Multi-scale Physical Representations for Approximating PDE Solutions with Graph Neural Operators

Léon Migus, Yuan Yin, Jocelyn Ahmed Mazari, Patrick Gallinari

International Conference on Learning Representations (ICLR) 2022 Workshop on Geometrical and Topological Representation Learning

April 29, 2021

Representing physical signals at different scales is among the most challenging problems in engineering. Several multi-scale modeling tools have been developed to describe physical systems governed by Partial Differential Equations (PDEs). These tools are at the crossroad of principled physical models and numerical schema. Recently, data-driven models have been introduced to speed-up the approximation of PDE solutions compared to numerical solvers. Among these recent data-driven methods, neural integral operators are a class that learn a mapping between function spaces. These functions are discretized on graphs (meshes) which are appropriate for modeling interactions in physical phenomena. In this work, we study three multi-resolution schema with integral kernel operators that can be approximated with Message Passing Graph Neural Networks (MPGNNs). To validate our study, we make extensive MPGNNs experiments with well-chosen metrics considering steady and unsteady PDEs.


1/ Sorbonne Université, CNRS, ISIR, F-75005 Paris, France

2/ INRIA Paris, ANGE Project-Team, 75589 Paris Cedex 12, France

3/ Sorbonne Université, CNRS, Laboratoire Jacques-Louis Lions, 75005 Paris, France

4/ Extrality, 75002 Paris, France

5/ Criteo AI Lab, Paris, France


/ Other published papers

An extensible Benchmarking Graph-Mesh dataset for studying Steady-State Incompressible Navier-Stokes Equations
Florent Bonnet, Jocelyn Ahmed Mazari, Thibaut Munzer, Pierre Yser, Patrick Gallinari
Learn more
Need to hit production with short go-to-market and better control?
Cookie Consent

By clicking “Accept”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts.