Building upon our recently developed partial Hessian fitting (PHF) method (Wang et al., J. Comput. Chem. 2016, 37, 2349), we formulated and implemented two other rapid force‐field parameterization schemes called full Hessian fitting (FHF) and internal Hessian fitting (IHF), and comparisons were made among these three parameterization schemes to assess their performance. FHF minimizes deviation between the Hessian matrices in Cartesian coordinates computed by quantum mechanics (QM) and molecular mechanics (MM), to determine the best possible MM force‐constant parameters. While PHF requires step‐by‐step fittings of 3 × 3 partial Hessian matrices, FHF compares the lower triangular part of the QM and MM Hessian matrices, which allows simultaneous determination of all force‐constant parameters. In addition to this simple FHF scheme, IHF was developed such that it considers the Hessian matrices in redundant internal coordinates, where all possible internal coordinates that arise from the user‐defined interatomic connectivity are utilized. The results show that IHF performs best overall, followed by PHF and then FHF. Python‐based programing codes were developed to automate various tedious steps involved in the parameterization processes. © 2017 Wiley Periodicals, Inc.

中文翻译：

---

用于有效确定分子力学中力常数参数的分析 Hessian 拟合方案

基于我们最近开发的部分 Hessian 拟合 (PHF) 方法 (Wang et al., J. Comput. Chem. 2016, 37, 2349)，我们制定并实施了另外两个快速力场参数化方案，称为完全 Hessian 拟合 (FHF)和内部 Hessian 拟合（IHF），并在这三种参数化方案之间进行比较以评估它们的性能。FHF 最大限度地减少了由量子力学 (QM) 和分子力学 (MM) 计算的笛卡尔坐标中 Hessian 矩阵之间的偏差，以确定可能的最佳 MM 力常数参数。虽然 PHF 需要逐步拟合 3 × 3 部分 Hessian 矩阵，但 FHF 比较 QM 和 MM Hessian 矩阵的下三角部分，这允许同时确定所有力常数参数。除了这个简单的 FHF 方案，IHF 的开发使其考虑了冗余内坐标中的 Hessian 矩阵，其中利用了用户定义的原子间连通性产生的所有可能的内坐标。结果表明，IHF 整体表现最好，其次是 PHF，然后是 FHF。开发了基于 Python 的编程代码来自动化参数化过程中涉及的各种繁琐步骤。© 2017 威利期刊公司。