Skip to content
Prev 14721 / 20628 Next

glmmPQL crashes on inclusion of corSpatial object

This example crashes for me too under Ubuntu 14.04 (virtual), r-devel,
with

fit5 <- glmmPQL(fo, random = ~1 | date, data = d,
+                 correl = correl, family = binomial)
iteration 1
*** Error in `/usr/local/lib/R/bin/exec/R': corrupted double-linked
list: 0x0000000007bef0d0 ***

  This was asked on Stack Overflow as well:

http://stackoverflow.com/questions/38509095/glmmpql-crashes-on-inclusion-of-corspatial-object

 I was going to suggest INLA as well.  I was working up to providing a
worked example, but it's non-trivial to set up (understatement): see
e.g.
https://www.math.ntnu.no/inla/r-inla.org/tutorials/spde/spde-tutorial.pdf

  Another non-trivial approach:
https://github.com/bbolker/mixedmodels-misc/blob/master/notes/corr_braindump.rmd
(use lme4's modular structure to incorporate nlme-based corStruct
information within models) -- this would be less efficient since it
would involve computing the Cholesky decomposition of a 1000x1000 matrix
at each step (and, indeed, might crash in the same place that glmmPQL
did), where INLA uses some very very clever computational tricks ...
On 16-07-25 07:36 AM, Thierry Onkelinx wrote: