Due to its removal from CRAN, QGglmm dropped R2Cuba as a dependency to solve multivariate integrals. It is now using the package cubature. By taking advantage of the "vectorised" version of the algorithm, the multivariate computations of QGglmm (QGmvparams, QGvcov, QGmvmean, QGmvpsi, QGmvicc, QGmvpred) are considerably faster. Most functions are 10x-50x faster, but especially QGmvicc is 100x-500x faster. A comparison between the old and new version of the example of the man page of QGmvicc showed a decreased in computation from 25 minutes to... 4 seconds!
Multivariate code is now complete, including multivariate predictions. The univariate predictions have been changed as well: it now uses the derivative of fitness (to be calculated by the user), which allows for much improved computation time!
As usual, the code has been tested but this is still a pre-release which might contain some bugs. This version, however, is the first feature-full one, so we are switching to a beta pre-release version.
Prepared for CRAN submission. This will be the first version submitted to CRAN.
This version has an almost complete multivariate code (only the QGmvpred is missing). This multivariate code has been tested, but some glitches might still lie there.
Also added a new Gaussian model (i.e. LMM), especially useful for multivariate analysis + updated all the doc and metadata. Several other minor bug/typos fixes.
This is the first "alpha" development version. The package should compile and install in R and all the functions should work without any critical issue.
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.