arXiv Science⌕ Search

arXiv · 2610.40384

Neural Network Approximation and Structural Complexity of Semilinear PDEs with Lower Bounds

Abstract

Neural networks are commonly used to learn the solution operators of PDEs such as the Navier-Stokes equations and viscous conservation laws. On standard benchmarks, moderately sized and often shallow networks achieve small errors that decay at algebraic rates with the network size. These observations are not well described by the current approximation theory, which treats the solution operator as a generic map and derives worst-case bounds over large classes of operators. We show that the structure of the equations and their properties explain the observed behavior. For semilinear parabolic equations, the approximation costs separate into two terms, where one is controlled by an invariant of the governing law and the other by the data. The constitutive law can be parameterized as a single block whose size for accuracy $η$ is $O(η^{-pχ}\logη^{-1})$ for every $p>p_f$, where the structural exponent $p_f$ only depends on the computation graph of $f$. This exponent does not depend on the dimension and is in fact zero for polynomial laws, so for the two-dimensional Navier-Stokes equations the nonlinear block costs only $O(\logη^{-1})$ parameters. The data enters through encoding layers with a bounded number of learned weights per mode, so neither term grows with the grid resolution or the time step. While the analysis only uses the max-norm error of the pointwise block, we show that one obtains approximation bounds in a stronger norm along the whole trajectory. For viscous Hamilton-Jacobi equations with a $C^s$ Hamiltonian, the approximation requires on the order of $η^{-1/s}$ parameters, and we prove a matching lower bound, up to logarithms. This shows that solution operators for this class of problems have low structural complexity, even though they are nonlinear maps between infinite-dimensional spaces.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Hayden Schaeffer. 2026-10-29. Neural Network Approximation and Structural Complexity of Semilinear PDEs with Lower Bounds. https://arxiv.science/abs/2610.40384

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite arXiv Science by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Adaptive mesh refinement for the Landau-Lifshitz-Gilbert equation

We propose a new adaptive algorithm for the approximation of the Landau--Lifshitz--Gilbert equation via a higher-order tangent plane scheme. We show that the adaptive approximation satisfies an energy inequality and demonstrate numerically that the adaptive algorithm outperforms uniform approaches.

math.NA↗

A one-dimensional model of catheter aspiration in blood vessels

Aspiration thrombectomy is a primary endovascular treatment option for acute ischemic stroke caused by large vessel occlusions, in which a catheter is navigated to the occlusion site and suction is applied to remove the thrombus. This work introduces a reduced one-dimensional model for aspiration in blood vessels that accounts for the elasticity of both the vessel wall and the catheter. The inclusion of vessel wall visco-elasticity transforms the governing equation for the flow rate into a parabolic form, enabling an accurate resolution of the diffuse pressure gradients near the catheter tip observed in CFD simulations. The numerical treatment is based on a relaxation of the hyperbolic subsystem that facilitates nodal solvers, enabling efficient coupling between catheterized and uncatheterized vessel segments, including bifurcations and the catheter tip. A splitting strategy ensures that the visco-elastic terms incur only negligible additional computational cost relative to the purely hyperbolic model. The model is benchmarked against three-dimensional CFD simulations and reference data from the literature. Numerical experiments investigating the role of catheter elasticity and suction force on the hemodynamics are presented, and an uncertainty quantification study demonstrates the suitability of the framework for efficient parameter studies. The results indicate that the proposed model resolves blood flow comparably accurate to three-dimensional simulations at a fraction of the computational cost. While further in vitro and in vivo validation is required, the framework represents a promising step toward computationally efficient preoperative planning and uncertainty-aware decision support in aspiration thrombectomy.

math.NA↗

Exploiting the nonzero diagonal pattern in matrix function computations

We consider the task of approximating a matrix function $f(A)$, where $A$ is a matrix in which only a relatively small number of (not necessarily consecutive) sub- and superdiagonals contain nonzero entries. Approximating $f$ by a low-degree polynomial $p$ allows us to obtain sparse approximations to $f(A)$, which one can efficiently work with (while, in general, $f(A)$ is a dense matrix, even when $A$ is sparse). Our approach is based on carefully inspecting the locations where nonzeros can occur in $p(A)$, and identifying the entries in $A$ that influence them. In particular, we illustrate how this approach can be used for efficiently approximating the trace of $f(A)$ and identify how this approach is related to established (stochastic) probing methods for trace estimation. Another application area in which our approach works particularly well is the computation of functions of Toeplitz matrices. Here, studying the sparsity pattern of $p(A)$ allows us to reduce the computation of the whole matrix polynomial to that of a single small-scale submatrix, yielding an algorithm that scales exceptionally well to large problem sizes.

math.NA↗