Gibbs Sampler is a popular statistical tool both in applied and theoretical work, by which we are able to avoid difficult calculations, replacing them instead with a sequence of easier calculations.
By broken the dependant relationships among the sequence elements, the stratege combined with function decomposition, global communication and simple mapping are explored to parallel the algorithm of Gibbs Sampler.
The evaluation of the parallel method is introducted by the result valification and performance evaluation.
Key Words: Gibbs Sampler, Functional Decomposition, Global Communication, MPI, Shared Memory.