Graph neural network structural limitation for thermal simulation and architecture optimization through rating system

Abstract Graph neural networks are well suited for physics based simulation. Among other features, graphs can accurately represent thermal effects, with energy conservation operating on the nodes (vertices) and heat flow coursing through edges. Moreover, graph neural networks incorporate the data to...

Full description

Saved in:
Bibliographic Details
Main Authors: Pierre Hembert, Chady Ghnatios, Julien Cotton, Francisco Chinesta
Format: Article
Language:English
Published: SpringerOpen 2025-08-01
Series:Advanced Modeling and Simulation in Engineering Sciences
Subjects:
Online Access:https://doi.org/10.1186/s40323-025-00306-5
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849226052528766976
author Pierre Hembert
Chady Ghnatios
Julien Cotton
Francisco Chinesta
author_facet Pierre Hembert
Chady Ghnatios
Julien Cotton
Francisco Chinesta
author_sort Pierre Hembert
collection DOAJ
description Abstract Graph neural networks are well suited for physics based simulation. Among other features, graphs can accurately represent thermal effects, with energy conservation operating on the nodes (vertices) and heat flow coursing through edges. Moreover, graph neural networks incorporate the data topology and allow simulation on a wide variety of geometries. In this paper, various graph neural networks are trained using analytical solutions to the transient heat equation. Several architectures and hyper-parameters are compared to empirically identify good practices when using a graph neural network for numerical simulation. In addition, a novel method to optimize the network architecture is proposed. It is based on rating systems used to evaluate players’ skill level in competitive games. The results show that graph neural network-based simulation shares mutual restrictions with classical discretizations techniques (such as the Von Neumann stability criterion) and also that graph neural networks can be a potent generalization tool for heat transfer problems.
format Article
id doaj-art-37fbae05984e48eab4ac960c03779984
institution Kabale University
issn 2213-7467
language English
publishDate 2025-08-01
publisher SpringerOpen
record_format Article
series Advanced Modeling and Simulation in Engineering Sciences
spelling doaj-art-37fbae05984e48eab4ac960c037799842025-08-24T11:41:00ZengSpringerOpenAdvanced Modeling and Simulation in Engineering Sciences2213-74672025-08-0112113010.1186/s40323-025-00306-5Graph neural network structural limitation for thermal simulation and architecture optimization through rating systemPierre Hembert0Chady Ghnatios1Julien Cotton2Francisco Chinesta3PIMM Laboratory, Arts et Metiers Institute of TechnologyDepartment of Mechanical Engineering, University of North FloridaAndra, French National Radioactive Waste Management AgencyPIMM Laboratory, Arts et Metiers Institute of TechnologyAbstract Graph neural networks are well suited for physics based simulation. Among other features, graphs can accurately represent thermal effects, with energy conservation operating on the nodes (vertices) and heat flow coursing through edges. Moreover, graph neural networks incorporate the data topology and allow simulation on a wide variety of geometries. In this paper, various graph neural networks are trained using analytical solutions to the transient heat equation. Several architectures and hyper-parameters are compared to empirically identify good practices when using a graph neural network for numerical simulation. In addition, a novel method to optimize the network architecture is proposed. It is based on rating systems used to evaluate players’ skill level in competitive games. The results show that graph neural network-based simulation shares mutual restrictions with classical discretizations techniques (such as the Von Neumann stability criterion) and also that graph neural networks can be a potent generalization tool for heat transfer problems.https://doi.org/10.1186/s40323-025-00306-5Graph neural networkThermal simulationMesh convergenceMesh parametersRating systemArchitecture optimization
spellingShingle Pierre Hembert
Chady Ghnatios
Julien Cotton
Francisco Chinesta
Graph neural network structural limitation for thermal simulation and architecture optimization through rating system
Advanced Modeling and Simulation in Engineering Sciences
Graph neural network
Thermal simulation
Mesh convergence
Mesh parameters
Rating system
Architecture optimization
title Graph neural network structural limitation for thermal simulation and architecture optimization through rating system
title_full Graph neural network structural limitation for thermal simulation and architecture optimization through rating system
title_fullStr Graph neural network structural limitation for thermal simulation and architecture optimization through rating system
title_full_unstemmed Graph neural network structural limitation for thermal simulation and architecture optimization through rating system
title_short Graph neural network structural limitation for thermal simulation and architecture optimization through rating system
title_sort graph neural network structural limitation for thermal simulation and architecture optimization through rating system
topic Graph neural network
Thermal simulation
Mesh convergence
Mesh parameters
Rating system
Architecture optimization
url https://doi.org/10.1186/s40323-025-00306-5
work_keys_str_mv AT pierrehembert graphneuralnetworkstructurallimitationforthermalsimulationandarchitectureoptimizationthroughratingsystem
AT chadyghnatios graphneuralnetworkstructurallimitationforthermalsimulationandarchitectureoptimizationthroughratingsystem
AT juliencotton graphneuralnetworkstructurallimitationforthermalsimulationandarchitectureoptimizationthroughratingsystem
AT franciscochinesta graphneuralnetworkstructurallimitationforthermalsimulationandarchitectureoptimizationthroughratingsystem