McmcParams-class {CNPBayes} | R Documentation |
An object to specify MCMC options for a later simulation
thin
A one length numeric to specify thinning. A value of n indicates that every nth sample should be saved. Thinning helps to reduce autocorrelation.
iter
A one length numeric to specify how many MCMC iterations should be sampled.
burnin
A one length numeric to specify burnin. The first $n$ samples will be discarded.
nstarts
A one length numeric to specify the number of chains in a simulation.
param_updates
Indicates whether each parameter should be updated (1) or fixed (0).
min_GR
minimum value of multivariate Gelman Rubin statistic for diagnosing convergence. Default is 1.2.
min_effsize
the minimum mean effective size of the chains. Default is 1/3 * iter.
max_burnin
The maximum number of burnin iterations before we give up and return the existing model.
min_chains
minimum number of independence MCMC chains used for assessing convergence. Default is 3.
McmcParams() McmcParams(iter=1000) mp <- McmcParams() iter(mp)