Expression Rates of Neural Operators for Linear Elliptic PDEs in Polytopes

2024/09/26 by Carlo Marcati, Christoph Schwab, Marcati, Carlo +1 · 1 citation
Engineering · #Advanced Numerical Analysis Techniques

paper · pdf · doi:10.48550/arxiv.2409.17552

Abstract

We study the approximation rates of a class of deep neural network approximations of operators which arise as data-to-solution maps S of linear elliptic partial differential equations (PDEs), and act between pairs X,Y of suitable infinite-dimensional spaces. We prove expression rate bounds for approximate neural operators G with the structure G = R ∘ A ∘ E, with linear encoders E and decoders R. We focus in particular on deepONets emulating the coefficient-to-solution maps for elliptic PDEs set in polygons and in some polyhedra. Exploiting the regularity of the solution sets of elliptic PDEs in polytopes, we show algebraic rates of convergence for problems with data with finite regularity, and exponential rates for analytic data.

Cited by

• One model to solve them all: 2BSDE families via neural operators

Related

• Deep Operator Network Approximation Rates for Lipschitz Operators
• Collocation approximation by deep neural ReLU networks for parametric\n elliptic PDEs with lognormal inputs
• Parametric Complexity Bounds for Approximating PDEs with Neural Networks
• Neural and spectral operator surrogates: unified construction and expression rate bounds
• Model Reduction and Neural Networks for Parametric PDEs