粗粒化分子动力学模拟被广泛应用于研究或预测物质的形态结构，并能够合理解释介观尺度复杂体系的动态变化和性质。然而，现有的诸多粗粒化分子动力学力场还难以同时准确重现（弱）极性物质的多项宏观性质，仍有较大的改进空间。本论文采用分段Morse势函数描述粗粒化分子间相互作用，通过运用“元-多线性插值算法”（英文简写Meta-MIP）进行参数优化，发展了一套准确、可移植的弱极性物质的粗粒化力场（简称OPT1.35力场）。根据分子的极性差异，本论文创造性地提出了直接优化和间接优化两种策略，并开发出一套便捷、高效、成熟的自动参数化程序（将上传至GitHub托管平台开源共享）。对于极性较弱的分子采用直接参数化策略；而对于极性较强的分子，因其在模拟中易于结晶导致难以进行直接参数化，则提出通过连接丁基（或丙基）构成其同系物来进行间接优化力场参数。另外，本论文还巧妙地通过引入“搬土距离算法”（英文简写EMD），实现了成键参数的自动优化。通过EMD算法匹配全原子OPLS-AA力场下分子的键长、键角、二面角等结构信息的系综分布，从而获得粗粒化OPT1.35力场的成键参数，这对于描述生物大分子或聚合物分子的结构细节具有重要意义。

本论文所发展的OPT1.35力场能够准确同时重现简单的醛类、酮类、酯类、胺类、醚类、含硫有机分子、硝基烷烃、卤代烷烃、烯烃和炔烃等多种物质的多个宏观性质，如密度、蒸发焓、表面张力、溶剂化自由能等，模拟结果与实验值之间的误差均不超过5%。另外，本论文还严格优化了这些分子与水分子之间的力场参数，能够同时准确重现出它们与水的水化自由能、混合密度、相互作用能等宏观性质（误差均不超过4%）。该方法的准确性和可移植性通过重现脂肪酸甲酯（FAMEs）和聚氧化乙烯（PEO）等高分子的结构细节和宏观性质得以验证。本论文还将优化方法拓展至糖环、氨基酸等生物分子，提出了具有建设性的参数化路径，开发了诸多环形分子如四氢呋喃、四氢吡喃、环烷烃类、芳香族类的粗粒化力场。另外，通过本论文提出的“三角形边长法则”，OPT1.35力场能够有效地区分在苯环邻、间、对位取代的同分异构体分子。

Coarse-grained (CG) molecular dynamics simulations has been widely used to study or predict morphological structures of substances, and can reasonably explain the dynamic changes and properties of complex systems on a mesoscopic scale. However, most current CG force fields (FFs) are not precise enough, especially for (weak) polar molecules or functional groups. This thesis described the interaction between the CG molecules in the segmentation Morse potential function and developed a set of accurate and transferable CG FFs (OPT1.35 force field) by using the "meta multilinear interpolation of parameter sets method" (Meta-MIP). According to the polarity difference between the molecules, two strategies of direct optimization and indirect optimization were created, and a set of easy, efficient, as well as mature automatic parameterization programs were put forward (will be uploaded to the GitHub managed platform open-source sharing). For molecules with weak polarity, a direct parameterization strategy was applied. While for the polar molecule, it is difficult to perform direct parameterization since the freezing problem met at room temperature. Therefore, this thesis proposed to constitute one of its homologs by connecting butyl (or propyl) to indirectly optimize the force field parameters. In addition, this thesis wisely introduced "Earth Mover’s Distances method" (EMD) to automatically optimize the bonded parameters. The EMD algorithm is used to match the ensemble distribution of molecular bond length, bond angle, dihedral angle and other structural information under the all-atom OPLS-AA FF, so as to obtain the bonded parameters of the coarse-grained OPT1.35 FF which is useful for describing organisms. This is very important to describe the structural details of macromolecules or polymer molecules.

        The developed OPT1.35 FF accurately reproduced multi-properties of simple aldehydes, ketones, esters, amines, ethers, sulfur-containing organic molecules, nitroalkanes, haloalkanes, olefins, and alkynes, including density, heat of vaporization, surface tension, gas-liquid solution free energy, etc.; the relative errors between the simulation results and the experimental data are less than 5%. The interaction parameters between these molecules and water were also strictly optimized herein, and they can simultaneously reproduce the properties of hydration free energy, mixed density, and mixing energy (relative errors within 4%). The accuracy and transferability of OPT1.35 FFs were further demonstrated by reproducing the structural details and thermodynamic properties of polymers like fatty acid methyl esters (FAMEs) and polyethylene oxide (PEO). This thesis also extends the optimization method to biomolecules, such as carbohydrates and amino acids. In addition, some constructive parameterization paths were proposed in this thesis, and the CG FF of many ring molecules were also developed, such as tetrahydrofuran, tetrahydropyran, cycloalkanes, and aromatics. What's more, according to the "triangle side length rule" which was first proposed in this thesis, OPT1.35 FF could effectively distinguish the benzene ring in the ortho, meta, para-substituted isomer molecules.