Metropolis sampler after a Metropolis within Gibbs Sampler

Metropolis sampler after a Metropolis within Gibbs Sampler

MCMC(
  niterMwG,
  niterMH,
  parwalk,
  parinit,
  Rexp,
  tdensD,
  alpha,
  parprior,
  adaptive,
  calibration
)

Arguments

niterMwG	number of iterations of the Metropolis withing Gibbss algorithm mainly used to infer a covariance matrix for the random walk in the upcoming MH algorithm
niterMH	number of iterations of the Metropolis Hastings algorithm
parwalk	vector of variances for the random walk
parinit	vector of initial parameters rhos, variance of the observation noise, variance of the discrepancy and (if calibration) thetas
Rexp	residual of field data vs simulator
tdensD	output of the function tensordist
alpha	power in the exponential kernel
parprior	prior parameters for the parameters
adaptive	boolean to adapt the random walk
calibration	list of needed field if a calibration is to be performed in this order 1) computer model, 2) field observation, 3) field variable, 4) boolean indicating whether calibration is to be performed or not.

Value

Posterior sample in the transformed space

Details

The MwG sampler is used to estimate a covariance matrix for the random walk in the upcoming Metropolis sampler.