[SciPy-user] exponential integral behaviour near -20

[David Cournapeau]

Robert Kern wrote:
>
> Ah, I think found it using this clue. It's a bug in SPECFUN. The
> "IMPLICIT DOUBLE PRECISION" statement is missing "A" so A0 is REAL
> rather than DOUBLE. Fixing that makes both of them go through the same
> code path. Can you change the line to this:
>
>           IMPLICIT DOUBLE PRECISION (A,D-H,O-Y)
>
> in your specfun.f file, and rebuild scipy?
>   

Sorry for the delay: you're right, this seems to fix the problem, at
least for me. The example now gives me:

(25615628.4058-3.14159265359j)
(25615630.8316-3.14159265359j)

cheers,

David