PROBABILITIES : AN ALTERNATIVE TO THE METROPOLIS – HASTINGS ALGORITHM
PROBABILITIES : AN ALTERNATIVE TO THE METROPOLIS – HASTINGS ALGORITHM
复制标题
概率：都市-黑斯廷斯算法的替代方案
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
S. Walker
中科院分区：
文献类型：
--
作者：
S. Walker;S. Walker
马尔可夫链蒙特卡罗方法现在非常流行，并用于科学学习的各个方面。其中使用最广泛和最有效的方法是Metropolis - Hastings算法。在本文中，当状态空间是可数无限且平稳或目标分布由一组非归一化概率表示时，我们引入了Metropolis-Hastings采样器的替代方案。我们用大都会-黑斯廷斯采样器的比较来说明。
Markov chain Monte Carlo methods are now hugely popular and are used in all aspects of scientific learning. One of the most widely used and efficient methods is the Metropolis– Hastings algorithm. In this note, we introduce an alternative to the Metropolis–Hastings sampler when the state space is countably infinite and the stationary or target distribution is represented by a set of unnormalized probabilities. We illustrate with a comparison of the Metropolis–Hastings sampler.