Professors Christophe Andrieu, Anthony Lee and Drs. Sam Power and Andi Q. Wang have had a recent paper accepted into the Annals of Statistics.
Their paper, titled ‘Comparison of Markov chains via weak Poincaré inequalities with application to pseudo-marginal MCMC’, studies subgeometric convergence rates of Markov chains. These results are particularly applicable to a class of computational methods known as pseudo-marginal MCMC methods, which are applicable to modern Bayesian inference problems with intractable likelihoods. By introducing a new class of functional inequalities in this context, the paper gives new convergence bounds which are substantially stronger and more transparent than previous approaches.
The research was funded by EPSRC grants CoSInES and Bayes4Health, and the paper can be found online.
