From: hadley wickham <h.wickham_at_gmail.com>

Date: Sat 14 Jan 2006 - 09:10:01 EST

}

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Sat Jan 14 09:16:48 2006

Date: Sat 14 Jan 2006 - 09:10:01 EST

One small problem with apply is that it is not idempotent - it always
puts the result vector in the first dimension of the result array. For
example:

a <- array(1:24, c(2,3,4))

dim(apply(a, c(1,2), I))

dim(apply(a, c(1,2), diff))

I've written a wrapper around apply that rearranges the dimensions into the original order.

# Idempotent apply

iapply <- function (X, MARGIN, FUN, ...) {

dims <- c((1:length(dim(X)))[!(1:length(dim(X)) %in% MARGIN)], MARGIN) res <- apply(X, MARGIN, FUN) if (length(dim(res)) == length(dims)) { aperm(res, order(dims)) } else { res }

}

dim(iapply(a, c(1,2), I))

dim(iapply(a, c(1,2), diff))

Hopefully this may be of use to someone else. The function isn't completely satisfying as apply doesn't deal nicely with functions that return higher-dimensional arrays (eg. apply(a, 1, I)) and reduces 1-D arrays to vectors (eg. apply(a,3,sum)). Any comments on the code would be greatfully appreciated.

Hadley

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Sat Jan 14 09:16:48 2006

*
This archive was generated by hypermail 2.1.8
: Sat 14 Jan 2006 - 14:06:25 EST
*