> -----Original Message-----
> From: Richard Fateman [mailto:fateman at cs.berkeley.edu]
>
> Raymond Toy wrote:
>
> >
> >
> >Yeah, I was lazy. Plus, I wasn't sure how to do directed rounding
> >when you mix a single-float with a double-float.
> >
> >
> Also, it is hard to tell if you have x-x or x-y when
> someone types in
>
> interval(-1,1)-interval(-1,1), and simplifying that to 0 is
> probably
> wrong.
>
> RJF
>
I have "Interval Methods for Circuit Analysis" by L V Kolev. I think it's
a good book. Of course it is dealing with circuit tolerances rather than
floating point calculation errors.
But interval analysis is limited.
When you enter an equation for Interval Analysis each calculation is
independent; two different 10 ohm resistors have different values. Now in a
lot of circuit cases (parameters of linear equations) you can accomplish
isolation of part calculations; but consider (my case)
int((x+a)/(x+b)),x=[c,d]) , you can't determine the answer by straight
forward interval analysis since the maximum could be at an intermediate
value of v (say x=-b), and "interval analysis" is really endpoint analysis.
Ray