This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: testing besjn


On Fri, Feb 26, 2010 at 01:23:33AM +0100, Tobias Burnus wrote:
> Dear Steve,
> 
> If I run
>    real(4) :: r4  = 2.4048255576957729_8
>    real(8) :: r8  = 2.4048255576957729_8
>    real(10):: r10 = 2.4048255576957729_8

You might need additional decimal digits in the r10 value
to get the infinity.

>    print *, besjn(3,r4)
>    print *, besjn(3,r8)
>    print *, besjn(3,r10)
>    print *, besjn(3,2.4048255576957729_4)
>    print *, besjn(3,2.4048255576957729_8)
>    print *, besjn(3,2.4048255576957729_10)
> with GLIBC 2.11.1 (20100118) of openSUSE, I get the following result
> for -m64 (x86-64):
>   0.23056939
>                  -Infinity
>   0.19899961853217685017

This, however, suggest that the problem is in jnf, jn, and jnl.

>   0.19899988
>   0.19899990535769083
>   0.19899990535769083403

These are the expected values, which are generated by MPFR
(which I think you already know but others might not).

Thanks for running the test.

-- 
Steve


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]