ChibML: Marginal Likelihood by Chib & Jeliazikov's Method for...
In erlisR/iLaplaceExamples: Extra Utilities and Examples for iLaplace

Description Usage Arguments Details Value References See Also

The function computes the marginal likelihood, i.e. the posterior normalising constant, with the method of Chib & Jeliazikov (2001) for user-written functions, from which an MCMC posterior sample is available.

1	ChibML(logfun, theta.star, tune, V, mcmcsamp, df, verbose)

`logfun`	The logarithm of the objective function
`theta.star`	The starting value of the inner MCMC sampling and the value required by the Chib & Jeliazikov's method. This must be a high denstiy point, such as the posterior mean, median or mode.
`tune`	The tunning value to be used to achieve the desired efficiency
`V`	The proposal scale matrix
`mcmcsamp`	The MCMC sample from the joint posterior
`df`	The degrees of freedom of the proposal
`verbose`	A switch which determines whether or not the progress of the sampler is printed to the screen. If verbose is greater than 0 the iteration number, and the Metropolis acceptance rate are sent to the screen every `verbose`th iteration

The function produce an approximation of the posterior normalizing constant via the Chib & Jeliazikov method in a single block sampling. The proposal distribution for the block is a Student's t-density with df degrees of freedom. The proposal is centered at the current value of theta and has scale matrix H. H is calculated as: H = T*V*T, where T is a the diagonal positive definite matrix formed from the tune.

double, the logarithm of the posterior normalising constant

Chib S. & Jeliazikov I. (2001). Marginal likelihood from the Metropolis-Hastings output. Journal of the American Statistical Association, 46, 270–281.

Robert C. P. & Casella G. (2004). Monte Carlo Statistical Methods. 2nd Edition. New York: Springer.

nlpost_gomp and nlpost_bod2 for examples; MHmcmc, isML

erlisR/iLaplaceExamples documentation built on May 16, 2019, 8:48 a.m.