[R] sparse matrix

[Prof Brian Ripley]

On Mon, 8 Jan 2001, Martin Gotz wrote:

> On Mon, 8 Jan 2001, Prof Brian Ripley wrote:
> 
> > A matrix in R is a vector plus a dim atttribute.  I presume you mean a
> > numeric matrix (there are other types).  Then it needs a 8byte space per
> > entry.
> 
> Yes. Numeric matrix.
> 
> > There are other ways to handle matrices: look at package Matrix, for
> > example.
> > 
> > One obvious representation is to store the non-zero elements and their
> > locations.
> > 
> > Just how large is this matrix, and does it have a pattern to its sparsity?
> 
> 10000 x 10000 elements and 50 000 x 50 000 elements
> There is no pattern of sparsity. Values in matrix means some distance
> between words (for 10000 and 5 0000 words). 
> 98% of elements in matrix are the same (with value 7500)
> 
> > And what do you want to do with it?  Doing things with sparse matrices has
> > a tendency to make them less sparse.
> 
> Hierarchical clustering. Input matrix is distance matrix for hierarchical
> clustering with function hclust.   

No chance. You need to find an algorithm that does not store the
distance matrix.  I think *any* clustering algorithm on 50 000
elements is going to be pretty pointless, but other low-storage algorithms
do exist.


-- 
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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._