On Mon, Oct 31, 2011 at 9:22 AM, Stavros Macrakis <macrakis at alum.mit.edu>wrote:
> Re: What's missing with bfloats?
>
>
> solve and allroots don't support bfloats. romberg doesn't support bfloats
>
There is bfallroots. How would you tell romberg to use bfloats? The
limits are bfloats? The tolerance is a bfloat? Some other way?
> Of course, none of the external math packages (quadpack etc.) support
> bfloats.
>
For all of the special functions that use slatec (including quadpack), I
think the main issue would be the constants that are contained in the
algorithms. It would be fairly difficult to recompute them to the correct
precision when using bfloats.
>
> That's off the top of my head. I'm sure there are others.
>
>
Yes, I"m sure there are others, now that you've pointed out some of the
obvious cases.
I also notice that ratepsilon is still 2e-8, even though maxima has been
using double precision for years (decades?) now.
Ray