124, 103601 (2020)

    work page 2020

  42. [42]

    A. Kalz, A. Honecker, S. Fuchs, and T. Pruschke, The European Physical Journal B 65, 533 (2008)

    work page 2008

  43. [43]

    Kashima and M

    T. Kashima and M. Imada, J. Phys. Soc. Jpn. 70, 2287 (2001)

    work page 2001

  44. [44]

    P. J. Reynolds, J. Tobochnik, and H. Gould, Computer in Physics 4, 662 (1990)

    work page 1990

  45. [45]

    Trivedi and D

    N. Trivedi and D. M. Ceperley, Physical Review B 41, 4552 (1990)

    work page 1990

  46. [46]

    Hida, Physical Review B 45, 2207 (1992)

    K. Hida, Physical Review B 45, 2207 (1992)

    work page 1992

  47. [47]

    Waintal, Physical Review B 73, 075417 (2006)

    X. Waintal, Physical Review B 73, 075417 (2006)

    work page 2006

  48. [48]

    Mora and X

    C. Mora and X. Waintal, Physical Review Letters 99, 030403 (2007)

    work page 2007

  49. [49]

    Nomura, J

    Y. Nomura, J. Phys.: Condens. Matter 33, 174003 (2021)

    work page 2021

  50. [50]

    Cheng, L

    S. Cheng, L. Wang, T. Xiang, and P. Zhang, Physical Review B 99, 155131 (2019)

    work page 2019

  51. [51]

    Shi, L.-M

    Y.-Y. Shi, L.-M. Duan, and G. Vidal, Phys. Rev. A 74, 022320 (2006)

    work page 2006