Abstract:Although very successfully used in conventional machine learning, convolution based neural network architectures -- believed to be inconsistent in function space -- have been largely ignored in the context of learning solution operators of PDEs. Here, we present novel adaptations for convolutional neural networks to demonstrate that they are indeed able to process functions as inputs and outputs. The resulting architecture, termed as convolutional neural operators (CNOs), is designed specifically to preserve its underlying continuous nature, even when implemented in a discretized form on a computer. We prove a universality theorem to show that CNOs can approximate operators arising in PDEs to desired accuracy. CNOs are tested on a novel suite of benchmarks, encompassing a diverse set of PDEs with possibly multi-scale solutions and are observed to significantly outperform baselines, paving the way for an alternative framework for robust and accurate operator learning. Our code is publicly available at <a class="link-external link-https" href="https://github.com/bogdanraonic3/ConvolutionalNeuralOperator" rel="external noopener nofollow">this https URL</a>

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to achieve robust and accurate performance in the learning of solution operators of partial differential equations (PDEs) using convolutional neural network (CNN) architectures. Although CNNs have achieved great success in traditional machine - learning tasks, their performance in handling PDEs solution operators has been overlooked because it is believed that CNNs are inconsistent in function space. The paper proposes a new adaptation method that enables CNNs to handle functions as inputs and outputs, and designs convolutional neural operators (CNOs) to preserve their underlying continuous properties, even when implemented in discrete form on a computer. The paper proves that CNOs can approximate the operators in PDEs with the required accuracy, and verifies the performance of CNOs through a series of benchmark tests, indicating that CNOs significantly outperform the baseline models and provide a new framework for robust and accurate operator learning. Specifically, the main contributions of the paper include: 1. **Proposing a new modification**: Modify the CNN to enforce structure - preserving continuous - discrete equivalence, thereby achieving aliasing - free true operator learning. The resulting architecture is called the convolutional neural operator (CNO), which is implemented as a new variant of the widely used U - Net architecture. 2. **Theoretical proof**: Prove that the CNO is a representation - equivalent neural operator, and strictly show that the CNO can approximate the operators corresponding to a large class of PDEs to the required accuracy. 3. **Experimental verification**: Test the CNO on a new set of benchmark tests (called representative PDE benchmark tests, RPB), which cover a variety of cases from linear elliptic and hyperbolic to nonlinear parabolic and hyperbolic PDEs, and may have multi - scale solutions. The experimental results show that the CNO performs excellently in all benchmark tests, both in - distribution and out - of - distribution tests. Through these contributions, the paper shows that the CNN - based operator learning model has good theoretical properties and excellent empirical performance, and has the potential for wide application in PDEs learning.

Convolutional Neural Operators for robust and accurate learning of PDEs

Neural Inverse Operators for Solving PDE Inverse Problems

A foundational neural operator that continuously learns without forgetting

Operator Learning with Neural Fields: Tackling PDEs on General Geometries

Enhancing Solutions for Complex PDEs: Introducing Complementary Convolution and Equivariant Attention in Fourier Neural Operators

CoNO: Complex Neural Operator for Continous Dynamical Physical Systems

Pseudo-Differential Neural Operator: Generalized Fourier Neural Operator for Learning Solution Operators of Partial Differential Equations

Render Unto Numerics : Orthogonal Polynomial Neural Operator for PDEs with Non-periodic Boundary Conditions.

Neural Operators with Localized Integral and Differential Kernels

Diffeomorphic Latent Neural Operators for Data-Efficient Learning of Solutions to Partial Differential Equations

Neural Operator: Learning Maps Between Function Spaces

Reference Neural Operators: Learning the Smooth Dependence of Solutions of PDEs on Geometric Deformations

Learning Partial Differential Equations with Deep Parallel Neural Operator

Dilated convolution neural operator for multiscale partial differential equations

Diffeomorphism Neural Operator for various domains and parameters of partial differential equations

Kernel Neural Operator for Efficient Solving PDEs

Regularity-Conforming Neural Networks (ReCoNNs) for solving Partial Differential Equations

Nonlinear Reconstruction for Operator Learning of PDEs with Discontinuities

BENO: Boundary-embedded Neural Operators for Elliptic PDEs

Neural Green's Operators for Parametric Partial Differential Equations

Koopman neural operator as a mesh-free solver of non-linear partial differential equations