Marginal likelihood computation paper
In the last summer, we (Martin, Neal and I) have tested various methods to improve the accuracy of marginal likelihood for Bayes factor computation. What we’ve found then came out as a paper submitted to a journal and has been posted on astroph today (http://arxiv.org/abs/1301.3156). The methods is simple but provides a remarkably accurate marginal likelihood value. It has been tested up to 16 dimensional space and put a solid foundation for GALPHAT galaxy model selection in the coming papers.