DFT 力量的精度有多高? 在分子数据集中出乎意料地存在很大的不确定性
Domantas Kuryla1,2, Fabian Berger1, Gábor Csányi2
1Yusuf Hamied Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge, United Kingdom.
The Journal of chemical physics
|December 10, 2025
概括
准确的机器学习原子间潜力 (MLIP) 需要来自密度函数理论 (DFT) 计算的高质量数据. 这项研究在几个流行的DFT数据集中发现了显著的非零净力和力误差,影响了MLIP的准确性.
科学领域:
- 计算材料科学科学 计算材料科学
- 机器学习在化学中的应用
- 量子力学就是量子力学.
背景情况:
- 机器学习原子间潜力 (MLIP) 越来越多地用于材料模拟.
- 训练MLIP需要大量的数据集,通常是使用密度函数理论 (DFT) 生成的.
- MLIPs的准确性取决于底层的DFT数据的质量和趋同.
研究的目的:
- 评估用于MLIP培训的DFT生成数据的质量.
- 在常用的 DFT 数据集中识别潜在的错误来源.
- 量化DFT不准确性对力组件的影响.
主要方法:
- 在几个著名的DFT数据集 (SPICE,Transition1x,ANI-1x,ANI-1xbb,AIMNet2,QCML,OMol25) 中分析了净力.
- 使用更接近的DFT设置来量化错误的力量的重新计算.
- 原始和重新计算的数据集之间的力成分的比较.
主要成果:
- 一些分析的DFT数据集显示出显著的非零净力,表明潜在的趋同问题.
- 量化了单个力组件的误差,平均差异范围从1.7 meV/Å (SPICE) 到33.2 meV/Å (ANI-1x).
- 在多个数据集中观察到强力组件的实质性差异.
结论:
- DFT数据的质量是制定准确的MLIPs的关键瓶.
- 非汇聚的DFT计算在力组件中引入了重大错误,损害了MLIP的可靠性.
- 确保良好融合的DFT数据对于推动MLIP开发和应用至关重要.
更多相关视频
12:11Computation of Atmospheric Concentrations of Molecular Clusters from ab initio Thermochemistry
Published on: April 8, 2020
8.6K
10:52Multiscale Sampling of a Heterogeneous Water/Metal Catalyst Interface using Density Functional Theory and Force-Field Molecular Dynamics
Published on: April 12, 2019
13.3K
相关概念视频
Propagation of Uncertainty from Systematic Error
1.2K
The atomic mass of an element varies due to the relative ratio of its isotopes. A sample's relative proportion of oxygen isotopes influences its average atomic mass. For instance, if we were to measure the atomic mass of oxygen from a sample, the mass would be a weighted average of the isotopic masses of oxygen in that sample. Since a single sample is not likely to perfectly reflect the true atomic mass of oxygen for all the molecules of oxygen on Earth, the mass we obtain from this...
1.2K
Molecular Geometry and Dipole Moments
17.8K
The VSEPR theory can be used to determine the electron pair geometries and molecular structures as follows:
17.8K
IR Spectroscopy: Hooke's Law Approximation of Molecular Vibration
2.7K
A covalently bonded heteronuclear diatomic molecule can be modeled as two vibrating masses connected by a spring. The vibrational frequency of the bond can be expressed using an equation derived from Hooke's law, which describes how the force applied to stretch or compress a spring is proportional to the displacement of the spring. In this case, the atoms behave like masses, and the bond acts like a spring.
According to Hooke's law, the vibrational frequency is directly proportional to...
According to Hooke's law, the vibrational frequency is directly proportional to...
2.7K
