We propose a new Monte Carlo method for efficiently sampling trajectories with fixed initial and final conditions in a system with discrete degrees of freedom. The method can be applied to any stochastic process with local interactions, including systems that are out of equilibrium. We combine the proposed path-sampling algorithm with thermodynamic integration to calculate transition rates. We demonstrate our method on the well studied 2D Ising model with periodic boundary conditions, and show agreement with other results both for large and small system sizes. The method scales well with the system size, allowing one to simulate systems with many degrees of freedom, and providing complementary information with respect to other algorithms.
Persistence of a Continuous Stochastic Process with Discrete-Time Sampling
我们引入了“离散时间持续性”的概念,它涉及在离散时间T = n ΔT测量的连续随机过程X(T)的零交叉。对于松弛率为μ的高斯马尔可夫过程,我们证明了对于大的n,持续(无交叉)概率随ρ(a)^n衰减,其中a = \exp(-μΔT),并且我们高精度地计算了ρ(a).我们还定义了“交替持续性”的概念,它对应于<0. For a>1,对应于不稳定势中的运动(μ<0),在无限时间内没有零交叉的非零概率,我们展示了如何计算它。
Discrete Equilibrium Sampling with Arbitrary Nonequilibrium Processes
D-Wave Systems Inc.(D-Wave系统公司)
我们提出了一个新的框架进行统计抽样,期望估计,分区函数近似使用\emph{arbitrary}启发式随机过程定义在离散状态空间。使用一个高度并行的结构,我们称之为\emph{sequential constraining process},我们能够同时产生的启发式过程的状态,并准确地估计其概率,即使它们太小,实际上是通过直接计数推断。在表明理论上正确的重要性抽样和马尔可夫链蒙特卡罗是可能的顺序约束过程中,我们将其集成到一个方法称为\emph{state space sampling},从计算机科学的抽样范围内的状态空间搜索的想法。该方法包括动态数据结构,该动态数据结构构造由受精度约束的启发式过程生成的统计的鲁棒贝叶斯模型,该精度约束为后验库尔贝克-莱布勒散度。从动态结构中采样通常会产生部分状态,这是通过递归调用启发式算法来细化结构并恢复采样来完成的。我们的各种伊辛模型的实验表明,状态空间采样,使启发式状态生成与准确的概率估计,证明了模拟退火过程的收敛到玻尔兹曼分布的运行长度增加。因此,迄今为止,前所未有的直接重要性抽样使用一般随机过程的\emph{final}(边际)分布是允许的,潜在地增加了蒙特卡罗从业者的处置的算法的范围。
Scalable Multitemperature Free Energy Sampling of Classical Ising Spin States
Bakar Institute of Digital Materials for the Planet, University of California, Berkeley(加州大学伯克利分校巴卡尔数字材料星球研究所);Institute of Science and Technology Austria(奥地利科学技术研究所);Department of Chemistry, University of California, Berkeley(加州大学伯克利分校化学系)