It is not a bug. The ",numer" at the end of the line means "evaluate using
floating-point arithmetic".
Compare with:
fpprec:25$
bfloat(1/3),numer => 3.333333333333333148296163b-1
This is equivalent to bfloat(float(1/3)).
To evaluate using bfloat arithmetic, use ", bfloat"
1/3, bfloat
=> 3.333333333333333333333333b-1
Now let's try with bessel.
bessel_k(-1/6,6^3/3), bfloat =>
bessel_k(1.666666666666666666666667b-1, 7.2b1)
bessel_k does not return a bfloat result because there is no bfloat bessel
implementation.
-s
On Tue, Jul 22, 2008 at 3:04 PM, Richard Hennessy <rvh2007 at comcast.net>
wrote:
> I tried this. I don't know if it is a bug.
>
> (%i1) fpprec:100$
> (%i2) bfloat(bessel_k(-1/6,6^3/3)),numer;
> (%o2) 7.9346061457579466752056762821[45
> digits]48713376730743163101067239b-33
>