[R] Efficient marginal sums?

[David Winsemius]

On Nov 13, 2010, at 10:49 AM, (Ted Harding) wrote:

> Hi Folks,
> [This is not unrelated to the current "vector of vectors" thread,
> but arises quite independently]
>
> Say I have a function f(x,y) which computes a value for scalar
> x and y; or, if either x=X or y=Y is a vector, a corresponding
> vector of values f(X,y) or f(x,Y) (with the usual built-in
> vectorisation of operations).
>
> Now I have X=(x.1,x.2,...,x.m) and Y=(y.1,y.2,...,y.n).
> I'm seeking a fast and efficient method to compute (say)
>
>  sum[over elements of Y](f(X,Y))

?sweep

-- 
David.


>
> returning an m-vector in which, for each x.i in X, I have
>
>  sum(f(x.i,Y))
>
> I know I can do this by constructing matrices say M.X and M.Y
>
>  M.X <- matrix(rep(X,length(Y)),nrow=length(Y),byrow=TRUE)
>  M.Y <- matrix(rep(Y,length(X)),nrow=length(Y))
>
> then F <- f(M.X,M.Y), then colSums(F). But that doesn't strike me
> as particularly "fast and efficient", because of the preliminary
> spadework to make M.X and M.Y.
>
> Is there any much better way? Or some function somewhere that
> does it? Something like "marg.sum(X,Y,function=f,margin=2)";
> therefore with scope for generalisation to more than 2 variables,
> e.g. marg.sum(X,Y,Z,function=f,margins=2)
> or   marg.sum(X,Y,Z,function=f,margins=c(2,3))
>
> Or, even more general, like:
>
>  marg.apply(X,Y,Z,fun1=f,fun2=sum,margins=c(2,3))
>
> (Such a question must have been asked before, but I haven't
> located it).
>
> With thanks,
> Ted.
>
> --------------------------------------------------------------------
> E-Mail: (Ted Harding) <ted.harding at wlandres.net>
> Fax-to-email: +44 (0)870 094 0861
> Date: 13-Nov-10                                       Time: 15:49:22
> ------------------------------ XFMail ------------------------------
>
> ______________________________________________
> 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.

David Winsemius, MD
West Hartford, CT