[R-sig-ME] MCMCglmm_2.13

[Jarrod Hadfield]

Hi,

A new version of MCMCglmm (MCMCglmm_2.13) has been submitted to CRAN  
and should be on-line soon.

NEW FEATURES:

a) Covariance structures for the random effects of the form  
kronecker(V,A) where A is some arbitrary matrix can now be fitted.  
Previous versions only allowed A matrices defined by a pedigree or  
phylogeny, and even then only single phylogenies/pedigrees could be  
passed. The A matrices (actually their sparse inverses) are passed via  
the argument ginverse.

b) c. 10% increase in speed for phylogenetic and pedigree models.

BUG FIXES:

i) If a column sum of Z was zero the associated term could be  
incorrectly omitted from the mixed model equations. This is unlikely  
to affect the vast majority of users.

ii) If a prior was specified with too many structures then the  
redundant prior structures were ignored. An error message is now issued.

iii) MCMCglmm failed with a cryptic error message if the internal  
nodes of a phylogeny were labelled but not uniquely. A more helpful  
error message is now given.

Cheers,

Jarrod




-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.