From: Ben Tilly <ben_tilly@...>
Date: 2001-01-03T06:49:50+09:00
Subject: [ruby-talk:8543] Re: speedup of anagram finder

"Christoph Rippel" <crippel@primenet.com> wrote:
>
> > This allows us to come up with a hashing function using floating
> > point numbers.  The only issue is what point we will get into
> > pain from round-off errors.
>Actually round-off errors are a major issue (in theory and practice) -
>floating point arithmetic is not associative - this idea breaks down  if
>words.lenght > 2.

In theory, yes.  In practice if you round reasonably, it will
work.  Of course the more you round, the sooner you get to a
false positive.

I said the idea was sick?  Sick but likely to work in a
practical situation.   (That said, *I* would not want to rely
on such code in production...)

Cheers,
Ben
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com