On Sun, 18 Jul 1999, Geoffrey Faivre-Malloy wrote:
> I was reading Fermat's Enigma today and something occurred to me...would it
> be possible to work with a number quicker if we used a higher base? I.E.
> Use base 32 instead of base 10? Thus, you're doing less math. Of course,
> this would have to be in software because the floats can't be in that base.
If the floats you're talking about are used in the FFT, they are integers to
start with, the digits of the number being squared in some base. After the
multiplication and inverse FFT they are rounded back to integers. The base is
chosen so that the roundoff error of the FFT is less than 1/2 when the digits
are about squared as big as they were to start.
phma
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
