X2-PEC:基于原子对能量校正的神经网络模型
Minghong Jiang1, Zhanfeng Wang1, Yicheng Chen1
1Collaborative Innovation Center of Chemistry for Energy Materials, Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, MOE Key Laboratory of Computational Physical Sciences, Department of Chemistry, Fudan University, Shanghai, China.
Journal of computational chemistry
|March 18, 2025
概括
新的X2-PEC深度学习方法提高了低密度函数理论 (DFT) 计算的高准确性. 这种人工神经网络 (ANN) 方法改善了对分子性质的预测,例如原子化能量和形成的度.
科学领域:
- 计算化学计算化学
- 机器学习在化学中的应用
- 量子化学 是一个量子化学.
背景情况:
- 人工神经网络 (ANN) 在化学中越来越多地用于预测分子性质.
- 现有的方法经常面临准确性的局限性,特别是在低流密度函数理论 (DFT) 近似中.
- 需要一种方法来弥合低级别和高级别的DFT计算之间的准确差距.
研究的目的:
- 介绍X2-PEC方法,一个先进的人工神经网络 (ANN) 模型用于分子性质预测.
- 通过使用深度学习,提高低级别的DFT计算的准确性,达到高级别的DFT方法的水平.
- 为了证明X2-PEC对原子化能,形成和反应障碍的预测能力.
主要方法:
- 开发了X2-PEC方法,这是ANN方法X1系列的概括,包含对能量校正 (PEC).
- 使用重叠积分和核心哈密尔顿积分来捕获原子相互作用信息的构建特征向量.
- 在QM9数据集上训练了X2-PEC模型,并在各种标准化学数据集上评估了其性能.
主要成果:
- X2-PEC准确地预测了C6H8和C4H4N2O等同体的原子化能量.
- 该模型显示了多个数据集 (G2-HCNOF,PSH36,ALKANE28,BIGMOL20,HEDM45) 形成的标准度表现值得称赞.
- 在BH9的HCNOF子集上,X2-PEC还对反应障碍物具有很好的准确性,这表明它具有很强的概括能力.
结论:
- 通过深度学习,X2-PEC方法有效地将低级别的DFT计算的准确性提高到高级别的DFT.
- 该模型的特征向量结构成功地结合了原子相互作用的关键物理和化学信息.
- X2-PEC显示出显著的实用意义和计算化学进一步发展的潜力.
关键词:
B3LYPYP 在线播放BLYP BLYP 在线播放X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1 X1XYGJ-OSOS 的时间.密度函数理论密度函数理论形成的凝聚力形成的凝聚力.神经网络的神经网络的神经网络更多相关视频
相关概念视频
The Quantum-Mechanical Model of an Atom
41.7K
Shortly after de Broglie published his ideas that the electron in a hydrogen atom could be better thought of as being a circular standing wave instead of a particle moving in quantized circular orbits, Erwin SchrÃÂödinger extended de BroglieâÃÂÃÂs work by deriving what is now known as the SchrÃÂödinger equation. When SchrÃÂödinger applied his equation to hydrogen-like atoms, he was able to reproduce BohrâÃÂÃÂs...
41.7K
Hybridization of Atomic Orbitals II
31.5K
sp3d and sp3d 2 Hybridization
31.5K
Nuclear Binding Energy
12.1K
The difference between the calculated and experimentally measured masses is known as the mass defect of the atom. In the case of helium-4, the mass defect indicates a “loss” in mass of 4.0331 amu – 4.0026 amu = 0.0305 amu. The loss in mass accompanying the formation of an atom from protons, neutrons, and electrons is due to the conversion of that mass into energy that is evolved as the atom forms. The nuclear binding energy is the energy produced when the atoms’ nucleons...
12.1K
VSEPR Theory and the Effect of Lone Pairs
41.6K
Effect of Lone Pairs of Electrons on Molecule Geometry
41.6K
¹H NMR: Pople Notation
1.7K
The Pople nomenclature system classifies spin systems based on the difference between their chemical shifts. Coupled spins are denoted by capital letters with subscripts indicating the number of equivalent nuclei. When the coupled nuclei have well-separated chemical shifts, they are assigned letters that are far apart in the alphabet, such as A and X. When the difference in chemical shifts is small, coupled nuclei are named using adjacent letters of the alphabet (AB, MN, or XY).
A proton...
A proton...
1.7K
Molecular Orbital Theory II
18.9K
Molecular Orbital Energy Diagrams
18.9K


