Learning Solution Operators for Partial Differential Equations via Monte Carlo-Type Approximation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Choutri, Salah Eddine, Chauhan, Prajwal, Mazhar, Othmane, Jabari, Saif Eddin
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911283568181248
author Choutri, Salah Eddine
Chauhan, Prajwal
Mazhar, Othmane
Jabari, Saif Eddin
author_facet Choutri, Salah Eddine
Chauhan, Prajwal
Mazhar, Othmane
Jabari, Saif Eddin
contents The Monte Carlo-type Neural Operator (MCNO) introduces a lightweight architecture for learning solution operators for parametric PDEs by directly approximating the kernel integral using a Monte Carlo approach. Unlike Fourier Neural Operators, MCNO makes no spectral or translation-invariance assumptions. The kernel is represented as a learnable tensor over a fixed set of randomly sampled points. This design enables generalization across multiple grid resolutions without relying on fixed global basis functions or repeated sampling during training. Experiments on standard 1D PDE benchmarks show that MCNO achieves competitive accuracy with low computational cost, providing a simple and practical alternative to spectral and graph-based neural operators.
format Preprint
id arxiv_https___arxiv_org_abs_2511_18930
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning Solution Operators for Partial Differential Equations via Monte Carlo-Type Approximation
Choutri, Salah Eddine
Chauhan, Prajwal
Mazhar, Othmane
Jabari, Saif Eddin
Machine Learning
Artificial Intelligence
The Monte Carlo-type Neural Operator (MCNO) introduces a lightweight architecture for learning solution operators for parametric PDEs by directly approximating the kernel integral using a Monte Carlo approach. Unlike Fourier Neural Operators, MCNO makes no spectral or translation-invariance assumptions. The kernel is represented as a learnable tensor over a fixed set of randomly sampled points. This design enables generalization across multiple grid resolutions without relying on fixed global basis functions or repeated sampling during training. Experiments on standard 1D PDE benchmarks show that MCNO achieves competitive accuracy with low computational cost, providing a simple and practical alternative to spectral and graph-based neural operators.
title Learning Solution Operators for Partial Differential Equations via Monte Carlo-Type Approximation
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.18930