[R] why is object.size is more for constant numeric vector?

[Prof Brian Ripley]

On Wed, 6 Jan 2010, Utkarsh Singhal wrote:

> Hi All,
>
> I ran the following lines in R:
>
> print(object.size(a <- rep(1,10^6)),units="Mb")
> print(object.size(a <- rep(3.542,10^6)),units="Mb")
>
> print(object.size(b <- rep("x",10^6)),units="Mb")
> print(object.size(b <- rep("xyzxyz xyz",10^6)),units="Mb")
> print(object.size(b <- 1:10^6),units="Mb")
> print(object.size(b <- rep(1:10,each=10^5)),units="Mb")
> print(object.size(b <- rep(TRUE,10^6)),units="Mb")
>
> The object size from first two lines is 7.6 MB, but from the last five it is
> 3.8 MB, although the length of vector is same.
>
> Apparently, the size of any vector of a given length is twice if the vector
> is numeric constant than if it is not.
>
> Why is it so? Is my observation wrong? Or, is there some catch with
> 'object.size'?

Your observation is faulty.  The first two are type "double", and a C 
double takes 8 bytes.  The last three are type "integer" or "logical" 
with values stored in C int, 4 bytes each.  Character strings are 
harder to compute storage for as identical strings share storage.  On 
a 32-bit machine identical strings take an extra 4 bytes per string, 
on a 64-bit machine an extra 8 bytes.  If you look at the values in 
bytes you will see that 'twice' is an approximation.

So

- the storage needed depends on the type of the vector as well as the 
length.

- for character vectors it depends on the architecture and the 
content.

Please do consult the R manuals rather than expect others to read them 
for you: this is all in the 'R Internals' manual.

>
> Thanks in advance.
> Regards
> Utkarsh
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595