Analysis of Convergence Diagnostics for MCMC methods
Palavras-chave:
Convergence, MCMC, Hellinger distance, StrategiesResumo
The most used method in Bayesian inference is the Markov Chain Monte Carlo (MCMC). But they
are expensive, not only because they require a lot of simulations of the model to explore the posterior distribution
but also because there are no clear criteria to determine if the method has converged to ensure the quality of the
obtained parameters and error estimates. This work objective is to explore a set of convergence diagnostics or
MCMC methods and their strategies for a convection-diffusion model and an epidemiological model, a SIR-type
model.
