1: \begin{abstract}

2: The objective of this paper is to study the Gibbs sampling for

3: computing the mean of observable in very high dimension -- a powerful

4: Markov chain Monte Carlo method. Under the Dobrushin's uniqueness

5: condition, we establish some explicit and sharp estimate of the

6: exponential convergence rate and prove some Gaussian concentration

7: inequalities for the empirical mean.

8: \end{abstract}