Density Functional Theory for Microwave Spectroscopy of Noncovalent Complexes: A Benchmark Study

In this work, we compare the results obtained with 89 computational methods for predicting noncovalent bond lengths in weakly bound complexes. Evaluations for the performance in noncovalent interaction energies and covalent bond lengths obtained from five other data sets are included. The overall best performing density functional is the ωB97M-V method, achieving balanced results across all three categories. For noncovalent geometries, the best methods include B97M-V, B3LYP-D3­(BJ) and DSD-PBEPBE-D3­(BJ). The effects of systematic improvement of the density functional approximation and of dispersion corrections are also discussed.本研究对比了89种用于预测弱结合复合物（weakly bound complexes）中非共价键长（noncovalent bond lengths）的计算方法所得结果。本研究同时纳入了来自另外5个数据集的、针对非共价相互作用能（noncovalent interaction energies）与共价键长（covalent bond lengths）计算性能的评估结果。全局表现最优的密度泛函（density functional）为ωB97M-V方法，其在全部三类评估维度中均取得了均衡的结果。针对非共价几何结构（noncovalent geometries）的预测任务，表现优异的计算方法包括B97M-V、B3LYP-D3(BJ)以及DSD-PBEPBE-D3(BJ)。本研究还探讨了密度泛函近似（density functional approximation）与色散校正（dispersion corrections）的系统性改进所带来的影响。