预测具有分段常数相互作用的流体结构：比较五种有效积分方程理论的精度
Predicting the structure of fluids with piecewise constant interactions: Comparing the accuracy of five efficient integral equation theories
摘要：
我们使用分子动力学模拟来测试积分方程理论预测的球形颗粒的流体结构与八个不同的分段常数对相互作用的形式，包括一个硬核和两个肩膀和/或井的组合。由于像这样的模型对势对于复杂流体中有效相互作用的离散化或粗粒度表示（例如，对于计算密集型逆优化问题），我们在这里集中评估如何准确地预测其属性可以通过分析或简单的数值闭包，包括Percus-Yevick，超网链，参考超网链，一阶平均球面近似，和修改的一阶平均球面近似。为了使预测和模拟的径向分布函数之间的定量比较，我们引入了累积结构误差度量。对于这些模型的平衡流体状态点，我们发现，参考超网链封闭是最准确的测试近似，其特征在于此度量或相关的热力学量。
英文摘要：
We use molecular dynamics simulations to test integral equation theory predictions for the structure of fluids of spherical particles with eight different piecewise-constant pair interaction forms comprising a hard core and a combination of two shoulders and/or wells. Since model pair potentials like these are of interest for discretized or coarse-grained representations of effective interactions in complex fluids (e.g., for computationally intensive inverse optimization problems), we focus here on assessing how accurately their properties can be predicted by analytical or simple numerical closures including Percus-Yevick, hypernetted chain, reference hypernetted chain, first-order mean spherical approximation, and a modified first-order mean spherical approximation. To make quantitative comparisons between the predicted and simulated radial distribution functions, we introduce a cumulative structural error metric. For equilibrium fluid state points of these models, we find that the reference hypernetted chain closure is the most accurate of the tested approximations as characterized by this metric or related thermodynamic quantities.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
