Markov Chain Monte Carlo (MCMC) methods have greatly influenced the applied statistics community

by enabling them to simulate more efficiently from complex high dimensional distributions. However, the

efficiency of the algorithm is very much dependent on the choice of the proposal distribution and bad choice

of the proposal distribution will slow down the convergence of the algorithm. In this talk I will describe the

MCMC algorithm and discuss a way to address the above problem. This talk will be at a very basic level

requiring only a little knowledge of Markov chain theory.