These are built-in functions to be used by ran.CMData for generating
random multinomial data.
mg.Resample(n, clustersizes, param)mg.DirMult(n, clustersizes, param)
mg.LogitNorm(n, clustersizes, param)
mg.MixMult(n, clustersizes, param)
number of independent clusters to generate
an integer vector specifying the sizes of the clusters
a list of parameters for each specific generator
For mg.Resample: the param list should be list(param=...), in which
the CMData object to be resampled is passed.
For mg.DirMult: the param list should be list(shape=...), in which
the parameter vector of the Dirichlet distribution is passed
(see rdirichlet).
For mg.LogitNorm: the param list should be list(mu=...,sigma=...),
in which the mean vector and covariance matrix of the underlying Normal distribution
are passed. If sigma is NULL (or missing), then an identity matrix is assumed.
They should have K-1 dimensions for a K-variate multinomial.
For mg.MixMult: the param list should be list(q=...,m=...),
in which the vector of mixture probabilities q and the matrix m
of logit-transformed means of each component are passed.
For a K-variate multinomial, the matrix m should have K-1 columns
and length(q) rows.