Abstract:Combining the efficiency of semi-empirical potentials with the accuracy of quantum mechanical methods, machine-learning interatomic potentials (MLIPs) have significantly advanced atomistic modeling in computational materials science and chemistry. This necessitates the continual development of MLIP models with improved accuracy and efficiency, which enable long-time scale molecular dynamics simulations to unveil the intricate underlying mechanisms that would otherwise remain elusive. Among various existing MLIP models, the moment tensor potential (MTP) model employs a highly descriptive rotationally-covariant moment tensor to describe the local atomic environment, enabling the use of even linear regression for model fitting. Although the current MTP model has achieved state-of-the-art efficiency for similar accuracy, there is still room for optimizing the contraction process of moment tensors. In this work, we propose an effective genetic algorithm based optimization scheme that can significantly reduce the number of independent moment tensor components and intermediate tensor components. This leads to a speedup of nearly one order of magnitude in efficiency and also improved accuracy compared to the traditional MTP model for intricate basis sets. We have applied our improved MTP model to predicting the energetic and dynamical properties of various point and planar defects in Ni-Al alloys, showing overall good performances and in general outperforming the semi-empirical potentials. This work paves the way for fast and accurate atomistic modeling of complex systems and provides a useful tool for modeling defects in Ni-Al alloys.

The MLIP package: moment tensor potentials with MPI and active learning

The MLIP package: Moment Tensor Potentials with MPI and Active Learning

MLIP-3: Active learning on atomic environments with Moment Tensor Potentials

Learning Interatomic Potentials at Multiple Scales

Active learning of linearly parametrized interatomic potentials

Accelerating the Training and Improving the Reliability of Machine-Learned Interatomic Potentials for Strongly Anharmonic Materials through Active Learning

Cartesian Atomic Moment Machine Learning Interatomic Potentials

Robust Training of Machine Learning Interatomic Potentials with Dimensionality Reduction and Stratified Sampling

Machine Learning Interatomic Potentials and Long-Range Physics

Uncertainty-biased molecular dynamics for learning uniformly accurate interatomic potentials

Evidential Deep Learning for Interatomic Potentials

Machine learning interatomic potential: Bridge the gap between small-scale models and realistic device-scale simulations

Geometry-enhanced pretraining on interatomic potentials

i-PI 3.0: a flexible and efficient framework for advanced atomistic simulations

Enhancing the Quality and Reliability of Machine Learning Interatomic Potentials through Better Reporting Practices

Geometry-enhanced Pre-training on Interatomic Potentials

Learning together: Towards foundation models for machine learning interatomic potentials with meta-learning

Efficient moment tensor machine-learning interatomic potential for accurate description of defects in Ni-Al Alloys