首页> 外文OA文献 >On the use of Markov chain Monte Carlo methods for the sampling of mixture models
【2h】

On the use of Markov chain Monte Carlo methods for the sampling of mixture models

机译:关于马尔可夫链蒙特卡罗方法的应用   混合模型

摘要

In this paper we study asymptotic properties of differentdata-augmentation-type Markov chain Monte Carlo algorithms sampling frommixture models comprising discrete as well as continuous random variables. Ofparticular interest to us is the situation where sampling from the conditionaldistribution of the continuous component given the discrete component isinfeasible. In this context, we cast Carlin & Chib's pseudo-prior method intothe framework of mixture models and discuss and compare different variants ofthis scheme. We propose a novel algorithm, the FCC sampler, which is lesscomputationally demanding than any Metropolised Carlin & Chib-type algorithm.The significant gain of computational efficiency is however obtained at thecost of some asymptotic variance. The performance of the algorithm vis-\`a-visalternative schemes is investigated theoretically, using some recent resultsobtained in [3] for inhomogeneous Markov chains evolving alternatinglyaccording to two different reversible Markov transition kernels, as well asnumerically.
机译:在本文中,我们研究了从包含离散和连续随机变量的混合模型中采样的不同数据增强型马尔可夫链蒙特卡罗算法的渐近性质。我们特别感兴趣的是这样一种情况,即在给定离散分量的情况下,从连续分量的条件分布进行采样是不可行的。在这种情况下,我们将Carlin&Chib的伪先验方法转换为混合模型的框架,并讨论和比较了该方案的不同变体。我们提出了一种新颖的算法,即FCC采样器,它比任何Metropolised Carlin&Chib型算法都具有较低的计算要求,但是以一些渐近方差为代价获得了可观的计算效率。理论上研究了算法相对于另类方案的性能,使用[3]中获得的一些最新结果,证明了非均质马尔可夫链根据两个不同的可逆马尔可夫转移核交替演化,并且在数值上有所变化。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号