- Physical Review Journals
All JournalsPhysics Magazine

Physical Review D

Reuse & Permissions

It is not necessary to obtain permission to reuse this article or its components as it is available under the terms of the Creative Commons Attribution 4.0 International license. This license permits unrestricted use, distribution, and reproduction in any medium, provided attribution to the author(s) and the published article's title, journal citation, and DOI are maintained. Please note that some figures may have been included with permission from other third parties. It is your responsibility to obtain the proper permission from the rights holder directly for these figures.

Export citation

Export citation

Choose format for download:

Download Citation
  • Letter
  • Open Access

Optimal architecture and fundamental bounds in neural network field theory

Zhengkang Zhang

PDF

Phys. Rev. D 114, L061704 – Published 30 September, 2026

DOI: https://doi.org/10.1103/4fzf-tpsv

Abstract

Neural network field theory (NNFT) represents fields as neural networks and samples field configurations by drawing network parameters from a probability distribution. We identify a previously unexplored architectural freedom in NNFT, parametrized by α, that leaves the infinite-width theory invariant but dramatically affects finite-width errors in the calculation of correlation functions. For a massive scalar field, we show that α=0, corresponding to propagator-weighted neuron momenta and constant neuron amplitudes, is optimal: it minimizes finite-width variance and uniquely removes IR-sensitive corrections in the interacting theory. Even at α=0, relative errors from both bias and variance grow exponentially with distance beyond the correlation length. The bias can be removed by extrapolating to infinite width, which we demonstrate numerically, while the variance imposes a fundamental bound on the achievable signal-to-noise ratio as in lattice field theory. These results chart a path toward developing NNFT into a practical tool for the numerical study of field theories.

View figure in article

Physics Subject Headings (PhySH)

Article Text

Supplemental Material

References (33)

  1. J. Halverson, A. Maiti, and K. Stoner, Neural networks and quantum field theory, Mach. Learn. Sci. Tech. 2, 035002 (2021).
  2. J. Halverson, Building quantum field theories out of neurons, arXiv:2112.04527.
  3. M. Demirtas, J. Halverson, A. Maiti, M. D. Schwartz, and K. Stoner, Neural network field theories: Non-Gaussianity actions, and locality, Mach. Learn. Sci. Tech. 5, 015002 (2024).
  4. J. Halverson, J. Naskar, and J. Tian, Conformal fields from neural networks, J. High Energy Phys. 10 (2025) 039.
  5. P. Capuozzo, B. Robinson, and B. Suzzoni, Conformal defects in neural network field theories, J. High Energy Phys. 05 (2026) 124.
  6. B. Robinson, Virasoro symmetry in neural network field theories, arXiv:2512.24420.
  7. G. Huang and K. Zhou, The neural networks with tensor weights and emergent fermionic Wick rules in the large-width limit, Phys. Lett. B 873, 140146 (2026).
  8. S. Frank, J. Halverson, A. Maiti, and F. Ruehle, Fermions and supersymmetry in neural network field theories, Mach. Learn. Sci. Tech. 7, 045070 (2026).
  9. S. Frank and J. Halverson, String theory from infinite width neural networks, arXiv:2601.06249.
  10. D. S. Ageev and Y. A. Ageeva, Excited string states and D-branes from infinite width neural networks, arXiv:2602.10214.
  11. H. Erbin, V. Lahoche, and D. O. Samary, Non-perturbative renormalization for the neural network-QFT correspondence, Mach. Learn. Sci. Tech. 3, 015027 (2022).
  12. H. Erbin, V. Lahoche, and D. O. Samary, Renormalization in the neural network-quantum field theory correspondence, arXiv:2212.11811.
  13. J. N. Howard, M. S. Klinger, A. Maiti, and A. G. Stapleton, Bayesian RG flow in neural network field theories, SciPost Phys. Core 8, 027 (2025).
  14. J. Halverson, TASI lectures on physics for machine learning, arXiv:2408.00082.
  15. C. Ferko and J. Halverson, Quantum mechanics and neural networks, Mach. Learn. Sci. Tech. 7, 015002 (2026).
  16. C. Ferko, J. Halverson, and A. Mutchler, Universality of neural network field theory, arXiv:2601.14453.
  17. D. S. Ageev and Y. A. Ageeva, Neural network quantum field theory from transformer architectures, arXiv:2602.10209.
  18. C. Ferko, J. Halverson, V. Jejjala, and B. Robinson, Topological effects in neural network field theory, arXiv:2604.02313.
  19. A. Maiti, K. Stoner, and J. Halverson, Symmetry-via-duality: Invariant neural network densities from parameter-space correlators, arXiv:2106.00694.
  20. R. M. Neal, Bayesian Learning for Neural Networks, Lecture Notes in Statistics, Vol. 118 (Springer, New York, 1996).
  21. C. K. I. Williams, Computing with infinite networks, in Advances in Neural Information Processing Systems, Vol. 9, edited by M. Mozer, M. Jordan, and T. Petsche (MIT Press, Cambridge, MA, 1996).
  22. J. Lee, Y. Bahri, R. Novak, S. S. Schoenholz, J. Pennington, and J. Sohl-Dickstein, Deep neural networks as gaussian processes, arXiv:1711.00165.
  23. A. G. d. G. Matthews, M. Rowland, J. Hron, R. E. Turner, and Z. Ghahramani, Gaussian process behaviour in wide deep neural networks, arXiv:1804.11271.
  24. G. Yang, Tensor programs I: Wide feedforward or recurrent neural networks of any architecture are gaussian processes, arXiv:1910.12478.
  25. B. Hanin, Random neural networks in the infinite width limit as gaussian processes, arXiv:2107.01562.
  26. S. Sen and V. Vaidya, Viability of perturbative expansion for quantum field theories on neurons, Mach. Learn. Sci. Tech. 7, 035044 (2026).
  27. A. Rahimi and B. Recht, Random features for large-scale kernel machines, in Advances in Neural Information Processing Systems, Vol. 20, edited by J. Platt, D. Koller, Y. Singer, and S. Roweis (Curran Associates, Inc., Red Hook, NY, 2007), pp. 1177–1184.

  28. This freedom was noted in passing in Ref. [9] in the context of the 2D free boson.

  29. See Supplemental Material at http://link.aps.org/supplemental/10.1103/4fzf-tpsv for the derivation of the correlation functions for i.i.d. neurons, the minimization of κnnoise, and the full O(λ) result for the two-point function in ϕ4 theory.

  30. Sampling |ai| from a Gaussian as in Refs. [2, 3] instead gives ⟨|ai|2n⟩=(2n−1)!!⟨|ai|2⟩n, which amplifies the bias and variance by numerical prefactors but does not change the parametric dependence on α.

  31. G. Parisi, The strategy for computing the hadronic mass spectrum, Phys. Rep. 103, 203 (1984).
  32. G. P. Lepage, The analysis of algorithms for lattice field theory, in Theoretical Advanced Study Institute in Elementary Particle Physics (World Scientific, Singapore, 1989).
  33. Z. Zhang, nnft-alpha: Simulation code and data for Optimal Architecture and Fundamental Bounds in Neural Network Field Theory, https://github.com/zzkevin2019/nnft-alpha (2026).

Outline

Information

Phys. Rev. D 114, L061704– Published 30 September, 2026
  • Received 8 July 2026
  • Accepted 10 September 2026
Crossmark - Check for updates button
Creative Commons Logo

Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. Funded by SCOAP3.

Published by the American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation