Center for Computational Molecular Science and Technology, School of Chemistry and Biochemistry and School of Computational Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332-0400, USA.
Molecular Structure and Design, Bristol-Myers Squibb Company, P.O. Box 5400, Princeton, New Jersey 08543, USA.
J Chem Phys. 2020 Feb 21;152(7):074103. doi: 10.1063/1.5142636.
Accurate prediction of intermolecular interaction energies is a fundamental challenge in electronic structure theory due to their subtle character and small magnitudes relative to total molecular energies. Symmetry adapted perturbation theory (SAPT) provides rigorous quantum mechanical means for computing such quantities directly and accurately, but for a computational cost of at least O(N), where N is the number of atoms. Here, we report machine learned models of SAPT components with a computational cost that scales asymptotically linearly, O(N). We use modified multi-target Behler-Parrinello neural networks and specialized intermolecular symmetry functions to address the idiosyncrasies of the intermolecular problem, achieving 1.2 kcal mol mean absolute errors on a test set of hydrogen bound complexes including structural data extracted from the Cambridge Structural Database and Protein Data Bank, spanning an interaction energy range of 20 kcal mol. Additionally, we recover accurate predictions of the physically meaningful SAPT component energies, of which dispersion and induction/polarization were the easiest to predict and electrostatics and exchange-repulsion are the most difficult.
准确预测分子间相互作用能是电子结构理论中的一个基本挑战，因为它们的性质微妙，且相对于总分子能量来说数值较小。对称适应微扰理论（SAPT）为直接、准确地计算这些数量提供了严格的量子力学手段，但计算成本至少为 O(N)，其中 N 是原子的数量。在这里，我们报告了 SAPT 分量的机器学习模型，其计算成本呈渐近线性缩放，即 O(N)。我们使用改进的多目标 Behler-Parrinello 神经网络和专门的分子间对称函数来解决分子间问题的特殊性，在包括从剑桥结构数据库和蛋白质数据库中提取的结构数据的氢键复合物测试集上实现了 1.2 kcal mol 的平均绝对误差，其相互作用能范围为 20 kcal mol。此外，我们还恢复了 SAPT 分量能量的准确预测，其中色散和诱导/极化最容易预测，静电和交换排斥最难预测。
