Retention-constrained fine-tuning of a pre-trained aluminium MACE potential for non-equilibrium defect dynamics and plasticity
Abstract
Adapting a pre-trained machine-learning interatomic potential for non-equilibrium defect dynamics and plasticity requires improving defect accuracy without degrading the near-equilibrium elastic and vibrational response. Here we achieve this balance for the aluminium MACE potential QP17 by integrating conditional readout uncertainty, iterative full-parameter updates and protocol-compatible relative-energy anchoring. Sensitivities over 144 linear readout weights guide acquisition, with within-batch covariance updates and evolving neural representations. After four active-learning rounds (32 acquired configurations and 400 gradient updates), force-uncertainty acquisition reduces mean development force error from 22.43 to 21.52 meV/Angstrom, outperforming random sampling (22.70), geometry-diversity selection (22.46) and uniform training across all 104 candidate labels (22.71) under the same update budget. All three force-uncertainty runs satisfy bulk-mechanical retention criteria: historical relative-energy error decreases from 0.375 to 0.207 meV/atom, while harmonic phonon root-mean-square drift remains below 0.05 THz. Matched objective ablations reveal that relative-energy anchoring suppresses harmonic drift. Out-of-sample evaluations across planar faults, twin boundaries and grain boundaries demonstrate transferable gains in defect energies and deviatoric stresses, even though virial stresses are never fitted. A closed-loop molecular dynamics and density functional theory (MD-DFT) pilot confirms that automated acceptance gating intercepts updates that violate historical elastic-group energy or vibrational tolerances. Together, these components establish a retention-constrained fine-tuning protocol for improving plastic-deformation observables while preserving the tested bulk mechanical response.
Related
- Retention-Constrained Post-Training Quantization of Cellpose-SAM for Stem Cell Microscopy
- A Transferable Machine-Learned Potential Enabling Large-Scale Copper-Carbon Simulations from Equilibrium to Extreme Conditions
- Study of cathodic reactions in defects of thermal spray aluminium coatings on steel in artificial seawater
- Planar energetics do not determine twin persistence in nanocrystalline aluminium
- Flame-Exposed Aluminium Foil as a Low-Cost, Customizable TEM Grid for Nanomaterial Imaging
BibTeX
@misc{wu2026retentionconstrained,
title = {{Retention-constrained fine-tuning of a pre-trained aluminium MACE potential for non-equilibrium defect dynamics and plasticity}},
author = {Yue Wu and Yuan Tian and Deng Pan and Mingwei Chen},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.35653},
eprint = {2610.35653},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.35653},
}
RIS
TY - UNPB
TI - Retention-constrained fine-tuning of a pre-trained aluminium MACE potential for non-equilibrium defect dynamics and plasticity
AU - Yue Wu
AU - Yuan Tian
AU - Deng Pan
AU - Mingwei Chen
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2610.35653
UR - https://arxiv.org/abs/2610.35653
AB - Adapting a pre-trained machine-learning interatomic potential for non-equilibrium defect dynamics and plasticity requires improving defect accuracy without degrading the near-equilibrium elastic and vibrational response. Here we achieve this balance for the aluminium MACE potential QP17 by integrating conditional readout uncertainty, iterative full-parameter updates and protocol-compatible relative-energy anchoring. Sensitivities over 144 linear readout weights guide acquisition, with within-batch covariance updates and evolving neural representations. After four active-learning rounds (32 acquired configurations and 400 gradient updates), force-uncertainty acquisition reduces mean development force error from 22.43 to 21.52 meV/Angstrom, outperforming random sampling (22.70), geometry-diversity selection (22.46) and uniform training across all 104 candidate labels (22.71) under the same update budget. All three force-uncertainty runs satisfy bulk-mechanical retention criteria: historical relative-energy error decreases from 0.375 to 0.207 meV/atom, while harmonic phonon root-mean-square drift remains below 0.05 THz. Matched objective ablations reveal that relative-energy anchoring suppresses harmonic drift. Out-of-sample evaluations across planar faults, twin boundaries and grain boundaries demonstrate transferable gains in defect energies and deviatoric stresses, even though virial stresses are never fitted. A closed-loop molecular dynamics and density functional theory (MD-DFT) pilot confirms that automated acceptance gating intercepts updates that violate historical elastic-group energy or vibrational tolerances. Together, these components establish a retention-constrained fine-tuning protocol for improving plastic-deformation observables while preserving the tested bulk mechanical response.
ER -