ABSTRACT
The early chapters of the book describe simulation approaches that work in lowdimensional problems. With complicated models, it is rare that samples from the posterior distribution can be obtained directly, and Chapter 11 describes iterative simulation algorithms that can be used with such models. In this chapter we describe an approximate approach using joint or marginal mode finding, and multivariate normal or t approximations about the modes. These approximations are useful for quick inferences and as starting points for iterative simulation algorithms such as described in Chapter 11. The approximations that we describe are relatively simple to compute, easy to understand, and can provide valuable information about the fit of the model.
