Xi'an's Og

This morning I received the following email

(…) I have a question regarding an algorithm in one of your papers, “Bayesian Modelling and Inference on Mixtures of Distributions“.  On page 33, in the Metropolis-Hastings algorithm for the mixture you accept the proposal if r < u.  As I understand the MH algorithm you accept the proposal with probability r (technically min(r,1)), so I would expect that you accept if u < r.  I cannot see or find a reason elsewhere why r < u works?  If you could clarify why r < u works for the MH algorithm I would really appreciate it. (…)

which rightly points out an embarrassing typo in our mixture survey, published in the Handbook of Statistics, volume 25. Indeed, the inequality should be the reverse, , as in the other algorithmic boxes of the survey.

This entry was posted on September 19, 2010 at 10:45 am and is filed under Books, Statistics with tags Bayesian inference, handbook, mixture estimation, typo. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.