On 14/04/11 09:48, baptiste auguie wrote:
Thanks, it is cleaner indeed. However, it's still painfully slow on
large data sets (typically a couple hundred thousand points). Looking
at the source of various functions, it looks like the place to look at
would be deldir::tile.list(). To my naive eyes it doesn't look
particularly efficient, given that the "meat" of the calculation that
I expected more costly was performed by deldir() in much less time.
You are very likely correct; tile.list() was an add-on, which was appended
to the deldir package in a rather cobbled-together manner at the request
of a user. Its functionality probably should be incorporated into the
output
process of the basic deldir structure, in the Fortran code.
I've been vaguely thinking about doing something along these lines for a
while. But this would require that I revive my understanding of the Fortran
code which I wrote lo, these many years ago. (Like about thirty (!) for the
original pass at the code.) So it would be a non-trivial undertaking.
I don't
know when I'll be able to get at it, but it certainly won't be less than
3 or 4
months from now.
Feel free to nudge me about it, from time to time! :-)
cheers,
Rolf Turner
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
