V the simulated chains should be converted to class “mcmc”

V summary should be more flexible

V add functions for checking convergence (see the coda package)

V plot methods should include a logical argument 'burnin' to include or not the burnin period in the graphs

V use the vector library or try using Rcpp

- add other relabel algorithms (e.g. Fruhwirth-Schnatter; see also the label.switching package)

V include thinning

V use a single .h file

V write check_control

X include the possibility to draw the z_ijh in random order [BETTER NOT!]

- simulating more than one chain produces results that appear to be "mixed up". should me perform another Procrustes analysis across chains to make the results more comparable?

- convert the bmds() function to C

- redo all examples involving dmbc_ic() but specifying post.proc = FALSE in the control list