> RSiteSearch("cross product")
> library(pracma)
> ?cross
>
> Speed is usually desired in the context of many similar computations, and
is
> normally achieved in R by vectorizing computation, so storing the large
> number of 3d vectors together in a structure like a Nx3 matrix so the
code
> can be vectorized is the logical approach. The cross() function takes
> inputs in this form, but the current implementation (0.6-3) then fails to
> take advantage of that storage since it iterates with a for loop. A
better
> core implementation of cross() might be:
>
> vcrossp <- function( a, b ) {
> result <- matrix( NA, nrow( a ), 3 )
> result[,1] <- a[,2] * b[,3] - a[,3] * b[,2]
> result[,2] <- a[,3] * b[,1] - a[,1] * b[,3]
> result[,3] <- a[,1] * b[,2] - a[,2] * b[,1]
> result
> }
>
> which is about 20 times faster than cross() on my machine.
Thanks, Jeff, for the hint. I think most of the functions in package
'pracma'
are vectorized (if appropriate), but this one was a real oversight. I has
been
corrected in version 0.7-1 (on R-Forge).
-- Hans Werner
[[alternative HTML version deleted]]
______________________________________________
R-help@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.
