1: \begin{abstract}

2: \noindent

3: We develop an Evolutionary Markov Chain Monte Carlo (EMCMC) algorithm

4: for sampling spatial partitions that lie within a large and complex

5: spatial state space.  Our algorithm combines the advantages of

6: evolutionary algorithms (EAs) as optimization heuristics for state

7: space traversal and the theoretical convergence properties of Markov

8: Chain Monte Carlo algorithms for sampling from unknown distributions.

9: Local optimality information that is identified via a directed search

10: by our optimization heuristic is used to adaptively update a Markov

11: chain in a promising direction within the framework of a Multiple-Try

12: Metropolis Markov Chain model that incorporates a generalized

13: Metropolis-Hasting ratio.  We further expand the reach of our EMCMC

14: algorithm by harnessing the computational power afforded by massively

15: parallel architecture through the integration of a parallel EA

16: framework that guides Markov chains running in parallel.

17: \end{abstract}

18: