Geometric ergodicity of Metropolis-Hastings algorithms for conditional simulation in generalized linear mixed models (Q5959768)

The authors consider random walk Metropolis and Langevin-Hastings algorithms for conditional simulation in generalized linear mixed model with correlated Gaussian effects. In particular, they study geometric ergodicity that ensures the validity of central limit theorems for Monte Carlo estimates, and justify the assessment of the precision of a Monte Carlo estimate by estimation of the asymptotic variance in the limiting normal distribution. Aside that, in an empirical study they compare asymptotic variances for random walk Metropolis and Langevin-Hastings algorithms and demonstrate the advantage of using the Langevin-Hastings algorithm.