Replica exchange and expanded ensemble simulations as Gibbs sampling: Simple improvements for enhanced mixing
Replica exchange and expanded ensemble simulations as Gibbs sampling: Simple improvements for enhanced mixing
复制标题
DOI：
10.1063/1.3660669
复制
发表时间：
2011-11-21
影响因子：
4.4
通讯作者：
Shirts, Michael R.
中科院分区：
文献类型：
--
作者：
Chodera, John D.;Shirts, Michael R.
副本交换和扩展系综算法在化学和生物物理学中用于模拟复杂分子系统的广泛流行，引起了人们对发现新的方法来增强这些协议的相空间混合以改善对不相关构型的采样的兴趣。在这里，我们演示了这两类算法如何被视为马尔可夫链蒙特卡罗框架内Gibbs抽样的特例。Gibbs抽样是统计推断领域中研究得很好的一种方法，在这种方法中，不同的随机变量从条件分布交替更新。虽然Metropolis蒙特卡罗或分子动力学对构象自由度的更新不可避免地会产生相关的样本，但我们展示了如何明智地更新热力学状态指数-对应于热力学参数，如温度或炼金术耦合变量-可以显著增加混合，同时仍从期望的分布中采样。我们展示了常用的状态更新方法如何导致次优混合，并提出了一些简单、廉价的替代方案，它们可以增加整个马尔可夫链的混合，减少获得期望精度估计所需的模拟时间。这些改进的方案被展示在几个常见的应用中，包括炼金术扩展集成模拟、并行回火和多维副本交换伞采样。C2011美国物理研究所。[DOI：10.1063/1.3660669]
The widespread popularity of replica exchange and expanded ensemble algorithms for simulating complex molecular systems in chemistry and biophysics has generated much interest in discovering new ways to enhance the phase space mixing of these protocols in order to improve sampling of uncorrelated configurations. Here, we demonstrate how both of these classes of algorithms can be considered as special cases of Gibbs sampling within a Markov chain Monte Carlo framework. Gibbs sampling is a well-studied scheme in the field of statistical inference in which different random variables are alternately updated from conditional distributions. While the update of the conformational degrees of freedom by Metropolis Monte Carlo or molecular dynamics unavoidably generates correlated samples, we show how judicious updating of the thermodynamic state indices-corresponding to thermodynamic parameters such as temperature or alchemical coupling variables-can substantially increase mixing while still sampling from the desired distributions. We show how state update methods in common use can lead to suboptimal mixing, and present some simple, inexpensive alternatives that can increase mixing of the overall Markov chain, reducing simulation times necessary to obtain estimates of the desired precision. These improved schemes are demonstrated for several common applications, including an alchemical expanded ensemble simulation, parallel tempering, and multidimensional replica exchange umbrella sampling. c 2011 American Institute of Physics. [doi: 10.1063/1.3660669]