A proper treatment of electrostatic interactions is crucial for the accurate calculation of forces in computer simulations. Electrostatic interactions are typically modeled using Ewald based methods, which have become one of the cornerstones upon which many other methods for the numerical computation of electrostatic interactions are based. However, their use with charge distributions rather than point charges requires the inclusion of ansatz for the solutions of the Poisson equation, since there is no exact solution known for smeared out charges. The interest for incorporating electrostatic interactions at the scales of length and time that are relevant for the study the physics of soft condensed matter has increased considerably. Using mesoscale simulation techniques, such as dissipative particle dynamics (DPD), allows us to reach longer time scales in numerical simulations, without abandoning the particulate description of the problem. The main problem with incorporating electrostatics into DPD simulations is that DPD particles are soft and those particles with opposite charge can form artificial clusters of ions. Here we show that one can incorporate the electrostatic interactions through Ewald sums with point charges in DPD if larger values of coarse graining degree are used, where DPD is truly mesoscopic. Using point charges with larger excluded volume interactions the artificial formation of ionic pairs with point charges can be avoided, and one obtains correct predictions. We establish ranges of parameters useful for detecting boundaries where artificial formation of ionic pairs occurs. Lastly, using point charges we predict the scaling properties of polyelectrolytes in solvents of varying quality and obtain predictions that are in agreement with calculations that use other methods, and with recent experimental results.
An O(N) quasi-Ewald splitting method for nanoconfined electrostatics
The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；The Hong Kong University of Science and Technology（香港科技大学）；Flatiron Institute（平顿研究所）；East China Normal University（华东师范大学）
Long-range interactions & parallel scalability in molecular simulations
典型的生物分子系统如细胞膜、DNA和蛋白质复合物是高度带电的。因此，有效和准确地处理静电相互作用是非常重要的计算建模这样的系统。我们采用GROMACS仿真软件包，在一系列计算机体系结构上对不同的常用静电方案进行了广泛的基准测试（Pentium-4、IBM Power 4和Apple/IBM G5），单处理器和并行性能最高可达8个节点-我们还测试了四种不同网络上的可扩展性，即Infiniband、GigaBit以太网、快速以太网和几乎统一的内存架构，即，通过直接从其它CPU的本地存储器读取或写入其它CPU的本地存储器，CPU之间的通信成为可能。事实证明，粒子网格埃瓦尔德方法（PME）的表现令人惊讶的好，并提供有竞争力的性能，除非并行运行的PC硬件与旧的网络基础设施是必要的。使用大小为128、512和2048个脂质分子的脂质双层作为代表生物分子模拟中遇到的典型情况的测试系统。我们的研究结果使一个准确的预测计算速度在大多数当前的计算系统，无论是串行和并行运行。这些结果应有助于例如为一个小型部门计算机集群选择最合适的配置。
