What Neural Network Field Theory Can and Cannot Realise on a Computer
Abstract
One aim of neural network field theory is to put a quantum or effective field theory on a computer, with the network ensemble itself as the theory. We ask how far that aim can be pushed for a function class regular enough to be computed with. Our main result is a no-go theorem with assumptions that hold for standard network architectures. We use it to separate four versions of neural network field theory, according to whether the defining object is the finite width ensemble or its infinite width limit, and whether the target we want to compute is a quantum or an effective field theory. Neither finite width interpretation is straightforwardly consistent. For finite width ensembles with finite variance at each point, the QFT interpretation fails reflection positivity, while the EFT interpretation establishes no scale separation by which the positivity violation can be placed outside its domain of validity. Of the two limit versions, one can be simulated in full and the other only in part, as only its smeared correlators are computable with a controlled error. As such, at the level of a controlled numerical computation, the QFT and EFT versions cannot be distinguished. One dimension escapes the obstruction, yet reflection positivity is shown to still fail there at every finite width for the cosine network. Two escapes from the theorem remain, giving up either finite variance at a point or exact rotation invariance, and we discuss both of these possibilities.
Related
- Why Multi-Layer Message Passing Works: Completeness Theory for Graph Neural Network Interatomic Potentials
- Constructive Neural Network Field Theory: ϕ24 in Finite Volume
- Graded Representation Theory of Equivariant Neural Networks
- A diffusion-free multi-layer neural-network method for multidimensional nonlinear hyperbolic equations
- Bridging Denotator Theory and Sheaf Neural Networks for Symbolic Music Generation
BibTeX
@misc{harvey2026what,
title = {{What Neural Network Field Theory Can and Cannot Realise on a Computer}},
author = {Thomas R. Harvey},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2608.21523},
eprint = {2608.21523},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.21523},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - What Neural Network Field Theory Can and Cannot Realise on a Computer
AU - Thomas R. Harvey
PY - 2026
DA - 2026/08/27
PB - arXiv
DO - 10.48550/arxiv.2608.21523
UR - https://arxiv.org/abs/2608.21523
AB - One aim of neural network field theory is to put a quantum or effective field theory on a computer, with the network ensemble itself as the theory. We ask how far that aim can be pushed for a function class regular enough to be computed with. Our main result is a no-go theorem with assumptions that hold for standard network architectures. We use it to separate four versions of neural network field theory, according to whether the defining object is the finite width ensemble or its infinite width limit, and whether the target we want to compute is a quantum or an effective field theory. Neither finite width interpretation is straightforwardly consistent. For finite width ensembles with finite variance at each point, the QFT interpretation fails reflection positivity, while the EFT interpretation establishes no scale separation by which the positivity violation can be placed outside its domain of validity. Of the two limit versions, one can be simulated in full and the other only in part, as only its smeared correlators are computable with a controlled error. As such, at the level of a controlled numerical computation, the QFT and EFT versions cannot be distinguished. One dimension escapes the obstruction, yet reflection positivity is shown to still fail there at every finite width for the cosine network. Two escapes from the theorem remain, giving up either finite variance at a point or exact rotation invariance, and we discuss both of these possibilities.
ER -