arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.10652cond-mat.stat-mech
收藏

一种用于粗粒度分子动力学模拟中广义朗之万恒温器记忆核迭代优化的高斯-牛顿方法

A Gauss-Newton method for iterative optimization of memory kernels for generalized Langevin thermostats in coarse-grained molecular dynamics simulations

V. Klippenstein, N. Wolf, N. F. A. van der Vegt

AI总结:

针对粗粒度分子动力学模拟中非马尔可夫广义朗之万恒温器记忆核提取的逆问题,提出基于IOMK的IOMK-GN高斯-牛顿方案,实现马尔可夫嵌入模型的直接参数化,解决原有方法误差风险高、计算成本大的问题。

摘要:

在分子动力学模拟中,动力学一致的粗粒度(CG)模型通常使用随机恒温器来模拟粗粒度描述中丢失的摩擦与涨落。尽管这类恒温器的马尔可夫(即时间局域)公式能够准确表示扩散系数/长时间动力学,但要正确描述所有时间尺度上的动力学,通常需要非马尔可夫(即非时间局域)恒温器。这类恒温器通常采用由记忆核确定的广义朗之万方程(GLE)形式。在本研究中,我们使用了一种独立作用于每个粗粒度自由度的、与位置无关的GLE恒温器的马尔可夫嵌入公式。从原子参考数据中提取该粗粒度模型的记忆核需要若干近似,因此该任务最好被理解为一个逆问题。我们近期提出的近似牛顿方案——记忆核迭代优化(IOMK)——能够实现记忆核的迭代优化,但马尔可夫嵌入仍然存在潜在的误差倾向且计算成本高昂。在本研究中,我们提出了一种基于IOMK的IOMK-高斯-牛顿方案(IOMK-GN),该方案可实现马尔可夫嵌入模型的直接参数化。

英文摘要:

In molecular dynamics simulations, dynamically consistent coarse-grained (CG) models commonly use stochastic thermostats to model friction and fluctuations that are lost in a CG description. While Markovian, i.e., time-local, formulations of such thermostats allow for an accurate representation of diffusivities/long-time dynamics, a correct description of the dynamics on all time scales generally requires non-Markovian, i.e., non-time-local, thermostats. These thermostats are typically in the form of a Generalized Langevin Equation (GLE) determined by a memory kernel. In this work, we use a Markovian embedded formulation of a position-independent GLE thermostat acting independently on each CG degree of freedom. Extracting the memory kernel of this CG model from atomistic reference data requires several approximations. Therefore, this task is best understood as an inverse problem. While our recently proposed approximate Newton scheme, Iterative Optimization of memory kernels (IOMK), allows for the iterative optimization of a memory kernel, Markovian embedding remained potentially error-prone and computationally expensive. In this work, we present a IOMK-Gauss-Newton scheme (IOMK-GN) based on IOMK, that allows for the direct parameterization of a Markovian embedded model.

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。

↑