[Rd] matrix operations

[Prof Brian Ripley]

On Sun, 20 Feb 2005, Melanie Vida wrote:

> In R, I'm imported a data frame of  2,321,123 by 4 called "dataF".
> I converted the data frame "dataF" to a matrix
>
> dataM <- as.matrix(dataF)

I am assuming this is a numeric matrix, and "inf" means Inf and in places 
that there are no NAs.

> Does R have an efficient routine to treat the special elements that contain 
> "inf" in them. For example, can you separate the rows that have "inf" 
> elements from the matrix into a separate matrix without iterating over the 
> entire matrix?

Not really, but is.infinite will do a C-level iteration, as will ==

dataM[rowSums(is.infinite(dataM)) > 0, ] # rows containing Inf or -Inf
dataM[rowSums(dataM == Inf), ]           # rows containing Inf  (no NAs)
A <- !is.na(dataM) & dataM == Inf
dataM[rowSums(A), ]                      # rows containing Inf (possible NAs)

> Also, does R have an efficient way to sort columns in a matrix?

How about

for(i in 1:4) dataM[, i] <- sort(dataM[, i], method ="quick")

?

-- 
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