Skip to content
Prev 7635 / 20628 Next

Bleeding edge lme4 (or lme4a) plus DF estimation

On Mon, Feb 27, 2012 at 4:48 PM, Joehanes, Roby (NIH/NHLBI) [F]
<roby.joehanes at nih.gov> wrote:
The SVN revision number is stored in the DESCRIPTION file for recent
versions of lme4Eigen.  It is actually the revision number for the
last check-in of the DESCRIPTION file but that gets updated fairly
frequently so the revision number in there is a pretty tight lower
bound.
The current revision (1623) should compile.  The breakage was
something I did on Friday and needed to commit because it was only on
my laptop.  I realized later that I had a modified version of
RcppEigen on my laptop and without those modifications the compilation
croaked. I created a version that doesn't use those particular
modifications, at least for now.
Yes, check out the refit function.  I just saw that the documentation
suffers from cut-and-paste errors but the general idea is to give a
fitted model a new response and run only the optimization step.
Linear mixed model fit by REML ['lmerMod']
Formula: Yield ~ 1 | Batch
   Data: Dyestuff

REML criterion at convergence: 319.6543

Random effects:
 Groups   Name        Variance Std.Dev.
 Batch    (Intercept) 1764     42.00
 Residual             2451     49.51
Number of obs: 30, groups: Batch, 6

Fixed effects:
            Estimate Std. Error t value
(Intercept)  1527.50      19.38    78.8
Linear mixed model fit by REML ['lmerMod']
Formula: Yield ~ 1 | Batch
   Data: Dyestuff

REML criterion at convergence: 161.8283

Random effects:
 Groups   Name        Variance Std.Dev.
 Batch    (Intercept)  0.00    0.000
 Residual             13.81    3.716
Number of obs: 30, groups: Batch, 6

Fixed effects:
            Estimate Std. Error t value
(Intercept)   5.6656     0.6784   8.352