R help - Mar 2003 - Question about Error due to memory allocation issue

Hi, 

I'm new to using R and I'm having problems in trying to do some quantile
regressions with the package quantreg. When I try to use rq [specifically, I 
type in: rq(lnwagehr~lnmin1 +  lncpi +  ue03 + ue35 + ue57 + ue79, tau=.05, 
data=mydata, weights=pworwgt, method="fn", na.omit) ] , the computer
churns for
about 15 minutes and then spits out the following error message:

Error: cannot allocate vector of size 109035 Kb

I'm not sure what it going on but here is some (possibly) relevant info:

-My computer has 512Mb of memory
-My data set was originally in Stata and I read it into R using the package 
foreign
-My data set is VERY LARGE: about 1.4 million obs. The data set was 90Mb in 
Stata. The model I've tried to evaluate only has 5 variables, but I'd
like to
expand this to about 43 (I have lots of state and year dummies). 

I have tried using the command: memory.size(max=TRUE):
> memory.size(max=TRUE)
[1] 2076487680

but I then get (after submitting another rq command):

Error: cannot allocate vector of size 76324 Kb

Thanks very much for any help you can provide!

Jeannette Wicks-Lim
Department of Economics
University of Massachusetts, Amherst
Amherst, MA 01003

On Thu, 20 Mar 2003, Jeannette Wicks-Lim wrote:
> Hi,
>
> I'm new to using R and I'm having problems in trying to do some
quantile
> regressions with the package quantreg. When I try to use rq [specifically,
I
> type in: rq(lnwagehr~lnmin1 +  lncpi +  ue03 + ue35 + ue57 + ue79, tau=.05,
> data=mydata, weights=pworwgt, method="fn", na.omit) ] , the
computer churns for
> about 15 minutes and then spits out the following error message:
>
> Error: cannot allocate vector of size 109035 Kb
>
This means that it tried to allocate 109035k on top of what was already
allocated, and couldn't.

You probably can't fit this model in the memory.  Fortunately you probably
don't want to.

You can almost certainly either aggregate to a smaller number of
observations or analyse subgroups of the data and get a more useful result
as well as one that R can compute.

	-thomas