Subj : Re: ratio approximation algorithm
To   : comp.programming
From : Alex Fraser
Date : Tue Aug 16 2005 12:19 pm

"Mark Maglana" <mmaglana@gmail.com> wrote in message
news:1124181130.920740.314610@g47g2000cwa.googlegroups.com...
> I also got the same response from another mailing list I posted to. A
> minor correction though. The method of comparison should probably be
>
> -t.b.d < b.d.x - a.c < t.b.d

Is the tolerance relative or absolute?

If it is absolute, the comparison should be: -t <= bdx - ac <= t
If it is relative, then I think you need: -bdtx <= bdx - ac <= bdtx

In either case though, it is probably better to compute the range of
acceptable values of ac.

> I was wondering if it would be better to have the B*D products computed
> in advance and saved inside of a database.

If that is appropriate, you may as well compute every possible value of x in
advance.

Alex

.