Machine learning configuration interaction for ab initio potential energy curves
J. P. Coe
备注：
This document is the unedited Author's version of a Submitted Work that was subsequently accepted for publication in The Journal of Chemical Theory and Computation, copyright American Chemical Society after peer review. To access the final edited and published work see https://pubs.acs.org/articlesonrequest/AOR-7xICeRwJFp8nKTCDecNs
The concept of machine learning configuration interaction (MLCI) [J. Chem. Theory Comput. 2018, 14, 5739], where an artificial neural network (ANN) learns on the fly to select important configurations, is further developed so that accurate ab initio potential energy curves can be efficiently calculated. This development includes employing the artificial neural network also as a hash function for the efficient deletion of duplicates on the fly so that the singles and doubles space does not need to be stored and this barrier to scalability is removed. In addition configuration state functions are introduced into the approach so that pure spin states are guaranteed, and the transferability of data between geometries is exploited. This improved approach is demonstrated on potential energy curves for the nitrogen molecule, water, and carbon monoxide. The results are compared with full configuration interaction values, when available, and different transfer protocols are investigated. It is shown that, for all of the considered systems, accurate potential energy curves can now be efficiently computed with MLCI. For the potential curves of N$_{2}$ and CO, MLCI can achieve lower errors than stochastically selecting configurations while also using substantially less processor hours.
Intensities of all fine-structure resolved rovibrational electric quadrupole absorption lines in $^{16}$O$_2$($X^{3}Σ^{-}_{g}$) calculated with a new $\textit{ab initio}$ quadrupole moment curve
Universal Interatomic Potentials as Configuration-Space Generators for One-Shot and Iterative Fine-Tuning of Ab Initio-Accurate Material-Specific Models
Formation and properties of astrophysical carbonaceous dust. I: ab-initio calculations of the configuration and binding energies of small carbon clusters
Ab initio studies of ionization potentials of hydrated hydroxide and hydronium
用多体方法计算了氢氧化物和水合氢离子的电离势分布，电子激发构型由{ \it ab initio}分子动力学产生。实验结果得到了很好的再现，并与分子激发态密切相关。在稳定构型中，电离势主要受第一溶剂化层内水分子的扰动。另一方面，在质子转移过程中，电子激发在质子接收复合物和质子供给复合物上都是离域的，这使得两种水合离子的激发能发生移动，光谱变宽。
A fully {\it ab initio} potential curve of near-spectroscopic quality for the OH^- anion: importance of connected quadruple excitations and scalar relativistic effects
对羟基阴离子OH ^{-}的基态电势曲线进行了基准研究，包括1粒子和n粒子基组的详细校准。CCSD（T）基组极限高估了$ω_e$约10 cm ^{-1}，这只能通过在耦合团簇展开中包含连接的四重激发--或者等效地，在多参考计算的有效空间中包含$2π$轨道来补救。在包含标量相对论效应（$ω_e$上的~3 cm ^{-1}）后，获得光谱质量（亚cm ^{-1}精度）的势曲线。我们最好的计算EA（OH），1.828 eV，同意小数点后三位的最佳实验值。我们计算的最佳离解能D_0（OH ^-）= 4.7796 eV和D_0（OH）= 4.4124 eV表明实验D_0（OH）= 4.392 eV可能偏低了0.02 eV。
Interactions of Hydrogen Molecules with Halogen-Containing Diatomics from Ab Initio Calculations: Spherical-Harmonics Representation and Characterization of the Intermolecular Potentials
CH$_4\cdot$F$^-$ revisited: Full-dimensional ab initio potential energy surface and variational vibrational states
采用ROBOSURFER程序包，对CH $_4\cdot$ F $^-$络合物进行了从头算全维势能面（PES）的自动化研究.新的潜力提供了一个近光谱质量的描述在广泛的配置范围内，包括甲烷离子的解离，以及孤立的甲烷振动。特别是，它改进了早期的[Czakó，Braams，Bowman（2008）] PES在中间甲烷-氟化物距离。全维（12维）变分振动计算，使用新的PES和GENIUSH-Sjerak算法表明，隧穿分裂大于0.1厘米$^{-1}$下面出现的四个等效的最小值的复杂的相互转换障碍的顶部。
Enabling ab initio geometry optimization of strongly correlated systems with transferable deep quantum Monte Carlo
FU Berlin, Department of Mathematics and Computer Science（柏林自由大学数学与计算机科学系）；Microsoft Research AI4Science（微软研究院AI4Science）；FU Berlin, Department of Physics（柏林自由大学物理系）；Rice University, Department of Chemistry（莱斯大学化学系）
Potential energy curve for the $a^3Σ_u^+$ state of lithium dimer with Slater-type orbitals
我们报告了最先进的从头计算的势能曲线的$a^3Σ_u^+$状态的锂二聚体进行，以实现光谱的准确性（$<$ 1厘米$^{-1}$）没有任何事先调整，以适应相应的实验数据。非相对论性的钳核分量的相互作用能计算的复合方法，涉及六电子耦合集群和全组态相互作用理论结合Slater型轨道的质量范围从双到sextuple-zeta的基组。为了超越非相对论性的伯恩-奥本海默图像，我们包括领先的相对论和绝热修正，并发现这两种影响是不可忽略的，在目前的精度标准。我们开发的势能曲线允许计算分子参数（$D_e$、$D_0$、$ω_e$等）。对于这个系统，以及相应的振动能级，误差只有十分之几个波数（$0.2-0.4\,$ cm $^{-1}$）。我们还报告了一个从头计算值的散射长度的两个$^2S$锂原子，决定了相关的玻色-爱因斯坦凝聚的稳定性。
