Small-supercell and Small-dataset Training Strategy of Machine Learning Interatomic Potentials for Point Defects
Abstract
Machine learning interatomic potentials (MLIPs) can treat large-scale material systems with near first-principles accuracy and have been widely used to accelerate point-defect simulations. However, the training of MLIPs usually relies on large amounts of DFT data. This issue is particularly pronounced for charged defects, for which DFT calculations of large supercells are required to avoid long-range Coulomb interactions and finite-size effects, making the construction of datasets computationally expensive. In this work, we propose an efficient MLIP training scheme for neutral and lowly charged point defects based on small supercells (less than 100 atoms) and limited number of DFT calculations. The scheme requires only four DFT structural relaxations to construct the training dataset and the trained MLIPs dedicated for the defect can predict the defect formation energies in larger supercells (over 200 atoms) with small errors (mostly smaller than 0.3 eV). Using defects in GaN, SiO2, and Cu2ZnSnS4 as representative examples, we evaluate the extrapolation capability of this scheme for predicting defect total energies and structural relaxations across different supercell sizes. The results show that an MLIP trained only on single small defect supercell produces severe errors when treating large supercells. Incorporating defect data from multiple supercell sizes improves the predictive performance of the MLIP, while adding pristine defect-free bulk supercells further enhances the accuracy. These results provide practical guidance for training MLIP models of defect systems with low computational costs.
Related
- Benchmarking machine-learning interatomic potentials using classical force fields
- MLIP Detective: Active Failure Mode Discovery Beyond Benchmark Scores for Machine-Learning Interatomic Potentials
- Cluster-based Structural Similarity for Dataset Visualization and Data Selection for Machine Learning Interatomic Potentials
- BranchIP: Learning Adaptive Equivariant Computation for Interatomic Potentials
- Leveraging Machine Learning Interatomic Potentials into Fitting Fast and Accurate AMBER-Compatible Precise Force Field
BibTeX
@misc{dai2026smallsupercell,
title = {{Small-supercell and Small-dataset Training Strategy of Machine Learning Interatomic Potentials for Point Defects}},
author = {Zhenxing Dai and Mingjue Ni and Xinpeng Li and Menglin Huang and Anderson Janotti and Shiyou Chen},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.24293},
eprint = {2609.24293},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.24293},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Small-supercell and Small-dataset Training Strategy of Machine Learning Interatomic Potentials for Point Defects
AU - Zhenxing Dai
AU - Mingjue Ni
AU - Xinpeng Li
AU - Menglin Huang
AU - Anderson Janotti
AU - Shiyou Chen
PY - 2026
DA - 2026/09/21
PB - arXiv
DO - 10.48550/arxiv.2609.24293
UR - https://arxiv.org/abs/2609.24293
AB - Machine learning interatomic potentials (MLIPs) can treat large-scale material systems with near first-principles accuracy and have been widely used to accelerate point-defect simulations. However, the training of MLIPs usually relies on large amounts of DFT data. This issue is particularly pronounced for charged defects, for which DFT calculations of large supercells are required to avoid long-range Coulomb interactions and finite-size effects, making the construction of datasets computationally expensive. In this work, we propose an efficient MLIP training scheme for neutral and lowly charged point defects based on small supercells (less than 100 atoms) and limited number of DFT calculations. The scheme requires only four DFT structural relaxations to construct the training dataset and the trained MLIPs dedicated for the defect can predict the defect formation energies in larger supercells (over 200 atoms) with small errors (mostly smaller than 0.3 eV). Using defects in GaN, SiO_2, and Cu_2ZnSnS_4 as representative examples, we evaluate the extrapolation capability of this scheme for predicting defect total energies and structural relaxations across different supercell sizes. The results show that an MLIP trained only on single small defect supercell produces severe errors when treating large supercells. Incorporating defect data from multiple supercell sizes improves the predictive performance of the MLIP, while adding pristine defect-free bulk supercells further enhances the accuracy. These results provide practical guidance for training MLIP models of defect systems with low computational costs.
ER -