Active Learning of Uniformly Accurate Inter-atomic Potentials for Materials Simulation
By Linfeng Zhang, D. Lin, Han Wang, Roberto Car, E. Weinan. PHYSICAL REVIEW MATERIALS, 2018
422 citations recorded by the source corpus.
An active learning procedure called Deep Potential Generator (DP-GEN) is proposed for the construction of accurate and transferable machine learning-based models of the potential energy surface (PES) for the molecular modeling of materials. This procedure consists of three main components: exploration, generation of accurate reference data, and training. Application to the sample systems of Al, Mg and Al-Mg alloys demonstrates that DP-GEN can produce uniformly accurate PES models with a minimal number of reference data.
Subfields: Materials Chemistry, Statistical and Nonlinear Physics, Structural Biology
View Paper: https://www.semanticscholar.org/paper/847c98c3e89e58d45a03a483f37059e22bc08cff
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Materials Chemistry
- Evgeny V. Podryabinkin
- Linfeng Zhang
- Konstantin Gubaev
- Xianglin Liu
- Jiaxin Zhang
- Zongrui Pei
- Han Wang
- David J. Srolovitz
In Statistical and Nonlinear Physics
Citations in the corpus
- Deep Potentials for Materials Science (2022, methodology): “For other properties, not included in the training datasets, DP is in better agreement with DFT than MEAM [10] for phonon dispersion relations, equations of state, and the liquid state radial distribution function [78].”
- The MLIP package: moment tensor potentials with MPI and active learning (2020, background): “It is the algorithm of choice for other neural-network-based potentials as well [70, 80].”
- Machine Learning for High-entropy Alloys: Progress, Challenges and Opportunities (2022, methodology): “Similarly, in MD simulations, the trajectory-based iterative sampling algorithms [124, 123] can be used to sample the dynamically important regions in the configuration space and add new data points where the trajectories visit most frequently.”