By default, all model parameters are sampled from a multivariate normal
distribution, whose characteristics are specified by the variance-covariance matrix.
OMEGAs and SIGMAs can be sampled from scaled inverse chi-squared or Wishart distributions
by setting the wishart argument to TRUE. In that case, THETAs are still sampled
from a multivariate normal distribution, while OMEGAS and SIGMAs are sampled from
scaled inverse chi-squared (univariate OMEGA/SIGMA distribution) and Wishart (block of
OMEGAs/SIGMAs) distribution, respectively. When wishart is set to TRUE, the degrees
of freedom of the distribution must be specified, respectively, odf for the OMEGAs and
sdf for the SIGMAs.
Usage
AutoReplicationSettings(
wishart = FALSE,
odf = NA,
sdf = NA,
checkMinMax = TRUE,
checkPosDef = FALSE,
quiet = NA
)Arguments
- wishart
logical, sample OMEGAs and SIGMAs from scaled inverse chi-squared (univariate OMEGA distribution) or Wishart distribution (block of OMEGAs)
- odf
the degrees of freedom for the scaled inverse chi-squared/Wishart distribution with regards to the OMEGAs, single integer value (the same degrees of freedom for all OMEGA blocks) or integer vector (one value per OMEGA block)
- sdf
the degrees of freedom for the scaled inverse chi-squared/Wishart distribution with regards to the SIGMAs, single integer value (the same degrees of freedom for all SIGMA blocks) or integer vector (one value per SIGMA block)
- checkMinMax
logical, check for min/max values when sampling the parameters, default is TRUE
- checkPosDef
logical, check for positive definiteness when sampling the OMEGA/SIGMA parameters from the variance-covariance matrix (i.e. when
wishart=FALSE), default is FALSE (requires extra time)- quiet
logical, suppress info messages, default is NA. By default, messages will be printed out when the success rate of sampling the parameters is below 95%.