Abstract:
The lateral stress profile of a lipid bilayer constitutes a valuable link between molecular simulation and mesoscopic elastic theory. Even though it is frequently calculated in simulations, its statistical precision (or that of observables derived from it) is often left unspecified. This omission can be problematic, as uncertainties are prerequisite to assessing statistical significance. In this chapter, we provide a comprehensive yet accessible overview of the statistical error analysis for the lateral stress profile. We detail two relatively simple but powerful techniques for generating error bars: block-averaging and bootstrapping. Combining these methods allows us to reliably estimate uncertainties, even in the presence of both temporal and spatial correlations, which are ubiquitous in simulation data. We illustrate these techniques with simple examples like stress moments, but also more complex observables such as the location of stress profile extrema and the monolayer neutral surface.
摘要:
脂质双层的横向应力分布构成了分子模拟和介观弹性理论之间的宝贵联系。即使它经常在模拟中计算,它的统计精度(或从中得出的可观测值的精度)通常不确定。这种遗漏可能会有问题,因为不确定性是评估统计显著性的先决条件。在这一章中,我们提供了横向应力分布的统计误差分析的全面而可访问的概述。我们详细介绍了两种相对简单但功能强大的生成错误条的技术:块平均和自举。结合这些方法,我们可以可靠地估计不确定性,即使存在时间和空间相关性,在模拟数据中无处不在。我们用简单的例子来说明这些技术,比如压力力矩,但也有更复杂的观测,如应力分布极值和单层中性表面的位置。