Residual neural networks overcome the curse of dimensionality for semilinear heat equations
Abstract
Rigorous results show that feedforward neural networks can overcome the curse of dimensionality in the numerical approximation of high-dimensional partial differential equations (PDEs), but comparatively little is known about residual neural networks (ResNets) in the nonlinear PDE setting. We prove that ResNets overcome the curse of dimensionality in the numerical approximation of solutions of semilinear heat equations with globally Lipschitz continuous, gradient-independent nonlinearities: under polynomial growth and network approximability hypotheses on the PDE data, there exist η∈(0,∞) and ResNets Ψd,ε, d∈ℕ, ε∈(0,1], with at most ηdηε−η parameters whose realizations approximate the solution in dimension d with an L2-error of at most ε. The proof represents one deterministic realization of a multilevel Picard estimator by a ResNet whose shortcut connections transmit the spatial variable and a scalar accumulator, while the residual branches successively add the summands of the estimator. For ridge-sum initial conditions, admissible sigmoidal activations, and globally Lipschitz truncations of the nonlinearity, we obtain, for every ξ>0, the explicit bound Cξd4+ξε−(3+ξ) on the number of parameters.
Related
- Direct and Indirect Physics-Informed Neural Networks for Dirichlet Boundary Control of Semilinear Parabolic Equations: A Conditional Error Analysis
- Oversimplified dislocation dynamics from semilinear homogenization. Heat and damped wave equations
- An exact traveling waves solution for a special class of semilinear equations
- Interior regularity of mixed local-nonlocal parabolic semilinear equations
- Liouville Theorems for Semilinear Parabolic Equations Involving the Nonlinear Gradient Term
BibTeX
@misc{mukhammadiev2026residual,
title = {{Residual neural networks overcome the curse of dimensionality for semilinear heat equations}},
author = {Ilkhom Mukhammadiev and Diyora Salimova},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.03626},
eprint = {2609.03626},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.03626},
}
RIS
TY - UNPB
TI - Residual neural networks overcome the curse of dimensionality for semilinear heat equations
AU - Ilkhom Mukhammadiev
AU - Diyora Salimova
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2609.03626
UR - https://arxiv.org/abs/2609.03626
AB - Rigorous results show that feedforward neural networks can overcome the curse of dimensionality in the numerical approximation of high-dimensional partial differential equations (PDEs), but comparatively little is known about residual neural networks (ResNets) in the nonlinear PDE setting. We prove that ResNets overcome the curse of dimensionality in the numerical approximation of solutions of semilinear heat equations with globally Lipschitz continuous, gradient-independent nonlinearities: under polynomial growth and network approximability hypotheses on the PDE data, there exist η∈(0,∞) and ResNets Ψ_d,ε, d∈ℕ, ε∈(0,1], with at most ηd^ηε^−η parameters whose realizations approximate the solution in dimension d with an L^2-error of at most ε. The proof represents one deterministic realization of a multilevel Picard estimator by a ResNet whose shortcut connections transmit the spatial variable and a scalar accumulator, while the residual branches successively add the summands of the estimator. For ridge-sum initial conditions, admissible sigmoidal activations, and globally Lipschitz truncations of the nonlinearity, we obtain, for every ξ>0, the explicit bound C_ξd^4+ξε^−(3+ξ) on the number of parameters.
ER -