[R] Solutions for memory problems (packages ff, bigmemory)?

[hhafner at statistik-hessen.de]

Hello,

I want to impute missing values using the package mi. My file has around 
28 MB (50.000 observations, 120 variables). My computer is a Windows XP 
32-bit machine, 4 GB Duo Core processor.
When I run mi, already after some minutes during iteration 1, I get the 
message that a vector allocation fails. Even with a 10% sample of the 
dataset, it doesn't work.

I've read about the packages ff and bigmemory. Can they help me that the 
mi package doesn't use the RAM for temporary files but the hard drive 
instead? Since I'm not a R expert, I don't understand completely how these 
packages work. Or are there other packages for a better memory management? 
Or is really a 64-bit OS the only solution for this problem? I hope that 
there are some suggestions for my problem since many R procedures seem to 
be very promising but they don't work on my machine because of the memory 
limitation.

Thanks,
Hans-Peter