Performance of two complementary machine-learned potentials in modelling chemically complex systems
Performance of two complementary machine-learned potentials in modelling chemically complex systems
复制标题
DOI：
10.1038/s41524-023-01073-w
复制
发表时间：
2023-07
影响因子：
9.7
通讯作者：
Konstantin Gubaev;Viktor Zaverkin;P. Srinivasan;A. Duff;Johannes Kästner;B. Grabowski
中科院分区：
文献类型：
--
作者：
Konstantin Gubaev;Viktor Zaverkin;P. Srinivasan;A. Duff;Johannes Kästner;B. Grabowski
化学复杂的多组分合金具有来自无穷无尽的成分空间的特殊性能。然而，这种复杂性使得原子间势的开发具有挑战性。我们探索了两种互补的机器学习势-矩张量势（MTP）和高斯矩神经网络（GM-NN）-同时描述Ta-V-Cr-W合金族的构型和振动自由度。这两种模型同样准确，具有针对密度泛函理论评估的出色性能。他们实现了根均方误差（RMSE）的能量小于几个meV/原子在0 K有序和高温无序配置包括在训练中。即使对于不在训练中的组合物，在高温下的相对能量RMSE也在几meV/原子内。高温分子动力学力具有类似的约0.15 eV/cm 2的小RMSE，用于包括在训练中的无序四元和不属于训练的三元。MTP在训练规模的情况下实现更快的收敛; GM-NN在执行方面更快。主动学习是部分有益的，应该与传统的基于人类的训练集生成相补充。
Chemically complex multicomponent alloys possess exceptional properties derived from an inexhaustible compositional space. The complexity however makes interatomic potential development challenging. We explore two complementary machine-learned potentials—the moment tensor potential (MTP) and the Gaussian moment neural network (GM-NN)—in simultaneously describing configurationalandvibrational degrees of freedom in the Ta-V-Cr-W alloy family. Both models are equally accurate with excellent performance evaluated against density-functional-theory. They achieve root-mean-square-errors (RMSEs) in energies of less than a few meV/atom across 0 K ordered and high-temperature disordered configurations included in the training. Even for compositions not in training, relative energy RMSEs at high temperatures are within a few meV/atom. High-temperature molecular dynamics forces have similarly small RMSEs of about 0.15 eV/Å for the disordered quaternary included in, and ternaries not part of training. MTPs achieve faster convergence with training size; GM-NNs are faster in execution. Active learning is partially beneficial and should be complemented with conventional human-based training set generation.