j.hadfield at ed.ac.uk
Sat Dec 19 18:29:46 CET 2009
I've released a new version of MCMCglmm which should appear on CRAN
soon. The main updates are:
I) Parameter expanded models now implemented which a) speed up mixing
when variances are close to zero and b) allow non-central F-
distributed priors for the variance components
||) Latent variables in binary models can be updated using slice
sampling (slice=TRUE) - this is usually more efficient
|||) Known bugs fixed - all but one (for ordinal models with more than
2 categories) resulted in failure to fit a model rather than fitting
an incorrect model.
||||) Extended set of course notes vignette("CourseNotes",
"MCMCglmm"). These are incomplete, but better than not having them,
hopefully. There will undoubtedly be errors in there - if you could
let me know I'd be grateful.
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
More information about the R-sig-mixed-models