Precision and CD's

[Bruce R Miller]

I think I'm doing a rather poor job of expressing myself here,
for which I apologize;
  To reiterate: I had no intention of getting into the precision
& accuracy of numbers from a engineering or measurement POV!


On Jul 15, 10:21am, Richard M. Timoney wrote:
> Subject: Re: Precision and CD's
> On Wed, Jul 14, 1999 at 05:02:53PM -0400, Bruce R Miller wrote:
[...]
> I think
> bigfloat(x1234567890,16,30) => (1 x 16^10 + ..... + x 16^2 + 9 x 16 + 0) x
16^30
> makes more sense.

Are you suggesting the mantissa OMI should be required to be in the same
base as the radix?  (I hope you aren't proposing that the bigfloat should be
presented that way by a formatter!)

I think that would be an awkward rule to enforce, since bigfloat's
signature simply says it gets 3 OMI; they can be encoded in either
base (10 or 16) and the bigfloat probably would never `see' the OMI
encoding. ie:
   bigfloat(65535,xA,x20) == bigfloat(xFFFF,10,32).

Anyway, John Davenport's bigfloat CD has a lot going for it, but I'm
still wondering what a formatter, or converter to MathML, would
do with this type of bigfloat.

-- 
--
bruce.miller at nist.gov
http://math.nist.gov/~BMiller/