Rates of convergence for Markov chains associated with Dirichlet processes.

TL;DRAbstract

We argue that Monte Carlo algorithms are ideally suited to parallel computing, and that “parallel Monte Carlo” should be more widely used. We consider a number of issues that arise, including dealing with slow or unreliable computers. We also discuss the possibilities of parallel Markov chain Monte Carlo. We illustrate our results with actual computer experiments.

Chat with Paper

AI Agents for this Paper