Zhang Yaolong, Hu Ce, Jiang Bin
Hefei National Laboratory for Physical Science at the Microscale, Key Laboratory of Surface and Interface Chemistry and Energy Catalysis of Anhui Higher Education Institutes, Department of Chemical Physics, University of Science and Technology of China, Hefei, Anhui 230026, China.
Phys Chem Chem Phys. 2021 Jan 28;23(3):1815-1821. doi: 10.1039/d0cp05089j.
Recently, machine learning methods have become easy-to-use tools for constructing high-dimensional interatomic potentials with ab initio accuracy. Although machine-learned interatomic potentials are generally orders of magnitude faster than first-principles calculations, they remain much slower than classical force fields, at the price of using more complex structural descriptors. To bridge this efficiency gap, we propose an embedded atom neural network approach with simple piecewise switching function-based descriptors, resulting in a favorable linear scaling with the number of neighbor atoms. Numerical examples validate that this piecewise machine-learning model can be over an order of magnitude faster than various popular machine-learned potentials with comparable accuracy for both metallic and covalent materials, approaching the speed of the fastest embedded atom method (i.e. several μs per atom per CPU core). The extreme efficiency of this approach promises its potential in first-principles atomistic simulations of very large systems and/or in a long timescale.
最近，机器学习方法已成为构建具有从头算精度的高维原子间势的易用工具。尽管机器学习的原子间势通常比第一性原理计算快几个数量级，但以使用更复杂的结构描述符为代价，它们仍然比经典力场慢得多。为了弥合这种效率差距，我们提出了一种基于简单分段切换函数描述符的嵌入原子神经网络方法，从而实现与相邻原子数量成良好线性比例的计算效率。数值示例验证了这种分段机器学习模型对于金属和共价材料，在具有可比精度的情况下，比各种流行的机器学习势快一个数量级以上，接近最快的嵌入原子方法的速度（即每个CPU核心每个原子几微秒）。这种方法的极高效率使其在超大型系统的第一性原理原子模拟和/或长时间尺度模拟中具有潜力。
