Saved in:
Bibliographic Details
Main Authors: Moustafa, Heisam, Kovacs, Alexander, Fischbacher, Johann, Gusenbauer, Markus, Ali, Qais, Breth, Leoni, Schrefl, Thomas, Oezelt, Harald
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.23615
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911029805449216
author Moustafa, Heisam
Kovacs, Alexander
Fischbacher, Johann
Gusenbauer, Markus
Ali, Qais
Breth, Leoni
Schrefl, Thomas
Oezelt, Harald
author_facet Moustafa, Heisam
Kovacs, Alexander
Fischbacher, Johann
Gusenbauer, Markus
Ali, Qais
Breth, Leoni
Schrefl, Thomas
Oezelt, Harald
contents Graph neural networks (GNN) are a promising tool to predict magnetic properties of large multi-grain structures, which can speed up the search for rare-earth free permanent magnets. In this paper, we use our magnetic simulation data to train a GNN to predict coercivity of hard magnetic microstructures. We evaluate the performance of the trained GNN and quantify its uncertainty. Subsequently, we reuse the GNN architecture for predicting the maximum energy product. Out-of-distribution predictions of coercivity are also performed, following feature engineering based on the observed dependence of coercivity on system size.
format Preprint
id arxiv_https___arxiv_org_abs_2506_23615
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Graph Neural Networks to Predict Coercivity of Hard Magnetic Microstructures
Moustafa, Heisam
Kovacs, Alexander
Fischbacher, Johann
Gusenbauer, Markus
Ali, Qais
Breth, Leoni
Schrefl, Thomas
Oezelt, Harald
Computational Physics
Graph neural networks (GNN) are a promising tool to predict magnetic properties of large multi-grain structures, which can speed up the search for rare-earth free permanent magnets. In this paper, we use our magnetic simulation data to train a GNN to predict coercivity of hard magnetic microstructures. We evaluate the performance of the trained GNN and quantify its uncertainty. Subsequently, we reuse the GNN architecture for predicting the maximum energy product. Out-of-distribution predictions of coercivity are also performed, following feature engineering based on the observed dependence of coercivity on system size.
title Graph Neural Networks to Predict Coercivity of Hard Magnetic Microstructures
topic Computational Physics
url https://arxiv.org/abs/2506.23615