Catalytic perfect simulation
The aim of the present paper is to introduce a general methodology for perfect simulation technique. The paper describes how to convert an existing Markov chain implementation with prescribed stationary distribution, so that perfect samples from the target density to be read off the sample path (at random time intervals). Three important approaches are competing to accomplish this perfect sampling methodology within a general Markov chain: (1) The Read-Once CFTP (Coupling From The Past) algorithm of \textit{D. B. Wilson} [Random Struct. Algorithms 16, No. 1, 85-113 (2000; Zbl 0952.60072)] (2) The second approach used is due to D. Murdoch (1999), and is aimed to obtain a uniformly ergodic Markov chain by utilizing iterations from a merely positive recurrent chain. This is done by inserting regularly a step from an independence sampler with a sufficiently heavy tailed proposal distribution (usually the prior in a Bayesian posterior simulation problem). The independence sampler is one of the simplest types of Metropolis-Hastings algorithms, where the sequence of proposed transitions is merely made up of an independent, identically distributed sequence of random variables. (3) The third involved method belongs to the authors, and provides a coupling construction which allows a computationally straightforward approach to the implementation of Wilson's Read-Once CFTP blueprint. The proposed methodology for the perfect simulation is illustrated on two examples of Bayesian posterior distributions: a hierarchical model example, and a finite mixture example.
- On the exact simulation of functionals of stationary Markov chains
- Perfect sampling from independent Metropolis-Hastings chains
- Small sets and Markov transition densities.
- Geometric ergodicity and perfect simulation
- A mixture representation of \(\pi\) with applications in Markov chain Monte Carlo and perfect sampling.
- Perfect sampling for Bayesian variable selection in a linear regression model
- Perfect simulation for models of industry dynamics
- One-shot CFTP; application to a class of truncated Gaussian densities
- Perfect simulation for mixtures with known and unknown number of components
- Practical perfect sampling using composite bounding chains: the Dirichlet-multinomial model
- Perfect simulation
- Perfect posterior simulation for mixture and hidden Markov models
- Notes on perfect simulation
- Using a Markov Chain to Construct a Tractable Approximation of an Intractable Probability Distribution
- Perfect sampling for doeblin chains
- scientific article; zbMATH DE number 1560253 (Why is no real title available?)
- scientific article; zbMATH DE number 1560256 (Why is no real title available?)
- Perfect Samplers for Mixtures of Distributions
- Monotonic multigamma coupling for perfect sampling
- Extensions of Fill's Algorithm for Perfect Simulation
- How to couple from the past using a read-once source of randomness
- Nonasymptotic bounds on the mean square error for MCMC estimates via renewal techniques
- Perfect Simulation for Image Restoration
- Perfect Forward Simulation via Simulated Tempering
This page was built for publication: Catalytic perfect simulation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1610835)
