MetropolisHastingsCppD: C++ implementation of the algorithm for parameter calibration...

In CaliCo: Code Calibration in a Bayesian Framework

Description

Run a Metropolis Hastings within Gibbs algorithm and a Metropolis Hastings algorithm with the covariance matrix estimated on the the sample set generated in the Metropolis within Gibbs. This algorithm is suitable only for models with discrepancy.

Usage

MetropolisHastingsCppD(Ngibbs, Nmh, theta_init, r, SIGMA, Yf, binf, bsup,
  LogTest, stream)

Arguments

Ngibbs	the number of iteration in the Metropolis within Gibbs
Nmh	the number of iteration in the Metropolis Hastings
theta_init	the starting point
r	regulation percentage in the modification of the k
SIGMA	the covariance of the proposition distribution
Yf	the vector of recorded data
binf	the lower bound of the parameters to calibrate
bsup	the upper bound of the parameters to calibrate
LogTest	the log posterior density distribution
stream	(default=1) if stream=0 the progress bar is disabled

Value

list of outputs:

• PHIwg the points of the Metropolis within Gibbs algorithm in the transformed space

• PHI the points of the Metropolis Hastings algorithm in the transformed space

• THETAwg the points of the Metropolis within Gibbs algorithm in the real space

• THETA the points of the Metropolis Hastings algorithm in the real space

• AcceptationRatioWg the vector of the acceptance ratio for each parameter in the Metropolis within Gibbs

• AcceptationRatio the acceptance ratio in the Metropolis Hastings

• S the covariance computed after the Metropolis within Gibbs

• LikeliWG the likelihood computed at each iteration of the Metropolis within Gibbs algorithm

• Likeli the likelihood computed at each iteration of the Metropolis Hastings algorithm

CaliCo documentation built on May 2, 2019, 4:05 p.m.