Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: rjmcmc
Posted:
Nov 26, 2012 9:57 AM


On Sunday, November 25, 2012 1:41:43 AM UTC+5:30, David Jones wrote: > "majeti dinesh" wrote in message > > news:356ece8f0b884a3d88d2b3bd23bab14b@googlegroups.com... > > > > It is a method to samples from a posterior distribution. I am applying it to > > a gaussian mixture. I am sampling the mixture parameters and weights and the > > number of mixtures using this method. The sampling is done using the > > Metropolis method. To accept a move, we need to compute the probability of > > accepting it. This value is becoming very large negative value for me. > > > >  > > > > Given that clue, I found > > http://en.wikipedia.org/wiki/Reversiblejump_Markov_chain_Monte_Carlo and > > there are many other online pages to be found using an online search. You > > might try > > http://www1.imperial.ac.uk/resources/8b3cf549039e4f968beccab969a0695c/eph200501.pdf > > if you have not seen it, as it discusses some practical issues. Also > > www.springerlink.com/index/w72n45j33w754g94.pdf seems ontopic.
thank you very much



