branch cut fixup for atanh(bfloat(x))



I'm not sure there is no need... minus zero might not be
related to underflow.  Now a bigfloat zero is 0*2^0.  but
it could also be 0*2^(1) or 0*2^(-1).
Maybe Stavros has another idea..

I think I've suggested this at least once... that bfloats could be replaced
(at least on certain lisps) by MPFR, associated with the GMP
project.  This would also tie us in with (other peoples') efforts
to have arbitrary precision routines for Bessel, Erf, etc.
And maybe complex and interval too.
RJF

  ----- Original Message -----
  From: Stavros Macrakis
  To: Raymond Toy
  Cc: Maxima list mailing
  Sent: Monday, February 27, 2006 7:32 AM
  Subject: Re: [Maxima] branch cut fixup for atanh(bfloat(x))


    Maxima's bigfloats don't support signed-zeroes.

  1) They could be implemented.

  2) They probably don't *need* to be implemented, because Maxima
  bigfloats never underflow (try 2.0b0^-(2^1000) for example).

            -s