1: \begin{abstract} 

2: 

3: In this paper, we consider the implications of the fact that parallel raw-power can be exploited by a generic

4: Metropolis--Hastings algorithm if the proposed values are independent from the current value of the Markov

5: chain. In particular, we present improvements to the independent Metropolis--Hastings algorithm that

6: significantly decrease the variance of any estimator derived from the MCMC output, at a null 

7: computing cost since those improvements are based on a fixed number of target density evaluations that can be

8: produced in parallel. The techniques developed in this paper do not jeopardize the Markovian convergence

9: properties of the algorithm, since they are based on the Rao--Blackwell principles of

10: \cite{gelfand:smith:1990}, already exploited in \cite{casella:robert:1996}, \cite{atchade:perron:2005} and

11: \cite{douc:robert:2010}. We illustrate those improvements both on a toy normal example and on a classical

12: probit regression model, but stress the fact that they are applicable in any case where the independent

13: Metropolis--Hastings is applicable. 

14: %The code used in this paper is available as supplementary material.

15: 

16: \noindent {\bf Keywords:} MCMC algorithm, independent Metropolis--Hastings, parallel computation,

17: Rao\--Black\-well\-ization, permutation.

18: \end{abstract}

19: