Recently, an increasingly amount of literature focused on Bayesian computationalmethods to address problems with intractable likelihood. These algorithms are known asApproximate Bayesian Computational (ABC) methods. One of the problems of thesealgorithms is that the performance depends on the tuning of some parameters, such asthe summary statistics, distance and tolerance level.To bypass this problem, an alternative method based on empirical likelihood wasintroduced by Mengersen et al. (2013), which can be easily implemented when a set ofconstraints, related with the moments of the distribution, is known.However, the choice of the constraints is crucial and sometimes challenging in the sensethat it determines the convergence property of the empirical likelihood. To overcomethis problem, we propose an alternative method based on a bootstrap likelihoodapproach. The method is easy to implement and in some cases it is faster than the otherapproaches. The performance of the algorithm is illustrated with examples in PopulationGenetics, Time Series and a recent non-explicit bivariate Beta distribution. Finally, wetest the method on simulated and real data random fields.
展开▼
