[R-sig-ME] Summarizing the fitted model takes more RAM than fitting it

[Jan Pablo Burgard]

I encountered the same Problem and I solved it by inserting the summary into a 
database created by the filehash package.

The nicer solution though (iff the platform is accessible) is to use a 64bit 
Linux with a huge swap. As the summary is only dumped and not much computation 
is done on it it won't cost very much more time. Save it and load it again for 
further computations.

Pablo Burgard