arxiv: 2604.15888 · v1 · submitted 2026-04-17 · ❄️ cond-mat.str-el · cond-mat.dis-nn· physics.comp-ph· quant-ph

Recognition: unknown

Enhancing Neural-Network Variational Monte Carlo through Basis Transformation

Zhixuan Liu , Dongheng Qian , Jing Wang

Authors on Pith no claims yet

Pith reviewed 2026-05-10 07:36 UTC · model grok-4.3

classification ❄️ cond-mat.str-el cond-mat.dis-nnphysics.comp-phquant-ph

keywords neural-network variational Monte Carlobasis transformationGaussian basishomogeneous electron gasvariational energyFermiNetWigner crystallocality parameter

0 comments

The pith

A learnable Gaussian basis with one locality parameter α makes neural-network variational Monte Carlo represent the ground state more accurately without enlarging the network.

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

The paper sets out to show that a physically motivated change of basis can improve the performance of neural-network variational Monte Carlo by turning the target ground-state wave function into a form that existing networks learn more easily. A single adjustable parameter α controls the spatial decay of Gaussian orbitals, and optimizing it lowers the variational energy for the three-dimensional homogeneous electron gas while adding almost no extra cost. The same change also sharpens the predicted location of the Fermi-liquid to Wigner-crystal transition when a message-passing network is used. Because the transformation leaves the neural-network architecture untouched, the method can be added to FermiNet, message-passing networks, or any other continuous-space ansatz. If the claim holds, accuracy gains in variational Monte Carlo no longer need to come only from bigger or more elaborate networks.

Core claim

By writing the many-body wave function in a Gaussian basis controlled by a single learnable locality parameter α, the authors reshape the target ground state so that standard neural-network ansatzes achieve lower variational energies on the three-dimensional homogeneous electron gas. The transformation preserves the variational upper bound while introducing negligible overhead, and it yields a sharper estimate of the density at which the Fermi liquid gives way to the Wigner crystal when message-passing networks are employed.

What carries the argument

a Gaussian basis transformation controlled by one learnable locality parameter α that rescales the spatial extent of the orbitals and thereby alters the representation of the ground-state wave function

If this is right

The variational energy is lowered for both FermiNet and message-passing neural networks on the three-dimensional homogeneous electron gas.
Message-passing networks locate the Fermi-liquid to Wigner-crystal transition density more precisely when the basis parameter is optimized.
The method adds only one extra variational parameter and combines directly with any existing neural-network architecture.
Accuracy improvements can be obtained by making the target state easier to represent rather than by increasing the complexity of the neural ansatz.

Where Pith is reading between the lines

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

The same single-parameter basis change could be tested on molecular systems to check whether optimal α tracks physical length scales such as bond distances.
Joint optimization of α together with neural-network weights may produce larger gains than optimizing either alone.
Analogous locality parameters might be introduced in other basis sets, such as plane waves or Slater-type orbitals, for continuous-space variational calculations.

Load-bearing premise

The optimized Gaussian basis must still span the exact ground state and must not introduce any uncontrolled bias into the variational energy estimate.

What would settle it

An independent variational Monte Carlo run on the three-dimensional homogeneous electron gas at fixed density that finds the energy with optimized α to be higher than the energy obtained with the identical neural network in the original basis.

Figures

Figures reproduced from arXiv: 2604.15888 by Dongheng Qian, Jing Wang, Zhixuan Liu.

**Figure 2.** Figure 2: FIG. 2. Comparison of ground-state energies with and with [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Figure 3: FIG. 3. Ground-state energy as a function of [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

**Figure 4.** Figure 4: FIG. 4. Electron-electron correlations after basis transforma [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

read the original abstract

Neural-network variational Monte Carlo (NNVMC) has emerged as a powerful tool for solving quantum many-body problems, yet systematic pathways for improving its accuracy remain largely heuristic. Here, we introduce a physically motivated basis transformation for NNVMC that enhances variational expressivity without increasing the complexity of the neural-network ansatz itself. By formulating the many-body wave function in a Gaussian basis, we introduce a single learnable locality parameter, $\alpha$, that reshapes the target ground state into a more learnable representation. This approach introduces minimal computational overhead and can be readily combined with existing neural-network architectures. Using the three-dimensional homogeneous electron gas as a benchmark, we show that the optimized basis transformation consistently lowers the variational energy for both FermiNet and message-passing neural-network architectures. Notably, for the latter, it enables a more precise determination of the Fermi liquid to Wigner crystal phase transition. More broadly, our results highlight basis transformation as a new route to improving NNVMC in continuous space, showing that accuracy can be enhanced not only by refining the ansatz but also by making the target ground state easier to represent.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note private letter to a colleague

Optimizing a Gaussian locality parameter inside NNVMC lowers energies on the 3D electron gas for FermiNet and message-passing nets.

read the letter

The paper's main point is that a single learnable parameter alpha, which sets the width in a Gaussian basis for the many-body wave function, improves the accuracy of neural-network variational Monte Carlo on the three-dimensional homogeneous electron gas. This approach is new because it applies the basis transformation inside the NNVMC framework for continuous space problems and optimizes alpha variationally along with the network parameters. The authors show lower energies for both the FermiNet architecture and a message-passing neural network. They also report that the phase transition from Fermi liquid to Wigner crystal is resolved more clearly with the message-passing net when alpha is included. The extra cost is minimal, and they derive the necessary derivatives for the kinetic energy term to keep the calculation correct. The results support the claim that reshaping the target state this way makes it easier for the neural ansatz to represent without adding complexity to the network. The variational upper bound is maintained, as confirmed by the proper handling of the operators. One soft spot is that all the tests are on the uniform electron gas. It is not yet clear how well the method carries over to systems with external potentials or different particle statistics. The paper would be stronger with at least one additional example. This is useful for anyone working on neural variational methods for quantum many-body problems in continuous space. Readers looking for practical ways to enhance existing codes without major redesigns will find value here. The work is grounded enough and the benchmark is standard, so it deserves to go through peer review. I recommend sending it out for review. The idea is clean, the implementation is described, and the gains are shown on a relevant system.

Referee Report

2 major / 3 minor

Summary. The manuscript proposes a basis transformation for neural-network variational Monte Carlo (NNVMC) in which the many-body wave function is expressed in a Gaussian basis controlled by a single learnable locality parameter α. This transformation is applied to the 3D homogeneous electron gas and is shown to lower variational energies for both FermiNet and message-passing neural-network ansatzes while improving resolution of the Fermi-liquid to Wigner-crystal transition, all without increasing the complexity of the neural network itself.

Significance. If the reported energy reductions and improved phase-transition resolution hold under the stated conditions, the work supplies a lightweight, physically motivated route to enhancing NNVMC expressivity by reshaping the target state rather than enlarging the ansatz. The approach preserves the variational upper bound because α is optimized jointly inside the Monte Carlo loop, introduces only one extra parameter, and is compatible with existing architectures. These features constitute a genuine addition to the toolkit for continuous-space quantum many-body calculations.

major comments (2)

[§2.3, Eq. (12)] §2.3, Eq. (12): the chain-rule expression for the kinetic-energy operator after the Gaussian transformation is given, but the manuscript does not explicitly verify that the resulting local energy remains real-valued and that the Monte Carlo estimator remains unbiased when α is updated on the fly; a short numerical check or analytic argument confirming this would strengthen the central claim.
[§4.1, Table 1] §4.1, Table 1: the energy differences with and without the basis transformation are reported to several decimal places, yet no breakdown is provided of the statistical uncertainty arising from finite Monte Carlo sampling versus the uncertainty in the optimized α itself; this makes it hard to judge whether the quoted improvements exceed the combined error bars.

minor comments (3)

[Abstract] The abstract states that energies are lowered and the phase transition is better resolved but supplies no numerical values or error bars; a single sentence quantifying the typical energy gain (e.g., “∼0.001 Ha per electron”) would improve readability.
[Figure 3] Figure 3 caption: the color scale for the order-parameter plot is not labeled with units or a numerical range, making it difficult to compare the sharpness of the transition with and without the transformation.
[§3.2] §3.2: the optimization schedule for α (learning rate, update frequency, initialization) is described only qualitatively; a concise table or paragraph listing the hyper-parameters used for all reported runs would aid reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive assessment and the recommendation for minor revision. The comments identify useful opportunities to strengthen the presentation of the kinetic-energy implementation and the error analysis. We address each point below and have incorporated revisions to the manuscript.

read point-by-point responses

Referee: [§2.3, Eq. (12)] §2.3, Eq. (12): the chain-rule expression for the kinetic-energy operator after the Gaussian transformation is given, but the manuscript does not explicitly verify that the resulting local energy remains real-valued and that the Monte Carlo estimator remains unbiased when α is updated on the fly; a short numerical check or analytic argument confirming this would strengthen the central claim.

Authors: We thank the referee for highlighting this point. The Gaussian transformation is a real multiplicative prefactor, and the neural-network ansatz for the homogeneous electron gas is real-valued by construction. Application of the chain rule therefore yields a strictly real local energy. When α is optimized jointly with the network parameters inside the Monte Carlo loop, the estimator remains unbiased because each sample is drawn from the instantaneous wave function and the variational principle applies at every step, exactly as in standard VMC with variational parameters. To make this explicit we have added a short analytic paragraph in the revised §2.3 proving that the imaginary part vanishes identically, together with a brief numerical check (now included in the main text) confirming that the local-energy variance is consistent with real sampling and that the energy estimator converges without detectable bias. revision: yes
Referee: [§4.1, Table 1] §4.1, Table 1: the energy differences with and without the basis transformation are reported to several decimal places, yet no breakdown is provided of the statistical uncertainty arising from finite Monte Carlo sampling versus the uncertainty in the optimized α itself; this makes it hard to judge whether the quoted improvements exceed the combined error bars.

Authors: We agree that separating the two sources of uncertainty improves interpretability. The original Table 1 reported Monte Carlo statistical errors obtained via blocking analysis, but did not propagate the uncertainty arising from the finite optimization of α. In the revised manuscript we have updated Table 1 to display combined error bars: the Monte Carlo error is retained, while the uncertainty in α is estimated from the curvature of the energy surface at the optimum and added in quadrature. A new paragraph in §4.1 now describes the full error-propagation procedure. With these changes the reported energy reductions are shown to lie outside the combined uncertainties. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper's central construction introduces an explicit Gaussian basis transformation with one additional variational parameter α that is optimized jointly with the neural-network weights inside the standard NNVMC loop. The energy lowering follows directly from the enlarged variational manifold while the physical Hamiltonian and the variational upper-bound property remain unchanged; the coordinate change and chain-rule kinetic-energy evaluation are given explicitly. No load-bearing self-citation, self-definitional reparameterization, or fitted quantity renamed as a prediction appears in the derivation. The benchmark results on the 3D HEG are therefore independent evidence rather than a tautology.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The claim rests on one free parameter alpha and the domain assumption that a Gaussian-basis change leaves the variational principle intact.

free parameters (1)

alpha
Single learnable locality parameter that controls the Gaussian basis width and is optimized during the variational procedure.

axioms (1)

domain assumption The transformed wave function remains a valid variational ansatz whose energy is an upper bound to the true ground-state energy.
Invoked when claiming that lower variational energy after basis change constitutes an improvement.

pith-pipeline@v0.9.0 · 5505 in / 1154 out tokens · 39454 ms · 2026-05-10T07:36:24.020747+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

77 extracted references · 13 canonical work pages · 2 internal anchors

[1]

Hohenberg and W

P. Hohenberg and W. Kohn, Inhomogeneous electron gas, Phys. Rev.136, B864 (1964)

1964
[2]

Kohn and L

W. Kohn and L. J. Sham, Self-consistent equations in- cluding exchange and correlation effects, Phys. Rev.140, A1133 (1965)

1965
[3]

R. O. Jones and O. Gunnarsson, The density functional formalism, its applications and prospects, Rev. Mod. Phys.61, 689 (1989)

1989
[4]

S. R. White, Density matrix formulation for quantum renormalization groups, Phys. Rev. Lett.69, 2863 (1992)

1992
[5]

Schollw¨ ock, The density-matrix renormalization group, Rev

U. Schollw¨ ock, The density-matrix renormalization group, Rev. Mod. Phys.77, 259 (2005)

2005
[6]

Schollw¨ ock, The density-matrix renormalization group in the age of matrix product states, Ann

U. Schollw¨ ock, The density-matrix renormalization group in the age of matrix product states, Ann. Phys.326, 96 (2011)

2011
[7]

Or´ us, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Ann

R. Or´ us, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Ann. Phys.349, 117 (2014)

2014
[8]

J. I. Cirac, D. P´ erez-Garc´ ıa, N. Schuch, and F. Ver- straete, Matrix product states and projected entangled pair states: Concepts, symmetries, theorems, Rev. Mod. Phys.93, 045003 (2021)

2021
[9]

B. L. Hammond, W. A. Lester, and P. J. Reynolds,Monte Carlo methods in ab initio quantum chemistry, Vol. 1 (World Scientific, 1994)

1994
[10]

W. M. C. Foulkes, L. Mitas, R. J. Needs, and G. Ra- jagopal, Quantum monte carlo simulations of solids, Rev. Mod. Phys.73, 33 (2001)

2001
[11]

E. Y. Loh, J. E. Gubernatis, R. T. Scalettar, S. R. White, D. J. Scalapino, and R. L. Sugar, Sign problem in the nu- merical simulation of many-electron systems, Phys. Rev. B41, 9301 (1990)

1990
[12]

Li and H

Z.-X. Li and H. Yao, Sign-problem-free fermionic quan- tum monte carlo: Developments and applications, Annu. Rev. Condens. Matter Phys.10, 337 (2019)

2019
[13]

W. L. McMillan, Ground state of liquid he 4, Phys. Rev. 138, A442 (1965)

1965
[14]

Toulouse and C

J. Toulouse and C. J. Umrigar, Optimization of quantum monte carlo wave functions by energy minimization, J. Chem. Phys.126, 084102 (2007)

2007
[15]

Carleo and M

G. Carleo and M. Troyer, Solving the quantum many- body problem with artificial neural networks, Science 355, 602 (2017)

2017
[16]

D.-L. Deng, X. Li, and S. Das Sarma, Quantum entan- glement in neural network states, Phys. Rev. X7, 021021 (2017)

2017
[17]

Nomura, A

Y. Nomura, A. S. Darmawan, Y. Yamaji, and M. Imada, 6 Restricted boltzmann machine learning for solving strongly correlated quantum systems, Phys. Rev. B96, 205152 (2017)

2017
[18]

X.-Q. Sun, T. Nebabu, X. Han, M. O. Flynn, and X.- L. Qi, Entanglement features of random neural network quantum states, Phys. Rev. B106, 115138 (2022)

2022
[19]

K. Choo, G. Carleo, N. Regnault, and T. Neupert, Sym- metries and many-body excitations with neural-network quantum states, Phys. Rev. Lett.121, 167204 (2018)

2018
[20]

Ferrari, F

F. Ferrari, F. Becca, and J. Carrasquilla, Neural gutzwiller-projected variational wave functions, Phys. Rev. B100, 125131 (2019)

2019
[21]

Luo and B

D. Luo and B. K. Clark, Backflow transformations via neural networks for quantum many-body wave functions, Phys. Rev. Lett.122, 226401 (2019)

2019
[22]

K. Choo, A. Mezzacapo, and G. Carleo, Fermionic neural-network states for ab-initio electronic structure, Nat. Commun.11, 2368 (2020)

2020
[23]

Hibat-Allah, M

M. Hibat-Allah, M. Ganahl, L. E. Hayward, R. G. Melko, and J. Carrasquilla, Recurrent neural network wave func- tions, Phys. Rev. Res.2, 023358 (2020)

2020
[24]

J. R. Moreno, G. Carleo, A. Georges, and J. Stokes, Fermionic wave functions from neural-network con- strained hidden states, Proc. Natl. Acad. Sci. U.S.A.119, e2122059119 (2022)

2022
[25]

Pescia, J

G. Pescia, J. Han, A. Lovato, J. Lu, and G. Carleo, Neural-network quantum states for periodic systems in continuous space, Phys. Rev. Res.4, 023138 (2022)

2022
[26]

X. Li, Z. Li, and J. Chen, Ab initio calculation of real solids via neural network ansatz, Nat. Commun.13, 7895 (2022)

2022
[27]

L. L. Viteritti, R. Rende, and F. Becca, Transformer vari- ational wave functions for frustrated quantum spin sys- tems, Phys. Rev. Lett.130, 236401 (2023)

2023
[28]

B. Fore, J. M. Kim, G. Carleo, M. Hjorth-Jensen, A. Lovato, and M. Piarulli, Dilute neutron star matter from neural-network quantum states, Phys. Rev. Res.5, 033062 (2023)

2023
[29]

Hermann, J

J. Hermann, J. Spencer, K. Choo, A. Mezzacapo, W. M. C. Foulkes, D. Pfau, G. Carleo, and F. No´ e, Ab initio quantum chemistry with neural-network wavefunc- tions, Nat. Rev. Chem.7, 692 (2023)

2023
[30]

Wilson, S

M. Wilson, S. Moroni, M. Holzmann, N. Gao, F. Wu- darski, T. Vegge, and A. Bhowmik, Neural network ansatz for periodic wave functions and the homogeneous electron gas, Phys. Rev. B107, 235139 (2023)

2023
[31]

J. Lin, G. Goldshlager, and L. Lin, Explicitly antisym- metrized neural network layers for variational monte carlo simulation, J. Comput. Phys.474, 111765 (2023)

2023
[32]

Lange, A

H. Lange, A. Van de Walle, A. Abedinnia, and A. Bohrdt, From architectures to applications: a review of neural quantum states, Quantum Sci. Technol.9, 040501 (2024)

2024
[33]

Chen and M

A. Chen and M. Heyl, Empowering deep neural quantum states through efficient optimization, Nat. Phys.20, 1476 (2024)

2024
[34]

R. Li, H. Ye, D. Jiang, X. Wen, C. Wang, Z. Li, X. Li, D. He, J. Chen, W. Ren,et al., A computational frame- work for neural network-based variational monte carlo with forward laplacian, Nat. Mach. Intell.6, 209 (2024)

2024
[35]

Sprague and S

K. Sprague and S. Czischek, Variational monte carlo with large patched transformers, Commun. Phys.7, 90 (2024)

2024
[36]

Rende, S

R. Rende, S. Goldt, F. Becca, and L. L. Viteritti, Fine- tuning neural network quantum states, Phys. Rev. Res. 6, 043280 (2024)

2024
[37]

Zhang, R.-S

Q. Zhang, R.-S. Wang, and L. Wang, Neural canonical transformations for vibrational spectra of molecules, J. Chem. Phys.161, 024103 (2024)

2024
[38]

Y. Qian, T. Zhao, J. Zhang, T. Xiang, X. Li, and J. Chen, Describing landau level mixing in fractional quantum hall states with deep learning, Phys. Rev. Lett.134, 176503 (2025)

2025
[39]

C. Roth, A. Chen, A. Sengupta, and A. Georges, Superconductivity in the two-dimensional hubbard model revealed by neural quantum states (2025), arXiv:2511.07566 [cond-mat.supr-con]

work page arXiv 2025
[40]

Y. Gu, W. Li, H. Lin, B. Zhan, R. Li, Y. Huang, D. He, Y. Wu, T. Xiang, M. Qin, L. Wang, and D. Lv, Solving the hubbard model with neural quantum states (2025), arXiv:2507.02644 [cond-mat.str-el]

work page arXiv 2025
[41]

Valenti, Y

A. Valenti, Y. Vituri, Y. Yang, D. E. Parker, T. Soe- jima, J. Dong, M. A. Morales, A. Vishwanath, E. Berg, and S. Zhang, Quantum geometry driven crystallization: A neural-network variational monte carlo study (2025), arXiv:2512.07947 [cond-mat.str-el]

work page arXiv 2025
[42]

J. A. Sobral, M. Perle, and M. S. Scheurer, Physics- informed transformers for electronic quantum states, Nat. Commun.16, 10811 (2025)

2025
[43]

Zhang and D

L. Zhang and D. Luo, Neural transformer backflow for solving momentum-resolved ground states of strongly correlated materials (2025), arXiv:2509.09275 [cond- mat.str-el]

work page arXiv 2025
[44]

Chen, Z.-Q

A. Chen, Z.-Q. Wan, A. Sengupta, A. Georges, and C. Roth, Neural network-augmented pfaffian wave- functions for scalable simulations of interacting fermions (2025), arXiv:2507.10705 [cond-mat.str-el]

work page arXiv 2025
[45]

Gerard, M

L. Gerard, M. Scherbela, H. Sutterud, W. M. C. Foulkes, and P. Grohs, Transferable neural wavefunctions for solids, Nat. Comput. Sci.5, 1147–1157 (2025)

2025
[46]

Zaklama, M

T. Zaklama, M. Geier, and L. Fu, Large electron model: A universal ground state predictor (2026), arXiv:2603.02346 [cond-mat.str-el]

work page arXiv 2026
[47]

D. Pfau, J. S. Spencer, A. G. D. G. Matthews, and W. M. C. Foulkes, Ab initio solution of the many-electron schr¨ odinger equation with deep neural networks, Phys. Rev. Res.2, 033429 (2020)

2020
[48]

J. S. Spencer, D. Pfau, A. Botev, and W. M. C. Foulkes, Better, faster fermionic neural networks (2020), arXiv:2011.07125 [physics.comp-ph]

work page arXiv 2020
[49]

Cassella, H

G. Cassella, H. Sutterud, S. Azadi, N. D. Drummond, D. Pfau, J. S. Spencer, and W. M. C. Foulkes, Discov- ering quantum phase transitions with fermionic neural networks, Phys. Rev. Lett.130, 036401 (2023)

2023
[50]

Hermann, Z

J. Hermann, Z. Sch¨ atzle, and F. No´ e, Deep-neural- network solution of the electronic schr¨ odinger equation, Nat. Chem.12, 891 (2020)

2020
[51]

von Glehn, J

I. von Glehn, J. S. Spencer, and D. Pfau, A self- attention ansatz for ab-initio quantum chemistry (2023), arXiv:2211.13672 [physics.chem-ph]

work page arXiv 2023
[52]

Y. Teng, D. D. Dai, and L. Fu, Solving the fractional quantum hall problem with self-attention neural network, Phys. Rev. B111, 205117 (2025)

2025
[53]

Geier, K

M. Geier, K. Nazaryan, T. Zaklama, and L. Fu, Self- attention neural network for solving correlated electron problems in solids, Phys. Rev. B112, 045119 (2025)

2025
[54]

C.-T. Li, T. Ong, M. Geier, H. Lin, and L. Fu, Attention is all you need to solve chiral superconductivity (2025), arXiv:2509.03683 [cond-mat.supr-con]

work page internal anchor Pith review Pith/arXiv arXiv 2025
[55]

Geier, K

M. Geier, K. Nazaryan, T. Zaklama, and L. Fu, Self- 7 attention neural network for solving correlated electron problems in solids, Phys. Rev. B112, 045119 (2025)

2025
[56]

S. Dash, L. Gravina, F. Vicentini, M. Ferrero, and A. Georges, Efficiency of neural quantum states in light of the quantum geometric tensor, Commun. Phys.8, 92 (2025)

2025
[57]

M. S. Moss, A. Orfi, C. Roth, A. M. Sengupta, A. Georges, D. Sels, A. Dawid, and A. Valenti, Double de- scent: When do neural quantum states generalize?, Phys. Rev. E113, 045303 (2026)

2026
[58]

J. R. Moreno, J. Cohn, D. Sels, and M. Motta, Enhanc- ing the expressivity of variational neural, and hardware- efficient quantum states through orbital rotations (2023), arXiv:2302.11588 [quant-ph]

work page arXiv 2023
[59]

R. S. Cortes, A. S. Shankar, M. Dalmonte, R. Verdel, and N. Niggemann, Basis dependence of neural quan- tum states for the transverse field ising model (2026), arXiv:2512.11632 [quant-ph]

work page internal anchor Pith review Pith/arXiv arXiv 2026
[60]

S. B. Koˇ zi´ c, V. Zlati´ c, F. Franchini, and S. M. Giampaolo, Exploring the effect of basis rotation on nqs performance (2025), arXiv:2512.17893 [quant-ph]

work page arXiv 2025
[61]

M. S. Moss, R. Wiersema, M. Hibat-Allah, J. Car- rasquilla, and R. G. Melko, Leveraging recurrence in neu- ral network wavefunctions for large-scale simulations of heisenberg antiferromagnets on the square lattice, Phys. Rev. B112, 134450 (2025)

2025
[62]

Pescia, J

G. Pescia, J. Nys, J. Kim, A. Lovato, and G. Carleo, Message-passing neural quantum states for the homoge- neous electron gas, Phys. Rev. B110, 035108 (2024)

2024
[63]

Smith, Y

C. Smith, Y. Chen, R. Levy, Y. Yang, M. A. Morales, and S. Zhang, Unified variational approach description of ground-state phases of the two-dimensional electron gas, Phys. Rev. Lett.133, 266504 (2024)

2024
[64]

Wigner, On the interaction of electrons in metals, Phys

E. Wigner, On the interaction of electrons in metals, Phys. Rev.46, 1002 (1934)

1934
[65]

N. D. Drummond, Z. Radnai, J. R. Trail, M. D. Towler, and R. J. Needs, Diffusion quantum monte carlo study of three-dimensional wigner crystals, Phys. Rev. B69, 085116 (2004)

2004
[66]

Giuliani and G

G. Giuliani and G. Vignale,Quantum theory of the elec- tron liquid(Cambridge University Press, 2008)

2008
[67]

Azadi, N

S. Azadi, N. D. Drummond, and S. M. Vinko, Correlation energy of the paramagnetic electron gas at the thermo- dynamic limit, Phys. Rev. B107, L121105 (2023)

2023
[68]

Sorella, Generalized lanczos algorithm for variational quantum monte carlo, Phys

S. Sorella, Generalized lanczos algorithm for variational quantum monte carlo, Phys. Rev. B64, 024512 (2001)

2001
[69]

Goldshlager, N

G. Goldshlager, N. Abrahamsen, and L. Lin, A kaczmarz- inspired approach to accelerate the optimization of neural network wavefunctions, J. Comput. Phys.516, 113351 (2024)

2024
[70]

See Supplemental Material for more details
[71]

G. D. Mahan,Many-particle physics(Springer Science & Business Media, 2013)

2013
[72]

Y. Kwon, D. M. Ceperley, and R. M. Martin, Effects of backflow correlation in the three-dimensional electron gas: Quantum monte carlo study, Phys. Rev. B58, 6800 (1998)

1998
[73]

P. P. Ewald, Die berechnung optischer und elektro- statischer gitterpotentiale, Annalen der Physik369, 253 (1921)

1921
[74]

L. M. Fraser, W. M. C. Foulkes, G. Rajagopal, R. J. Needs, S. D. Kenny, and A. J. Williamson, Finite- size effects and coulomb interactions in quantum monte carlo calculations for homogeneous systems with periodic boundary conditions, Phys. Rev. B53, 1814 (1996)

1996
[75]

A. Y. Toukmaji and J. A. Board, Ewald summation tech- niques in perspective: a survey, Comput. Phys. Commun. 95, 73 (1996)

1996
[76]

X. Li, C. Fan, W. Ren, and J. Chen, Fermionic neural network with effective core potential, Phys. Rev. Res.4, 013021 (2022)

2022
[77]

W. Fu, R. Fujimaru, R. Li, Y. Liu, X. Wen, X. Li, K. Hongo, L. Wang, T. Ichibha, R. Maezono, J. Chen, and W. Ren, Local pseudopotential unlocks the true po- tential of neural network-based quantum monte carlo (2025), arXiv:2505.19909 [physics.chem-ph]

work page arXiv 2025