MCMC methods with Gibbs sampling and non-informative flat prior, using posterior distribution