AI 中文总结
本研究开发了结合身份切换与AVB移动的RxVB蒙特卡罗试探方法，用于受限空间化学吸附采样，在单狭缝孔模型中效率提升显著，已集成于FEASST软件包。
AI 中文摘要
采用蒙特卡罗方法对化学吸附进行分子建模时，需要开发新的试探移动以高效采样复杂流体，比如布朗斯特酸沸石中的水。本研究开发了一种反应体积偏倚（RxVB）蒙特卡罗试探方法，用于建模化学吸附，该方法结合了身份切换与聚集体积偏倚（AVB）移动。RxVB试探通过选择处于任意指定反应体积内的反应对，以促进反应采样。在单狭缝孔模型中，RxVB试探相比无偏试探，可使已接受反应事件的数量提升高达90倍；在考虑计算开销后，统计效率提升对应为70倍。但将相同试探应用于布朗斯特酸沸石中的水且无定向偏倚时，单个MFI晶胞未观测到可测量的加速效果。本研究推导了一个简单表达式，可在选择彼此靠近的反应物为主要采样瓶颈的最简情形下，预测最大效率提升值。稠密水系统可能需要额外的构型偏倚或定向偏倚，以改善氢键网络的采样，进而提高接受率。该新型RxVB试探方法已在开源自由能与高级采样模拟工具包（FEASST）模拟软件包中提供，并附带示例。
英文摘要
Molecular modeling of chemisorption with Monte Carlo requires the development of new trial moves to efficiently sample complex fluids such as water in Bronsted acid zeolites. Here, we develop a reaction volume bias (RxVB) Monte Carlo trial for modeling chemisorption by combining identity-switch and aggregation-volume-bias (AVB) moves. This method aims to promote the sampling of reactions by choosing reactive pairs that are within an arbitrarily specified reaction volume. The RxVB move achieves up to a 90-fold increase in accepted reaction events over unbiased moves in a single-site slit-pore model, corresponding to a 70-fold gain in statistical efficiency after accounting for computational overhead. But when the same move is applied to water in a Bronsted acid zeolite without orientational bias, there is no measurable speedup for a single MFI unit cell. We demonstrate a simple expression that predicts the maximum efficiency increase in the simplest case where selecting reactants that are near each other is the major sampling bottleneck. Dense water systems may require additional configuration-bias or orientational bias to improve sampling of the hydrogen bond network in order to increase acceptance. This new RxVB trial was made available with examples in the open-source Free Energy and Advanced Sampling Simulation Toolkit (FEASST) simulation package.
Comments15 pages, 9 figures, 3 tables
