PreprintCC BY-NC-SA 4.0arXiv
Small-data physics-informed ML model for magnetic exchange based on chemical composition and structure
Abstract
In this paper we propose an ensemble model based on a swarm of dense neural networks, able to predict Heisenberg exchange interactions within less than 1% error. The model uses the inputs of the ab-initio Schrödinger equation, i.e. the compound structure and chemical composition. In this first work, we trained the model on a few compounds of the hexagonal manganite family and were able to infer, with similar accuracy, the J values for other compounds of the same family at randomly chosen geometries, thus showing that we accurately learned the effect of the chemical composition and structure on effective magnetic integrals J.
Related
- Role of composition-tuned disorder on magnetic exchange interactions in NiRuMnGa
- Interplay of chemical structure and magnetic order coupling at the interface between Cr2O3 and Fe3O4 in hybrid nanocomposites
- Interplay of Composition, Crystallinity, and Chemical Structure in CoHCF and NiHCF Thin Films Prepared at Different Temperatures
- Corridor Bonding Theory: A QTAIM-Based Density-Informed Graph Language for Teaching and Interpreting Chemical Structure
- Transport of organic solutes in ion-exchange membranes: Mechanisms and influence of solvent ionic composition
BibTeX
@misc{zanguim2026smalldata,
title = {{Small-data physics-informed ML model for magnetic exchange based on chemical composition and structure}},
author = {Brice Kengni Zanguim and Thibaut Véry and Marie-Bernadette Lepetit},
howpublished = {arXiv},
year = {2026},
month = jul,
doi = {10.48550/arxiv.2611.04681},
eprint = {2611.04681},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.04681},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Small-data physics-informed ML model for magnetic exchange based on chemical composition and structure
AU - Brice Kengni Zanguim
AU - Thibaut Véry
AU - Marie-Bernadette Lepetit
PY - 2026
DA - 2026/07/30
PB - arXiv
DO - 10.48550/arxiv.2611.04681
UR - https://arxiv.org/abs/2611.04681
AB - In this paper we propose an ensemble model based on a swarm of dense neural networks, able to predict Heisenberg exchange interactions within less than 1% error. The model uses the inputs of the ab-initio Schrödinger equation, i.e. the compound structure and chemical composition. In this first work, we trained the model on a few compounds of the hexagonal manganite family and were able to infer, with similar accuracy, the J values for other compounds of the same family at randomly chosen geometries, thus showing that we accurately learned the effect of the chemical composition and structure on effective magnetic integrals J.
ER -