Xi'an's Og

Dennis Prangle and Richard Everitt arXived a note today where they point out the identity between the auxiliary variable approach of Møller et al. (2006) [or rather its multiple or annealed version à la Murray] and [exact] ABC (as in our 2009 paper) in the case of Markov random fields. The connection between the two appears when using an importance sampling step in the ABC algorithm and running a Markov chain forward and backward the same number of steps as there are levels in the annealing scheme of MAV. Maybe more a curiosity than an indicator of a large phenomenon, since it is so rare that ABC can be use in its exact form.

This entry was posted on May 9, 2016 at 12:16 am and is filed under Books, pictures, Statistics, University life with tags ABC, auxiliary variable, doubly intractable problems, Markov random field, Newcastle-upon-Tyne, unbiased estimation. 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.