03-09-2013 09:58 PM
When a parameter is constrained in some range, say positive, we can choose it's piror distribution to keep it's positivity. The posivity of parameters sample from posterir can not be guaranteed because of the complexity of posteior distributon. In Proc MCMC, we can control the acceptance rate of M-H sampling by setting options.How can we contral the acceptance of sampling results, say only the positive samples are accept? Thanks alot!