| Contents |
| Girolami and Calderhead suggested the use of variable masses to improve the performance of Monte Carlo algorithms to sample from a target density. Recently explicit methods for simulating diffusions whose generator is self-adjoint with respect to a known density have been proposed in the context of Molecular Dynamics. Furthermore, these methods do not require the evaluation of the divergence of the mass matrix.
In this talk we investigate the applicability and efficiency of these methods when dealing with sampling problems arising in Bayesian Statistics. |
|