Barton Willis writes:
> Look towards the end of
> http://www.unk.edu/acad/math/people/willisb/maa-maxima-talk.pdf
> It's not a general scheme for floating point error analysis, but it's a
> start (and I think amusing).
I agree, and it looks like it was a neat talk besides the error
analysis, too. I hope there was a big audience for it.
(When I just got the file, by the way, it was at http://ww2... not
http://www....)
Jay